Advancing Gene Fusion Detection Towards Personalized Cancer Nanodiagnostics Kevin Maisheng Koo Bachelor of Science (Hons I)

Advancing Gene Fusion Detection towards Personalized Cancer Nanodiagnostics Kevin Maisheng Koo Bachelor of Science (Hons I)

A thesis submitted for the degree of Doctor of Philosophy at The University of Queensland in 2017 Australian Institute for Bioengineering and Nanotechnology

Abstract Prostate cancer (PCa) is one of the most prevalent non-cutaneous cancers in men, and is also one of the most lethal oncogenic diseases that accounts for a vast majority of male cancer-related deaths. Currently, widespread PCa screening is reliant on the prevailing usage of the FDA-approved blood- based prostate specific antigen (PSA) biomarker. Yet, landmark clinical trials in recent years have indicated that serum PSA screening holds a substantial risk of over-diagnosing low grade indolent PCa which are unlikely to result in mortality. Consequently, this paucity of accurate disease risk stratification during PCa screening has led to a variety of health burden associated with unnecessary biopsies, and over-treatment in a considerable fraction of patient population. Given that the screening shortcoming of the PSA test is outweighing its benefit, there is a clear need for better strategies to improve PCa risk stratification and accurately detect high-grade aggressive PCa molecular subtypes at an early stage for timely personalized treatment.

To address this PCa screening conundrum, the research work described in this thesis primarily embodies a bipartite strategy which pairs together the use of next-generation PCa-specific molecular biomarkers, and the development of innovative nanodiagnostic technologies to target these superior biomarkers. In recent years, massive advances in next-generation sequencing techniques have led to the discoveries of novel PCa molecular targets which possess excellent PCa- specificity (i.e. next-generation PCa biomarkers); and have been shown to improve PCa diagnosis, prognosis, and prediction of therapeutic resistance in a multitude of seminal clinical studies. The most tantalizing prospect among such next-generation PCa biomarkers is the chromosomal gene fusion mutation between the TMPRSS2 and ERG genes (TMPRSS2:ERG). TMPRSS2:ERG is a highly-recurrent biomarker in the majority of clinically-detected PCa tumor foci, and its exclusive presence in cancer cells facilitates the accurate identification of a major PCa subtype. Through combinatorial detection of TMPRSS2:ERG with other next-generation PCa biomarkers, it is anticipated that PCa molecular subtyping and better disease risk stratification could be achieved. Furthermore, these next-generation biomarkers are detectable in patient urine samples as a form of liquid biopsy, thus opening up new avenues of truly non-invasive PCa diagnostics.

Presently, urinary biomarker detection is achieved through conventional laboratory-based methods which are not ideal for use in a clinical setting. Hence, to realize the use of PCa diagnostics for clinical screening, this thesis also details the use of nanotechnology to design various cutting-edge diagnostic platforms for next-generation PCa urinary biomarker screening. It is envisioned that such PCa nanodiagnostics which exhibits unique physical properties at the nanoscale; could result in accurate early PCa detection with unprecedented detection sensitivity, speed, and reliability.

This thesis first describes the design of miniaturized solution-based biosensing systems which coupled rapid isothermal RNA target amplification with innovative detection readouts that are enabled by nanoscaled mechanisms. These lab-in-a-drop systems integrated the detection of urinary TMPRSS2:ERG transcripts within a fluidic microdroplet, and exhibited single-cell detection sensitivity through visual and/or quantitative colorimetric readouts. Next, this thesis reports the pioneering use of label-free surface-enhanced Raman scattering (SERS) for RNA detection in clinical samples. Label-free SERS is a spectroscopic technique for direct detection of adsorbed nucleic acid on metallic nanoparticle surfaces but has prior been limited to detection of short-length oligonucleotides. Herein, label-free SERS detection of TMPRSS2:ERG RNA in patient urinary specimens was shown using a novel combination of isothermal RNA amplification and/or chemometrics.

In addition to the use of isothermal amplification, this thesis also investigated the development of enzymatic amplification-free nanobiosensors for detection of various RNA biomarker species. By exploiting adenine-bare gold affinity interactions, the detection of different species of next- generation PCa RNA biomarkers was shown. Further, to attain a more comprehensive screening outcome to aid in clinical decision-making, this thesis has also devised new nanobioassays for concurrent analysis of multiple next-generation PCa biomarkers.

Lastly, to facilitate clinical translation, clinical evaluation of both biomarkers and nanodiagnostic technologies (developed through this thesis) was carried out using a cohort of well-annotated patient samples. This represents an effort to progress the developed nanotechnologies from an academic research phase to patient usage by assessment of clinical performance metrics.

Today, there is an assortment of PCa treatment options such as active surveillance, radical prostatectomy, radiation or hormonal therapies; and each treatment has differing molecular subtype- dependent effectiveness and necessity. It is envisioned that research in the fusion of powerful new- age biomarkers and avant-garde detection strategies (as showcased in this thesis) could progress the field of PCa subtyping and risk stratification in the clinic, therefore granting a more personalized treatment approach that is tailored to the needs of individual patients.

Declaration by author This thesis is composed of my original work, and contains no material previously published or written by another person except where due reference has been made in the text. I have clearly stated the contribution by others to jointly-authored works that I have included in my thesis.

I have clearly stated the contribution of others to my thesis as a whole, including statistical assistance, survey design, data analysis, significant technical procedures, professional editorial advice, financial support and any other original research work used or reported in my thesis. The content of my thesis is the result of work I have carried out since the commencement of my higher degree by research candidature and does not include a substantial part of work that has been submitted to qualify for the award of any other degree or diploma in any university or other tertiary institution. I have clearly stated which parts of my thesis, if any, have been submitted to qualify for another award.

I acknowledge that an electronic copy of my thesis must be lodged with the University Library and, subject to the policy and procedures of The University of Queensland, the thesis be made available for research and study in accordance with the Copyright Act 1968 unless a period of embargo has been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright holder(s) of that material. Where appropriate I have obtained copyright permission from the copyright holder to reproduce material in this thesis and have sought permission from co-authors for any jointly authored works included in the thesis.

Publications during candidature

Peer-Reviewed Articles  Koo, K. M.; Wang, J.; Richards, R. S.; Farrell, A.; Yaxley, J. W.; Samaratunga, H.; Teloken, P. E.; Roberts, M. J.; Coughlin, G. D.; Lavin, M. F.; Mainwaring, P. N.; Wang, Y.; Gardiner, R. A. & Trau, M. (2018) Design and clinical verification of surface-enhanced Raman spectroscopy diagnostic technology for individual cancer risk prediction. ACS Nano 12: 8362-8371.

 Koo, K. M.; Dey, S. & Trau, M. (2018) Amplification-free multi-RNA type profiling for cancer risk stratification via alternating current electrohydrodynamic nano-mixing. Small 14: 1704025.

 Koo, K. M.; Carrascosa, L. G. & Trau, M. (2018) DNA-directed assembly of copper nanoblocks with inbuilt fluorescent and electrochemical properties: Application in simultaneous amplification-free analysis of multiple RNA species. Nano Res. 11: 940-952.

 Koo, K. M.; Wee, E. J. H.; Wang, Y. & Trau, M. (2017) Enabling miniaturised personalised diagnostics: From lab-on-a-chip to lab-in-a-drop. Lab Chip 17: 3193-3340. *Featured as Issue Front Cover Image

 Wang, J.; Koo, K.M.; Wee, E. J. H.; Wang, Y. & Trau, M. (2017) Nanoplasmonic label-free surface-enhanced Raman scattering strategy for non-invasive cancer genetic subtyping in patient samples. Nanoscale 9: 3496-3503. (co-first author)

 Koo, K. M.; Wee, E. J. H. & Trau, M. (2017) High-speed biosensing strategy for non-invasive profiling of multiple cancer fusion genes in urine. Biosens. Bioelectron. 89: 715-720.

 Koo, K. M.; Wee, E. J. H.; Mainwaring, P. N.; Wang, Y. & Trau, M. (2016) Toward precision medicine: A cancer molecular subtyping nano-strategy for RNA biomarkers in tumor and urine. Small 12: 6233-6242. *Featured as Issue Cover Image

 Koo, K. M.; McNamara, B.; Wee, E. J. H.; Wang, Y. & Trau, M. (2016) Rapid and sensitive fusion gene detection in prostate cancer urinary specimens by label-free SERS. J. Biomed. Nanotech. 12: 1798-1805.

 Koo, K. M.; Wee, E. J. H.; Mainwaring, P. N. & Trau, M. (2016) A simple, rapid, low-cost technique for naked-eye detection of urine-isolated TMPRSS2:ERG gene fusion RNA. Sci. Rep. 6: 30722.

 Koo, K. M.; Wee, E. J. H. & Trau, M. (2016) Colorimetric TMPRSS2-ERG gene fusion detection in prostate cancer urinary samples via recombinase polymerase amplification. Theranostics 6: 1415-1424. *Featured in NanoTheranostics Special Issue

 Koo, K. M.; Carrascosa, L. G.; Shiddiky, M. J. A. & Trau, M. (2016) Poly(A) extensions of miRNAs for amplification-free electrochemical detection on screen-printed gold electrodes. Anal. Chem. 88: 2000-2005.

 Koo, K. M.; Sina, A. A. I.; Carrascosa, L. G.; Shiddiky, M. J. A. & Trau, M. (2015) DNA-bare gold affinity interactions: mechanism and applications in biosensing. Anal. Mtds. 7: 7042-7054.

 Koo, K. M.; Sina, A. A. I.; Carrascosa, L. G.; Shiddiky, M. J. A. & Trau, M. (2014) eMethylsorb: rapid quantification of DNA methylation in cancer cells on screen-printed gold electrodes. Analyst 139: 6178-6184.

Book Chapters  Sina, A. A. I; Koo, K. M.; Ahmed, M.; Carrascosa, L. G. & Trau, M. (2017) Interfacial biosensing: Direct biosensing of biomolecules at the bare metal interface. Encyclopedia of Interfacial Chemistry: Surface Science and Electrochemistry: Elsevier (In Press) (co-first author)

Conference Abstracts  Koo, K. M.; Dey, S. & Trau, M. (2018) Amplification-free multi-RNA type profiling for cancer risk stratification via alternating current electrohydrodynamic nano-mixing. 9th International Nanomedicine Conference, Sydney (Oral Presentation - Selected)

 Koo, K. M.; Wee, E. J. H.; Mainwaring, P. N.; Wang, Y. & Trau, M. (2017) Towards precision medicine: A cancer molecular subtyping nano-strategy for RNA biomarkers in tumor and urine. 2017 International Conference on Bio-Nano Innovation (ICBNI 2017), Brisbane (Oral Presentation - Selected)

 Koo, K. M.; Wee, E. J. H.; Mainwaring, P. N.; Wang, Y. & Trau, M. (2017) Toward precision medicine: A cancer molecular subtyping nano-strategy for RNA biomarkers in tumor and urine. 8th International Nanomedicine Conference, Sydney (Oral Presentation - Selected)

 Koo, K. M.; Wee, E. J. H.; Mainwaring, P. N.; Wang, Y. & Trau, M. (2016) Towards precision medicine: A cancer molecular subtyping nano-strategy for RNA biomarkers in tumor and urine. Eur. J. Cancer 68: S33-S33. EORTC-NCI-AACR Molecular Targets & Cancer Therapeutics Symposium 2016, Munich (Poster Presentation)

 Koo, K. M.; Wee, E. J. H.; Mainwaring, P. N.; Wang, Y. & Trau, M. (2016) Toward precision medicine: A cancer molecular subtyping nano-strategy for RNA biomarkers in tumor and urine. 18th EMBL PhD Symposium 2016: Life by Numb3rs - Towards Quantitative Biology, Heidelberg (Flash talk & Poster Presentation)

 Koo, K. M.; Wee, E. J. H.; Mainwaring, P. N.; Wang, Y. & Trau, M. (2016) Toward precision medicine: A cancer molecular subtyping nano-strategy for RNA biomarkers in tumor and urine. The 6th NanoBio Conference and 1st Symposium on Minimally Invasive & Image Guided Surgery (NanoBio China 2016), Nanjing (Oral Presentation - Selected)

 Koo, K. M.; Carrascosa, L. G.; Shiddiky, M. J. A. & Trau, M. (2016) Amplification-free detection of gene fusions in prostate cancer urinary samples using mRNA-gold affinity interactions. 7th Australia and New Zealand Nano-Microfluidic Symposium (ANZNMF 2016), Brisbane (Poster Presentation) *Best Poster Prize

 Wee, E. J. H.; Koo, K. M.; Mainwaring, P. N. & Trau, M. (2015) Towards rapid and cost- effective point-of-care detection of TMPRSS2: ERG fusion transcripts in urine via a novel methodology. J. Clin. Oncol. 33: 11063-11063. American Society of Clinical Oncology Conference 2015, Chicago (Poster Presentation) Publications included in this thesis I declared that I have obtained permission from all co-authors to include the following publications directly into my thesis.

1. Koo, K. M.; Wee, E. J. H.; Wang, Y. & Trau, M. (2017) Enabling miniaturised personalised diagnostics: From lab-on-a-chip to lab-in-a-drop. Lab Chip 17: 3193-3340. Incorporated as Chapter 3.

Contributor Statement of contribution Kevin M Koo (Candidate) Drafting and production (65%) Eugene JH Wee Drafting and production (15%) Yuling Wang Drafting and production (15%) Matt Trau Drafting and production (5%)

2. Koo, K. M.; Wee, E. J. H.; Mainwaring, P. N. & Trau, M. (2016) A simple, rapid, low- cost technique for naked-eye detection of urine-isolated TMPRSS2:ERG gene fusion RNA. Sci. Rep. 6: 30722. Incorporated as Chapter 3.

Contributor Statement of contribution Kevin M Koo (Candidate) Conception and design (75%) Analysis and interpretation (80%) Drafting and production (80%) Eugene JH Wee Conception and design (15%) Analysis and interpretation (10%) Drafting and production (10%) Paul N Mainwaring Conception and design (5%) Drafting and production (5%) Matt Trau Conception and design (5%) Analysis and interpretation (10%) Drafting and production (5%)

3. Koo, K. M.; Wee, E. J. H. & Trau, M. (2016) Colorimetric TMPRSS2-ERG gene fusion detection in prostate cancer urinary samples via recombinase polymerase amplification. Theranostics 6: 1415-1424. Incorporated as Chapter 3.

Contributor Statement of contribution Kevin M Koo (Candidate) Conception and design (75%) Analysis and interpretation (90%) Drafting and production (85%) Eugene JH Wee Conception and design (25%) Analysis and interpretation (10%) Drafting and production (10%) Matt Trau Drafting and production (5%)

4. Koo, K. M.; McNamara, B.; Wee, E. J. H.; Wang, Y. & Trau, M. (2016) Rapid and sensitive fusion gene detection in prostate cancer urinary specimens by label-free SERS. J. Biomed. Nanotech. 12: 1798-1805. Incorporated as Chapter 4.

Contributor Statement of contribution Kevin M Koo (Candidate) Conception and design (80%) Analysis and interpretation (75%) Drafting and production (75%) Benjamin McNamara Analysis and interpretation (5%) Eugene JH Wee Conception and design (10%) Analysis and interpretation (10%) Drafting and production (10%) Yuling Wang Conception and design (10%) Analysis and interpretation (10%) Drafting and production (10%) Matt Trau Drafting and production (5%)

5. Wang, J.; Koo, K. M.; Wee, E. J. H.; Wang, Y. & Trau, M. (2017) Nanoplasmonic label- free surface-enhanced Raman scattering strategy for non-invasive cancer genetic subtyping in patient samples. Nanoscale 9: 3496-3503. (co-first author). Incorporated as Chapter 4.

Contributor Statement of contribution Jing Wang Conception and design (40%) Analysis and interpretation (40%) Drafting and production (30%) Kevin M Koo (Candidate) Conception and design (40%) Analysis and interpretation (40%) Drafting and production (45%) Eugene JH Wee Conception and design (10%) Analysis and interpretation (10%) Drafting and production (10%) Yuling Wang Conception and design (10%) Analysis and interpretation (10%) Drafting and production (10%) Matt Trau Drafting and production (5%)

6. Koo, K. M.; Carrascosa, L. G.; Shiddiky, M. J. A. & Trau, M., (2016) Amplification- free detection of gene fusions in prostate cancer urinary samples using mRNA-gold affinity interactions. Anal. Chem. 88: 6781-6788. Incorporated as Chapter 5.

Contributor Statement of contribution Kevin M Koo (Candidate) Conception and design (70%) Analysis and interpretation (75%) Drafting and production (75%) Laura G Carrascosa Conception and design (15%) Analysis and interpretation (10%) Drafting and production (10%) Muhammad JA Shiddiky Conception and design (15%) Analysis and interpretation (10%) Drafting and production (10%) Matt Trau Analysis and interpretation (5%) Drafting and production (5%)

7. Koo, K. M.; Carrascosa, L. G.; Shiddiky, M. J. A. & Trau, M. (2016) Poly(A) extensions of miRNAs for amplification-free electrochemical detection on screen-printed gold electrodes. Anal. Chem. 88: 2000-2005. Incorporated as Chapter 5.

8. Koo, K. M.; Carrascosa, L. G. & Trau, M. (2018) DNA-directed assembly of copper nanoblocks with inbuilt fluorescent and electrochemical properties: Application in simultaneous amplification-free analysis of multiple RNA species. Nano. Res. 11: 940- 952. Incorporated as Chapter 5.

Contributor Statement of contribution Kevin M Koo (Candidate) Conception and design (75%) Analysis and interpretation (80%) Drafting and production (80%) Laura G Carrascosa Conception and design (25%) Analysis and interpretation (15%) Drafting and production (15%) Matt Trau Analysis and interpretation (5%) Drafting and production (5%)

9. Koo, K. M.; Wee, E. J. H.; Mainwaring, P. N.; Wang, Y. & Trau, M. (2016) Toward precision medicine: A cancer molecular subtyping nano-strategy for RNA biomarkers in tumor and urine. Small 12: 6233-6242. Incorporated as Chapter 6.

Contributor Statement of contribution Kevin M Koo (Candidate) Conception and design (75%) Analysis and interpretation (80%) Drafting and production (70%) Eugene JH Wee Conception and design (10%) Analysis and interpretation (10%) Drafting and production (10%) Paul N Mainwaring Conception and design (5%) Drafting and production (5%) Yuling Wang Conception and design (10%) Analysis and interpretation (10%) Drafting and production (10%) Matt Trau Drafting and production (5%)

10. Koo, K. M.; Wee, E. J. H. & Trau, M. (2017) High-speed biosensing strategy for non- invasive profiling of multiple cancer fusion genes in urine. Biosens. Bioelectron. 89: 715- 720. Incorporated as Chapter 6.

11. Koo, K. M.; Dey, S. & Trau, M. (2018) Amplification-free multi-RNA type profiling for cancer risk stratification via alternating current electrohydrodynamic nano-mixing. Small 14: 1704025. Incorporated as Chapter 6.

Contributor Statement of contribution Kevin M Koo (Candidate) Conception and design (70%) Analysis and interpretation (55%) Drafting and production (75%) Shuvashis Dey Conception and design (30%) Analysis and interpretation (45%) Drafting and production (20%) Matt Trau Drafting and production (5%) Manuscripts included in this thesis I declared that I have obtained permission from all co-authors to include the following manuscripts directly into my thesis.

 Koo, K. M.; Mainwaring, P. N.; Tomlins, S. A. & Trau, M. Merging next-generation biomarkers and nanodiagnostic platforms for precision prostate cancer management. Nat. Rev. Urol. (Under editorial revision) Incorporated as Chapter 2.

Contributor Statement of contribution Kevin M Koo (Candidate) Drafting and production (70%) Paul N Mainwaring Drafting and production (10%) Scott A Tomlins Drafting and production (10%) Matt Trau Drafting and production (10%)

 Koo, K. M.; Wee, E. J. H.; Richards, R. S.; Lavin, M. F.; Gardiner, R. A.; Mainwaring, P. N. & Trau, M. TMPRSS2-ERG status in pre- and post-prostatectomy urine: A novel first-line triage nanoassay for non-invasive visual identification of significant prostate cancer. Submitted. Incorporated as Chapter 7.

Contributor Statement of contribution Kevin M Koo (Candidate) Conception and design (60%) Analysis and interpretation (80%) Drafting and production (80%) Eugene JH Wee Conception and design (10%) Renée S Richards Analysis and interpretation (5%) Martin F Lavin Analysis and interpretation (5%) Robert A Gardiner Conception and design (10%) Analysis and interpretation (5%) Drafting and production (10%) Paul N Mainwaring Conception and design (10%) Drafting and production (10%) Matt Trau Conception and design (10%) Analysis and interpretation (5%) Contributions by others to the thesis The research work being described in this thesis has led to a succession of published/submitted peer-reviewed articles with me named as the lead author. As stated under the previous section “Publications included in this thesis”, my fellow contributing authors on these publications have contributed in part to the experimental design, data analysis, and manuscript preparation processes.

Statement of parts of the thesis submitted to qualify for the award of another degree None.

Research involving human or animal subjects No animal or human subjects were involved in this research.

Acknowledgements This pursuit of my doctorate has been a thrilling ride that has no doubt been made so much more enjoyable due to the presence of so many wonderful individuals. With my following words in this thesis, I wish to express my heartfelt gratitude and appreciation in acknowledgement of your influence in the culmination of my dissertation.

I have been fortunate to benefit from various folks who are the best mentors I could have ever asked for. Matt - Thank You for always being so supportive and encouraging in moulding us as researchers. The research resources and environment you provided in the Trau group, together with your bubbly enthusiasm, give such an enviably unrestricted platform to drive my research further. Shiddiky - Thank You for recognizing my suitability for research right from the start, and being such a momumental figure in helping me through my formative beginnings. It is a pleasure to learn so much off you and I am always grateful of what you have spurred me on to achieve. Eugene - Thank You for patiently showing me the ropes of fundamental research skills at both the lab bench and the workdesk keyboard. It has been a pleasure working with us and I have definitely benefitted a lot from your generous sharing of knowledge and experience. Paul - Thank You for opening my eyes to the intriguing world of clinical research in the real world. You have encouraged and motivated the purpose of my work in this thesis and beyond. I also like to acknowledge so many others (Yuling, Laura, and the entire Trau group; 'Frank' and the lovely UQCCR collaborators; Scott and his awesome UMich group members) whom I have the joy of working alongside with during the course of my doctoral studies - Thank You All, I am humbled by all your positive influence and constructive pieces of advice.

I will also like to express my gratitude to the many friends and companions whom I have the blessing of knowing over the last few years. To my fellow colleagues in AIBN and UQ - Thank You all for being the reason why I wake up every morning and look forward to stepping into the lab - It is a joy to learn off everyone and be part of this stimulating research environment. To all the amazing mates both past and present - Thank You all for being part of my life in Brisbane, our treasured friendships have allowed me to hit a pause button on research, and take a breather to create so many great memories to look back upon.

Last but not least, I will like to dedicate this thesis to my loving family. Especially to Dad and Mum, your unreserved support has encouraged me to give my absolute best. I know how much you all miss having me away from your side, and words cannot express my deepest appreciation for your unconditional love in enabling me to pursue my endeavours - I am truly blessed and proud to be your son.

Finanacial support This research was supported by an Australian Government Research Training Program Scholarship/International Postgraduate Research Scholarship, and an Australian Postgraduate Award.

Keywords prostate cancer, precision medicine, early cancer detection, liquid biopsy, fusion genes, TMPRSS2:ERG, nanodiagnostics, electrochemical detection, surface-enhanced Raman spectroscopy, clinical translation

Australian and New Zealand Standard Research Classifications (ANZSRC) ANZSRC code: 100402, Medical Biotechnology Diagnostics (incl. Biosensors), 30% ANZSRC code: 100703, Nanobiotechnology, 40% ANZSRC code: 111202 Cancer Diagnosis, 30%

Fields of Research (FoR) Classification FoR code: 1004, Medical Biotechnology, 30% FoR code: 1007, Nanotechnology, 40% FoR code: 1112, Oncology and Carcinogenesis, 30%

Table of Contents

Chapter 1: Thesis Introduction ...... 1 1.1 Background ...... 1 1.2 Research aim and objectives ...... 3 1.3 Significance of project ...... 4 1.4 Structure of the thesis...... 4 1.5 References ...... 7 Chapter 2: Thesis Literature Review ...... 8 2.0 Merging Next-Generation Biomarkers and Nanodiagnostic Platforms for Precision Prostate Cancer Management ...... 9 Chapter 3: Colorimetric Gene Fusion Diagnostics for Visual Binary Readout ...... 39 3.1 Enabling miniaturised personalised diagnostics: From lab-on-a-chip to lab-in-a-drop ...... 40 3.2 A simple, rapid, low-cost technique for naked-eye detection of urine-isolated TMPRSS2:ERG gene fusion RNA ...... 82 3.3 Colorimetric TMPRSS2:ERG gene fusion detection in prostate cancer urinary samples via recombinase polymerase amplification ...... 96 Chapter 4: Label –Free Surface-Enhanced Raman Scattering Detection Systems for Clinical Biomarker Targets ...... 115 4.1 Rapid and sensitive fusion gene detection in prostate cancer urinary specimens by label-free SERS ...... 116 4.2 Nanoplasmonic label-free surface-enhanced Raman scattering strategy for non- invasive cancer genetic subtyping in patient samples ...... 132 Chapter 5: Amplification-Free Electrochemical RNA Biomarker Sensing ...... 149 5.1 Amplification-free detection of gene fusions in prostate cancer urinary samples using mRNA-gold affinity interactions ...... 150 5.2 Poly(A) extensions of miRNAs for amplification-free electrochemical detection on screen-printed gold electrodes ...... 170 5.3 DNA-directed assembly of copper nanoblocks with inbuilt fluorescent and electrochemical properties: Application in simultaneous amplification-free analysis of multiple RNA species...... 186 Chapter 6: Simultaneous Analysis of Multiple Biomarkers via High-Throughput Parallel/Multiplexing Assays ...... 210 6.1 Toward precision medicine: A cancer molecular subtyping nano-strategy for RNA biomarkers in tumor and urine ...... 211 6.2 High-speed biosensing strategy for non-invasive profiling of multiple cancer fusion genes in urine ...... 238 6.3 Amplification-free multi-RNA type profiling for cancer risk stratification via alternating current electrohydrodynamic nano-mixing ...... 259 Chapter 7: Clinical Evaluation of Non-Invasive Nanodiagnostics for PCa Risk Stratification ...... 281 7.0 TMPRSS2-ERG status in pre- and post-prostatectomy urine: A novel first-line triage nanoassay for non-invasive visual identification of significant prostate cancer ...... 282 Chapter 8: Conclusions and Future Work ...... 292

List of Abbreviations used in the thesis Prostate cancer PCa Food and Drug Administration FDA Prostate specific antigen PSA Surface-enhanced Raman scattering SERS Transmembrane protease, serine 2 TMPRSS2 V-ets avian erythroblastosis virus E26 oncogene homolog ERG Gene fusion between TMPRSS2 and ERG TMPRSS2:ERG Ribonucleic Acid RNA Quantitative polymerase chain qPCR Fluorescence in-situ hybridization FISH Chapter 1

Thesis Introduction

1.1 Background A detailed research background of this thesis is given in Chapter 2 “Merging Next- Generation Biomarkers and Nanodiagnostic Platforms for Precision Prostate Cancer Management”. The following is a summary of Chapter 2.

Prostate cancer (PCa) is one of the most commonly diagnosed cancers and a leading cause of cancer-related deaths in men with a projected 161, 360 (with 26, 730 deaths) and 16, 665 (with 3452 deaths) newly-diagnosed cases in the United States and Australia respectively for 2017.1 PCa is a clinically heterogeneous disease which ranges from indolent molecular subtypes which are insignificant to patient’s health to fatal metastatic tumor growths. Since the introduction of prostate-specific antigen (PSA) as a PCa screening biomarker in the 1980s, the PSA test has seen widespread usage to become the current standard test for PCa early detection. However, the PSA test is one with marked deficiency because of its poor PCa-specificity in identifying high-grade clinically-significant PCa.2,3 This has resulted in high incidences of PCa overdiagnosis and overtreatment of men with low-grade diseases which would never have a significant impact on mortality. Therefore, this highlights the importance of precision PCa screening that enables specific detection of aggressive PCa subtypes which require immediate treatment (i.e. risk stratification). To this end, this thesis seeks to engage a cross-disciplinary approach which combines the clinical evaluation of next- generation PCa biomarkers of superior PCa-specificity, along with the technological development of rapid and accurate nanodiagnostic techniques to detect these biomarkers in the clinic.

The rise of next-generation sequencing has led to exciting advances in PCa molecular subtyping. Specifically, there have been potential game-changing discoveries of next- generation PCa biomarkers that are associated with clinically-significant PCa. Among these next-generation PCa biomarkers, the gene fusion between transmembrane protease, serine 2 (TMPRSS2) and v-ets avian erythroblastosis virus E26 oncogene homolog (ERG) genes -

Chapter 1

TMPRSS2:ERG - is a frequent occurrence found in around half of all clinically-detected PCa foci.4 The TMPRSS2:ERG gene fusion mutation between two independent genes typically results in androgen-regulated overexpression of oncogenic ERG to drive tumorgenesis. This led to the hypothesis that TMPRSS2:ERG is an early PCa initiation event that cooperates with other genomic mutations for transition into high-grade invasive PCa.5 Given that TMPRSS2:ERG is generally absent in normal prostate tissue and benign prostatic hyperplasia, it can potentially function as a highly PCa-specific diagnostic biomarker to aid in accurate PCa risk stratification and reduction of unnecessary biopsies/treatments. The prognostic utility of TMPRSS2:ERG has been demonstrated in several reports with TMPRSS2:ERG being associated with aggressive PCa.6-8 This provides additional translational value to TMPRSS2:ERG as a biomarker to predict PCa aggressiveness and monitor post-treatment biochemical recurrence. Furthermore, TMPRSS2:ERG is detectable in urine, hence offering an attractive biomarker candidate for non-invasive PCa early detection.9 All in all, TMPRSS2:ERG is strongly implicated in the biology of PCa development and serves as an attractive target (along with other promising next-generation PCa biomarkers) for designing a new era of precision PCa diagnostics.

Presently, detection of next-generation PCa biomarkers is primarily limited to conventional laboratory techniques such as quantitative polymerase chain reaction (qPCR), microarray- based methods, fluorescence in-situ hybridization (FISH), and next-generation sequencing.4,9 Whilst these methodologies are generally efficient and reliable, they are limited for use in rapid clinical screening due to prolonged sample-to-outcome turnaround time, specialized equipment requirement, and need for trained personnel to perform the procedures. Nanotechnology is a promising avenue to overcome these limitations and improve patient diagnoses through desirable and unique biomarker-detection properties that exist at the nanoscale.10 Hence, this thesis seeks to harness the exciting possibilities and benefits of nanotechnologies for fast, cost-effective, and reliable PCa early detection and risk stratification in the clinic.

Overall, in a bid to progress towards a new era of precision PCa, this thesis details a two- pronged approach which fuses (i) the design of various novel nano-strategies with different readout platforms and applications, and (ii) the investigation into the use of next-generation PCa biomarkers in patient urine samples. Both analytical and clinical aspects of the approach are rigorously evaluated to accelerate laboratory-to-clinic translation.

Chapter 1

1.2 Research aim and objectives This thesis aims to merge the development of innovative nanotechnological approaches and the detection of next-generation PCa urinary biomarkers, for better non-invasive PCa risk stratification in the clinic. The specific research objectives to achieve this aim are as follows:

1. Development of gene fusion (TMPRSS2:ERG) diagnostics towards non-invasive clinical use. The TMPRSS2:ERG gene fusion is one of the most common driver mutations in PCa, making it a promising PCa- specific biomarker target for clinical diagnostics development. This objective involves the novel design of rapid, simple, and low-cost nanobiosensors for non-invasive TMPRSS2:ERG RNA detection in patient urine samples by the integration of an isothermal amplification technique.

2. Design of amplification-free nanobiosensors for universal detection of various RNA biomarker species. The direct detection of native nucleic acid biomarkers without requiring enzymatic amplification is ideal for avoiding amplification bias/artifacts. By using the detection of urinary biomarkers in PCa as a model, this objective covers the design of a new amplification-free methodology which is amenable for different RNA species.

3. Expansion into simultaneous detection of multiple biomarkers. This objective builds upon the outcomes of the previous objectives, and progresses towards innovative nanodiagnostics for the simultaneous analysis of various next-generation PCa urinary biomarkers (including TMPRSS2:ERG) to further improve clinical PCa risk stratification.

4. Clinical evaluation of developed nanotechnological assays. To evaluate the translational potential for use in the real-world clinical setting, this objective investigates the clinical metrics of developed PCa nanodiagnostics through testing in a cohort of well-annotated PCa patient urinary specimens.

5. Investigation of biomarker value via pilot clinical studies. Lateral to the clinical evaluation of the nanodiagnostic assays, this objective concurrently probes the clinical value of different next-generation PCa biomarkers in disease diagnosis, prognosis, and treatment decision-making.

Chapter 1

1.3 Significance of project There is a current unmet need for better strategies in comprehensive early PCa detection, and accurate discrimination of aggressive from indolent disease. The achievement of the stated research aim and objectives in this thesis could have a significant clinical impact in advancing these areas. The development of a suite of novel PCa analytical nanodiagnostics targeting next-generation PCa urinary biomarkers may enable different distinct methodologies for accurate non-invasive PCa early detection. The further clinical analysis of both nanotechnological assays and biomarker targets will aid in translating the thesis outcomes from the laboratory into clinical use. Together, the blend of avant-garde detection methods and superior biomarkers might reduce unneeded biopsies, overdiagnosis, and overtreatment of PCa; and lead to more effective treatment decisions of men with, or at risk of developing PCa. Finally, from a research perspective, the studies in this thesis could raise awareness to the detection of next-generation PCa biomarkers (especially fusion genes) for nanotechnologists; as well as communicate the vast potential of nanodiagnostics to clinicians.

1.4 Structure of the thesis This thesis comprises of eight chapters, and is presented as a combination of peer-reviewed published articles and submitted manuscripts.

Chapter 1 – Thesis Introduction This chapter provides a brief research background; aims and objectives; and significance.

Chapter 2 – Thesis Literature Review This chapter provides a detailed literature review of research background that is presented in this thesis. Specifically, it will encompass the advent of next-generation PCa biomarkers for personalized risk stratification and the role that nanotechnology can play in translating these biomarkers for clinical screening usage. Chapter 2 is presented as an accepted review manuscript under editorial revision:  Koo, K. M.; Mainwaring, P. N.; Tomlins, S. A. & Trau, M. (2017) Merging next- generation biomarkers and nanodiagnostic platforms for precision prostate cancer management. Nat. Rev. Urol. (Under editorial revision)

Chapter 1

Chapter 3 – Colorimetric Gene Fusion Diagnostics for Visual Binary Readout This chapter will describe the development of a rapid microtube-based assay which exploits the binary nature of gene fusion biomarkers for visual cancer detection, and allows quantitative detection of gene fusion transcripts for potential PCa staging/monitoring. Chapter 3 is presented as one peer-reviewed review article, followed by two published research articles in the following order:  Koo, K. M.; Wee, E. J. H.; Wang, Y. & Trau, M. (2017) Enabling miniaturised personalised diagnostics: From lab-on-a-chip to lab-in-a-drop. Lab Chip 17: 3193-3340.  Koo, K. M.; Wee, E. J. H.; Mainwaring, P. N. & Trau, M. (2016) A simple, rapid, low- cost technique for naked-eye detection of urine-isolated TMPRSS2:ERG gene fusion RNA. Sci. Rep. 6: 30722.  Koo, K. M.; Wee, E. J. H. & Trau, M. (2016) Colorimetric TMPRSS2-ERG gene fusion detection in prostate cancer urinary samples via recombinase polymerase amplification. Theranostics 6: 1415-1424.

Chapter 4 – Label-Free Surface-Enhanced Raman Scattering Detection Systems for Clinical Biomarker Targets This chapter will describe the integration of isothermal amplification with nanoscaled label- free surface-enhanced Raman scattering technique for RNA biosensing in PCa patient samples. Chapter 4 is presented as two published research articles in the following order:  Koo, K. M.; McNamara, B.; Wee, E. J. H.; Wang, Y. & Trau, M. (2016) Rapid and sensitive fusion gene detection in prostate cancer urinary specimens by label-free SERS. J. Biomed. Nanotech. 12: 1798-1805.  Wang, J.; Koo, K. M.; Wee, E. J. H.; Wang, Y. & Trau, M. (2017) Nanoplasmonic label-free surface-enhanced Raman scattering strategy for non-invasive cancer genetic subtyping in patient samples. Nanoscale 9: 3496-3503. (co-first author)

Chapter 1

Chapter 5 – Amplification-Free Electrochemical RNA Biomarker Sensing This chapter will include a series of amplification-free electrochemical RNA nanobiosensors based on adenine-bare gold affinity interactions. Chapter 5 is presented as three published research articles in the following order:  Koo, K. M.; Carrascosa, L. G.; Shiddiky, M. J. A. & Trau, M., (2016) Amplification-free detection of gene fusions in prostate cancer urinary samples using mRNA-gold affinity interactions. Anal. Chem. 88: 6781-6788.  Koo, K. M.; Carrascosa, L. G.; Shiddiky, M. J. A. & Trau, M. (2016) Poly(A) extensions of miRNAs for amplification-free electrochemical detection on screen- printed gold electrodes. Anal. Chem. 88: 2000-2005.  Koo, K. M.; Carrascosa, L. G. & Trau, M. (2018) DNA-directed assembly of copper nanoblocks with inbuilt fluorescent and electrochemical properties: Application in simultaneous amplification-free analysis of multiple RNA species. Nano Res. 11: 940- 952.

Chapter 6 – Simultaneous Analysis of Multiple Biomarkers via High-Throughput Parallel/Multiplexing Assays This chapter will describe the development of novel bioassays which enable simultaneous detection of multiple biomarkers for a more comprehensive PCa diagnosis. Chapter 6 is presented as three published research articles in the following order:  Koo, K. M.; Wee, E. J. H.; Mainwaring, P. N.; Wang, Y. & Trau, M. (2016) Toward precision medicine: A cancer molecular subtyping nano-strategy for RNA biomarkers in tumor and urine. Small 12: 6233-6242.  Koo, K. M.; Wee, E. J. H. & Trau, M. (2017) High-speed biosensing strategy for non-invasive profiling of multiple cancer fusion genes in urine. Biosens. Bioelectron. 89: 715-720.  Koo, K. M.; Dey, S. & Trau, M. (2018) Amplification-free multi-RNA type profiling for cancer risk stratification via alternating current electrohydrodynamic nano-mixing. Small 14: 1704025.

Chapter 1

Chapter 7 – Clinical Evaluation of Non-Invasive Nanodiagnostics for PCa Risk Stratification This chapter will describe the clinical evaluations of a nanodiagnostic assay which has been developed during candidature and the initial results of investigating gene fusion biomarker at different PCa stages. Chapter 7 is presented as one submitted research manuscript.  Koo, K. M.; Wee, E. J. H.; Richards, R. S.; Lavin, M. F.; Gardiner, R. A.; Mainwaring, P. N.; Trau, M. TMPRSS2-ERG status in pre- and post-prostatectomy urine: A novel first-line triage nanoassay for non-invasive visual identification of significant prostate cancer. Submitted.

Chapter 8 – Conclusions and Future Work This chapter summarizes the preceding thesis chapters and outlines recommended future work based on the current progress showcased through this thesis.

1.5 References 1. Siegel, R. L.; Miller, K. D.; Jemal, A. Ca-A Cancer Journal for Clinicians 2017, 67, 7- 30. 2. Pinsky, P. F.; Prorok, P. C.; Kramer, B. S. New England Journal of Medicine 2017, 376, 1285-1289. 3. Tsodikov, A.; Gulati, R.; Heijnsdijk, E. M.; et al. Annals of Internal Medicine 2017. 4. Tomlins, S. A.; Rhodes, D. R.; Perner, S.; Dhanasekaran, S. M.; Mehra, R.; Sun, X. W.; Varambally, S.; Cao, X. H.; Tchinda, J.; Kuefer, R.; Lee, C.; Montie, J. E.; Shah, R. B.; Pienta, K. J.; Rubin, M. A.; Chinnaiyan, A. M. Science 2005, 310, 644-648. 5. Adamo, P.; Ladomery, M. R. Oncogene 2016, 35, 403-414. 6. Wang, J. H.; Cai, Y.; Ren, C. X.; Ittmann, M. Cancer Research 2006, 66, 8347-8351. 7. Young, A.; Palanisamy, N.; Siddiqui, J.; Wood, D. P.; Wei, J. T.; Chinnaiyan, A. M.; Kunju, L. P.; Tomlins, S. A. American Journal of Clinical Pathology 2012, 138, 685-696. 8. Wang, J. H.; Cai, Y.; Yu, W. D.; Ren, C. X.; Spencer, D. M.; Ittmann, M. Cancer Research 2008, 68, 8516-8524. 9. Laxman, B.; Tomlins, S. A.; Mehra, R.; Morris, D. S.; Wang, L.; Helgeson, B. E.; Shah, R. B.; Rubin, M. A.; Wei, J. T.; Chinnaiyan, A. M. Neoplasia 2006, 8, 885-888. 10. Ferrari, M. Nature Reviews Cancer 2005, 5, 161-171.

Chapter 2

Thesis Literature Review

Introduction Chapter 2 provides a detailed thesis literature review of the advent of next-generation prostate cancer biomarkers for personalized risk stratification, and the role that nanotechnology can play in translating these biomarkers for clinical screening usage. Specifically, it firstly introduces the biology of prostate cancer, as well the importance of disease risk stratification. Next, current prostate cancer screening assays and several promising next-generation prostate cancer biomarkers are highlighted. Following that, the use and potential of nanotechnological approaches in precision prostate cancer early detection and monitoring are reviewed. Finally, insights and future outlook towards the clinical translation of precision prostate cancer nanodiagnostics for next-generation biomarker detection are provided.

Chapter 2 is based on one review manuscript under editorial revision in Nature Reviews Urology.

Chapter 2

Merging Next-Generation Biomarkers and Nanodiagnostic Platforms for Precision Prostate Cancer Management

Kevin M. Koo, Paul N. Mainwaring, Scott A. Tomlins, Matt Trau

ABSTRACT The accurate identification and stratified treatment of clinically-significant early prostate cancer (PCa) has been an ongoing concern since the outcomes of large international PCa screening trials were reported. The controversy surrounding clinical and cost-benefits of PCa screening has highlighted the lack of strategies for discriminating high-risk disease requiring treatment as opposed to watchful waiting. To address this conundrum, there have been evolving advances in molecular subtyping and multi-omic nanotechnology-based PCa risk delineation to enable refinement of PCa molecular taxonomy into clinically meaningful and druggable subtypes. Due to inter- and intra-tumoural heterogeneity, there is an urgent need for novel rapid, cost-effective, accurate PCa nanodiagnostic technologies that distinguish clinically-significant PCa for precision oncology approaches. Herein we review the next generation of PCa biomarkers and the latest complementary nano-strategies, as well as discuss fundamental clinical translation challenges yet to be overcome.

INTRODUCTION Prostate cancer (PCa) is the most commonly-diagnosed solid malignancy in men, and aggressive sub-variants account for a significant proportion of male cancer-related deaths1. PCa typically affects men over 65 years of age, with a higher incidence in men of Caucasian or African descent as compared to other races and ethnicities. The majority of PCa is predominantly hormone-driven, and men with aggressive PCa will generally progress towards hormone-refractory/metastatic castration resistant PCa (CRPC) that is still largely influenced initially by androgen receptor activity. In recent years, comprehensive genomic, epigenetic, transcriptomic and proteomic analysis of the PCa landscape has revealed the extent of heterogeneity within the disease. These efforts have revealed common oncogenic drivers, such as point mutations in TP53, SPOP, and FOXA1; amplifications and copy number variations in AR (in CRPC); aberrations in DNA repair genes such as BRCA1/2,

Chapter 2

PIK3CA/B, PTEN, MYC; gene fusions involving the ETS gene family; and germline variants in susceptibility loci associated with predisposition to PCa development and metastatic progression, including ATM2-16. These endeavours have also refined our understanding of the PCa aetiology and progression, leading to potential advancements in precision treatment strategies17.

A significant portion of PCa is made up of localized, low grade, indolent disease that is unlikely to result in patient death. Ideally, PCa screening should be able to differentiate between these different aggressive and indolent PCa subtypes18, 19, in order to avoid over diagnosis as well unnecessary and potentially morbid therapy. However, to date, accurate definition and treatment of high-grade aggressive PCa remains a challenging and controversial conundrum in oncology. The present state of clinical PCa screening and management is heavily reliant on a blood-based biomarker-driven approach that measures serum prostate specific antigen (PSA) levels. Also known as kallikrein-3 (KLK3), PSA is a glycoprotein enzyme produced almost exclusively by the prostate gland but is not PCa- specific with elevated levels in a variety of non-PCa states such as increased age, benign prostatic hyperplasia (BPH), or inflamed/infected prostate (prostatitis)20. Since FDA-approval in 1986, an elevated serum PSA level has been used for opportunistic PCa screening; and this has, arguably, led to a revolution in PCa management by reducing PCa mortality through enabling early disease detection and treatment intervention. Yet, the use of PSA as a PCa screening biomarker is becoming increasingly controversial, due to its high tendency to provide false-positive and –negative diagnoses, as well as conflicting clinical screening trial results.

Two major large-scale randomized trials that have assessed the potential benefit of PSA screening – the European Randomized Study of Screening for Prostate Cancer (ERSPC) trial21 and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening (PLCO) trial22. The ERSPC trial investigated PSA screening in a largely unselected population drawn from seven different countries, and reported a 21% decrease in PCa mortality in men aged 55-60 years after 11 years of follow-up. The PLCO trial was conducted in the USA, and initially found no reduction in PCa mortality as well as no benefit of PSA screening22-24. However, a recent PLCO data reanalysis has found that PSA screening conferred up to 32% lower PCa mortality risk25, and suggested that initial conclusions were influenced by PSA testing contamination in the PLCO trial control group (approximately 85% of men underwent a PSA test at least

Chapter 2 once26, 27). Importantly, this contamination may have been a major cause for the reduction in statistical significance for the initial PLCO finding summary. Nonetheless, a general consensus of both trials is that widespread PSA-based PCa screening leads to PCa overdiagnosis and overtreatment with serious implications on men’s health and well-being28. In 2012, the United States Preventive Task Force (UPSTF) recommended against population- wide PSA screening for PCa, regardless of age29. This recommendation was associated with a significant decline in PCa screening and incidence30, 31, thus reducing the harmful widespread screening effects, but also simultaneously eliminating the known screening benefits for patients with the likelihood of developing advanced-stage PCa. As a result of vigorous international debate, UPSTF revised its PCa screening guidelines in 2016 to recommend clinicians informing men aged 55 – 69 years about the benefits and harms of PSA screening, and allowing for personalized shared decision making. In sum, an alternative approach to PCa screening might not be to abandon PSA testing altogether, but to design improved screening methodologies with better risk stratification capabilities32, 33.

Optimal PCa screening risk stratification calls for molecular subtyping to yield information on disease biology, prognosis, and treatment benefits. Yet, the molecular classification of PCa into disease subtypes for effective targeted therapies (such as that being currently performed for HER2 status-breast cancer subtype classification), is still an ongoing effort34-41. One notable example is the Stockholm 3 (STHLM3) study42 for early PCa screening, the first proof-of-concept population-based study of using a combination of plasma protein biomarker test panel, genetic polymorphism and clinical information for individualized risk-prediction of high-grade PCa. The study enrolled male participants aged 50-69 years without PCa, and performed both STHLM3 and PSA screening on each participant. Men with a PSA concentration of ≥3 ng/mL or being scored as high-risk on the STHLM3 model were referred for PCa examination and biopsies. The STHLM3 study reported superior performance in predicting aggressive PCa (GS≥7) with an area-under-the-curve (AUC) of 0.74 as compared with 0.56 for PSA levels alone, and resulted in a 32% reduction in prostate biopsies. Using the existing data, the STHLM3 model has currently been updated and shown to reduce biopsies by 34% when being used as a reflex test for men with PSA ≥3 ng/mL43.

To predict prognosis after prostatectomy, Spratt et al. have recently undertaken the intrinsic molecular subtyping study of 4236 primary PCa samples through unbiased high-throughput sequencing analyses of 100 genes. Through this study, three novel intrinsic PCa subtypes

Chapter 2 associated with known PCa drivers such as AR and ERG were described, and validated within both retrospective and prospective cohorts. Following radical prostatectomy, 10-year distant metastasis-free survival rates are 73.6%, 64.4%, and 57.1% across the three subtypes. This study also reported a molecular taxonomy describing patients who would benefit from postoperative radiotherapy. Hence, it is anticipated other large-scale studies of primary PCa molecular subtyping35-38 and multi-institutional research collaborations into personalized metastatic PCa treatments (eg. Stand Up To Cancer-Prostate Cancer Foundation Dream Team) will result in novel discoveries of biomarker-driven precision PCa management approaches44.

In light of the potential translational value of next-generation PCa biomarkers, there is a need for complementary cost-effective as well as easy-to-implement advanced detection technologies. At present, PCa biomarkers are being characterized with laboratory-based techniques which are fitting within the research setting or individual specialized clinical laboratories; but unsuitable for rapid, highly cost-effective, and point-of-care clinical diagnostics. The attractive potential of nanotechnology may alleviate the current limitations of molecular testing in a single reference laboratory or need for expensive proprietary equipment, thence enabling point-of-care development for broad cost-effective cancer diagnostics. Furthermore, the use of nanoparticles for nucleic acid or protein target detection has attained single molecule-level detection sensitivity, and are capable for use in the promising field of liquid biopsies whereby high analytical sensitivity is required for low target copies in circulation. In this Review, we discuss the progress of biomarker-driven PCa molecular subtyping and the development of companion nano-strategies to refine clinical biomarker detection. We also underscore the existing challenges of merging both aspects for precision PCa management, and provide our insights on possible resolutions.

NEXT-GENERATION PCa BIOMARKERS In recent years, molecular profiling advances in microarray profiling and next-generation sequencing have uncovered novel PCa-specific biomarkers with better disease-informing abilities (i.e. next generation PCa biomarkers)45-53 (TABLE 1; FIG. 1) that may be rapidly translated into the clinic to aid better diagnostic, prognostic and predictive algorithms. To feature non-invasive circulating next-generation PCa biomarkers, this section will categorically summarize urinary/blood-based biomarkers with potential clinical utility

Chapter 2 encompassing early diagnosis (TMPRSS2:ETS, PCA3); prognosis (SChLAP1); or targeted treatment prediction (PTEN, ARV7) in CRPC.

Table 1 Next-generation PCa biomarkers Next-Gen PCa Description Biological Samplin Potential Clincial Utility and Biomarkers Function g Source Performance TMPRSS2:ETS Chromosomal Drives the Tissue, Diagnosis and potentially fusion genes rearrangement overexpression Blood, prognosis based on transcript s of TMPRSS2 of oncogenic Urine levels. gene and ETS ERG protein AUC = 0.75-0.77 transcription under androgen (TMPRSS2:ERG+PCA3+PCP factors control in T risk calculator)172 tumorigenesis PCA3 Long non- Postulated to be Tissue, Diagnosis. coding RNA involved in PCa Blood, AUC = 0.750.77 exclusively cell survival, in Urine (TMPRSS2:ERG+PCA3+PCP expressed in part through T risk calculator)172 prostate tissue, modulating AR and signalling overexpressed in PCa. SChLAP1 Long non- Associated with Tissue, Prognosis. coding RNA cancer cell Urine AUC = 0.68 for metastatic PCa highly invasiveness prediction100 overexpressed and metastasis in a subset of by antagonizing aggressive the functions of PCa patients the SW1/SNF and solely chromatin- expressed in modifying PCa complex PTEN Most Activates the Tissue, Prognosis. frequently phosphoinositid Blood mutated and e-3-kinase deleted (PI3K) pathway suppressor to promote gene tumor initiation and growth. ARV7 Splice variant Maintain Blood Predictive of treatment of androgen androgen (CTCs) efficacy. receptor receptor- lacking ligand- regulated binding transcription in domain treatment- resistant PCa models

Chapter 2

Figure 1 Molecular biomarkers for precision PCa management. Progressive next- generation sequencing and bioinformatic analysis has yielded a suite of next-generation biomarkers and assays for PCa molecular subtyping from blood, urine, and tissue samples.

TMPRSS2:ETS fusion genes Chromosomal rearrangements have been found to be common aberrations in PCa, and through the carcinogenic processes of kaetegis and chromothripsis, lead to fusions of distant genes54-57. In 2005, by using a novel bioinformatics analysis, recurrent fusions between the 5’ promoter sequence of the androgen-regulated TMPRSS2 gene and ETS transcription factors (in particularly the coding sequence of the ERG gene) were described to be commonly present in PCa.58 TMPRSS2 encodes for a membrane-bound serine protease59 found expressed on prostate cells; and involved in a signal transduction pathway associated with PCa metastasis and invasion60. ERG translates into an oncogenic protein which is overexpressed in PCa to drive PIN to carcinoma transition61. TMPRSS2:ERG (T2:ERG) fusion genes are a result of chromosomal translocations or interstitial deletions between the two genes which are about 3 Mb apart on chromosome 2162-66. As TMPRSS2 contains androgen-sensitive elements, it was originally conceptualized that the fusion event puts ERG expression under androgen control, and drives the overexpression of oncogenic ERG protein in tumorigenesis67. However, breakthrough studies of late have revealed that that there are additional layers of complexities to this process68-74.

Altogether, T2:ERG is present approximately 50% (ranging from about 10% in Chinese75 to about 70% in northern European populations) of clinically-detected PCa cases, with rarer occurrences of fusions involving other ETS family members such as ETV1 or ETV558. Several

Chapter 2

T2:ERG isoforms may exist through alternate splicing and differing fusion junctions between the two genes, with the most common (>90%) isoform being the T1E4 gene fusion between exon 1 of TMPRSS2 and exon 4 of ERG76, 77. T2:ERG has displayed superior PCa specificity with general absence in non-PCa conditions, and largely detected in the precursor high-grade prostate intraepithelial neoplasia (HGPIN) in close proximity to malignant carcinoma. Hence, as one of the most PCa-specific biomarkers currently available, T2:ERG is highly attractive as a target for next-generation PCa diagnostic development. Recent studies have also suggested that T2:ERG-positive PCa is a distinct subtype which could be targeted by T2:ERG- or ERG-based therapeutics78-84.

Since the original T2:ERG description, numerous international studies85 have investigated the clinical utility of urinary T2:ERG detection. The prognostic potential of urinary T2:ERG is a source of debate, with some studies associating T2:ERG to aggressive PCa86-88 with other studies showing conflicting outcomes89. However, as T2:ERG is specifically expressed by tumor cells only; higher detected urinary T2:ERG levels could be linked to increased tumor volume which is indicative of high-grade PCa90. To further increase the performance of T2:ERG for non-invasive PCa stratification, a strategy utilizing a combination of next- generation urinary PCa biomarkers such as: T2:ERG and PCA3 for the Mi Prostate Score (MiPS) test, or ERG with PCA3 for the ExoDx Prostate IntelliScore assay, may be explored.

PCA3 PCA3 is a highly PCa-specific long non-coding RNA biomarker which is exclusively overexpressed in cancerous prostate tissues and HGPIN91, and has been postulated to be involved in PCa cell survival, in part through modulating AR signalling92. PCA3 was first described in 199993, and is one of the pioneering next-generation PCa screening biomarkers being incorporated into strategies to improve the diagnostic accuracy of PSA-based PCa screening. PCA3 is found to be detectable in urine and prostatic fluid as a non-invasive PCa biomarker, and numerous subsequent studies have investigated the clinical utility of urinary PCA394-96. The clinical sensitivity and specificity of urinary PCA3 for PCa detection is around 65% and 90% respectively. PCA3’s first major clinical contribution was through combination with serum PSA, resulting in a PCa detection improvement with an AUC of 0.75. This eventually led to an FDA-approved PCa diagnostic assay in 2012 for men with a previous negative prostate biopsy, and subsequently the assay was further refined upon

Chapter 2 through the incorporation of T2:ERG. This diagnostic evolutionary approach of combinations of next-generation PCa biomarkers is clearly improving PCa screening accuracy97.

SChLAP1 SChLAP1 is a lncRNA biomarker which is first discovered through bioinformatics analysis of a subset of cancers to identify selectively-upregulated lncRNAs associated with PCa recurrence and progression98. It has been found to be highly overexpressed in a subset of aggressive PCa patients and solely expressed in PCa99, and rarely in normal tissues or other cancer types. SChLAP1 is associated with cancer cell invasiveness and metastasis by antagonizing the tumor-suppressive functions of the SW1/SNF chromatin-modifying complex98. Separately, a large unbiased multi-institution analysis of genes related to metastasis and mortality after primary PCa radical prostatectomy, has revealed and validated SChLAP1 to be the top-ranked prognostic biomarker for metastasis development100. In this study, SChLAP1 was detectable non-invasively in PCa urine sediments and found to be a viable urinary biomarker candidate for discriminating high-risk aggressive PCa. The potential utility of SChLAP1 has been further shown in latter studies which associated SChLAP1 dysregulation as an aggressive feature in unfavorable intraductal and cribriform subpathologies101, and a demonstration of non-invasive SChLAP1 detection in PCa patient circulating tumor cells (CTCs)102.

PTEN Phosphatase and tensin homolog (PTEN) is the most frequently mutated and deleted (~30- 60%) tumor suppressor gene103-106 in high-grade PCa. PTEN loss is commonly used as a tissue biomarker in immunohistochemical detection, and is associated with poorer prognosis and aggressive metastatic PCa, as well as rapid development of hormonal treatment resistance. Recently, PTEN loss in CTCs and matched tumor tissue samples has shown strong correlation107, thus illustrating the potential use of liquid biopsies for detecting PTEN loss. Inactivation of PTEN, a tumour suppressor, in turn permits downstream signalling of the phosphoinositide-3-kinase (PI3K) pathway to cause increased phosphorylated AKT levels; promoting cancerous cell growth, proliferation, survival, and migration via multiple downstream promoters108. PTEN deletion has also been found to co-exist with T2:ERG presence to drive carcinogenesis in late-stage PCa109-114, and are considered to reflect particularly aggressive PCa phenotypes.

Chapter 2

ARV7 Androgen receptor splice variants lack the ligand-binding domain, and have been shown to be capable of maintaining androgen receptor-regulated transcription in treatment-resistant PCa models115-119. Androgen receptor splice variant 7 (ARV7), which has a deletion of exon 7120- 122, has been implicated in the development of castrate-resistant PCa123. A recent seminal study has suggested ARV7 transcript presence in CTCs of castrate-resistant PCa patients is associated with lower rate of PSA decline, as well as reduced progression-free and overall survival with androgen-mediated pathway targeted therapies such as enzalutamide and abiraterone124. Lately, in keeping with previous reports, a correlative study has found that patients with positive CTC nuclear ARV7 protein expression have superior overall survival with taxane therapy over enzalutamide and abiraterone125. Thus, ARV7 and other splice variants are being refined for their role as biomarkers for predicting treatment sensitivity in PCa patients. More recently, AR aberrations in circulating cell-free nucleic acids have been linked to enzalutamide and abiraterone resistance126-129, thus offering another minimally invasive approach for predicting therapeutic resistance in metastatic PCa. With promising data from the ongoing SPARTAN130, PROSPER131, STAMPEDE132, and LATITUDE133 trials investigating different therapeutic options for metastasis-free survival; studies into biomarkers (such as ARV7) associated with treatment-prediction will be invaluable and potentially practice-changing for high-risk PCa management.

EXISTING AND EMERGING BIOMARKER-DIRECTED PCa SCREENING DIAGNOSTICS Apart from the above prominent examples, there also exist several other promising biomarkers: other gene fusion variants involving other ETS genes such as ETV1, ETV4, ETV5; GSTP1- and EZH2- associated epigenetic changes; SPINK1; AMACR; and different variations of PSA proteins. A multi-omic approach which combines these different next- generation PCa biomarkers may provide information needed to define PCa therapy after initial diagnosis, and/or help define patient-specific characteristics that are important when determining therapy. As evident of the promise of biomarkers, there are presently several biomarker-based diagnostics (TABLE 2) clinically available to support PSA screening outcomes134, 135.

Chapter 2

Table 2 Biomarker-driven PCa assays for use on human clinical samples to support PSA testing Commercial Clinical Biomarkers Sampling Provider Certification Performance Molecular Utility Source PCa Tests PHI Improves Levels of Blood Beckman FDA AUC = 0.69- prediction of total PSA, Coulter, Inc 0.77 aggressive free PSA (CA, USA) disease and (fPSA), and risk of p2PSA (a progression PCa-specific during active fPSA surveillance isoform) 4Kscore® Discriminates Levels of Blood OPKO CLIA AUC = 0.81- aggressive total PSA, Laboratory 0.82 disease, and free PSA, (FL, USA) predict long- intact PSA, term risk of human cancer kallikrein 2 metastasis (KLK2) before biopsy or after negative biopsy ConfirmMDx Advises Methylation Tissue MDxHealth, CLIA Sensitivity = repeat biopsy levels of Inc. (CA, 0.68 decisions GSTP1, APC, USA) Specificity = and RASSF1 0.64 Oncotype Dx Advises Expression of Tissue Genomic CLIA NA appropriate 12 cancer- Health, Inc. treatment associated (CA, USA) such as active genes(and surveillance five reference or invasive genes) treatment representative during early of four diagnosis different biological pathways Prolaris Provides Expression Tissue Myraid CLIA NA disease levels of 31 Genetics, progression (and 15 Inc. (UT, risk in biopsy housekeeping USA) tissues genes) cell cycle progression genes Decipher® Predicts the Analysis of Tissue GenomeDx CLIA AUC = 0.82 metastatic 22 aggressive Biosciences, risk of PCa- Inc. (CA, metastatic associated USA) disease within RNA markers five years after radical

Chapter 2

prostatectomy MiPS Provides Levels of Urine MLabs (MI, CLIA AUC = 0.75- additional PCA3, USA) 0.77 information T2:ERG, for repeat KLK3 biopsy decision, and predicts risk of high-grade disease SelectMDx Predicts high- Levels of Urine MDxHealth, CLIA AUC = 0.71- grade disease, HOXC6, Inc. (CA, 0.77 and aids in DLX1, KLK3 USA) biopsy selection decisions ExoDx Improves Levels of Urinary Exosome CLIA AUC = 0.73- discrimination PCA3 and exosomes Diagnostics, 0.77 Prostate of GS≥7, T2:ERG Inc. (MA, IntelliScore GS=6, and USA) benign tumors

PHI and 4Kscore® In view of the poor sensitivity of total PSA alone, commercially available PCa screening blood tests such as the prostate health index (PHI) assay (Beckman Coulter, Inc) and four- kallikrein (4Kscore®) test (OPKO Laboratory) have been developed. Both tests utilize combinations of different serum PSA isoforms and/or related proteins for increased PCa- specific sensitivity.

PHI is a FDA-approved blood serum assay that combines the levels of total PSA, free PSA (fPSA), and p2PSA (a PCa-specific fPSA isoform). By using the formula: (p2PSA/fPSA)*PSA0.5, PHI has been shown to significantly outperform the use of total PSA and fPSA alone, with test outcomes demonstrating improved prediction of aggressive disease as well as the likelihood of progression during active surveillance136-138. Similar to PHI, the 4Kscore® test is a Clinical Laboratory Improvement Amendments (CLIA)-certified blood test which combines the levels of total PSA, free PSA, intact PSA, human kallikrein 2 (KLK2), and clinical information. The test can be administered before biopsy or after negative biopsy for better individualized PCa risk stratification. Several European and American studies have indicated that test outcomes could discriminate aggressive disease (GS ≥7), as well as predict long-term risk of cancer metastasis139-142.

Chapter 2

ConfirmMDx The ConfirmMDx test (MDxHealth, Inc.) is a CLIA-certified tissue-based assay which interrogates the methylation levels of a multi-gene panel (GSTP1, APC, and RASSF1) from biopsy samples143-145. The underlying biological concept behind the ConfirmMDx test lies in the occurrence of an epigenetic field effect whereby cells adjacent to cancer foci display DNA methylation changes which can be detected by the assay but undetectable by histopathology146-149. As there is a tendency for tumors to be missed by an initial round of transrectal ultrasound-guided biopsies, patients may often be subjected to multiple further biopsies. Through DNA methylation analysis, the ConfirmMDx test can advise repeat biopsy decisions to separate patients who have a true negative biopsy from those who may have occult cancer150-153. ConfirmMDx has an excellent 90% Negative Predictive Value (NPV), but a Positive Predictive Value (PPV) of only 28%.

Oncotype Dx The Oncotype Dx assay (Genomic Health, Inc.) is a CLIA-certified multi-gene early PCa detection assay based on a comparison between the individual’s tumor and healthy tissue at biopsy. The assay detects the expression of 12 cancer-associated genes representative of four different biological pathways, five reference genes and uses an algorithm to calculate a Genomic Prostate Score (GPS)154-156. For men recently diagnosed with early-stage PCa, the GPS have been demonstrated to risk stratify via prediction of probability of pathological characteristics, and advise the selection of appropriate treatment such as active surveillance or invasive treatment157.

Prolaris The Prolaris Score assay (Myraid) is CLIA-certified and directly measures cancer cell growth characteristics in biopsy tissues for disease prognosis and risk stratification. By utilizing the combined measured expression levels of 31 cell cycle progression and 15 housekeeping genes, the Prolaris Score assay is aimed at using proliferation to discriminate disease progression risk (i.e. low expression signature is correlated with low disease progression risk)158-160.

Decipher® The Decipher® test (GenomeDx) is a CLIA-certified tissue-based assay which aims to predict the risk of metastatic disease (independent of Gleason scoring or PSA test results)

Chapter 2 within five years after radical prostatectomy surgery161-164. This is achieved through the analysis of 22 aggressive PCa-associated RNA markers which were discovered from genome-wide search algorithms of more than a million markers, and has been extensively validated in collaboration with leading academic medical institutions. These 22 markers are associated with cell proliferation, migration, tumor motility, androgen signalling and immune system evasion.

Progensa and MiPS The Progensa test (Hologic) is the first FDA-approved urine-based molecular assay which quantifies the ratio of PCA3 to PSA mRNA levels non-invasively from urinary samples to assist in decisions for repeat-biopsies165-170. PCA3 is a unique non-PSA long non-coding RNA biomarker which has been found to be overexpressed in more than 95% of primary and metastatic PCa tissues, but not in normal or non-PCa tissues.93 Although the Progensa test has been shown for use in detecting PCa of higher-grade171, a lack of an optimal signal cutoff has thus far limited its prognostic ability.

PCA3 has also been combined with T2:ERG in a CLIA-certified urinary PCa early detection test called the MiPS that also incorporates serum PSA. MiPS provides additional information on whether a biopsy is needed by providing a risk estimate of PCa detection upon biopsy. Additionally, the combination of highly PCa-specific PCA3 and T2:ERG biomarkers in MiPs also predicts a patient’s risk of developing high-grade PCa172-177 and validation of alternative formulas combining T2:ERG and PCA3 has also been conducted178.

SelectMDx SelectMDx (MDx Health, Inc) is a CLIA-certified urinary PCa early detection assay which was first demonstrated in a Dutch study in 2015 to aid in selecting men for biopsy179. SelectMDx consists of the RT-PCR analysis of a three gene panel (HOXC6 and DLX1 overexpression, with KLK3 as an internal control) in combination with risk factors like PSA, PSA density, DRE, age and family history of prostate cancer. The HOXC6 and DLX1 biomarkers were selected from a biomarker discovery study based on gene expression profiling of tissue and urinary sediment samples179, and have been reported to be good predictors for high-grade PCa detection.180 SelectMDx has a reported NPV of 98% for aggressive PCa, and an AUC of 0.89.

Chapter 2

ExoDx Prostate IntelliScore Exosomes are nm-sized double-lipid membrane vesicles which are secreted from cells, and can be found in different biofluids such as blood or urine. A study in 2009 first reported that PCa cell-derived exosomes in urine contain both PCA3 and T2:ERG mRNA181. Using advances in exosomal RNA purification, this original discovery has subsequently been developed into the CLIA-certified ExoDx Prostate IntelliScore assay (Exosome Diagnostics). The assay assesses exosomal RNA levels of PCA3, ERG, and SPDEF genes by RT-PCR to derive an overall score for improved discrimination of GS≥7, GS=6, and benign PCa: AUC = 0.74 (95% CI 0.68-0.80) in 774 men. Given the increased research into the biomarker cargo in PCa cell-derived urinary exosomes, exosomes may potentially be a viable resource for PCa diagnosis and clinical management182, 183.

Multiparametric magnetic resonance imaging Advanced tissue imaging is currently surfacing as a pathological biomarker to supplement/enhance molecular biomarker testing in distinguishing aggressive tumor foci. Explicitly, multiparametric magnetic resonance imaging (mpMRI) is emerging as a promising biomarker for PCa screening, localization, staging, and risk stratification. A mpMRI combines three separate parameters (i.e. imaging techniques): T2-weighted (T2W) imaging, diffusion-weighted imaging (DWI), and dynamic contrast enhancement (DCE) imaging to provide detailed anatomical and functional prostate imaging.

In recent years, studies have indicated that mpMRI offers a biomarker for better discrimination between high- and low-risk PCa184-186, as well as more accurate mpMRI- guided biopsies to target tumors187-189. In the latest multicentre randomized PRECISION trial consisting of 500 men in total190, it was found that the use of mpMRI for pre-biopsy risk assessment and targeted biopsy was superior to standard transrectal ultrasonography-guided biopsy (TRUSGB). 95 (38%) men who underwent mpMRI-targeted biopsy were diagnosed with clinically-significant PCa, as compared to 64 men (26%) who went through TRUSGB. Concurrently, it was also observed that there is less overdetection of clinically-insignificant PC (9% vs. 22%), and fewer biopsy cores being required in mpMRI group than in the TRUSGB group. Together with molecular biomarker-driven approaches, mpMRI could aid in minimizing the over-diagnosis and –treatment of clinically insignificant PCa (thus reducing the number of unnecessary prostate biopsies).

Chapter 2

NANO-STRATEGIES FOR NEXT-GENERATION PCa BIOMARKER DETECTION The foregoing discussion provides an overview of biomarker-driven efforts in PCa molecular subtyping. To augment these advances in PCa biology, modern cancer nanotechnology research may yield next-generation PCa biomarker detection nano-strategies to revolutionize precision PCa management.

Nanotechnology applications in the field of cancer require cross-disciplinary research interfacing biology, chemistry, physics, engineering, and medicine191, 192. The basic rationale for utilizing nm-sized materials is to exploit the unique physical (eg. optical, magnetic, electronic, or structural) properties which transpire within nanoscale-sized range193. A familiar example of cancer nanotechnology application is the use of synthetic nanovectors such as liposome for therapeutic drug delivery into target cancerous tissues (e.g. liposomal doxorubicin). Beside cancer drug delivery applications, the advantage of a nanotechnological approach in being of similar dimensions to molecular biomarker targets is also beneficial for cancer diagnostics development.

Table 3 Examples of nanomaterials and nanoparticles used in the detection of different PCa biomarkers with enhanced limits-of-detection Nanotechnology PCa Biomarker Detection Limit Detection Medium Silicon nanowires PSA protein 0.8 pg/mL196 Serum

Carbon nanotubes PSA protein 4 pg/mL197 Serum, Tissue

Graphene PSA protein 8 pg/mL198 Serum Iron oxide Whole PCa tumor cell; 0.1 ng/mL (protein)204 Urine paramagnetic various isoforms of 1000 copies (RNA)208 nanoparticles PSA protein; T2:ERG, PCA3, SChLAP1 mRNA Quantum dots PSA protein 0.5 ng/mL213 Surface-enhanced Various isoforms of 12 pg/mL (protein)215 Serum, Urine Raman Scattering PSA protein; T2:ERG 100 copies (RNA)217 nanoparticles variants, PCA3, ARV7 mRNA

The emergence of nanotechnology-based approaches for PCa screening (TABLE 3) is extremely promising due to highly sensitive analytical detection features, easy clinical utility, and affordability. Essentially serving as general nucleic acid/protein/metabolite biomarker

Chapter 2 sensors, such nano-strategies impart remarkable detection capabilities without any specialized sample processing techniques. Increasingly, we have seen many instances of miniaturized platforms (eg. integrated diagnostics, wearables, implantables enabled by nano-components) for disease biomarker panel analysis after comprehensive genetic screening by next– generation sequencers (FIG. 2). This section will review the background and performance of nano-strategies (nanostructured materials, nanoparticles, and miniaturized integrated systems) that have been demonstrated for PCa biomarker detection.

Figure 2 A futuristic view of diagnostic miniaturization. Continuous innovations in the nanotechnology field have seen the evolution of clinical diagnostics from next-generation sequencers with massive genetic screening capabilities to miniaturized technologies for targeted biomarker analysis. We are currently on the cusp of witnessing the advent of nanoscaled diagnostics that could fundamentally change clinical practice though enhanced disease detection, treatment, and monitoring. Adapted from ref.219 with permission from the Royal Society of Chemistry.

Nanostructured materials Generally, nanostructured materials refer to constructions of nanoscaled dimensions and properties. Particularly, the aforementioned physical properties of such nanoscaled approaches are able to impart significant breakthroughs in detection speed and sensitivity. With regards to PCa nanodiagnostics which stemmed from nanostructured materials over the past decade, there have been developments that largely focus on PSA as a target for detection in clinical specimens194. A historical footnote is the first nanotechnology assay which used microcantilevers for PSA protein detection at low clinically-relevant detection limits195. This was followed by silicon-nanowire field-effect sensors which incorporated nanowires and surface PSA receptors into arrays for highly-sensitive PSA protein detection in undiluted serum samples196. In recent years, the rise of carbon nanotubes197 and graphene198 as new

Chapter 2 forms of super-conductive nanomaterials has also been reported for electrochemical PSA protein detection in human tissue and serum samples. Whilst such nanostructured materials have been demonstrated for breakthroughs in PSA detection sensitivity, there is a lack of successful clinical translation; likely due to the restricted academic research expertise in fabricating novel nanomaterials and a scarcity in progressing towards clinical studies from shortage of funding and/or diagnostic commercialization knowledge.

Nanoparticles The most commonly used nanoparticles for molecular diagnostic applications are metallic substrates of various shapes and transpire within the nm range. After a couple of decades of widespread research, the production and use of conventional nanoparticles are established and commercially available. Due to their high surface-to-volume ratio, nanoparticles serve as an ideal substrate for maximal loading of intended biological molecules onto the nanoparticle surface. In addition, as electron behaviours are constrained differently within nanostructures, this imbues the nanoscaled substrates with unique size-dependent magnetic, electronic, and optical properties which are useful for diagnostic applications. In terms of in vitro diagnostics, nanoparticles have been widely utilized in PCa nanodiagnostic applications199- 201. Iron oxide-core paramagnetic magnetic nanoparticles have mostly been used for isolation and purification of specific molecular targets prior to detection202. This has been demonstrated for the detection of different PSA protein isoforms and even whole PCa tumor cells203, 204. Aside from PSA detection, there has been progress in the last couple of years with the use of magnetic particles for detection of next-generation PCa RNA biomarkers such as T2:ERG, PCA3, and SChLAP1205-210. In particular, magnetic particle-based purification has enabled amplification-free electrochemical detection of these biomarkers at a fraction of PCR assay cost, and with improved sensitivity and speed. Lately, magnetic particles have also been showcased for dual functions in both the isolation and visual detection of next- generation PCa RNA biomarker (T2:ERG) non-invasively from urine samples211.

Quantum dots are nm-sized semiconducting particles with better enhanced fluorescence emissions212 than conventional organic fluorophores due to a quantum confinement effect of electron energy bands. With tunable size-dependent emission wavelengths and better photostability, quantum dots have mainly found applications in PSA protein detection213, 214. Although quantum dots are excellent for nucleic acid detection, the nature of multiple

Chapter 2 overlapping fluorescence emission spectra does limit their ability in multiplexed biomarker detection.

In this multiplexing regard, the use of surface-enhanced Raman scattering (SERS) nanoparticles might be more optimal. SERS nanoparticles generally consist of nanostructured metallic surfaces to greatly enhance the Raman scattering signals of surface-adsorbed biomolecules. Importantly, Raman spectra widths are much narrower than those of fluorescence, thus making SERS more suitable for multiplexed biomarker detection. SERS has been utilized in many studies for multiplexed detection of PCa biomarkers such as various PSA forms with excellent detection sensitivity215. Of late, SERS nanoparticles have also been employed in the detection of next-generation PCa biomarkers for potential molecular subtyping purposes216-218. A drawback of SERS nanoparticles is the degree of batch variability during bulk nanoparticle synthesis which could impact detection performance and accuracy. Thus, further research efforts into large-scale SERS nanoparticles with high reproducibility and stability are required to realize clinical SERS nanodiagnostics.

Miniaturized integrated systems Besides the synthesis of novel nanomaterials and nanoparticles, nanotechnology has also initiated the design of miniaturized device-based lab-on-a-chip or solution-based lab-in-a- drop systems. Such systems seek to downscale and integrate different aspects of biomarker analysis workflow (i.e patient sample preparation, target copy amplification, and target detection readout) onto a singular platform; thus minimizing patient sample requirement and sample-to-outcome turnaround time. A limitation of lab-on-a-chip systems is that a need for sophisticated and complex engineering is hampering the translation into widespread clinical use. Thence, the development of miniaturized systems which could similarly perform integrated biomarker analysis in absence of specialized fabrication is a potentially attractive alternative. Pointedly, the main conceived advantage of lab-in-a-drop diagnostics (FIG. 3)219 in avoiding the need for costly and specialized precise engineering; may be beneficial in the translation to practical commercial products for personalized disease applications.

Chapter 2

Figure 3 Lab-in-a-drop. A nanodiagnostic platform which miniaturizes an entire laboratory- based biomarker analysis workflow (crude sample preparation, target copy amplification, and biomarker detection readout) within a singular fluid droplet219. The above scheme illustrates a system in which different barcoded nanoparticles are used for multiplex quantification of various PCa biomarkers in patient urinary samples217. Lab-in-a drop nanodiagnostics could be a rapid and cost-effective tool for interrogating cancer biomarkers from liquid biopsies with high sensitivity and reliability at point-of-care.

CHALLENGES IN CLINICAL TRANSLATION OF NOVEL BIOMARKERS AND NANO-STRATEGIES The above-mentioned nano-sized technology solutions represent the field’s work towards the nano-optimization of PCa screening. However, it is worth noting that PCa nanodiagnostics is presently at an early stage of promise, and there are necessary steps to be taken for successful translation into viable precision clinical screening tools. This section covers our perspectives on the challenges and potential solutions for the clinical translation of both next-generation biomarkers and nanodiagnostic technologies within the next decade.

Validation of biomarker panel With the availability of high-throughput next-generation sequencing technologies today for whole-exome and transcriptomic sequencing of cancer specimens, the discovery of new and better biomarkers in PCa and other cancer types has progressed at an exponential rate but may be nearing saturation220. Given the heterogeneity of PCa, it is unlikely that a single

Chapter 2 biomarker is sufficient for disease risk stratification. It is more probable that a panel of several biomarkers is required, and the main challenge would be the assembly of such a molecular target panel for precision PCa management221. For PCa, the ideal biomarker-driven screening test needs to show superior clinical sensitivity and specificity, and demonstrate significant improvement to the ROC curve in comparison to the PSA test and routinely- available clinicopathological variables. To ensure maximal efficacy for clinical use, proper and in-depth biomarker validation needs to be performed through well-designed and - conducted biomarker clinical trials with validation across different institutions. Moreover, from a clinical perspective, a PCa diagnostic biomarker panel may need to be modified for patients of different germline heritage, especially in underrepresented populations (eg. different PCa drivers in men of African descent as compared to Caucasian descent)222. For treatment biomarkers, smarter trial designs such as adaptive enrichment approaches223, 224 may effectively help identify predictive PCa biomarkers to benefit from specific treatments.

Clinical verification of nanodiagnostics To date, the translation success of nanodiagnostics from research concept to clinical use has been limited. The foremost issue is that nanodiagnostics are presently being evaluated mostly based on their analytical performance in research laboratories, with lack of continued evaluation in a clinical setting. While the establishment of analytical detection sensitivity, selectivity, and repeatability is no doubt crucial; the subsequent clinical evaluation of these nanodiagnostics to test robustness, reproducibility, standardization, and applicability for actual patient use is generally being overlooked, thereby impeding most promising nanotechnologies from progressing towards clinical translation use. Thus, this is a significant obstacle to be overcome for promising PCa nanodiagnostic technologies to be adopted for clinical use. Clinical evaluation involves the testing of novel nanotechnologies on a cohort of patient samples (of appropriate size and context) to establish clinical sensitivity, specificity, PPV, NPV, and AUC. Lately, progress on this front has been made by employing a relevant and proven MiPS PCa clinical biomarker model to comprehensively verify the clinical performance of SERS nanodiagnostic technology225. The evaluation of these clinical parameters will enable the generation of clinically-relevant biomarker detection limits, which is notably different from its analytical counterparts. For instance, analytical detection limit of single-molecule level is an outstanding limit-of-detection analytically, however, might not be of great clinical utility if there is no disease implication (i.e. no clinical action required) when

Chapter 2 an oncogenic biomarker is present at a single copy. This may especially be of significance in PCa, where indolent disease is extremely common and there is a need to avoid overtreatment.

Acceleration of biomarker and nano-strategy validation process Comprehensive clinical validation and translation of novel PCa biomarkers and new biomarker-based nanotechnologies needs to be effectively accelerated (typically 5-10 years for each) to provide faster patient benefit. A feasible way to streamline this process might be the simultaneous combined evaluation of new nanodiagnostic technologies on novel biomarkers to investigate potential clinical benefits, thus leading to a shorter overall timeframe. This move may be realized by carefully-designed clinical studies with well- established pre-study research objectives regarding the choice of nano-strategy and biomarker panel, and relevant patient cohorts and samples.

Standardization of protocols It is of a paramount concern that protocols related to PCa biomarker analysis and nanodiagnostic development are now varied across different research institutions. Under the circumstances of collaborative biomarker studies within larger consortia, care must be taken to minimize systematic as well as random variation with proper control conditions to allow proper head-to-head comparisons and cross-validation of related studies. Particularly, the standardization of patient sample processing and molecular testing processes in biomarker analysis are pivotal. Recently, a combination of standardized whole urine sample collection protocol, automated urinary RNA extraction and PCR platform has been demonstrated to enhance the overall analytical performance of urinary RNA assays such as SelectMDx226. For nanodiagnostic development, there needs to be standardization of manufacturing and synthesis processes and storage conditions to minimize batch-to-batch variability and ensure optimal functions during applications.

Multidisciplinary collaboration The successful realization of a novel clinical PCa diagnostic test involving next-generation PCa biomarker detection requires complementary cutting-edge detection approaches. In this regard, the role of nanotechnology in enabling rapid, highly sensitive, accurate, and affordable biomolecular target sensing is greatly encouraging. Due to relatively poor awareness of next-generation PCa biomarkers in the nanotechnology research field, we have

Chapter 2 seen the emergence of many innovative nanotechnology-based biosensing techniques from research groups around the world which have only focused on improving PSA detection as a PCa diagnosis model. This calls for more frequent interactions between the fields of clinical and nanodiagnostic research in the form of multidisplinary conferences and forums, so that nanotechnologists could be kept up to date with the most relevant clinical information on biomarker discovery and advances (and clinicians, in turn, with nanodiagnostic revolutions). This will allow state-of-the-art diagnostic nanotechnologies to be designed for the most promising and exciting cancer biomarkers available, thus directing efforts to address the most relevant and pressing need in clinics.

CONCLUSIONS AND FUTURE OUTLOOK In conclusion, to allow accurate PCa risk stratification, a single biomarker is unlikely to be sufficient to achieve the required diagnostic sensitivity and specificity. Emerging evidence from panels of various next-generation PCa biomarkers shows effective PCa classification into different molecular subtypes that are potentially amenable for personalized treatment strategies. Besides early detection purposes, the use of PCa molecular subtyping could extend to the guidance of PCa treatment decisions. It is anticipated that research efforts in the next decade will focus on the clinical evaluation and translation of different combinations of next- generation PCa biomarkers, in order to elucidate a multi-omic comprehensive panel which possess high discrimination ability for clinically-significant PCa subtypes.

Besides the translation of next-generation PCa biomarkers, the importance of developing superior detection technologies to target these biomarkers in human specimens is equally essential to justify changing clinical as well as regulatory screening paradigms. In this aspect, the unique benefits of nanotechnological approaches could be exploited for higher detection sensitivity, accuracy, and speed at a reduced cost. The urinary detection of PCa biomarkers could offer a truly non-invasive form of PCa liquid biopsy which may yet overcome the challenges of tumor heterogeneity associated with surgical tissue biopsies227. While preliminary results have been promising with the introduction of several urine-based PCa tests in the clinic, the possibility of more powerful urine-based diagnostics which could provide an accurate overall disease molecular snapshot, demands for further clinical evaluation before clinical translation. With the many innovative nanodiagnostic developments internationally, the potential of cancer nanotechnology in improving current

Chapter 2

PCa standard-of-care is optimistic. To translate these nanotechnologies for clinical use in the near future, researchers must properly evaluate their developed methodologies in relevant patient cohorts, established clinically-relevant detection limits, and comprehensively evaluate clinical performance parameters.

Lastly, it is of our view that optimal outcomes for PCa patients will be derived from the synergy of ideas and knowledge integrated from different fields brought together under the umbrella of cross-platform meetings. We hope that this review can serve to bridge the gap in knowledge between the PCa biological/clinical and nanodiagnostic fields: by informing urologic clinicians about future possibilities and benefits of PCa nanotechnology-based screening, and alerting nanotechnologists to the availability of next-generation PCa biomarkers for integration into their nanodiagnostic research.

REFERENCES 1. Siegel, R.L., Miller, K.D. & Jemal, A. Cancer statistics, 2017. Ca-a Cancer Journal for Clinicians 67, 7-30 (2017). 2. Robinson, D. et al. Integrative clinical genomics of advanced prostate cancer. Cell 161, 1215-1228 (2015). 3. Robinson, D.R. et al. Integrative clinical genomics of metastatic cancer. Nature 548, 297-303 (2017). 4. Gundem, G. et al. The evolutionary history of lethal metastatic prostate cancer. Nature 520, 353-357 (2015). 5. Beltran, H. et al. Divergent clonal evolution of castration-resistant neuroendocrine prostate cancer. Nature Medicine 22, 298-305 (2016). 6. Fraser, M. et al. Genomic hallmarks of localized, non-indolent prostate cancer. Nature 541, 359-364 (2017). 7. Hong, M.K.H. et al. Tracking the origins and drivers of subclonal metastatic expansion in prostate cancer. Nature Communications 6, 6605(2015). 8. Kumar, A. et al. Substantial interindividual and limited intraindividual genomic diversity among tumors from men with metastatic prostate cancer. Nature Medicine 22, 369-378 (2016).

Chapter 2

9. Smith, B.A. et al. A basal stem cell signature identifies aggressive prostate cancer phenotypes. Proceedings of the National Academy of Sciences of the United States of America 112, E6544-E6552 (2015). 10. Carreira, S. et al. Tumor clone dynamics in lethal prostate cancer. Science Translational Medicine 6, 254ra125 (2014). 11. Barbieri, C.E. et al. The mutational landscape of prostate cancer. European Urology 64, 567-576 (2013). 12. Taylor, R.A. et al. Germline BRCA2 mutations drive prostate cancers with distinct evolutionary trajectories. Nature Communications 8, 13671 (2017). 13. Spratt, D.E., Zumsteg, Z.S., Feng, F.Y. & Tomlins, S.A. Translational and clinical implications of the genetic landscape of prostate cancer. Nature Review Clinical Oncology 13, 597-610 (2016). 14. Schumacher, F.R. et al. Association analyses of more than 140,000 men identify 63 new prostate cancer susceptibility loci. Nature Genetics 50, 928-936 (2018). 15. Quigley, D.A. et al. Genomic hallmarks and structural variation in metastatic prostate cancer. Cell doi:10.1016/j.cell.2018.06.039 (2018). 16. Mijuskovic, M. et al. Rare germline variants in DNA repair genes and the angiogenesis pathway predispose prostate cancer patients to develop metastatic disease. British Journal of Cancer 119, 96-104 (2018). 17. Wedge, D.C. et al. Sequencing of prostate cancers identifies new cancer genes, routes of progression and drug targets. Nature Genetics 50, 682-692 (2018). 18. Attard, G. & Beltran, H. Prioritizing precision medicine for prostate cancer. Annals of Oncology 26, 1041-1042 (2015). 19. Irshad, S. et al. A molecular signature predictive of indolent prostate cancer. Science Translational Medicine 5, 202ra122 (2013). 20. Polascik, T.J., Oesterling, J.E. & Partin, A.W. Prostate specific antigen: A decade of discovery - What we have learned and where we are going. Journal of Urology 162, 293-306 (1999). 21. Schroeder, F.H. et al. Screening and prostate-cancer mortality in a randomized european study. New England Journal of Medicine 360, 1320-1328 (2009). 22. Andriole, G.L. et al. Mortality results from a randomized prostate-cancer screening trial. New England Journal of Medicine 360, 1310-1319 (2009).

Chapter 2

23. Andriole, G.L. et al. Prostate cancer screening in the randomized Prostate, Lung, Colorectal, and Ovarian cancer screening trial: mortality results after 13 years of follow-up. Journal of the National Cancer Institute 104, 125-132 (2012). 24. Pinsky, P.F. et al. Extended mortality results for prostate cancer screening in the PLCO trial with median follow-up of 15 years. Cancer 123, 592-599 (2017). 25. Tsodikov, A., Gulati, R., Heijnsdijk, E.M. & et al. REconciling the effects of screening on prostate cancer mortality in the ERSPC and PLCO trials. Annals of Internal Medicine (2017). 26. Shoag, J.E., Mittal, S. & Hu, J.C. Reevaluating PSA testing rates in the PLCO trial. New England Journal of Medicine 374, 1795-1796 (2016). 27. Pinsky, P.F. et al. Assessing contamination and compliance in the prostate component of the Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer screening trial. Clinical Trials 7, 303-311 (2010). 28. Pinsky, P.F., Prorok, P.C. & Kramer, B.S. Prostate cancer screening - a perspective on the current state of the evidence. New England Journal of Medicine 376, 1285-1289 (2017). 29. Moyer, V.A. & Force, U.S.P.S.T. screening for prostate cancer: US Preventive Services Task Force recommendation statement. Annals of Internal Medicine 157, 120-134 (2012). 30. Jemal, A. et al. Prostate cancer incidence and psa testing patterns in relation to USPSTF screening recommendations. JAMA-Journal of the American Medical Association 314, 2054-2061 (2015). 31. Sammon, J.D. et al. Prostate-specific antigen screening After 2012 US Preventive Services Task Force recommendations. JAMA-Journal of the American Medical Association 314, 2075-2077 (2015). 32. Carter, H.B. et al. Early detection of prostate cancer: AUA guideline. Journal of Urology 190, 419-426 (2013). 33. Gandaglia, G. et al. The problem is not what to do with indolent and harmless prostate cancer-the problem is how to avoid finding these cancers. European Urology 70, 547- 548 (2016). 34. Carlsson, S.V. & Kattan, M.W. Prostate cancer: Personalized risk - stratified screening or abandoning it altogether? Nature Review Clinical Oncology 13, 140-142 (2016).

Chapter 2

35. Grasso, C.S. et al. Integrative molecular profiling of routine clinical prostate cancer specimens. Annals of Oncology 26, 1110-1118 (2015). 36. Abeshouse, A. et al. The molecular taxonomy of primary prostate cancer. Cell 163, 1011-1025 (2015). 37. Tomlins, S.A. et al. Characterization of 1577 primary prostate cancers reveals novel biological and clinicopathologic insights into molecular subtypes. European Urology 68, 555-567 (2015). 38. Kaffenberger, S.D. & Barbieri, C.E. Molecular subtyping of prostate cancer. Current Opinion in Urology 26, 213-218 (2016). 39. Paulo, P. et al. Molecular subtyping of primary prostate cancer reveals specific and shared target genes of different ETS rearrangements. Neoplasia 14, 600-611 (2012). 40. Smith, S.C. & Tomlins, S.A. Prostate cancer SubtyPINg BiomarKers and outcome: is Clarity EmERGing? Clinical Cancer Research 20, 4733-4736 (2014). 41. Na, R. et al. Germline mutations in ATM and BRCA1/2 distinguish risk for lethal and indolent prostate cancer and are associated with early age at death. European Urology 71, 740-747 (2017). 42. Grönberg, H. et al. Prostate cancer screening in men aged 50-69 years (STHLM3): a prospective population-based diagnostic study. Lancet Oncology 16, 1667-1676 (2015). 43. Ström, P. et al. The Stockholm-3 model for prostate cancer detection: algorithm update, biomarker contribution, and reflex test potential. European Urology 74, 204- 210 (2018). 44. Barbieri, C.E., Chinnaiyan, A.M., Lerner, S.P., Swanton, C. & Rubin, M.A. The emergence of precision urologic oncology: a collaborative review on biomarker- driven therapeutics. European Urology 71, 237-246 (2017). 45. Prensner, J.R., Rubin, M.A., Wei, J.T. & Chinnaiyan, A.M. Beyond PSA: The next generation of prostate cancer biomarkers. Science Translational Medicine 4, 127rv3 (2012). 46. Dijkstra, S., Mulders, P.F.A. & Schalken, J.A. Clinical use of novel urine and blood based prostate cancer biomarkers: a review. Clinical Biochemistry 47, 889-896 (2014). 47. Wu, D. et al. Urinary biomarkers in prostate cancer detection and monitoring progression. Critical Reviews in Oncology/Hematology 118, 15-26 (2017).

Chapter 2

48. Velonas, V.M., Woo, H.H., dos Remedios, C.G. & Assinder, S.J. Current status of biomarkers for prostate cancer. International Journal of Molecular Sciences 14, 11034-11060 (2013). 49. Jedinak, A. et al. Novel non-invasive biomarkers that distinguish between benign prostate hyperplasia and prostate cancer. BMC Cancer 15, 259(2015). 50. Cuzick, J. et al. Prevention and early detection of prostate cancer. Lancet Oncology 15, E484-E492 (2014). 51. Hessels, D. & Schalken, J.A. Urinary biomarkers for prostate cancer: a review. Asian Journal of Andrology 15, 333-339 (2013). 52. Narayan, V.M., Konety, B.R. & Warlick, C. Novel biomarkers for prostate cancer: an evidence-based review for use in clinical practice. Internation Journal of Urology 24, 352-360 (2017). 53. Ploussard, G. & de la Taille, A. Urine biomarkers in prostate cancer. Nature Reviews Urology 7, 101-109 (2010). 54. Prensner, J.R. & Chinnaiyan, A.M. Oncogenic gene fusions in epithelial carcinomas. Current Opinions in Genetic & Development 19, 82-91 (2009). 55. Kumar-Sinha, C., Kalyana-Sundaram, S. & Chinnaiyan, A.M. Landscape of gene fusions in epithelial cancers: seq and ye shall find. Genome Medicine 7, 129 (2015). 56. Kumar-Sinha, C., Tomlins, S.A. & Chinnaiyan, A.M. Recurrent gene fusions in prostate cancer. Nature Review Cancer 8, 497-511 (2008). 57. Edwards, P.A.W. Fusion genes and chromosome translocations in the common epithelial cancers. Journal of Pathology 220, 244-254 (2010). 58. Tomlins, S.A. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science 310, 644-648 (2005). 59. PaoloniGiacobino, A., Chen, H.M., Peitsch, M.C., Rossier, C. & Antonarakis, S.E. Cloning of the TMPRSS2 gene, which encodes a novel serine protease with transmembrane, LDLRA, and SRCR domains and maps to 21q22.3. Genomics 44, 309-320 (1997). 60. Wilson, S. et al. The membrane-anchored serine protease, TMPRSS2, activates PAR- 2 in prostate cancer cells. Biochemistry Journal 388, 967-972 (2005). 61. Carver, B.S. et al. ETS rearrangements and prostate cancer initiation. Nature 457, E1 (2009).

Chapter 2

62. Linn, D.E., Penney, K.L., Bronson, R.T., Mucci, L.A. & Li, Z. Deletion of interstitial genes between TMPRSS2 and ERG promotes prostate cancer progression. Cancer Research 76, 1869-1881 (2016). 63. Clark, J. et al. Diversity of TMPRSS2-ERG fusion transcripts in the human prostate. Oncogene 26, 2667-2673 (2007). 64. Clark, J.P. & Cooper, C.S. ETS gene fusions in prostate cancer. Nature Reviews Urology 6, 429-439 (2009). 65. Barbieri, C.E. & Rubin, M.A. Genomic rearrangements in prostate cancer. Current Opinion in Urology 25, 71-76 (2015). 66. Mani, R.S. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Reports 17, 2620-2631 (2016). 67. Adamo, P. & Ladomery, M.R. The oncogene ERG: a key factor in prostate cancer. Oncogene 35, 403-414 (2016). 68. Kron, K.J. et al. TMPRSS2-ERG fusion co-opts master transcription factors and activates NOTCH signaling in primary prostate cancer. Nature Genetics 49, 1336- 1345 (2017). 69. Bose, R. et al. ERF mutations reveal a balance of ETS factors controlling prostate oncogenesis. Nature 546, 671-675 (2017). 70. Kim, J., Wu, L., Zhao, J.C., Jin, H.J. & Yu, J. TMPRSS2-ERG gene fusions induce prostate tumorigenesis by modulating microRNA miR-200c. Oncogene 33, 5183- 5192 (2014). 71. Wu, L.T. et al. ERG is a critical regulator of Wnt/LEF1 signaling in prostate cancer. Cancer Research 73, 6068-6079 (2013). 72. Kedage, V. et al. An interaction with Ewing's Sarcoma breakpoint protein EWS defines a specific oncogenic mechanism of ETS factors rearranged in prostate cancer. Cell Reports 17, 1289-1301 (2016). 73. Lucas, J.M. et al. The androgen-regulated protease TMPRSS2 activates a proteolytic cascade involving components of the tumor microenvironment and promotes prostate cancer metastasis. Cancer Discovery 4, 1310-1325 (2014). 74. Tian, T.V. et al. Identification of novel TMPRSS2: ERG mechanisms in prostate cancer metastasis: involvement of MMP9 and PLXNA2. Oncogene 33, 2204-2214 (2014).

Chapter 2

75. Ren, S.C. et al. Whole-genome and transcriptome sequencing of prostate cancer identify new genetic alterations driving disease progression. European Urology 73, 322-339 (2018). 76. St. John, J., Powell, K., Conley-LaComb, M.K. & Chinni, S.R. TMPRSS2-ERG fusion gene expression in prostate tumor cells and its clinical and biological significance in prostate cancer progression. Journal of Cancer Science and Therapy 4, 94-101 (2012). 77. Hu, Y. et al. Delineation of TMPRSS2-ERG splice variants in prostate cancer. Clinical Cancer Research 14, 4719-4725 (2008). 78. Attard, G. et al. Characterization of ERG, AR and PTEN gene status in circulating tumor cells from patients with castration-resistant prostate cancer. Cancer Research 69, 2912-2918 (2009). 79. Wang, X.J. et al. Development of peptidomimetic inhibitors of the ERG gene fusion product in prostate cancer. Cancer Cell 31, 532-548 (2017). 80. Chatterjee, P. et al. The TMPRSS2-ERG gene fusion blocks XRCC4-mediated nonhomologous end-joining repair and radiosensitizes prostate cancer cells to PARP inhibition. Molecular Cancer Therapeutics 14, 1896-1906 (2015). 81. Furusato, B. et al. ERG oncoprotein expression in prostate cancer: clonal progression of ERG-positive tumor cells and potential for ERG-based stratification. Prostate Cancer Prostatic Diseases 13, 228-237 (2010). 82. White, N.M., Feng, F.Y. & Maher, C.A. Recurrent rearrangements in prostate cancer: causes and therapeutic potential. Current Drug Targets 14, 450-459 (2013). 83. Hessels, D. & Schalken, J.A. Recurrent gene fusions in prostate cancer: their clinical implications and uses. Current Urology Reports 14, 214-222 (2013). 84. Kissick, H.T., Sanda, M.G., Dunn, L.K. & Arredouani, M.S. Development of a peptide-based vaccine targeting TMPRSS2:ERG fusion-positive prostate cancer. Cancer Immunology Immunotherapy 62, 1831-1840 (2013). 85. Sanguedolce, F. et al. Urine TMPRSS2: ERG fusion transcript as a biomarker for prostate cancer: literature review. Clinical Genitourinary Cancer 14, 117-121 (2016). 86. Wang, J.H., Cai, Y., Ren, C.X. & Ittmann, M. Expression of variant TMPRSS2/ERG fusion messenger RNAs is associated with aggressive prostate cancer. Cancer Research 66, 8347-8351 (2006). 87. Wang, J.H. et al. Pleiotropic biological activities of alternatively spliced TMPRSS2/ERG Fusion gene transcripts. Cancer Research 68, 8516-8524 (2008).

Chapter 2

88. Park, K. et al. TMPRSS2:ERG gene fusion predicts subsequent detection of prostate cancer in patients with high-grade prostatic intraepithelial Neoplasia. Journal of Clinical Oncology 32, 206-211 (2014). 89. Culig, Z. TMPRSS:ERG fusion in prostate cancer: from experimental approaches to prognostic studies. European Urology 66, 861-862 (2014). 90. Young, A. et al. Correlation of urine TMPRSS2:ERG and PCA3 to ERG plus and total prostate cancer burden. American Journal of Clinical Pathology 138, 685-696 (2012). 91. Day, J.R., Jost, M., Reynolds, M.A., Groskopf, J. & Rittenhouse, H. PCA3: From basic molecular science to the clinical lab. Cancer Letters 301, 1-6 (2011). 92. Ferreira, L.B. et al. PCA3 noncoding RNA is involved in the control of prostate- cancer cell survival and modulates androgen receptor signaling. BMC Cancer 12, 507 (2012). 93. Bussemakers, M.J.G. et al. DD3: A new prostate-specific gene, highly overexpressed in prostate cancer. Cancer Research 59, 5975-5979 (1999). 94. Cui, Y. et al. Evaluation of prostate cancer antigen 3 for detecting prostate cancer: a systematic review and meta-analysis. Scientific Reports 6, 25776 (2016). 95. Lee, G.L., Dobi, A. & Srivastava, S. Diagnostic performance of the PCA3 urine test. Nature Reviews Urology 8, 123-124 (2011). 96. Deras, I.L. et al. PCA3: A molecular urine assay for predicting prostate biopsy outcome. Journal of Urology 179, 1587-1592 (2008). 97. Laxman, B. et al. A first-generation multiplex biomarker analysis of urine for the early detection of prostate cancer. Cancer Research 68, 645-649 (2008). 98. Prensner, J.R. et al. The long noncoding RNA SChLAP1 promotes aggressive prostate cancer and antagonizes the SWI/SNF complex. Nature Genetics 45, 1392- 1398 (2013). 99. Mehra, R. et al. Overexpression of the long non-coding RNA SChLAP1 independently predicts lethal prostate cancer. European Urology 70, 549-552 (2016). 100. Prensner, J.R. et al. RNA biomarkers associated with metastatic progression in prostate cancer: a multi-institutional high-throughput analysis of SChLAP1. The Lancet Oncology 15, 1469-1480 (2014). 101. Chua, M.L.K. et al. A prostate cancer "nimbosus'': genomic instability and SChLAP1 dysregulation underpin aggression of intraductal and cribriform subpathologies. European Urology 72, 665-674 (2017).

Chapter 2

102. Yin, C.Q. et al. Molecular profiling of pooled circulating tumor cells from prostate cancer patients using a dual-antibody-functionalized microfluidic device. Anal. Chem. 90, 3744-3751 (2018). 103. Di Cristofano, A. & Pandolfi, P.P. The multiple roles of PTEN in tumor suppression. Cell 100, 387-390 (2000). 104. Yoshimoto, M. et al. Interphase FISH analysis of PTEN in histologic sections shows genomic deletions in 68% of primary prostate cancer and 23% of high-grade prostatic intra-epithelial neoplasias. Cancer Genetics and Cytogenetics 169, 128-137 (2006). 105. Murphy, S.J. et al. Integrated analysis of the genomic instability of PTEN in clinically insignificant and significant prostate cancer. Modern Pathology 29, 143-156 (2016). 106. Phin, S., Moore, M.W. & Cotter, P.D. Genomic Rearrangements of PTEN in Prostate Cancer. Frontiers in Oncology 3, 240 (2013). 107. Punnoose, E.A. et al. PTEN loss in circulating tumour cells correlates with PTEN loss in fresh tumour tissue from castration-resistant prostate cancer patients. British Journal of Cancer 113, 1225-1233 (2015). 108. Carnero, A., Blanco-Aparicio, C., Renner, O., Link, W. & Leal, J.F.M. The PTEN/PI3K/AKT signalling pathway in cancer, therapeutic implications. Current Cancer Drug Targets 8, 187-198 (2008). 109. Krohn, A. et al. Genomic deletion of PTEN is associated with tumor progression and early PSA recurrence in ERG fusion-positive and fusion-negative prostate cancer. American Journal of Pathology 181, 401-412 (2012). 110. Yoshimoto, M. et al. Absence of TMPRSS2:ERG fusions and PTEN losses in prostate cancer is associated with a favorable outcome. Modern Pathology 21, 1451- 1460 (2008). 111. Han, B. et al. Fluorescence in situ hybridization study shows association of PTEN deletion with ERG rearrangement during prostate cancer progression. Modern Pathology 22, 1083-1093 (2009). 112. Ahearn, T.U. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. JNCI-Journal of the National Cancer Institute 108, djv346 (2016). 113. King, J.C. et al. Cooperativity of TMPRSS2-ERG with PI3-kinase pathway activation in prostate oncogenesis. Nature Genetics 41, 524-526 (2009).

Chapter 2

114. Carver, B.S. et al. Aberrant ERG expression cooperates with loss of PTEN to promote cancer progression in the prostate. Nature Genetics 41, 619-624 (2009). 115. Haile, S. & Sadar, M.D. Androgen receptor and its splice variants in prostate cancer. Cellular and Molecular Life Sciences 68, 3971-3981 (2011). 116. Hu, R. et al. Ligand-independent androgen receptor variants derived from splicing of cryptic exons signify hormone-refractory prostate cancer. Cancer Research 69, 16-22 (2009). 117. Hornberg, E. et al. Expression of androgen receptor splice variants in prostate cancer bone metastases is associated with castration-resistance and short survival. PLoS One 6, e19059 (2011). 118. Ware, K.E., Garcia-Blanco, M.A., Armstrong, A.J. & Dehm, S.M. Biologic and clinical significance of androgen receptor variants in castration resistant prostate cancer. Endocrine-Related Cancer 21, T87-T103 (2014). 119. Dehm, S.M. & Tindall, D.J. Alternatively spliced androgen receptor variants. Endocrine-Related Cancer 18, R183-R196 (2011). 120. Sun, S.H. et al. Castration resistance in human prostate cancer is conferred by a frequently occurring androgen receptor splice variant. Journal of Clinical Investigation 120, 2715-2730 (2010). 121. Watson, P.A. et al. Constitutively active androgen receptor splice variants expressed in castration-resistant prostate cancer require full-length androgen receptor. Proceedings of the National Academy of Sciences of the United States of America 107, 16759-16765 (2010). 122. Liu, L.L. et al. Mechanisms of the androgen receptor splicing in prostate cancer cells. Oncogene 33, 3140-3150 (2014). 123. Guo, Z.Y. et al. A novel androgen receptor splice variant is up-regulated during prostate cancer progression and promotes androgen depletion-resistant growth. Cancer Research 69, 2305-2313 (2009). 124. Antonarakis, E.S. et al. AR-V7 and resistance to enzalutamide and abiraterone in prostate cancer. New England Journal of Medicine 371, 1028-1038 (2014). 125. Scher, H.I., Graf, R.P., Schreiber, N.A. & et al. Assessment of the validity of nuclear-localized androgen receptor splice variant 7 in circulating tumor cells as a predictive biomarker for castration-resistant prostate cancer. JAMA Oncology doi:10.1001/jamaoncol.2018.1621 (2018).

Chapter 2

126. Azad, A.A. et al. Androgen receptor gene aberrations in circulating cell-free DNA: biomarkers of therapeutic resistance in castration-resistant prostate cancer. Clinical Cancer Research 21, 2315-2324 (2015). 127. Salvi, S. et al. Circulating cell-free AR and CYP17A1 copy number variations may associate with outcome of metastatic castration-resistant prostate cancer patients treated with abiraterone. British Journal of Cancer 112, 1717-1724 (2015). 128. Todenhofer, T. et al. AR-V7 transcripts in whole blood RNA of patients with metastatic castration resistant prostate cancer correlate with response to abiraterone acetate. Journal of Urology 197, 135-142 (2017). 129. Wyatt, A.W. et al. Genomic alterations in cell-free DNA and enzalutamide resistance in castration-resistant prostate cancer. JAMA Oncology 2, 1598-1606 (2016). 130. Smith, M.R. et al. Apalutamide treatment and metastasis-free survival in prostate cancer. New England Journal of Medicine 378, 1408-1418 (2018). 131. Hussain, M. et al. PROSPER: A phase 3, randomized, double-blind, placebo (PBO)-controlled study of enzalutamide (ENZA) in men with nonmetastatic castration-resistant prostate cancer (M0 CRPC). Journal of Clinical Oncology 36, 3-3 (2018). 132. James, N.D. et al. Abiraterone for prostate cancer not previously treated with hormone therapy. New England Journal of Medicine 377, 338-351 (2017). 133. Fizazi, K. et al. Abiraterone plus prednisone in metastatic, castration-sensitive prostate cancer. New England Journal of Medicine 377, 352-360 (2017). 134. Sartori, D.A. & Chan, D.W. Biomarkers in prostate cancer: what's new? Current Opinion in Oncology 26, 259-264 (2014). 135. Feng, F., Schaich, M. & Hughes, L. Current and future applications of genetic prostate cancer screening in the urologic clinic. Urologic Nursing 34, 281-289 (2014). 136. Loeb, S. et al. The Prostate Health Index selectively identifies clinically significant prostate cancer. Journal of Urology 193, 1163-1169 (2015). 137. de la Calle, C. et al. Multicenter evaluation of the Prostate Health Index to detect aggressive prostate cancer in biopsy naive men. Journal of Urology 194, 65-72 (2015). 138. Catalona, W.J. et al. A multicenter study of -2 pro-prostate specific antigen combined with prostate specific antigen and free prostate specific antigen for prostate

Chapter 2

cancer detection in the 2.0 to 10.0 ng/ml prostate specific antigen range. Journal of Urology 185, 1650-1655 (2011). 139. Parekh, D.J. et al. A multi-institutional prospective trial in the USA confirms that the 4Kscore accurately identifies men with high-grade prostate cancer. European Urology 68, 464-470 (2015). 140. Braun, K., Sjoberg, D.D., Vickers, A.J., Lilja, H. & Bjartell, A.S. A four- kallikrein panel predicts high-grade cancer on biopsy: independent validation in a community cohort. European Urology 69, 505-511 (2016). 141. Vickers, A. et al. Reducing unnecessary biopsy during prostate cancer screening using a four-kallikrein panel: an independent replication. Journal of Clinical Oncology 28, 2493-2498 (2010). 142. Vickers, A.J. et al. A panel of kallikrein markers can reduce unnecessary biopsy for prostate cancer: data from the European Randomized Study of Prostate Cancer Screening in Goteborg, Sweden. BMC Medicine 6, 19 (2008). 143. Van Neste, L. et al. The epigenetic promise for prostate cancer diagnosis. Prostate 72, 1248-1261 (2012). 144. Henrique, R. et al. Epigenetic heterogeneity of high-grade prostatic intraepithelial neoplasia: Clues for clonal progression in prostate carcinogenesis. Molecular Cancer Research 4, 1-8 (2006). 145. Zhou, M., Tokumaru, Y., Sidransky, D. & Epstein, J.I. Quantitative GSTP1 methylation levels correlate with Gleason grade and tumor volume in prostate needle biopsies. Journal of Urology 171, 2195-2198 (2004). 146. Trujillo, K.A., Jones, A.C., Griffith, J.K. & Bisoffi, M. Markers of field cancerization: proposed clinical applications in prostate biopsies. Prostate Cancer 2012, 302894 (2012). 147. Slaughter, D.P., Southwick, H.W. & Smejkal, W. Field cancerization in oral stratified squamous epithelium - clinical implications of multicentric origin. Cancer 6, 963-968 (1953). 148. Braakhuis, B.J.M., Tabor, M.P., Kummer, J.A., Leemans, C.R. & Brakenhoff, R.H. A genetic explanation of Slaughter's concept of field cancerization: Evidence and clinical implications. Cancer Research 63, 1727-1730 (2003). 149. Mehrotra, J. et al. Quantitative, spatial resolution of the epigenetic field effect in prostate cancer. Prostate 68, 152-160 (2008).

Chapter 2

150. Partin, A.W. et al. Clinical validation of an epigenetic assay to predict negative histopathological results in repeat prostate biopsies. Journal of Urology 192, 1081-1087 (2014). 151. Van Neste, L. et al. Risk score predicts high-grade prostate cancer in DNA- methylation positive, histopathologically negative biopsies. Prostate 76, 1078-1087 (2016). 152. Stewart, G.D. et al. Clinical utility of an epigenetic assay to detect occult prostate cancer in histopathologically negative biopsies: results of the MATLOC study. Journal of Urology 189, 1110-1116 (2013). 153. Wojno, K.J. et al. Reduced rate of repeated prostate biopsies observed in ConfirmMDx clinical utility field study. American Health Drug Benefits 7, 129-134 (2014). 154. Cullen, J. et al. A biopsy-based 17-gene genomic prostate score predicts recurrence after radical prostatectomy and adverse surgical pathology in a racially diverse population of men with clinically low- and intermediate-risk prostate cancer. European Urology 68, 123-131 (2015). 155. Klein, E.A. et al. A 17-gene assay to predict prostate cancer aggressiveness in the context of Gleason grade heterogeneity, tumor multifocality, and biopsy undersampling. European Urology 66, 550-560 (2014). 156. Dall’Era, M.A. et al. Utility of the Oncotype DX® prostate cancer assay in clinical practice for treatment selection in men newly diagnosed with prostate cancer: a retrospective chart review analysis. Urology Practice 2, 343-348 (2015). 157. Canfield, S.E. et al. A guide for clinicians in the evaluation of emerging molecular diagnostics for newly diagnosed prostate cancer. Reviews in Urology 16, 172-180 (2014). 158. Cuzick, J. et al. Prognostic value of a cell cycle progression signature for prostate cancer death in a conservatively managed needle biopsy cohort. British Journal of Cancer 106, 1095-1099 (2012). 159. Whitfield, M.L. et al. Identification of genes periodically expressed in the human cell cycle and their expression in tumors. Molecular Biology of the Cell 13, 1977-2000 (2002). 160. Cuzick, J. et al. Prognostic value of an RNA expression signature derived from cell cycle proliferation genes in patients with prostate cancer: a retrospective study. Lancet Oncology 12, 245-255 (2011).

Chapter 2

161. Klein, E.A. et al. Decipher genomic classifier measured on prostate biopsy predicts metastasis risk. Urology 90, 148-152 (2016). 162. Badani, K.K. et al. Effect of a genomic classifier test on clinical practice decisions for patients with high-risk prostate cancer after surgery. BJU International 115, 419-429 (2015). 163. Badani, K. et al. Impact of a genomic classifier of metastatic risk on postoperative treatment recommendations for prostate cancer patients: a report from the DECIDE study group. Oncotarget 4, 600-609 (2013). 164. Klein, E.A. et al. A genomic classifier improves prediction of metastatic disease within 5 years after surgery in node-negative high-risk prostate cancer patients managed by radical prostatectomy without adjuvant therapy. European Urology 67, 778-786 (2015). 165. Haese, A. et al. Clinical utility of the PCA3 urine assay in european men scheduled for repeat biopsy. European Urology 54, 1081-1088 (2008). 166. Chun, F.K. et al. Prostate Cancer Gene 3 (PCA3): Development and internal validation of a novel biopsy nomogram. European Urology 56, 659-667 (2009). 167. Hessels, D. & Schalken, J.A. The use of PCA3 in the diagnosis of prostate cancer. Nature Review Urology 6, 255-261 (2009). 168. Ploussard, G. et al. Prostate cancer antigen 3 score accurately predicts tumour volume and might help in selecting prostate cancer patients for active surveillance. European Urology 59, 422-429 (2011). 169. Roobol, M.J. et al. Performance of the prostate cancer antigen 3 (PCA3) gene and prostate-specific antigen in prescreened men: exploring the value of PCA3 for a first-line diagnostic test. European Urology 58, 475-481 (2010). 170. Groskopf, J. et al. APTIMA PCA3 molecular urine test: development of a method to aid in the diagnosis of prostate cancer. Clinical Chemistry 52, 1089-1095 (2006). 171. Wei, J.T. et al. Can urinary PCA3 supplement PSA in the early detection of prostate cancer? Journal of Clinical Oncology 32, 4066-4072 (2014). 172. Tomlins, S.A. et al. Urine TMPRSS2:ERG fusion transcript stratifies prostate cancer risk in men with elevated serum PSA. Science Translational Medicine 3, 94ra72 (2011). 173. Tomlins, S.A. Urine PCA3 and TMPRSS2: ERG using cancer-specific markers to detect cancer. European Urology 65, 543-545 (2014).

Chapter 2

174. Leyten, G. et al. Prospective multicentre evaluation of PCA3 and TMPRSS2- ERG gene fusions as diagnostic and prognostic urinary biomarkers for prostate cancer. European Urology 65, 534-542 (2014). 175. Tomlins, S.A. et al. Urine TMPRSS2:ERG plus PCA3 for individualized prostate cancer risk assessment. European Urology 70, 45-53 (2015). 176. Merdan, S. et al. Assessment of long-term outcomes associated with urinary prostate cancer antigen 3 and TMPRSS2:ERG gene fusion at repeat biopsy. Cancer 121, 4071-4079 (2015). 177. Salami, S.S. et al. Combining urinary detection of TMPRSS2:ERG and PCA3 with serum PSA to predict diagnosis of prostate cancer. Urologic Oncology: Seminars and Original Investigations 31, 566-571 (2013). 178. Sanda, M.G. et al. Association between combined TMPRSS2:ERG and PCA3 RNA urinary testing and detection of aggressive prostate cancer. JAMA Oncology 3, 1085-1093 (2017). 179. Leyten, G. et al. Identification of a candidate gene panel for the early diagnosis of prostate cancer. Clinical Cancer Research 21, 3061-3070 (2015). 180. Van Neste, L. et al. Detection of high-grade prostate cancer using a urinary molecular biomarker-based risk score. European Urology 70, 740-748 (2016). 181. Nilsson, J. et al. Prostate cancer-derived urine exosomes: a novel approach to biomarkers for prostate cancer. British Journal of Cancer 100, 1603-1607 (2009). 182. Overbye, A. et al. Identification of prostate cancer biomarkers in urinary exosomes. Oncotarget 6, 30357-30376 (2015). 183. Duijvesz, D., Luider, T., Bangma, C.H. & Jenster, G. Exosomes as biomarker treasure chests for prostate cancer. European Urology 59, 823-831 (2011). 184. Mehralivand, S. et al. A magnetic resonance imaging-based prediction model for prostate biopsy risk stratification. JAMA Oncology 4, 678-685 (2018). 185. Ahmed, H.U. et al. Diagnostic accuracy of multi-parametric MRI and TRUS biopsy in prostate cancer (PROMIS): a paired validating confirmatory study. Lancet 389, 815-822 (2017). 186. Futterer, J.J. et al. Can clinically significant prostate cancer be detected with multiparametric magnetic resonance imaging? A systematic review of the literature. European Urology 68, 1045-1053 (2015). 187. Pokorny, M.R. et al. Prospective study of diagnostic accuracy comparing prostate cancer detection by transrectal ultrasound-guided biopsy versus magnetic

Chapter 2

resonance (MR) imaging with subsequent MR-guided biopsy in men without previous prostate biopsies. European Urology 66, 22-29 (2014). 188. Moore, C.M. et al. image-guided prostate biopsy using magnetic resonance imaging-derived targets: A systematic review. European Urology 63, 125-140 (2013). 189. Siddiqui, M.M. et al. Comparison of MR/ultrasound fusion-guided biopsy with ultrasound-guided biopsy for the diagnosis of prostate cancer. Jama-Journal of the American Medical Association 313, 390-397 (2015). 190. Kasivisvanathan, V. et al. MRI-targeted or standard biopsy for prostate-cancer diagnosis. New England Journal of Medicine 378, 1767-1777 (2018). 191. Ferrari, M. Cancer nanotechnology: Opportunities and challenges. Nature Reviews Cancer 5, 161-171 (2005). 192. Nie, S.M., Xing, Y., Kim, G.J. & Simons, J.W. Nanotechnology applications in cancer. Annual Review of Biomedical Engineering 9, 257-288 (2007). 193. Smith, S.J., Nemr, C.R. & Kelley, S.O. Chemistry-driven approaches for ultrasensitive nucleic acid detection. Journal of the American Chemical Society 139, 1020-1028 (2017). 194. Kang, B.J. et al. Diagnosis of prostate cancer via nanotechnological approach. International Journal of Nanomedicine 10, 6555-6569 (2015). 195. Wu, G.H. et al. Bioassay of prostate-specific antigen (PSA) using microcantilevers. Nature Biotechnology 19, 856-860 (2001). 196. Zheng, G.F., Patolsky, F., Cui, Y., Wang, W.U. & Lieber, C.M. Multiplexed electrical detection of cancer markers with nanowire sensor arrays. Nature Biotechnology 23, 1294-1301 (2005). 197. Yu, X. et al. Carbon nanotube amplification strategies for highly sensitive immunodetection of cancer biomarkers. Journal of the American Chemical Society 128, 11199-11205 (2006). 198. Xu, S.J., Liu, Y., Wang, T.H. & Li, J.H. Positive potential operation of a cathodic electrogenerated chemiluminescence immunosensor based on luminol and graphene for cancer biomarker detection. Analytical Chemistry 83, 3817-3823 (2011). 199. Grubisha, D.S., Lipert, R.J., Park, H.Y., Driskell, J. & Porter, M.D. Femtomolar detection of prostate-specific antigen: An immunoassay based on surface-enhanced Raman scattering and immunogold labels. Analytical Chemistry 75, 5936-5943 (2003).

Chapter 2

200. Mani, V., Chikkaveeraiah, B.V., Patel, V., Gutkind, J.S. & Rusling, J.F. Ultrasensitive immunosensor for cancer biomarker proteins using gold nanoparticle film electrodes and multienzyme-particle amplification. ACS Nano 3, 585-594 (2009). 201. Liu, X. et al. A one-step homogeneous immunoassay for cancer biomarker detection using gold nanoparticle probes coupled with dynamic light scattering. Journal of the American Chemical Society 130, 2780-2782 (2008). 202. Gupta, A.K., Naregalkar, R.R., Vaidya, V.D. & Gupta, M. Recent advances on surface engineering of magnetic iron oxide nanoparticles and their biomedical applications. Nanomedicine 2, 23-39 (2007). 203. Sarkar, P., Ghosh, D., Bhattacharyay, D., Setford, S.J. & Turner, A.P.F. Electrochemical immunoassay for free prostate specific antigen (f-PSA) using magnetic beads. Electroanalysis 20, 1414-1420 (2008). 204. Song, E.Q. et al. Fluorescent-magnetic-biotargeting multifunctional nanobioprobes for detecting and isolating multiple types of tumor cells. ACS Nano 5, 761-770 (2011). 205. Koo, K.M., Carrascosa, L.G., Shiddiky, M.J.A. & Trau, M. Poly(A) extensions of miRNAs for amplification-free electrochemical detection on screen- printed gold electrodes. Analytical Chemistry 88, 2000-2005 (2016). 206. Koo, K.M., Carrascosa, L.G., Shiddiky, M.J.A. & Trau, M. Amplification-free detection of gene fusions in prostate cancer urinary samples Using mRNA-gold affinity interactions. Analytical Chemistry 88, 6781-6788 (2016). 207. Koo, K.M., Carrascosa, L.G. & Trau, M. DNA-directed assembly of copper nanoblocks with inbuilt fluorescent and electrochemical properties: Application in simultaneous amplification-free analysis of multiple RNA species. Nano Research 11, 940-952 (2018). 208. Koo, K.M., Wee, E.J.H. & Trau, M. Colorimetric TMPRSS2-ERG gene fusion detection in prostate cancer urinary samples via recombinase polymerase amplification. Theranostics 6, 1415-1424 (2016). 209. Koo, K.M., Dey, S. & Trau, M. Amplification‐free multi‐RNA‐type profiling for cancer risk stratification via alternating current electrohydrodynamic nanomixing. Small 14, 1704025 (2018). 210. Labib, M. et al. Single-cell mRNA cytometry via sequence-specific nanoparticle clustering and trapping. Nature Chemistry 10, 489-495 (2018).

Chapter 2

211. Koo, K.M., Wee, E.J.H., Mainwaring, P.N. & Trau, M. A simple, rapid, low- cost technique for naked-eye detection of urine-isolated TMPRSS2: ERG gene fusion RNA. Scientific Reports 6, 30722 (2016). 212. Chinen, A.B. et al. Nanoparticle probes for the detection of cancer biomarkers, cells, and tissues by fluorescence. Chemical Reviews 115, 10530-10574 (2015). 213. Wang, J., Liu, G.D., Wu, H. & Lin, Y.H. Quantum-dot-based electrochemical immunoassay for high-throughput screening of the prostate-specific antigen. Small 4, 82-86 (2008). 214. Li, X. et al. Rapid and Quantitative detection of prostate specific antigen with a quantum dot nanobeads-based immunochromatography test strip. ACS Applied Materials & Interfaces 6, 6406-6414 (2014). 215. Cheng, Z. et al. Simultaneous detection of dual prostate specific antigens using surface-enhanced Raman scattering-based immunoassay for accurate diagnosis of prostate cancer. ACS Nano 11, 4926-4933 (2017). 216. Koo, K.M., McNamara, B., Wee, E.J.H., Wang, Y. & Trau, M. Rapid and sensitive fusion gene detection in prostate cancer urinary specimens by label-free surface-enhanced Raman scattering. Journal of Biomedical Nanotechnology 12, 1798- 1805 (2016). 217. Koo, K.M., Wee, E.J.H., Mainwaring, P.N., Wang, Y. & Trau, M. Toward precision medicine: a cancer molecular subtyping nano‐strategy for RNA biomarkers in tumor and urine Small 12, 6233-6242 (2016). 218. Wang, J., Koo, K.M., Wee, E.J.H., Wang, Y.L. & Trau, M. A nanoplasmonic label-free surface-enhanced Raman scattering strategy for non-invasive cancer genetic subtyping in patient samples. Nanoscale 9, 3496-3503 (2017). 219. Koo, K.M., Wee, E.J.H., Wang, Y. & Trau, M. Enabling miniaturised personalised diagnostics: from lab-on-a-chip to lab-in-a-drop. Lab on a Chip 17, 3200-3220 (2017). 220. Armenia, J. et al. The long tail of oncogenic drivers in prostate cancer. Nature Genetics 50, 645-651 (2018). 221. Biomarker tests for molecularly targeted therapies: Key to unlocking precision medicine. (The National Academies Press, Washington, DC; 2016). 222. Hessels, D. et al. Analytical validation of an mRNA-based urine test to predict the presence of high-grade prostate cancer. Translational Medicine Communications 2, 5 (2017).

Chapter 2

223. Yamoah, K. et al. Novel biomarker signature that may predict aggressive disease in african american men with prostate cancer. Journal of Clinical Oncology 33, 2789-U2108 (2015). 224. Simon, N. & Simon, R. Adaptive enrichment designs for clinical trials. Biostatistics 14, 613-625 (2013). 225. Koo, K.M. et al. Design and clinical verification of surface enhanced Raman spectroscopy diagnostic technology for individual cancer risk prediction. ACS Nano doi:10.1021/acsnano.8b03698 (2018). 226. Simon, R. & Simon, N. Inference for multimarker adaptive enrichment trials. Statistics in Medicine 36, 4083-4093 (2017). 227. Truong, M., Yang, B. & Jarrard, D.F. Toward the detection of prostate cancer in urine: a critical analysis. Journal of Urology 189, 422-429 (2013).

Chapter 3

Colorimetric Gene Fusion Diagnostics for Visual Binary Readout

Introduction The TMPRSS2:ERG is a highly-specific prostate cancer biomarker which has been found to occur in around half of all prostate cancer cases, potentially representing a major prostate cancer subtype for tailored treatment approaches. The binary nature of the TMPRSS2:ERG biomarker which renders it present in prostate cancer and absent in benign conditions, makes it an extremely attractive target for clinical diagnostics development. In this chapter, a new solution-based biosensing system is described for the non-invasive detection of TMPRSS2:ERG RNA in prostate cancer urine samples. This biosensing system aims to integrate an entire nucleic acid detection workflow within a miniscule fluid droplet (i.e. lab- in-a-drop (LID)) for point-of-care (POC) usage. Specifically, a combination of reverse- transcription recombinase polymerase amplification (RT-RPA) and different colorimetric readout methodologies within a downscaled fluid volume, has enabled the rapid convenient detection of urine-purified TMPRSS2:ERG RNA. Chapter 3.1 showcases the LID concept for POC nucleic acid biomarker detection with a comprehensive review of integrated miniaturized diagnostics. Chapter 3.2 introduces a novel coupling of isothermal RT-RPA with a flocculation-based visual readout, which exploits the binary nature of TMPRSS2:ERG for simplistic (yes/no), inexpensive, and non-invasive detection of urinary TMPRSS2:ERG RNA within a microdroplet. Chapter 3.3 describes the development of “FusBLU”, a microtube-based quantitative assay which coupled RT-RPA with an enzymatic substrate oxidation for flexible colorimetric, spectrophotometric or electrochemical TMPRSS2:ERG RNA detection.

Chapter 3 is based on three published articles in Lab on a Chip (2017) [Chapter 3.1], Scientific Reports (2016) [Chapter 3.2], and Theranostics (2016) [Chapter 3.3].

Chapter 3.1

Enabling Miniaturised Personalised Diagnostics: From Lab-on-a- Chip to Lab-in-a-Drop

Kevin M. Koo, Eugene J. H. Wee, Yuling Wang and Matt Trau

ABSTRACT The concept of personalised diagnostics is to direct accurate clinical decisions based on an individual’s unique disease molecular profile. Lab-on-a-chip (LOC) systems are prime personalised diagnostics examples which seek to perform an entire sample-to-outcome detection of disease nucleic acid (NA) biomarkers on a single miniaturised platform with minimal user handling. Despite the great potential of LOC devices in providing rapid, portable, and inexpensive personalised diagnosis at the point-of-care (POC), the translation of this technology for widespread use has still been hampered by the need for sophisticated and complex engineering. As an alternative miniaturised diagnostics platform free of precision fabrication, there have been recent developments towards a solution-based lab-in-a-drop (LID) system; by which an entire laboratory-based diagnostics workflow could be downscaled and integrated within a singular fluid droplet for POC detection of NA biomarkers. Contrary to existing excellent reviews on miniaturised LOC fabrication and individual steps of NA biomarker sensing, we herein focus on miniaturised solution-based NA biosensing strategies suited for integrated LID personalised diagnostics development. In this review, we first evaluate the three fundamental bioassay steps for miniaturised NA biomarker detection: crude sample preparation, isothermal target amplification, and detection readout of amplicons. Then, we provide insights on research advancements towards a functional LID system which integrates all three of the above-mentioned fundamental steps. Finally, we discuss perspectives and future directions of LID diagnostic platforms in personalised medicine applications.

1 INTRODUCTION Personalised diagnostics refers to the detection of biological disease markers in individual patients in order to guide decisions for patient-specific diagnosis, therapy, and monitoring.11- 14 With the advent of sequencing technologies which enable a comprehensive outlook of a

Chapter 3.1 person’s genetic makeup, it is now possible to innovatively tailor specific diagnoses/treatments accordingly to the individual’s genetic profile. Hence, instead of a one- size-fits-all treatment approach, this stratification of disease subtypes can improve clinical outcomes in a variety of human diseases. While personalised diagnostics is applicable for many diseases, the field of cancer is one which is in great need for personalised diagnostics to have a significant impact. The complex and heterogeneous nature of cancer-driving mechanisms often lead to patients exhibiting tumour genetic variabilities that differ from patient-to-patient and/or tumour-to-tumour within the same patient.15,16 Thus, molecular diagnostic technologies for simple and rapid analysis of cancer biomarkers from clinical patient samples such as tissue or biological fluids are particularly in demand. With the current emphasis on quick disease diagnosis and early treatment intervention, diagnostics are being preferably developed for clinician-orientated17 or patient-orientated18 point-of-care (POC) use within hospital or home settings respectively. An ideal vision for personalised cancer diagnostics is one which incorporates several workflows (from patient sample preparation to assay readout) onto a singular platform that could conveniently be performed at POC with a rapid and accurate analysis result.

A prime example of a single-platform POC personalised diagnostic system is the concept of lab-on-a-chip (LOC) devices. A LOC generally utilizes microfluidics and nanotechnology to perform one or several typical laboratory processes on a single device.19-21 The main benefits of integrating different laboratory functions on a small-scaled LOC are: (i) use of miniscule amount of fluid volumes which reduces required sample/reagent amount and costs; (ii) shorter sample-to-answer turnaround time due to smaller diffusion distances, higher surface- to-volume ratios, and faster heating (i.e. smaller heating capacities); (iii) potential for low- cost mass production. The exciting potential of LOC technologies for POC personalised diagnostics is highly promising but since their emergence decades ago, majority of LOC systems have faced difficulties in progressing to real-world products for practical applications. Currently, the constraints which impede LOC commercialization include high fabrication costs, complex precision engineering, and difficulties in controlling sophisticated on-chip fluid movement/biochemical reactions.22 Thus, in order to meet the pressing need for personalised diagnostics in the real-world, it is worthy to explore alternatives to LOC which are similarly able to combine different conventional laboratory workflows onto a smaller- scaled platform.

Chapter 3.1

Fig. 1 Precision fabrication-free lab-in-a-drop (LID) point-of-care diagnostics for combining different laboratory-based nucleic analysis steps onto a singular miniaturised platform. Adapted from ref. 76 with permission from Royal Society of Chemistry.

A feasible solution to current LOC limitations is a miniaturised microtube-based biosensing system which enables miniaturised analysis of disease biomarkers from a single µL- input volume of biological sample. As such, apart from circumventing the need for specialised chip fabrication and complicated microfluidic flow manipulation, such a solution-based system is attractive in allowing rapid and low-cost personalised diagnostics for diseases as traditional bioassays could essentially be downsized into a droplet - lab-in-a-drop (LID) - and be programmed for detection within a microtube conveniently with minimal instrumentation (Fig. 1).

Nucleic acids (NAs) such as DNA and RNA play vital roles in disease initiation/progression,23,24 and are key biomarkers for developing POC personalised diagnostics. Additionally, the target-specific isolation of NA biomarkers is easily achievable through well-designed oligonucleotide probes that are relatively cheap to manufacture, while sensitive detection of trace target amount is feasible through a range of NA amplification techniques. These factors are advantages over the detection of other biomarker classes such as proteins or metabolites, which are typically limited by the availabilities of specific antibodies and target amplification methods. In this review, we focus on miniaturised

Chapter 3.1 biosensors which target NAs as an important and ubiquitous class of biomarkers. The traditional biosensing of NA biomarkers from biological samples consists of 3 fundamental steps: (i) Sample preparation (i.e. NA biomarker isolation/purification) from raw biological samples such as blood, urine or saliva; (ii) Amplification of NA targets to increase detection sensitivity; (iii) Transduction of target levels into a readable qualitative or quantitative signal. The miniaturisation of these fundamental steps into simplistic and convenient downscaled fluid droplets could allow us to achieve a practical LID biosensor that is readily available for personalised diagnostics applications. Moreover, the successful miniaturisation of NA target extraction, amplification, and detection as demonstrated on many LOC platforms in literature are also applicable as model concepts for LID diagnostics development.

Whilst numerous reviews relating to each of these three individual aspects of NA biosensing have been written,25-30 an extensive review that considers all three aspects for the development of miniaturised personalised diagnostics is lacking. We acknowledge that extensive research and advances have been achieved at the individual aspects of NA diagnostics. However, more often than not, miniaturised diagnostics development neglect to consider sample preparation strategies for compatible use with downstream amplification and readout processes, thus overlooking the foremost biosensing aspect. Without a suitable upstream sample preparation strategy to isolate relevant biomarkers to begin the biosensing process, the emphasis on state-of-the-art downstream biosensing aspects may be ineffectual. Furthermore, vital factors of combining individual biosensing aspects, such as reagents of an upstream process inhibiting/affecting the performance of the next downstream process, are likely to be missed when integration is not considered.

In this review, we summarise the recent advancements of suitable techniques/technologies in the three fundamental biosensing aspects for miniaturised patient diagnostics. We also describe our own research group’s efforts in achieving a miniaturised LID system through consolidation of all these three aspects. Lastly, we discuss perspectives and future directions of LID diagnostic platforms in personalised medicine applications.

2 DOWNSCALED SAMPLE PREPARATION High-quality NA isolation from primary samples is crucial as the performance of any DNA/RNA molecular assay is dependent on the quality of the NA input.28 For POC

Chapter 3.1 personalised diagnostics applications, NA sample preparation is further complicated due to restrictions in availability of on-site resources. The first of two main steps during NA sample preparation is cellular lysis, where cell structures can be disrupted by mechanical, chemical, thermal, electrical, enzymatic or a combination of strategies to release genetic material for analysis. The second step is NA extraction which can be broadly classified into chemical or physical methods to concentrate and enrich for NA targets.31 Depending on the desired diagnostics application, an appropriate sampling approach must be considered as part of its design. The following section will discuss the three most practised strategies in laboratories and how they have or could be applied in miniaturised LOC/LID systems for POC personalised diagnostics applications.

2.1 Phenol:chloroform methods A classical chemical/physical sample preparation technique is the phenol:chloroform method.32 The advantages of this approach are combined sample lysis and NA extraction that has both high yield and purity. For improved efficiency, macerated tissue or crude lysate in a neutral aqueous buffer is mixed with the phenol:chloroform mixture and allowed to settle (centrifuge) into organic and aqueous phases. NA (and contaminating carbohydrates) is preferentially found in the aqueous phase while proteins and lipids are found in the organic phase. The aqueous phase is carefully removed and can be subjected to further rounds of phenol:cholorform extraction to improve NA purity. Next, NA is further purified by precipitation using a high salt and alcohol washes to remove the carbohydrates. RNases can be added at this point to degrade RNA and thus yielding relatively pure DNA.

Alternatively, by adjusting acidic conditions (pH 4-6) and with the addition of a chaotropic agent such as guanidinium thiocynade, the phenol:chloroform method can be modified to specifically partition RNA into the aqueous phase.33,34 Commercially, this mixture is marketed as TRIzol or TRI-reagent. RNA is finally recovered using alcohol precipitation similar to the classical phenol:chloroform methods. DNA and proteins can also be purified subsequently from the organic phase if desired, further underscoring the versatility of the method.

Notwithstanding the benefits, phenol:chloroform methods are relatively long and tedious (multiple centrifuge and wash steps) to perform as compared to other contemporary strategies and generates hazardous waste products that may not be suitable for POC diagnostics.

Chapter 3.1

Fig. 2 Schematic of the RNA purification process using the liquid-phase nucleic acid purification chip. (a) An aqueous phase containing DNA, RNA, and protein in the microwells. (b) Organic phase of pH 4.6 was introduced into the headspace channel with continuous forward and reverse flow. In the case of on-chip DNA extraction, an organic phase with a pH of 8.0 was introduced into the headspace channel instead. (c) Protein and DNA were transferred from the aqueous phase into the organic phase, while RNA was retained in the aqueous phase. (d) The organic phase was expelled and evaporated under vacuum while purified RNA in the microwell was concurrently dried (e, f). Residual organic phase was further decontaminated by repetitive washing and vacuum evaporating with 70% ethanol. (g) q-RT-PCR reaction mixture was loaded into the microwells. (h) Microwells were covered with mineral oil followed by on-chip q-RT-PCR amplification. Adapted from ref. 26 with permission from American Chemical Society.

Morales and Zahn first showed that microscaled droplet-based phenol extraction of bacterial DNA could be carried out on a microfluidic device.35 To resolve the issue of residual organic phase interference with downstream PCR amplification, Zhang et al. recently demonstrated a microfluidic liquid-phase nucleic acid purification chip to selectively isolate NA targets from

Chapter 3.1 single bacterial cells in nL sample volume for on-chip quantitative PCR in the same microwells (Fig. 2).36 While these miniaturised strategies have been developed for phenol:chloroform extraction within microfluidics-based systems, it will be challenging for translation into POC diagnostics due to poor compatibility of hazardous waste with routine disposable plastics.

Very recently, a spin column method (commercially known as DirectZol from Zymo Research) based on solid phase extraction (SPE) was developed to simplify and reduce extraction time to as little as 5 minutes. While in the right direction of enabling the methods for miniaturisation, its dependence on phenol, chloroform, and combined RNA/DNA isolation are its major limitations. In addition, due to its incompatibility with enzymes, phenol:chloroform methods are very unlikely to be used in direct POC sample-to-answer detection. A more likely strategy for enabling POC might be through miniaturised pneumatics (either manual or automated compressed air-driven) to enable SPE approaches, such as the SPE integration into a pipette tip that we will be reviewing in the ensuing section.

2.2 Solid phase extraction (SPE) Currently, most routine lab-based NA isolation protocols typically require the use of centrifuges (for example, silica spin column-based methods37,38) that are based on SPE or popularly known as the Boom method.38 Here, samples are first lysed enzymatically (usually sodium dodecyl sulphate with proteinase K method) or with guanidium-based buffers. Then, negative charges of the NA are manipulated in solution with positively charged salts (eg. sodium, magnesium, ammonium, guanidine, etc) to promote precipitation onto surfaces. After a wash step, highly pure NA is recovered at high yields by elution with a pH 7-8 buffer. However, access to centrifuges may not always be possible especially for POC or home- based personalised medicine. For this reason, numerous microfluidic LOC strategies39 have been proposed to implement SPE for POC.

One extension of SPE that avoids the need for centrifuges and has gained recent popularity is solid phase reversible immobilization (SPRI).40-44 SPRI typically precipitates NA onto magnetic microparticles and with the aid of a magnetic field, facilitates buffer solution changes, washes and elution in a single sample holder. Therefore, SPRI potentially requires simpler equipment and is more suited for miniaturised POC bioassays. In some iterations of SPRI, various microparticle surface modifications have been explored to improve extraction

Chapter 3.1 yield. These include carboxylic acid,40 cellulose,43,44 and chitosan.42 However, conventional SPRI, like classical methods, are limited by the need for multiple sample/liquid manipulations. Various strategies have since been developed to automate SPRI45 but most POC-tailored approaches still require some form of micro-equipment42,46-48 that may be miniaturised onto LOC platforms but may not be suitable for low resource settings.

Fig. 3 The TruTip nucleic acid extraction process, whereby a porous, monolithic binding matrix is inserted into a pipette tip. Adapted from ref. 39 with permission from MyJove Corporation.

A recent development in convenient SPE-based sample prep is the TruTip® system by Akonni Biosystems. The method involves engineering a highly porous silica frit within a pipette tip that function as the SPE platform (Fig. 3).49 This approach essentially avoids the laborious liquid handling and centrifuging to just a few pipetting steps, and significantly reducing preparation times.24,25 The TruTip® system is automatable, and also has comparable NA extraction performance with other market-leading spin column brands.50,51

In-built SPE platforms have been adopted numerous times for miniaturised LOC applications in literature.52,53 Hong et al. carried out mRNA purification on a single microfluidic chip in nL volumes by using functionalised magnetic beads and Akutsu et al. have developed an integrated system for automated NA extraction and purification using magnetic beads within a disposable pipette tip. Yet, these miniaturised platforms require specialised manufacturing which is not readily available to the community. Perhaps an SPE approach that is able to retrofit with existing common labware, such as manual pipettes, could be a more viable and cost-effective/affordable LID alternative for POC personalised diagnostics.

Chapter 3.1

2.3 Direct sampling The ideal POC personalised diagnostics test would be one that involves direct addition of crude samples to the NA detection reaction, and avoids prior lysis and extraction steps. However, achieving this goal has multiple challenges: For instance, the sample matrix (e.g. blood, mucus, saliva, urine, biofilm, etc.) may not be compatible with downstream bioassays due to the presence of inhibitors, hence needing purification approaches such as those discussed in the previous sections. In addition, without some form of cellular disruption (i.e. lysis), there may not be sufficient genetic material available for detection if raw sample is used, thus affecting assay analytical sensitivity.

The simplest direct sampling and detection approach is thermal disruption of samples during the initial 95°C phase of PCR. Directed-evolution techniques have now resulted in DNA polymerases that are tolerant of sample-derived inhibitors.54 55 While useful for a variety of applications, direct thermal disruption is not compatible with a variety of isothermal NA amplification methods that typically rely on thermophobic enzymes. The subsequent addition of thermophillic enzymes for isothermal amplification, or the bioengineering of novel thermostable enzymes, may be feasible for thermal sample preparation.

Direct sampling on miniaturised devices via electrical cell lysis has been shown in literature. Lam and co-workers demonstrated integrated on-chip bacterial lysis with an applied electric field and direct sensing of bacterial NA pathogens with nanostructured microsensors.56 Besant et al. also present an integrated device that leveraged electrochemistry-driven lysis in close proximity to a microelectrode for rapid analysis of bacteria mRNA at clinically relevant levels.57 For electrical lysis of mammalian cells, Lu et al. have showcased a micro- electroporation device for electrolysis of human carcinoma cells to release subcellular materials.58 Despite these successful demonstrations of downscaled electrical lysis and detection, it is challenging to apply the electrical lysis mechanism for clinical POC personalised diagnostics due to the need for careful voltage control and precise microelectrode fabrication. With advances in nanofabrication techniques and nanomaterial synthesis, electrical cell lysis may perhaps be applicable for eventual widespread sample preparation.

Chapter 3.1

Fig. 4 Schematic of a microdevice architecture highlighting the EA1 enzyme-containing ZyGEM solution and swab chambers. Also highlighted is the hydro-phobic valve that allows proper loading of the ZyGEM solution prior to rotation, and the two vents, which allow proper airflow for correct fluidic movement. Adapted from ref. 50 with permission from Royal Society of Chemistry.

Recently, a newly-characterised enzyme called EA159 has been used for closed-tube and microfluidic sample preparation in miniaturised PCR-based genetic assays (Fig. 4).60-62 While its mechanism of action is not well understood, it is likely similar to that of proteinase K- based lysis for NA isolation but with better thermostability and efficiency62,63 and thus more suited for trace NA applications. However, with EA1 being a proteinase, it is likely incompatible with downstream enzymatic processes without first performing a heat inactivation step. This thus effectively rules out its use in isothermal NA amplification techniques which are ideal for designing miniaturised personalised diagnostics (detailed review in the next section). Perhaps as new designer enzymes are being created through directed-evolution bioengineering techniques, it may be possible in the future to formulate isothermal assays with proteinase-resistant enzymes for POC personalised diagnostics.

3 MINIATURISED ISOTHERMAL AMPLIFICATION To enable small-scaled formats of traditional bioassays, high analytical sensitivity is needed to detect trace biomarker amounts in minuscule sample volumes. A way to increase low NA target copies to a detectable level from minute amounts of biological sample input is the employment of amplification techniques. Specifically, isothermal amplification methods are attractive due to the elimination of complex thermal cycling associated with tradition PCR. Isothermal amplification allows for rapid and efficient amplification at a constant temperature that is achievable with a simple heating element, thus rendering it ideal for integration onto miniaturised diagnostics systems. In addition, isothermal amplification is suited for use in

Chapter 3.1

POC settings in which constant NA amplification temperature is more easily achievable than more complicated temperature cycling protocol. The alternative working principles of different isothermal amplification techniques also typically render them more robust (less inhibitory) for use in biologically-complex matrix than traditional PCR; this is advantageous for use on patient samples which contain many non-target biological molecules. Many isothermal amplification techniques have emerged in recent years, and this section will summarise several promising isothermal techniques which have been shown to be functional in small-volume reactions on miniaturised systems.64,65

3.1 Loop-mediated isothermal amplification (LAMP)

Fig. 5 (A) Principle of LAMP method. (a) Primer design of the LAMP reaction. (b) Starting structure producing step. (c) Cycling amplification step. Adapted from ref. 57 with permission from Nature Publishing Group. (B) Schematic illustration of singleplex iμLAMP. Adapted from ref. 58 with permission from Royal Society of Chemistry. (C) Principle of HDA method. Adapted from ref. 64 with permission from EMBO Press. (D) Schematic layout of the microreactors and the reagent loading network for real-time isothermal helicase- dependent amplification. Adapted from ref. 65 with permission from Springer Publisher.

LAMP is an exponential amplification method66 which utilizes four target-specific primers: one forward inner primer (FIP), one backward inner primer (BIP), and two outer primers (F3 and B3), for hybridisation to six primer binding sites flanking the sequence region for amplification (Fig. 5A). The FIP and BIP both carry sequences which are essential for self-

Chapter 3.1 priming and creating fundamental LAMP dumbbell structures for a successful LAMP reaction.

The LAMP process firstly starts with FIP binding to template and extension, followed by F3 binding to displace the newly-synthesized strands.67 The newly-synthesized strands function as templates for BIP and B3 binding and extension, resulting in LAMP dumbbell structures. The cycling amplification proceeds on with priming and extension of FIP and BIP alternately on the dumbbell structures, and the resultant LAMP products consist of singular dumbbell and multi-looped cauliflower-like structures. With the inclusion of reverse transcriptase, reverse transcription LAMP (RT-LAMP) can be performed on RNA targets. LAMP can provide up to 109-fold amplification in approximately 60 min and at a constant temperature of 60-65 °C. The use of four primers for targeting six different sites aids in increasing amplification specificity but also results in constraints and complexity during primer design and multiplexing capability.

To date, the suitability of LAMP for integration into miniaturised diagnostics has been demonstrated for numerous bacterial and viral targets. Fang et al. have successfully reduced a LAMP reaction volume to 5 µL in an eight-channel microfluidic chip (µ-LAMP) for nucleic acid target detection (Fig. 5B).68 Oh et al. have reported the use of LAMP on a centrifugal microfluidic device for the detection of S. typhimurium and E. coli.69 The centrifugal device allows for simple reagents distribution to achieve 25 parallel LAMP assays on a single miniaturised platform. The end-point colourimetric readout provided a detection sensitivity of within 60 min. Lee et al. also reported the development of an integrated LAMP device for amplification and detection of hepatitis B virus (HBV) DNA.70 The key innovation of this work is the real-time visual detection of turbidity changes due to the precipitation of magnesium pyrophosphate, a by-product of the down-scaled isothermal LAMP reaction. The device was tested on clinical serum samples with 2 copies/µL detection sensitivity within 60 min. More recently, Song and co-workers also employed LAMP on miniaturised devices for the detection of Zika virus RNA in oral samples.71 By using leuco crystal violet dye, the LAMP products could be detected visually with a sensitivity of five plaque-forming units in under 40 min.

Besides visual detection, on-chip LAMP could also be suitably combined with more sensitive and quantitative electrochemical or fluorescent platforms. More recently, further

Chapter 3.1 miniaturisation of LAMP has been achieved through digital LAMP (dLAMP). These dLAMP assays involve the further partition of individual LAMP into minuscule droplets for absolute quantification of targets after isothermal amplification. Rane and co-workers introduced an integrated microfluidic device platform for LAMP amplification of N. gonorrhoeae DNA within droplets (~10 pL).72 Rodriguez-Manzano et al. have also achieved 5 nL-droplet LAMP reactions for hepatitis C virus RNA detection with a mobile phone camera-readout.73

3.2 Helicase-dependent amplification (HDA) Similar to many amplification methodologies, HDA is inspired by nature with a working principle which mimics the biological DNA replication process (Fig. 5C). Instead of using heat to separate DNA strands during amplification, HDA first uses a thermostable DNA helicase for strand separation.74 Next, single-stranded binding proteins maintain the opened DNA structure and allow conventional primer hybridisation and extension. This DNA helicase-dependent strand separation enables the HDA to occur at a uniform temperature of around 65°C for 30-90 min, and is particularly advantageous for amplifying kilobase-long target sequences. In a similar fashion to traditional PCR, a single HDA process consists of a pair of forward and reverse primers, and is capable of RNA amplification with the inclusion of reverse transcriptase. The simplistic workflow of HDA makes it amenable for downscaling for performance at small volumes.

There are several reports in literature of miniaturised platforms with integrated HDA for nucleic acid target detection. Ramalingam and co-workers designed a simple microfluidic system without complex pumps or valves (Fig. 5D).75 All solution flow was manipulated by capillary action and HDA primers were surface-bound in individual nL-microchambers to allow multiple biomarker analysis from a single sample. This miniaturised platform was shown to enable real-time HDA quantification of SARS cDNA at a constant 62 °C.

There are also several microfluidic devices that combined target extraction, isothermal HDA, and detection readout on a single platform. Mahalanabis et al. used micro-solid phase extraction to isolate DNA from bacteria cells for isothermal HDA at 65 °C.76 Using fluorescent reporters, real-time fluorescence readout of the HDA amplicons was achieved within 50 min. Zhang et al. utilised a droplet sample-to-answer microfluidic platform for the detection of disease biomarkers from crude biosamples.77 The platform made use of silica superparamagnetic particles and unique surface topographic features for cellular DNA

Chapter 3.1 purification, droplet HDA and real-time fluorescence readout. The detection of promising ovarian cancer biomarker Rsf-1 from human whole blood samples was achieved in 160 min.

3.3 Rolling circle amplification (RCA)

Fig. 6 (A) Principle of RCA method. Adapted from ref. 69 with permission from Nature Publishing Group. (B) Schematic illustration of working mechanism for amplified single- molecule detection by converting nanometer-scale specific molecular recognition events mediated by RCA to fluorescent micrometer-sized DNA molecules amenable to discrete optical detection. Adapted from ref. 71 with permission from Nature Publishing Group. (C) Principle of RPA method. Adapted from ref. 73 with permission from PLOS. (D) Schematic layout of a thermoformed lab-on-a-chip cartridge for fully automated analysis of nucleic acids based on isothermal RPA. Adapted from ref. 75 with permission from Royal Society of Chemistry.

RCA is an isothermal amplification technique which uses circular DNA templates to generate lengthy single-stranded DNA products that comprises of repeated sequences.78 In order to generate circular templates from target sequences, padlock probes bind specifically to DNA target sequences before being ligated, sealed and circularised. After primer hybridisation to a circular template, Φ29-DNA polymerase serves to extend the primer and the polymerase’s strand-displacing activity ensures continuous rolling amplification (Fig. 6A).

Chapter 3.1

RCA can provide linear 103-fold amplification at about 37 °C in 60 min. Although the low amplification temperature of RCA is attractive for integration into miniaturised platforms, it can also lead to non-specific amplification in addition to the need to generate circularised padlock probes from initial nucleic acid targets at different temperatures. The relatively low amplification efficiency of linear RCA can be improved by single-stranded DNA binding proteins addition. Another RCA modification to increase amplification efficiency to 109-fold is through the addition of a second primer (complementary to first-round amplification product) for exponential hyperbranched RCA (HRCA).79 HRCA generally uses Bst DNA polymerase at 60 °C for 90 min.

RCA has been integrated into many miniaturised diagnostics for detection of circular DNA that exist in plasmids and virus. Mahmoudian et al. reported a microchip platform which performed RCA in µL-level wells for the detection of V. cholerae from clinical samples.80 The on-chip amplification of padlock probes occurred at 37 °C and amplicons were immediately detected on the same platform via micro-electrophoresis in <65 min.

Single target molecule detection by miniaturised digital RCA has first been shown by Jarvius et al. (Fig. 6B).81 On this platform, nm-sized circular padlock probes are formed in target presence and amplified into µm-sized RCA amplicons. The amplicons are subsequently tagged with fluorescence probes and individually detected under a microscope for absolute enumeration. Multiplexed target detection is demonstrated by using different fluorescence probes for the detection of V. cholera and V. fisheri. Juul et al. have also shown a pL-droplet platform called rolling-circle enhanced enzyme activity detection (REEAD) for highly sensitive (< 1 parasite/µL) and non-invasive detection of all Plasmodium parasite species in crude patient saliva samples.82 This approach relies on the specific circularization of a RCA template by the endogenous Plasmodium enzyme topoimerase I prior to small-volume RCA.

However, despite these recent developments miniaturised RCA diagnostics, there is still room for improvement due to a current lack of RCA-based singular platforms which integrates the full sample extraction to target detection readout workflow.

3.4 Recombinase polymerase amplification (RPA) RPA is an isothermal exponential amplification technique which uses recombinase enzymes for strand separation, thereby removing the use of thermal cyclers (Fig. 6C).83 The

Chapter 3.1 recombinase enzymes form complexes with primers to locate complementary sequences on targets for primer hybridisation. Similar to HDA, single-stranded DNA binding proteins are used to maintain opened DNA structures during extension by DNA polymerase in RPA. RT- RPA is the modified RPA version for RNA templates by inclusion of reverse transcriptase enzyme to generate cDNA before amplification. RPA enables very rapid (~10-20 min) amplification at a constant temperature of 37-42 °C without compromising sensitivity and specificity, thus making it ideal for at point-of-care. Furthermore, RPA is commercially- available (TwistDx) in the form of preserved freeze-dried pellets, making the reaction stable for use outside of the laboratory environment. The simplicity, robustness, and versatility of RPA has recently been demonstrated in the design of a cutting-edge Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)-based NA detection platform;84 and used to detect specific viral strains, distinguish pathogenic bacteria, genotype human DNA, and identify circulating tumour DNA mutations with attomolar sensitivity.

Lutz et al. introduced the first miniaturised on-chip RPA with pre-stored RPA reagents on the chip (Fig. 6D).85 Each chip can performed up to 30 RPA assays concurrently in individual microchambers (10 µL each) by on-chip centrifugal force control. The successful detection of the antibiotic resistant gene mecA of S. aureus is achieved with less than < 10 copies detection sensitivity within 20 min. This chip-based RPA platform shows that downscaling of RPA reagent volume is possible and further developments in target multiplexing could enable analysis of several targets from a single patient sample.

Wee and Lau et al. have developed a miniaturised (µL-scale) tube-based RPA assays termed as Single Drop Genomics (SDG) in which a singular 1 µL-droplet of extracted NA sample is introduced into a downscaled RPA assay (12.5 µL each) for rapid (~15 min) isothermal target amplification. A drop of the resultant amplified products is then utilised to initiate a colourimetric bridging flocculation-based readout. SDG has been successfully demonstrated for a wide variety of pathogen and cancer biomarkers and consolidates the usual workflow of a centralised laboratory into a single miniaturised drop for rapid detection.86-89

Despite the high viscosity of the RPA buffer which complicates droplet formation, RPA has been successfully downscaled into droplet reactions as digital RPA. Shen et al. and Tsaloglou et al. have demonstrated a SlipChip device that confines miniaturised nL-scale RPA reactions onto a single platform.90,91 The RPA reactions on the SlipChip offer quantitative fluorescence

Chapter 3.1 detection of bacterial nucleic acid biomarkers down to a few hundred copies/mL in < 30 min. More recently, Li et al. showcased the downsizing of RPA reaction into thousands of pL- sized wells on an array chip.92 Within each well, RPA of DNA targets is completed rapidly within 30 min.

3.5 Nucleic acid sequence-based amplification (NASBA)/Transcription-mediated amplification (TMA)

Fig. 7 (A) Principle of NASBA method. Adapted from ref. 84 with permission from United States National Academy of Sciences. (B) Schematic illustration of a single device architecture showing the distinct functional microfluidic modules: RNA purification and real- time NASBA chambers. Adapted from ref. 86 with permission from Royal Society of Chemistry. (C) Principle of SDA method. For clarity, the initial SDA process is shown only in the forward direction. Adapted from ref. 90 with permission from United States National Academy of Sciences. (D) Strategy to realize cleavage-based RNA amplification. Adapted from ref. 93 with permission from Nature Publishing Group.

NASBA is an exponential amplification method which has been designed for RNA amplification at a constant temperature (Fig. 7A).93,94 In a NASBA reaction, reverse transcriptase firstly extends a target-bound forward primer (contains an overhang sequence complementary to T7 promoter) and synthesizes complementary DNA on the initial RNA target. Next, RNAse H degrades the initial RNA target before a reverse primer binds specifically to the remaining single-stranded DNA strand. Then, reverse transcriptase extends the reverse primer to form a double-stranded DNA template that contains both target and T7 promoter sequences. Using the double-stranded DNA template, T7 RNA polymerase transcribes many RNA strands (complementary to initial RNA target) that once again become

Chapter 3.1 templates for reverse transcriptase to form more double-stranded DNA templates. This continuous cycling of reverse transcription, degradation, and transcription results in exponential 109-fold amplification in 90-120 min at 41°C. TMA has an identical working principle to NASBA except for the use of an enzyme with dual reverse transcriptase and RNase H activities. Although the amplification process of NASBA/TMA proceeds at a constant temperature, it is worthy to note that pre-amplification steps at different temperatures are required to remove secondary RNA structures. The compatibility of NASBA with POC NA diagnostics has been shown by Pardee et al. recently through the development of a programmable diagnostic platform for low-cost femtomolar detection of Zika virus.95

Gulliksen et al. have shown the miniaturisation of NASBA on a microfluidic chip platform with real-time detection of human papillomavirus (HPV) in microchambers (Fig. 7B).96 The nL-scale reaction volume demonstrates the feasibility and performance of downscaled NASBA assays. To overcome the adsorption issue of NASBA reaction molecules on the surface of the microfluidic chip, a pre-surface treatment step is necessary.

The successful integration of NASBA with sample RNA extraction and real-time detection readout on a single miniaturised platform was first shown by Dimov et al.97 In this work, solid-phase RNA extraction was performed on chip using silica beads and fluorescence- tagged molecular beacons were used to quantitatively detect amplicons after NASBA in real- time. The whole sample-to-answer process was completed within 30 minutes in µL- chambers. Reinholt et al. also integrated target extraction and NASBA on a miniaturised device for the detection of C. parvum.98 The µL-scale NASBA reaction was completed within 90 min at 41 °C before the NASBA amplicons were detected off-chip. More recently, Chung et al. showcased the integration of Norovirus RNA target extraction via microbeads, small- volume NASBA, and fluorescence detection on a miniaturised device.99

3.6 Strand displacement amplification (SDA) SDA is an isothermal exponential amplification strategy for DNA targets (Fig. 7C).100 The first step involves thermal separation of the double-stranded T1 and T2 DNA templates, followed by hybridisation and extension of S1 forward and S2 reverse primer pair (carrying an overhang restriction enzyme site sequences). During S1 and S2 extension, B1 and B2 bumper primers (each located a few bases upstream from S1 and S2 respectively) extend

Chapter 3.1 enzymatically at the same time. These bumper primer extensions displace S1- and S2-primer extended products (S1-ext and S2- ext) from T1 and T2 templates. The displaced S1-ext and S2-ext serve as templates for S2 and S1 primers respectively to form double-stranded products containing active restriction enzyme sites. The restriction enzymes then create gaps on these active sites, and polymerase-catalysed extensions from these gaps displace downstream sequences. The displaced strands then once again serve as templates for S1 and S2 primers, and repeated cycling of this process result in approximately 107-fold exponential amplification in 120 min at 37 °C.

There are several reports of miniaturised SDA assays for NA detection. Burns and co- workers have successfully shown nL-volume SDA reactions and electrophoresis detection of resultant products on a device platform.101 The SDA reaction was completed within 17 min at 50 °C. Yang et al. integrated miniaturised dielectrophoretic extraction of DNA targets from E. coli cells, µL-scale SDA, and fluorescence readout on a single platform.102 The whole sample-to-answer process is completed within 2.5 hours at a constant SDA temperature of 60 °C. Zhao and co-workers created a µL-volume strategy named “cleavage-based RNA amplification” to simultaneously amplify and detect RNA within microtubes (Fig. 7D). The assay linked three SDA reactions together to achieve highly efficient signal amplification, and on-site detection can be realized by the DNAzyme-based colourimetric readout.103 A commercially-available SDA system (Becton Dickinson ProbeTecTM ET System, USA) has also shown successful detection of C. trachomatis and N. gonorrhoeae in miniaturised microchambers.104

More recently, Giuffrida et al. demonstrated nL-volume droplet SDA on a microfluidic platform for miRNA detection.105 The method was established to detect as little as 10-18 mol of microRNA target sequences that were compartmentalised in 20 nL droplets, and the ability to discriminate between full-matched, single-mismatched, and unrelated sequences was also investigated. Furthermore, the suitability of the method for biological samples was also tested by detecting microRNA-210 from transfected K562 cells.

Chapter 3.1

3.7 Overview of miniaturised isothermal amplification

Table 1 Overview of miniaturisable isothermal amplification techniques Technique Minimal Reaction Customary Approximate Estimated Limit of Miniaturised Refs. primer enzymes amplification amplification size of detection reaction quantity temperature duration (min) amplicons volume (°C) (bp) LAMP 4 DNA 60-65 60 Range 5 DNA/ pL 56, polymerase from RNA 57, 62 hundreds copies to thousands

HDA 2 Helicase and 65 30-90 ≤200 1 DNA/ nL 64, 65 DNA RNA polymerase copy

RCA 1 Ligase and 95 (optional 60 Range 10 DNA/ pL 68, 72 DNA initial heating), from RNA polymerase followed by 37 hundreds copies to thousands

RPA 2 Recombinase 37-42 10-20 ≤500 1 DNA/ pL 73, 82 and DNA RNA polymerase copy

NASBA/ 2 Reverse 65/95 (initial 90-120 ≤250 1 RNA nL 83, TMA transcriptase, heating), copy 84, 86 RNase H, followed by 41 and RNA polymerase

SDA 2 DNA 95 (initial 120 Range 10 DNA/ nL 90, 91 polymerase; heating), from RNA restriction followed by 37 hundreds copies enzyme to thousands

Chapter 3.1

As reviewed in this section, there have been many works of miniaturisable isothermal NA amplification techniques in literature. It is anticipated that as the mechanisms of more biological processes are being unravelled by continuing research efforts, the field of isothermal NA amplification will continue its progress towards more cutting-edge techniques. Each reviewed technique in this section has its own working mechanism for amplification without thermal cycling; thus leading to differing reaction requirements (eg. primers, enzyme, temperature) and product size (Table 1). Hence, an isothermal amplification technique if suitable characteristics can be employed in accordance with different designs and applications of a diagnostics test. The compatibility of these reviewed techniques with downsized reaction volumes, as generally demonstrated on LOC platforms, may similarly be utilised in miniaturised LID diagnostics. Yet, it is worthwhile to note the lack of integrated systems as sample preparation is commonly being performed independently of the isothermal amplification platform. Therefore, it would be helpful to focus future investigations on more successful integrations of miniaturised sample preparation methodologies with isothermal target amplification. On the other hand, integrated detection readout of amplification products on the same platform has been routinely shown over past years. The next section will review these different miniaturised detection readouts.

4 MICROSCALED DETECTION READOUT The design of a miniaturised readout system that could detect biomarker targets in a simple, effective and rapid way is essential towards designing POC personalised diagnostics. In line with the discussion on advancements in miniaturised sample preparation and isothermal amplification earlier in this review; compatible readout strategies become increasingly important as the final step to an integrated personalised diagnostics platform with high sensitivity, specificity, user-friendliness and cost-effectiveness. Over past years, various innovative detection readout systems have been reported for miniaturised LOC platforms, and this section will summarise these advances with a targeted focus on relevant detection technologies for solution-based LID platforms.

4.1 Microfluidics platforms A couple of decades ago, lateral flow devices were first developed as simple and low-cost bioassays for use in POC settings. The principle of such devices is generally based on the specific capture of amplified NA targets by surface-functionalised probes for a simplistic

Chapter 3.1 visual readout.106,107 The biggest limitation of lateral flow devices is the lack of multiplexing as it is difficult to analyse more than one target concurrently. To resolve this issue, a microfluidic approach is thus developed to offer advantages in (i) high-throughput parallel detection, (ii) small sample volume manipulation, (iii) portability; and (iv) high capacity workflow integration. These benefits offered by the combination of microfluidics and lateral flow assays are very attractive for personalised diagnostics.12 One recent example of such an assay is the development of a microfluidic single-molecule spectroscopy technique which allows one-step analysis and quantification of circulating DNA in serum volumes smaller than 1 pL without additional DNA isolation or enzymatic amplification steps.108 Several paper-based microfluidic approaches have also been shown to enable the simultaneous parallel detection of multiple analytes109-111 on a low-cost platform. As such paper microfluidic approaches are generally designed for POC testing in resource-poor settings, their quantitative performance might not be suited for accurate biomarker analysis within a clinical setting. Therefore, this has led to the emergence of more sophisticated microfluidic LOC devices for integrated NA biosensing. To date, both fluorescence and electrochemical- based detection readout have been integrated with microfluidic devices to provide a sample- to-answer format on a single platform.

Traditionally, the fluorescence-based readout has been commonly-employed for NA detection due to its high level of sensitivity and low background noise.112,113 Successful examples include the commercially-available GeneXpert platform (Cepheid) which integrates sample processing and PCR in a disposable plastic microfluidics cartridge which contains reagents for cell lysis, nucleic acid extraction, amplification and fluorescent amplicon detection. The DxN VERIS MDx system (Beckman Coulter) also uses a cartridge system for automated magnetic bead-based sample preparation, followed by PCR of NA targets with real-time fluorescence readout.114 Another commercial example is the cobas® Liat® PCR System (Roche) which utilises uniquely-engineered assay tubes for miniaturised POC “lab- in-a-tube” NA testing. This automated system is able to perform multiplexed NA biomarker detection with quantitative fluorescence readout, and consolidates single platform (samples preparation and qPCR) within 20 min.115 Despite these commercialised state-of-the-art technologies, there are still existing limitations (such as ionisation hazards of GeneXpert platform due to sonication; bulky and costly equipment of DxN VERIS MDx system) to be resolved in order to deploy these technologies for optimal use in POC diagnostics. The

Chapter 3.1 miniaturisation of fluorescence readout systems may aid in increasing the portability and affordability of such technologies for ideal POC diagnostics.

In the direction of miniaturised fluorescent NA detection readouts, there have been recent advancements in smartphone optical systems116,117 and portable fluorometers. The adaptation of smartphones into handheld fluorescence detectors offers the advantage of NA analysis on a widely-available platform in POC settings. Likewise, the Twista® (TwistDx) portable fluorometer has been shown by Teoh and co-workers for convenient and low-cost detection of dengue virus RNA with comparable outcomes to conventional laboratory qPCR technique. The good detection sensitivity of such portable fluorometers is useful for high-throughput small-volume assays for parallel detection of multiple NA targets, such as the miRPA developed by Wee and Trau.118 Additionally, Koo et al. have developed a simple and rapid gene fusion strategy which exploits the specificity of DNA ligase and the speed of isothermal amplification to simultaneously detect multiple fusion gene RNAs within a short sample-to- answer time frame of 60 min by using real-time fluorescent detection.119 These techniques are all suitable for downscaled NA target detection on miniaturised fluorescence readout systems.

Fig. 8 Droplet platform that incorporates sample extraction, washing, cell lysis and RT-PCR into a free-standing droplet into a flexible virtual laboratory with (sub)microliter volumes. Adapted from ref. 111 with permission from John Wiley & Sons.

The increased miniaturisation of fluorescence NA detection has been further enabled by the advent of droplet microfluidics. Droplet microfluidics is a recently-developed powerful NA assay which integrates sample preparation and genetic analysis within discrete droplets; including steps of cell lysis, DNA binding, washing, elution, amplification and detection (Fig. 8). As all reagents are stored in droplets, it eliminates the need for fluidic coupling to external reagent reservoirs.77,120,121 The successful demonstration of droplet microfluidics also serves to highlight the feasibility of LID systems which similarly aim to achieve miniaturised

Chapter 3.1 bioassays within small-volume fluid droplets. Integrated droplet NA assays have also allowed precise digital analysis of NA biomarker copy number but multiplexed target detection has so far been limited; largely due to the long-standing issue of multiple fluorescent spectral overlapping, and the challenging analysis of so many different droplets (i.e. fluidic bodies) within the solution. Hence, a LID platform such as the SDG assay, which integrates and miniaturises traditional biosensing into a singular droplet to maintain a single fluidic body (without need for droplet partitioning of a single reaction), might be more suited for high- throughput multiple biomarker analysis.

Electrochemical-based microfluidics have also displayed immense potential for POC personalised diagnostics due to its compatibility with low-cost, portable, and miniaturised analysis.122-124 Particularly, the development of nanostructured electrodes has made the electrochemical platform more powerful for ultrasensitive NA detection. As reported, chip- based nanostructured microelectrodes (NMEs) have been developed with a low 10 aM detection limit.57,125-128 Ölcer et al. developed a microfluidics- and nanoparticle-based amperometric biochip for cyanobacteria NA detection and all steps of the assay were performed during the reagent flow for fast and sensitive DNA detection.129 Electrode miniaturisation is a significant step towards integration of electrochemical detection with complementary upstream biosensing processes for personalised diagnostics.

Miniaturised electrochemical detection has also been incorporated with on-chip isothermal NA amplification such as LAMP to detect S. typhimurium with a detection limit of 16 copies of genomic DNA.130 To this end, Ng et al. recently demonstrated a portable electrochemical biosensor platform by combining with isothermal RPA of DNA targets, gold nanoparticles, and screen-printed electrodes for sensitive and rapid detection of M. tuberculosis DNA.88,131 The downscaled electrochemical analysis enabled an overall sample-to-answer timeframe of 90 min with a cost of under USD 10 for each test. The introduction of mass-produced screen- printed electrodes in electrochemical biosensing has led to new possibilities in miniaturised POC personalised diagnostics.132 Explicitly, the electrochemical detection of NA targets133,134 on inexpensive and disposable screen-printed electrodes could allow for rapid and cost- effective screening of multiple NA biomarkers in the clinic. Moreover, screen-printed electrodes are also readily suitable for existing POC instruments, such as personal glucometers. There are recent reports which exploit commercially-available glucometers for pM-level NA biosensing.135,136 Typically, the DNA targets are first bound to magnetic beads

Chapter 3.1 via covalently-attached capture strands, followed by the binding of secondary invertase- conjugated DNA strands to form a sandwich assay. Then, invertase enzymatically catalyzed the hydrolysis of sucrose to glucose and fructose, which can be subsequently detected using screen-printed electrodes on a personal glucometer. However, such methods have yet to undergo clinical evaluation, and future clinical trials with a relevant disease model will be useful to test the feasibility for widespread POC usage.

There are excellent reviews on microscaled detection readouts that have been successfully coupled to microfluidics devices for potential POC diagnostics use.137,138 Yet, widespread use of these devices is largely limited by high device manufacturing cost and reproducibility difficulties at massive production levels. Thus, there is a need for the development of alternative miniaturised technologies, such as LID systems, that could overcome these challenges.

4.2 Miniaturised lab-in-a drop (LID)-compatible readouts In this section, with a targeted focus on LID systems, we will summarise and overview miniaturised readouts approaches that are appropriate for solution-based LID personalized diagnostics. Specifically, we will highlight colourimetric as well as surface-enhanced Raman spectroscopy (SERS) readouts which are suitable for visual and multiplexed NA biomarker detection respectively.

4.2.1 Visual colourimetric readout Colourimetric detection is a very appealing for POC diagnostics development as it does not require the use of expensive and complex sensing equipment. One typical and widely-used colourimetric approach is based on enzymatic catalysis of substrate oxidation, such as horseradish peroxidase (HRP) for 3,3',5,5'-Tetramethylbenzidine (TMB).87,139,140 By combining isothermal NA amplification with colourimetric approaches, a quick turnaround time (including sample preparation) could be achieved. For instance, Koo and co-workers recently report a novel rapid (75 min), cost-efficient and minimal-equipment assay named “FusBLU”. The microtube-based FusBLU assay coupled isothermal target amplification with a HRP-based colourimetric readout to detect a prostate cancer fusion gene RNA biomarker.139 Despite the convenient and rapid nature of enzyme-catalysed colourimetric readouts, poor enzyme stability by denaturation, and costly enzyme purification are limiting factors for POC usage.141 Perhaps ongoing research into mass engineering and production of

Chapter 3.1 stable enzymes or enzyme storage strategies may enable the convenient use of enzyme-based colourimetric assays outside of research settings in the future.

Fig. 9 (A) Colourimetric assay that uses gold nanoparticles aggregation for DNA sensing. Adapted from ref. 132 with permission from Elsevier. (B) A combination of rolling circle amplification and nicking endonuclease-assisted nanoparticle amplification for rapid, colourimetric detection of DNA. Adapted from ref. 135 with permission from John Wiley & Sons.

Mirkin and colleagues pioneered the use of gold nanoparticles (AuNPs) for hybridisation- based DNA colourimetric detection (Fig. 9A).142-144 In presence of DNA targets, covalently- bound probes on AuNPs are cross-linked through hybridisation at each end of their DNA targets, thus produce cross-linked AuNPs aggregates to result in a reaction mixture colour change from red to blue. Recently, Xu et al. have further enhanced the AuNPs-based colourimetric approach for ultrasensitive DNA detection (1 pM) by combination with isothermal RCA of DNA targets (Fig. 9B).145 To date, the challenges for using AuNPs-based assays in clinical diagnostics include low AuNPs stability and need for specialised AuNPs

Chapter 3.1 preparation. The design of alternative nanomaterials and synthesis process could resolve these shortcomings and further the use of nanoparticle-based colourmetric detection.

Fig. 10 Single Drop Genomics assay that uses bridging flocculation of magnetic particles for rapid nucleic acid target detection. Adapted from ref. 76 with permission from Royal Society of Chemistry.

Recently, Trau and colleagues have developed the SDG assay, a novel bridging flocculation- based colourimetric approach for qualitative evaluation of isothermally amplified NA targets (Fig. 10).86 A key unique feature of the bridging flocculation mechanism (previously used in water purification for decades),146, is that the NA flocculation event is entirely sequence- independent and only relies on the presence/absence of amplicons (i.e. biomarker target). Hence, this renders SDG as an exceedingly simple and convenient way of detecting amplified targets with minimal conventional equipment. Wee and Lau et al. have demonstrated SDG on an isothermal RPA to generate double-stranded amplicons from plant pathogen targets, which in turn, induces flocculation of SPRI magnetic beads for simple naked-eye readout. In accordance to an ideal LID system, the entire SDG workflow is miniaturised for performance within microdroplets. The pivotal benefit of SDG is its simplistic and elegant detection of NA biomarkers, coupled with straightforward visual yes/no detection evaluation. The versatile nature of SDG has been shown for detection of a broad range of disease biomarkers including

Chapter 3.1

DNA methylation87; M . tuberculosis DNA;88 cancer gene fusion RNA;89 as well as different pathogenic NA biomarkers across the plant and animal kingdom.86 Thus, SDG shows great potential for further advancements into a true POC personalised LID diagnostics; enabling rapid, sensitive, and relatively inexpensive colourimetric NA biomarker detection in a resource-limited area. Yet, SDG is not yet a finished product with several areas for further improvement. Section 5 of this review will provide a detailed discussion of the required future advancements for SDG to realise an ideal LID platform.

4.2.2 Multiplexed surface-enhanced Raman spectroscopy (SERS) readout The high demand for comprehensive POC disease diagnosis call for a miniaturised “one-pot” multiplexed assay that can simultaneously detect and readout multiple biomarkers.147,148 Whilst colourimetric detection is an ideal diagnostics approach for qualitative yes/no readout (for a single biomarker), rapid and quantitative multiplexed biomarker detection is highly desired in diseases which require screening of a biomarker panel. Fluorescence-based detection is most commonly associated with multiplexed NA detection, and generally accomplished by hybridising each NA target with a different fluorescent dye molecule. Even though being largely reliable and effective, multiplexed fluorescence detection is constrained by its broad emission spectral peaks. This makes the interpretation of multiple overlapping spectral peaks (i.e. multiple biomarkers) a difficult task. Hence, surface-enhanced Raman spectroscopy (SERS), with its narrow and well-defined spectral peaks, could be an attractive option for multiplexed biomarker detection in personalised diagnostics applications.

SERS is an optical technique that utilises nanostructured metallic surfaces to enhance the Raman scattering of surface-adsorbed analytes.149,150 SERS has enabled the detection of various disease biomarkers with extremely low limit-of-detection, and even down to single molecule detection.151,152 SERS provides several notable benefits for “one-pot” multiplexed NA biomarker sensing, including: (i) the widths of Raman spectra are 10-100 times narrower than those of fluorescence, thus making it more suitable for extensive multiplexing; (ii) sensitivity with only one single laser excitation; and (iii) photostability to ensure signal reproducibility.

SERS has been utilised in various miniaturised multiplexed NA detection platforms involving designed nanostructures. Cao et al. have reported the use of microchip for DNA detection with a sandwich format by SERS, which was possible to detect six different target DNA

Chapter 3.1 sequences.153 Irudayaraj et al further developed the multiplexed assay by using the non- fluorescent Raman reporter molecules which were directly conjugated on the surface of the AuNPs. A low fmol-level detection limit was achieved to monitor gene expression in cancer cell lines.154,155 These platforms are spatial multiplexed approaches, which involve the physical separation of each NA target onto different spots on the microchip. The rationale for this spatial approach is to minimise nonspecific binding for rapid SERS scanning of individual spot. By using a solution-based LID detection platform, the spatial separation of targets can be avoided by “one-pot” SERS detection of a small-volume target mixture,156-158 thus enabling quicker analysis of multiple targets.

Fig. 11 “One-pot” surface-enhanced Raman scattering platform for simultaneous five-plexed detection of prostate cancer RNA biomarkers within 80 min. Adapted from ref. 147 with permission from John Wiley & Sons.

To enhance detection sensitivity, SERS detection readout can be augmented with a prior PCR amplification step to increase NA target copies.156,159,160 As reviewed earlier, isothermal amplification techniques are attractive for designing POC diagnostics. In order to achieve multiplexed SERS NA biomarker detection for POC use, it will be ideal to integrate sample preparation and isothermal amplification protocols prior to “one-pot” SERS readout. To realise this aim, our research group has designed a sensor platform based on isothermal RPA and SERS detection readout. Lau et al. recently demonstrated an integrated one-tube assay

Chapter 3.1

(combining sample preparation, target amplification, and detection readout) for simultaneous on-site (outside of a laboratory setting) detection of three different plant pathogen DNA within a 40 min sample-to-answer timeframe.158 For future POC personalised diagnostics, this miniaturised and integrated biosensing assay is currently being adapted for human clinical samples such as blood or urine. Additionally, the multiplexing capability of the RPA and SERS combination has been further investigated. Koo et al. have presented a five-plexed molecular subtyping assay for simultaneously detection of prostate cancer RNA biomarkers using multiplexed isothermal RPA prior to SERS readout of the different target amplicons (Fig. 11).157 This work shows distinct advantages for personalised diagnostics development, including (i) rapid detection owing to the isothermal RT-RPA with “one-pot” SERS readout; (ii) easy interpretation of detection outcomes from well-resolved SERS spectral peaks without the need for further data processing; (iii) applicable to NA targets in clinical samples such as tissue biopsy specimens and urine samples. As with SDG, this RPA and SERS combination strategy has been performed in small-volume reactions, making it amenable for future development into a LID personalised diagnostics platform for multiplexed NA biomarker screening. Despite the suitability of SERS for multiplexed biomarker detection in LID systems, the design of SERS-associated readout equipment and labels are not applicable for POC usage so far. Section 5 of this review will discuss these factors in detail and how current research directions enable SERS for LID personalised diagnostics.

5 TOWARDS TRUE LAB-IN-A-DROP (LID) SYSTEMS To meet the urgent need for POC personalised diagnostics, LID systems could harness similar advantages of rapid and miniaturised NA detection as offered by LOC platforms. Additionally, the main conceived benefit of LID systems is the downscaled integration of an entire diagnostics workflow into a droplet within a microtube and thus avoiding the need for costly precise device engineering. Although the fabrication of paper-based microfluidics or even more complex (such as the SlipChip)161 LOC platforms are becoming relatively simpler as compared to the past, it is inevitable that specialised skills and considerable costs are still required. LID could address these issues by miniaturising biosensing for compatibility with a mass-produced standard microtube format, without any form of fabrication. The ideal LID system would combine and integrate the 3 main aspects of NA diagnostics as reviewed in previous sections: sample preparation, target amplification and detection readout. In this section, we will discuss and offer insights on how to advance (with

Chapter 3.1 the SDG assay highlighted in section 4.2.1 as an example) towards a truly integrated LID system for POC use.

The SDG assay which currently utilizes a magnetic beads flocculation mechanism to produce a colourimetric change in presence of NA target. SDG has shown potential for development into simple, rapid and affordable LID diagnostics for use in homes by personnel with minimal training. In its present form, SDG has incorporated miniaturised isothermal RPA and flocculation-based colourimetric readout into droplet-sized reactions in a microtube for clinical sample testing. However, SDG is lacking a compatible upstream sample preparation workflow which could be downscaled and seamlessly integrated. A miniaturised microtube- based SPE technique could be a feasible sample preparation methodology for SDG. SPRI beads could be used for isolating NA from clinical samples before transferring a droplet of the extracted NA for isothermal amplification and subsequent naked-eye detection. Another plausible way is to functionalize the inner surface of the microtube with probes for NA targets capture from clinical samples and then performed a downscaled RPA and detection readout on the captured targets within the same microtube.162 Additionally, an ideal miniaturised sample preparation technique for SDG should be capable of shortening the current total assay time, and simplistic enough for self-testing by non-trained personnel.

Whilst the qualitative colourimetric readout of SDG is suited for providing a quick POC yes/no readout for a single NA target, it may not be suitable for immediate quantification of disease biomarker levels that are of interest to clinicians in hospitals. In this case, a quantitative readout could be achieved by changing the detection of RPA-amplified targets to an alternative colour change readout which could be measured spectrometrically or electrochemically from minimal liquid volume by trained personnel. Moreover, simultaneous quantitative analysis of several biomarkers from minuscule sample input via multiplexed SERS is also a viable alternative readout for SDG. At present, the biggest limitation of SERS for POC diagnostics is the high cost of portable/handheld Raman spectrometers. It is anticipated that the situation would change with the advancements in optical instruments miniaturisation into portable systems or mobile phone platforms. Additionally, the bulk synthesis of SERS labels with minimal batch variation and maximal stability is still challenging, and research in large-scale production of SERS labels for reproducible usage is highly desired. We anticipate that SDG could be coupled with such developments for quantitative and multiplexed NA biosensing at the POC.

Chapter 3.1

Hence, with further improvements by integrated miniaturised sample preparation and superior quantitative multiplexed readout, we envision that SDG has significant potential for evolving into a complete LID personalised diagnostics system with extensive applications.

6 SUMMARY AND PERSPECTIVES In summary, we have reviewed the three essential aspects of miniaturised biosensing, namely: downscaled sample preparation, miniaturised isothermal amplification, and microscaled detection readout. Additionally, we put forward the concept of a “lab-in-a-drop (LID)” detection platform, which integrates the three above-mentioned fundamental biosensing aspects within a miniaturised fluid droplet to achieve simple and rapid personalised diagnostics at POC.

We evaluated several different sample preparation methods using phenol:chloroform, SPE, and direct sampling. SPE is currently considered to be a suitable sample preparation technique for miniaturised diagnostics due to its simplicity, low cost of operation, and lack of hazardous waste. However, direct sampling from biological samples without any prior lysis and extraction would be most ideal and advancements in this area would be significant progress towards the translation of LID/LOC technologies for clinical use.

To detect trace amounts of NA targets from biological samples, a target amplification strategy is required to enhance the resultant readout signal to a detectable level. An isothermal amplification technique is extremely compatible for POC personalised diagnostics without the need for conventional PCR thermal cycling. We appraised several isothermal amplification methods which are compatible for use in miniaturised diagnostics. Each different method carries its own unique benefits and disadvantages; and the choice of which for use in LID/LOC platforms would be dependent on assay features assay such as nature of biomarker targets (i.e. DNA or RNA), required length of resultant amplicons, or amplification time.

As for types of detection readouts, we underlined various miniaturised fluorescence and electrochemical readouts which have been successfully demonstrated on microfluidic LOC systems. With a focus on the development of LID systems, we also provided a more in-depth

Chapter 3.1 review of colourimetric and SERS readouts which are compatible for providing rapid yes/no outcome of biomarker presence or measurements of several different biomarker targets from a single tube respectively.

Overall, we believe that the numerous recent displays of LID system innovations; which have now allow the possibility of integrating traditionally laboratory-based NA analysis procedures (preparation, amplification, detection) into a miniaturised fluid droplet, is currently resulting in a suite of highly-disruptive POC diagnostics technologies. LID systems could thus play a significantly useful role in personalised disease diagnosis. In this new era of precision medicine, rapid and affordable personalised diagnosis could pave the way for early detection and regular monitoring of disease for individual patients, thus allowing timely and highly-effective medical interventions. Through miniaturisation of typical NA detection assays into an integrated microtube-based platform, LID systems may allow for simple, convenient, and inexpensive NA biomarker detection. In essence, a LID assay could be employed for POC use in the clinic for quick clinical diagnoses; in homes for frequent individualised testing; or out in the field for community healthcare.

There still exist several hurdles towards the realisation of a LID system which could be performed at POC for personalised disease diagnosis. The main challenge is the integration of a suitable sample preparation technique into the current format of LID system (i.e. SDG assay). We envisage that increasing research effort into rapid and simple techniques for isolating NA from crude biological samples could resolve this conundrum in the near future. Another issue is the realisation of a detection readout which could be performed at POC to quantitatively analyse several NA biomarkers at the same time. We identify impending improvements to portable SERS instruments to be a possible answer to this problem.

Nonetheless, miniaturised biosensors are highly promising for the next generation of personalised diagnostics, and we anticipate that further research into the development of LID systems could play a role in the translation of personalised diagnostics into commercial use. Lastly, we hope that this review will provide readers with the crucial role of personalised diagnostics in this new era of precision disease diagnosis and treatment, as well as pique interest in the concept and potential of miniaturised LID systems.

Chapter 3.1

REFERENCES 1. M. A. Hamburg and F. S. Collins, N. Engl. J. Med., 2010, 363, 301-304. 2. M. U. Ahmed, I. Saaem, P. C. Wu and A. S. Brown, Crit. Rev. Biotechnol., 2014, 34, 180-196. 3. F. S. Collins and H. Varmus, N. Engl. J. Med., 2015, 372, 793-795. 4. N. B. La Thangue and D. J. Kerr, Nat. Rev. Clin. Oncol., 2011, 8, 587-596. 5. G. H. Heppner, Cancer Res., 1984, 44, 2259-2265. 6. I. J. Fidler, Cancer Res., 1978, 38, 2651-2660. 7. R. Junker, H. Schlebusch and P. B. Luppa, Deutsches Arzteblatt Int., 2010, 107, 561- U514. 8. C. Lehmann, Clin. Leadersh. Manag. Rev., 2002, 16, 27-31 9. Z. Zhang and S. Nagrath, Biomed. Microdevices, 2013, 15, 595-609. 10. M. L. Kovarik, P. C. Gach, D. M. Ornoff, Y. L. Wang, J. Balowski, L. Farrag and N. L. Allbritton, Anal. Chem., 2012, 84, 516-540. 11. M. L. Kovarik, D. M. Ornoff, A. T. Melvin, N. C. Dobes, Y. L. Wang, A. J. Dickinson, P. C. Gach, P. K. Shah and N. L. Allbritton, Anal. Chem., 2013, 85, 451- 472. 12. M. Boyd-Moss, S. Baratchi, M. Di Venere and K. Khoshmanesh, Lab Chip, 2016, 16, 3177-3192. 13. P. A. Sharp, Cell, 2009, 136, 577-580. 14. D. R. Robinson, Y.-M. Wu, R. J. Lonigro, P. Vats, E. Cobain, J. Everett, X. Cao, E. Rabban, C. Kumar-Sinha, V. Raymond, S. Schuetze, A. Alva, J. Siddiqui, R. Chugh, F. Worden, M. M. Zalupski, J. Innis, R. J. Mody, S. A. Tomlins, D. Lucas, L. H. Baker, N. Ramnath, A. F. Schott, D. F. Hayes, J. Vijai, K. Offit, E. M. Stoffel, J. S. Roberts, D. C. Smith, L. P. Kunju, M. Talpaz, M. Cieślik and A. M. Chinnaiyan, Nature, 2017, doi:10.1038/nature23306. 15. Y. X. Zhao, F. Chen, Q. Li, L. H. Wang and C. H. Fan, Chem. Rev., 2015, 115, 12491-12545. 16. L. Yan, J. Zhou, Y. Zheng, A. S. Gamson, B. T. Roembke, S. Nakayama and H. O. Sintim, Mol. Biosyst., 2014, 10, 970-1003. 17. M. Ritzi-Lehnert, Expert Rev. Mol. Diagnostics, 2012, 12, 189-206. 18. M. A. Dineva, L. Mahilum-Tapay and H. Lee, Analyst, 2007, 132, 1193-1199.

Chapter 3.1

19. C. A. Baker, C. T. Duong, A. Grimley and M. G. Roper, Bioanalysis, 2009, 1, 967- 975. 20. J. Wu and M. Gu, J. Biomed. Opt., 2011, 16, 080901. 21. J. Kim, M. Johnson, P. Hill and B. K. Gale, Interg. Biol., 2009, 1, 574-586. 22. J. Sambrook and D. W. Russell, Cold Spring Harbor Protoc., 2006, 2006, pdb.prot4455. 23. P. Chomczynski and N. Sacchi, Anal. Biochem., 1987, 162, 156-159. 24. R. P. Perry, J. La Torre, D. E. Kelley and J. R. Greenberg, Biochim. et Biophysica Acta (BBA) - Nucleic Acids and Protein Synth., 1972, 262, 220-226. 25. M. C. Morales and J. D. Zahn, Microfluidics Nanofluidics, 2010, 9, 1041-1049. 26. R. Zhang, H. Q. Gong, X. D. Zeng, C. P. Lou and C. Sze, Anal. Chem., 2013, 85, 1484-1491. 27. M. A. Marko, R. Chipperfield and H. C. Birnboim, Anal Biochem., 1982, 121, 382- 387. 28. R. Boom, C. J. Sol, M. M. Salimans, C. L. Jansen, P. M. Wertheim-van Dillen and J. van der Noordaa, J. Clin. Microbiol., 1990, 28, 495-503. 29. J. Wen, L. A. Legendre, J. M. Bienvenue and J. P. Landers, Anal. Chem., 2008, 80, 6472-6479. 30. M. M. DeAngelis, D. G. Wang and T. L. Hawkins, Nucleic Acids Res., 1995, 23, 4742-4743. 31. Y. C. Xu, B. Vaidya, A. B. Patel, S. M. Ford, R. L. McCarley and S. A. Soper, Anal. Chem., 2003, 75, 2975-2984. 32. K. A. Hagan, W. L. Meier, J. P. Ferrance and J. P. Landers, Anal Chem., 2009, 81, 5249-5256. 33. J. R. Moeller, N. R. Moehn, D. M. Waller and T. J. Givnish, Applications Plant Sci., 2014, 2, 1400048. 34. X. Su and A. M. Comeau, Anal. Biochem., 1999, 267, 415-418. 35. S. Fisher, A. Barry, J. Abreu, B. Minie, J. Nolan, T. M. Delorey, G. Young, T. J. Fennell, A. Allen, L. Ambrogio, A. M. Berlin, B. Blumenstiel, K. Cibulskis, D. Friedrich, R. Johnson, F. Juhn, B. Reilly, R. Shammas, J. Stalker, S. M. Sykes, J. Thompson, J. Walsh, A. Zimmer, Z. Zwirko, S. Gabriel, R. Nicol and C. Nusbaum, Genome Biol., 2011, 12, 1-15. 36. O. Mosley, L. Melling, M. D. Tarn, C. Kemp, M. M. Esfahani, N. Pamme and K. J. Shaw, Lab Chip, 2016, 16, 2108-2115.

Chapter 3.1

37. U. Lehmann, C. Vandevyver, V. K. Parashar and M. A. M. Gijs, Angew. Chem., Int. Edit., 2006, 45, 3062-3067. 38. K. Sur, S. M. McFall, E. T. Yeh, S. R. Jangam, M. A. Hayden, S. D. Stroupe and D. M. Kelso, J. Mol. Diagnostics : JMD, 2010, 12, 620-628. 39. D. P. Chandler, S. B. Griesemer, C. G. Cooney, R. Holmberg, N. Thakore, B. Mokhiber, P. Belgrader, C. Knickerbocker, J. Schied and K. St George, J. Virol. Methods, 2012, 183, 8-13. 40. R. C. Holmberg, A. Gindlesperger, T. Stokes, D. Brady, N. Thakore, P. Belgrader, C. G. Cooney and D. P. Chandler, J. Visualized Exp. : JoVE, 2013, 76, 50356. 41. S. B. Griesemer, R. Holmberg, C. G. Cooney, N. Thakore, A. Gindlesperger, C. Knickerbocker, D. P. Chandler and K. St.George, J. Clin. Virol., 2013, 58, 138-141. 42. J. W. Hong, V. Studer, G. Hang, W. F. Anderson and S. R. Quake, Nat. Biotechnol., 2004, 22, 435-439. 43. J. Akutsu, Y. Tojo, O. Segawa, K. Obata, M. Okochi, H. Tajima and M. Yohda, Biotechnol. Bioengineering, 2004, 86, 667-671. 44. A. Trombley Hall, A. McKay Zovanyi, D. R. Christensen, J. W. Koehler and T. Devins Minogue, PLOS ONE, 2013, 8, e73845. 45. M. B. Kermekchiev, L. I. Kirilova, E. E. Vail and W. M. Barnes, Nucleic Acids Res., 2009, 37, e40-e40. 46. B. Lam, Z. C. Fang, E. H. Sargent and S. O. Kelley, Anal. Chem., 2012, 84, 21-25. 47. J. D. Besant, J. Das, E. H. Sargent and S. O. Kelley, ACS Nano, 2013, 7, 8183-8189. 48. H. Lu, M. A. Schmidt and K. F. Jensen, Lab Chip, 2005, 5, 23-29. 49. T. Coolbear, J. M. Whittaker and R. M. Daniel, Biochem. J., 1992, 287, 367-374. 50. B. L. Thompson, C. Birch, J. Li, J. A. DuVall, D. Le Roux, D. A. Nelson, A.-C. Tsuei, D. L. Mills, S. T. Krauss, B. E. Root and J. P. Landers, Analyst, 2016, 141, 4667-4675. 51. J. A. Lounsbury, N. Coult, D. C. Miranian, S. M. Cronk, D. M. Haverstick, P. Kinnon, D. J. Saul and J. P. Landers, Forensic Sci. Int.: Genet., 2012, 6, 607-615. 52. D. Moss, S.-A. Harbison and D. J. Saul, Int. J. Legal Med., 2003, 117, 340-349. 53. T. Coolbear, C. W. Eames, Y. Casey, R. M. Daniel and H. W. Morgan, J. Applied Bacteriol., 1991, 71, 252-264. 54. M. C. Giuffrida and G. Spoto, Biosens. Bioelectron., 2017, 90, 174-186. 55. P. J. Asiello and A. J. Baeumner, Lab Chip, 2011, 11, 1420-1430.

Chapter 3.1

56. T. Notomi, H. Okayama, H. Masubuchi, T. Yonekawa, K. Watanabe, N. Amino and T. Hase, Nucleic Acids Res., 2000, 28, e63. 57. N. Tomita, Y. Mori, H. Kanda and T. Notomi, Nat. Protoc., 2008, 3, 877-882. 58. X. E. Fang, H. Chen, L. J. Xu, X. Y. Jiang, W. J. Wu and J. L. Kong, Lab Chip, 2012, 12, 1495-1499. 59. S. J. Oh, B. H. Park, J. H. Jung, G. Choi, D. C. Lee, D. H. Kim and T. S. Seo, Biosens. Bioelectron., 2016, 75, 293-300. 60. S. Y. Lee, J. G. Huang, T. L. Chuang, J. C. Sheu, Y. K. Chuang, M. Holl, D. R. Meldrum, C. N. Lee and C. W. Lin, Sens. Actuators, B, 2008, 133, 493-501. 61. J. Z. Song, M. G. Mauk, B. A. Hackett, S. Cherry, H. H. Bau and C. C. Liu, Anal. Chem., 2016, 88, 7289-7294. 62. T. D. Rane, L. B. Chen, H. C. Zec and T. H. Wang, Lab Chip, 2015, 15, 776-782. 63. J. Rodriguez-Manzano, M. A. Karymov, S. Begolo, D. A. Selck, D. V. Zhukov, E. Jue and R. F. Ismagilov, ACS Nano, 2016, 10, 3102-3113. 64. M. Vincent, Y. Xu and H. M. Kong, EMBO Rep., 2004, 5, 795-800. 65. N. Ramalingam, T. C. San, T. J. Kai, M. Y. M. Mak and H. Q. Gong, Microfluidics Nanofluidics, 2009, 7, 325-336. 66. M. Mahalanabis, J. Do, H. Almuayad, J. Y. Zhang and C. M. Klapperich, Biomed. Microdevices, 2010, 12, 353-359. 67. Y. Zhang, S. Park, K. Liu, J. Tsuan, S. Yang and T. H. Wang, Lab Chip, 2011, 11, 398-406. 68. J. Baner, M. Nilsson, M. Mendel-Hartvig and U. Landegren, Nucleic Acids Res., 1998, 26, 5073-5078. 69. P. M. Lizardi, X. H. Huang, Z. R. Zhu, P. Bray-Ward, D. C. Thomas and D. C. Ward, Nat. Genet., 1998, 19, 225-232. 70. L. Mahmoudian, N. Kaji, M. Tokeshi, M. Nilsson and Y. Baba, Anal. Chem., 2008, 80, 2483-2490. 71. J. Jarvius, J. Melin, J. Goransson, J. Stenberg, S. Fredriksson, C. Gonzalez-Rey, S. Bertilsson and M. Nilsson, Nat. Methods, 2006, 3, 725-727. 72. S. Juul, C. J. F. Nielsen, R. Labouriau, A. Roy, C. Tesauro, P. W. Jensen, C. Harmsen, E. L. Kristoffersen, Y. L. Chiu, R. Frohlich, P. Fiorani, J. Cox-Singh, D. Tordrup, J. Koch, A. L. Bienvenu, A. Desideri, S. Picot, E. Petersen, K. W. Leong, Y. P. Ho, M. Stougaard and B. R. Knudsen, ACS Nano, 2012, 6, 10676-10683.

Chapter 3.1

73. O. Piepenburg, C. H. Williams, D. L. Stemple and N. A. Armes, PLOS Biol., 2006, 4, e204. 74. J. S. Gootenberg, O. O. Abudayyeh, J. W. Lee, P. Essletzbichler, A. J. Dy, J. Joung, V. Verdine, N. Donghia, N. M. Daringer, C. A. Freije, C. Myhrvold, R. P. Bhattacharyya, J. Livny, A. Regev, E. V. Koonin, D. T. Hung, P. C. Sabeti, J. J. Collins and F. Zhang, Science, 2017, 356, 438-442. 75. S. Lutz, P. Weber, M. Focke, B. Faltin, J. Hoffmann, C. Muller, D. Mark, G. Roth, P. Munday, N. Armes, O. Piepenburg, R. Zengerle and F. von Stetten, Lab Chip, 2010, 10, 887-893. 76. E. J. H. Wee, H. Y. Lau, J. R. Botella and M. Trau, Chem. Commun., 2015, 51, 5828- 5831. 77. E. J. H. Wee, T. H. Ngo and M. Trau, Sci. Rep., 2015, 5, 15028. 78. B. Y. C. Ng, E. J. H. Wee, N. P. West and M. Trau, Sci. Rep., 2015, 5, 15027. 79. K. M. Koo, E. J. H. Wee, P. N. Mainwaring and M. Trau, Sci. Rep., 2016, 6, 30722. 80. F. Shen, E. K. Davydova, W. B. Du, J. E. Kreutz, O. Piepenburg and R. F. Ismagilov, Anal. Chem., 2011, 83, 3533-3540. 81. M. N. Tsaloglou, R. J. Watson, C. M. Rushworth, Y. Zhao, X. Niu, J. M. Sutton and H. Morgan, Analyst, 2015, 140, 258-264. 82. Z. Li, Y. Liu, Q. Q. Wei, Y. J. Liu, W. W. Liu, X. L. Zhang and Y. D. Yu, PLOS One, 2016, 11, e0153359. 83. J. Compton, Nature, 1991, 350, 91-92. 84. J. C. Guatelli, K. M. Whitfield, D. Y. Kwoh, K. J. Barringer, D. D. Richman and T. R. Gingeras, Proc. Natl. Acad. Sci. U. S. A., 1990, 87, 1874-1878. 85. K. Pardee, A. A. Green, M. K. Takahashi, D. Braff, G. Lambert, J. W. Lee, T. Ferrante, D. Ma, N. Donghia, M. Fan, N. M. Daringer, I. Bosch, D. M. Dudley, D. H. O'Connor, L. Gehrke and J. J. Collins, Cell, 2016, 165, 1255-1266. 86. A. Gulliksen, L. A. Solli, K. S. Drese, O. Sorensen, F. Karlsen, H. Rogne, E. Hovig and R. Sirevag, Lab Chip, 2005, 5, 416-420. 87. I. K. Dimov, J. L. Garcia-Cordero, J. O'Grady, C. R. Poulsen, C. Viguier, L. Kent, P. Daly, B. Lincoln, M. Maher, R. O'Kennedy, T. J. Smith, A. J. Ricco and L. P. Lee, Lab Chip, 2008, 8, 2071-2078. 88. S. J. Reinholt, A. Behrent, C. Greene, A. Kalfe and A. J. Baeumner, Anal. Chem., 2014, 86, 849-856.

Chapter 3.1

89. S. H. Chung, C. Baek, V. T. Cong and J. Min, Biosens. Bioelectron., 2015, 67, 625- 633. 90. G. T. Walker, M. C. Little, J. G. Nadeau and D. D. Shank, Proc. Natl. Acad. Sci. U. S. A., 1992, 89, 392-396. 91. M. A. Burns, B. N. Johnson, S. N. Brahmasandra, K. Handique, J. R. Webster, M. Krishnan, T. S. Sammarco, P. M. Man, D. Jones, D. Heldsinger, C. H. Mastrangelo and D. T. Burke, Science, 1998, 282, 484-487. 92. J. M. Yang, J. Bell, Y. Huang, M. Tirado, D. Thomas, A. H. Forster, R. W. Haigis, P. D. Swanson, R. B. Wallace, B. Martinsons and M. Krihak, Biosens. Bioelectron., 2002, 17, 605-618. 93. Y. Y. Zhao, L. Zhou and Z. Tang, Nat. Commun., 2013, 4, 1493. 94. V. Killick, P. Kell, H. Plumb, N. Hurd and P. Turner, Int. J. STD & AIDS, 2012, 23, e1-e3. 95. M. C. Giuffrida, L. M. Zanoli, R. D'Agata, A. Finotti, R. Gambari and G. Spoto, Anal. Bioanal. Chem., 2015, 407, 1533-1543. 96. G. A. Posthuma-Trumpie, J. Korf and A. van Amerongen, Anal. Bioanal. Chem., 2009, 393, 569-582. 97. A. Vassall, S. van Kampen, H. Sohn, J. S. Michael, K. R. John, S. den Boon, J. L. Davis, A. Whitelaw, M. P. Nicol, M. T. Gler, A. Khaliqov, C. Zamudio, M. D. Perkins, C. C. Boehme and F. Cobelens, PLOS Med., 2011, 8, e1001120. 98. K. J. Liu, M. V. Brock, L. M. Shin and T. H. Wang, J. Am. Chem. Soc., 2010, 132, 5793-5798. 99. A. W. Martinez, S. T. Phillips and G. M. Whitesides, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 19606-19611. 100. A. D. Warren, G. A. Kwong, D. K. Wood, K. Y. Lin and S. N. Bhatia, Proc. Natl. Acad. Sci. U. S. A., 2014, 111, 3671-3676. 101. B. W. Liu, D. Du, X. Hua, X. Y. Yu and Y. H. Lin, Electroanalysis, 2014, 26, 1214-1223. 102. C. Lui, N. C. Cady and C. A. Batt, Sensors, 2009, 9, 3713-3744. 103. P. N. Nge, C. I. Rogers and A. T. Woolley, Chem. Rev., 2013, 113, 2550- 2583. 104. P. Braun, R. Delgado, M. Drago, D. Fanti, H. Fleury, M. R. Gismondo, J. Hofmann, J. Izopet, S. Kuhn, A. Lombardi, M. A. Marcos, K. Saune, S. O'Shea, A.

Chapter 3.1

Perez-Rivilla, J. Ramble, P. Trimoulet, J. Vila, D. Whittaker, A. Artus and D. W. Rhodes, J. Clin. Microbiol., 2017, 55, 1186-1192. 105. M. J. Binnicker, M. J. Espy, C. L. Irish and E. A. Vetter, J. Clin. Microbiol., 2015, 53, 2353-2354. 106. J. E. Kong, Q. S. Wei, D. Tseng, J. Z. Zhang, E. Pan, M. Lewinski, O. B. Garner, A. Ozcan and D. Di Carlo, ACS Nano, 2017, 11, 2934-2943. 107. 98. K. Ming, J. Kim, M. J. Biondi, A. Syed, K. Chen, A. Lam, M. Ostrowski, A. Rebbapragada, J. J. Feld and W. C. W. Chan, ACS Nano, 2015, 9, 3060-3074. 108. E. J. H. Wee and M. Trau, ACS Sens., 2016, 1, 670–675. 109. K. M. Koo, E. J. H. Wee and M. Trau, Biosens. Bioelectron., 2017, 89, 715- 720. 110. J. Pipper, M. Inoue, L. F. P. Ng, P. Neuzil, Y. Zhang and L. Novak, Nat. Med., 2007, 13, 1259-1263. 111. J. Pipper, Y. Zhang, P. Neuzil and T. M. Hsieh, Angew. Chem.-Int. Edit., 2008, 47, 3900-3904. 112. T. M. H. Lee and I. M. Hsing, Ana. Chim. Acta, 2006, 556, 26-37. 113. J. Wang, Biosens. Bioelectron., 2006, 21, 1887-1892. 114. A. Kara, C. Rouillard, J. Mathault, M. Boisvert, F. Tessier, H. Landari, I. Melki, M. Laprise-Pelletier, E. Boisselier, M. A. Fortin, E. Boilard, J. Greener and A. Miled, Sensors, 2016, 16, 778. 115. B. Lam, J. Das, R. D. Holmes, L. Live, A. Sage, E. H. Sargent and S. O. Kelley, Nat. Commun., 2013, 4, 2001. 116. J. Das and S. O. Kelley, Anal. Chem., 2013, 85, 7333-7338. 117. L. Soleymani, Z. C. Fang, E. H. Sargent and S. O. Kelley, Nat. Nanotechnol., 2009, 4, 844-848. 118. L. Soleymani, Z. C. Fang, B. Lam, X. M. Bin, E. Vasilyeva, A. J. Ross, E. H. Sargent and S. O. Kelley, ACS Nano, 2011, 5, 3360-3366. 119. Z. Olcer, E. Esen, A. Ersoy, S. Budak, D. S. Kaya, M. Y. Gok, S. Barut, D. Ustek and Y. Uludag, Biosens. Bioelectron., 2015, 70, 426-432. 120. K. Hsieh, A. S. Patterson, B. S. Ferguson, K. W. Plaxco and H. T. Soh, Angew. Chem.-Int. Edit., 2012, 51, 4896-4900. 121. B. Y. C. Ng, W. Xiao, N. P. West, E. J. H. Wee, Y. Wang and M. Trau, Anal. Chem., 2015, 87, 10613-10618.

Chapter 3.1

122. Z. Taleat, A. Khoshroo and M. Mazloum-Ardakani, Microchim. Acta, 2014, 181, 865-891. 123. K. M. Koo, L. G. Carrascosa, M. J. A. Shiddiky and M. Trau, Anal. Chem., 2016, 88, 2000-2005. 124. K. M. Koo, L. G. Carrascosa, M. J. A. Shiddiky and M. Trau, Anal. Chem., 2016, 88, 6781-6788. 125. Y. Xiang and Y. Lu, Anal. Chem., 2012, 84, 1975-1980. 126. Y. Xiang and Y. Lu, Nat. Chem., 2011, 3, 697-703. 127. W. E. Jung, J. Han, J. W. Choi and C. H. Ahn, Microelectron. Engineering, 2015, 132, 46-57. 128. Y. J. Song, Y. Y. Huang, X. W. Liu, X. J. Zhang, M. Ferrari and L. D. Qin, Trends in Biotechnol., 2014, 32, 132-139. 129. K. M. Koo, E. J. H. Wee and M. Trau, Theranostics, 2016, 6, 1415-1424. 130. B. Y. C. Ng, E. J. H. Wee, N. P. West and M. Trau, ACS Sens, 2015, 1 , 173- 178. 131. L. Z. Gao, J. Zhuang, L. Nie, J. B. Zhang, Y. Zhang, N. Gu, T. H. Wang, J. Feng, D. L. Yang, S. Perrett and X. Yan, Nat. Nanotechnol., 2007, 2, 577-583. 132. C. S. Thaxton, D. G. Georganopoulou and C. A. Mirkin, Clin. Chim. Acta, 2006, 363, 120-126. 133. R. Elghanian, J. J. Storhoff, R. C. Mucic, R. L. Letsinger and C. A. Mirkin, Science, 1997, 277, 1078-1081. 134. R. A. Reynolds, C. A. Mirkin and R. L. Letsinger, J. Am. Chem. Soc., 2000, 122, 3795-3796. 135. W. Xu, X. J. Xie, D. W. Li, Z. Q. Yang, T. H. Li and X. G. Liu, Small, 2012, 8, 1846-1850. 136. R. H. Smellie and V. K. La Mer, J. Colloid Sci., 1958, 13, 589-599. 137. B. Laxman, D. S. Morris, J. J. Yu, J. Siddiqui, J. Cao, R. Mehra, R. J. Lonigro, A. Tsodikov, J. T. Wei, S. A. Tomlins and A. M. Chinnaiyan, Cancer Res., 2008, 68, 645-649. 138. A. Romeo, T. S. Leung and S. Sanchez, Lab Chip, 2016, 16, 1957-1961. 139. M. Fleischmann, P. J. Hendra and McQuilla.Aj, Chem. Phys. Lett., 1974, 26, 163-166. 140. M. G. Albrecht and J. A. Creighton, J. Am. Chem. Soc., 1977, 99, 5215-5217. 141. S. M. Nie and S. R. Emory, Science, 1997, 275, 1102-1106.

Chapter 3.1

142. K. Kneipp, Y. Wang, H. Kneipp, L. T. Perelman, I. Itzkan, R. Dasari and M. S. Feld, Phys. Rev. Lett., 1997, 78, 1667-1670. 143. Y. W. C. Cao, R. C. Jin and C. A. Mirkin, Science, 2002, 297, 1536-1540. 144. L. Sun, C. X. Yu and J. Irudayaraj, Anal. Chem., 2007, 79, 3981-3988. 145. L. Sun, C. X. Yu and J. Irudayaraj, Anal. Chem., 2008, 80, 3342-3349. 146. E. J. H. Wee, Y. Wang, S. C. H. Tsao and M. Trau, Theranostics, 2016, 6, 1506-1513. 147. K. M. Koo, E. J. H. Wee, P. N. Mainwaring, Y. Wang and M. Trau, Small, 2016, 12, 6233-6242. 148. H. Y. Lau, Y. Wang, E. J. H. Wee, J. R. Botella and M. Trau, Anal. Chem., 2016, 88, 8074-8081. 149. D. Graham, B. J. Mallinder, D. Whitcombe, N. D. Watson and W. E. Smith, Anal. Chem., 2002, 74, 1069-+. 150. A. MacAskill, D. Crawford, D. Graham and K. Faulds, Anal. Chem., 2009, 81, 8134-8140. 151. W. B. Du, L. Li, K. P. Nichols and R. F. Ismagilov, Lab Chip, 2009, 9, 2286- 2292. 152. M. N. Leiske, M. Hartlieb, C. Paulenz, D. Pretzel, M. Hentschel, C. Englert, M. Gottschaldt and U. S. Schubert, Adv. Func. Mater., 2015, 25, 2458-2466.

Chapter 3.2

A Simple, Rapid, Low-Cost Technique for Naked-Eye Detection of Urine-Isolated TMPRSS2:ERG Gene Fusion RNA

Kevin M. Koo, Eugene J. H. Wee, Paul N. Mainwaring, Matt Trau

ABSTRACT The TMPRSS2:ERG gene fusion is one of a series of highly promising prostate cancer (PCa) biomarker alternatives to the controversial serum PSA. Current methods for detecting TMPRSS2:ERG are limited in terms of long processing time, high cost and the need for specialized equipment. Thus, there is an unmet need for less complex, faster, and cheaper methods to enable gene fusion detection in the clinic. We describe herein a simple, rapid and inexpensive assay which combines robust isothermal amplification technique with a novel visualization method for evaluating urinary TMPRSS2:ERG status at less than USD 5 and with minimal equipment. The assay is sensitive, and rapidly detects as low as 105 copies of TMPRSS2:ERG transcripts while maintaining high levels of specificity.

INTRODUCTION Current prostate cancer (PCa) screening relies mainly on measuring serum prostate-specific antigen (PSA) levels which has clearly demonstrated improvements in patient survival163-165. Although PSA screening has been utilized widely for several decades, its value as a biomarker is controversial due to its poor sensitivity and specificity. As a consequence, this has frequently led to over diagnosis and unnecessary medical expenses163,166,167. Therefore, this has encouraged researchers to develop a new generation of more accurate PCa screening biomarkers. In 2005, the gene fusion between transmembrane protease, serine 2 (TMPRSS2) and v-ets avian erythroblastosis virus E26 oncogene homolog (ERG) was reported as a recurring event in PCa4. Further studies have confirmed TMPRSS2:ERG to be a highly- specific PCa biomarker that is present in at least 50% of PCa cases168 and is associated with PCa progression through inducing androgen-regulated ERG overexpression as well as partnering with other oncogenic events such as PTEN loss169. The prognostic potential of the TMPRSS2:ERG biomarker has been demonstrated in several studies in which TMPRSS2:ERG presence was associated with more aggressive forms of PCa and poorer

Chapter 3.2 clinical prognosis (higher PSA level, Gleason score and/or tumour stage)6,170. Additionally, TMPRSS2:ERG transcripts are also detectable in urine as a potential non-invasive and convenient biomarker for early PCa-detection. Most importantly, TMPRSS2:ERG is a binary biomarker which is commonly associated with premalignant PCa but not benign conditions, thus making it the most specific PCa biomarker to date171. For treatment purposes, recent studies have shown TMPRSS2:ERG to be a potential drug target for impeding tumor growth and metastasis172,173. Therefore, the detection of TMPRSS2:ERG presents as an unique opportunity for developing novel binary (positive/negative) strategies for PCa screening or treatment.

Reverse transcription-polymerase chain reaction (RT-PCR), and fluorescence in situ hybridization (FISH) are the most commonly-used techniques for TMPRSS2:ERG detection in clinical PCa specimens. While FISH assays are useful for detecting gene fusions on both DNA and RNA in tissue cells174, they have limited sensitivity due to high background fluorescence. As a consequence, FISH may be unsuitable for detecting trace amounts of TMPRSS2:ERG transcripts in clinical samples such as urine specimens. In contrast, RT-PCR is a sensitive assay which is routinely used in many studies for detecting low levels of TMPRSS2:ERG transcripts in PCa urine specimens9,147,175. However, PCR techniques are constrained to laboratory settings by the high number of thermal cycles needed to detect low TMPRSS2:ERG levels (typically requiring ≥120 min); need for highly trained personnel; and required specialized equipment. Recently, an isothermal transcription-mediated amplification (TMA)-based assay176 was developed to resolve issues faced by RT-PCR methods but the chemiluminescence readout of this assay still has limitations such as high reagent cost, prolonged experimental procedures as well as the need for specialized instrumentation. Hence, it is still of interest to the community to develop a rapid yet inexpensive novel point- of-care assay to address the shortcomings of current methodologies for routine TMPRSS2:ERG screening.

Colorimetric assays based on gold nanoparticles aggregation have been developed for sensitive and rapid detection of polynucleotide sequences177. However, the subtle colour changes of these assays frequently require the use of spectrophotometry for readout. Therefore, an opportunity presents itself to enable easier visual detection by naked-eye interpretation; and one such possibility could be polymer (eg. DNA)-mediated flocculation

Chapter 3.2 assays86,88,178 which typically possess a binary threshold for flocculation output and interpretation179-181. This sort of binary evaluation system would also complement the binary nature of the TMPRSS2:ERG biomarker.

Herein, we present a novel TMPRSS2:ERG detection technique by combining a robust isothermal amplification method with a flocculation-based visual readout. The assay is a relatively simple, rapid, and inexpensive methodology for evaluating urinary TMPRSS2:ERG status with good sensitivity and specificity.

MATERIALS RNA extraction from cell lines The cell lines DuCap and LnCap were generously donated by Matthias Nees (VTT, Finland); Gregor Tevz (APCRC, Australia); and Michelle Hill (UQDI, Australia). The cells were cultured in RPMI-1640 growth media (Life Technologies, Australia) supplemented with 10% fetal bovine serum (Life Technologies, Australia) in a humidified incubator containing 5% CO2 at 37 ºC. RNA was extracted using Trizol reagent (Life Technologies, Australia) and RNA integrity and purity were checked using a NanoDrop Spectrophotometer (Thermo Scientific, USA).

RNA extraction from clinical urinary specimens Ethics approval was obtained from The University of Queensland Institutional Human Research Ethics Committee (Approval No. 2004000047). Methods pertaining to clinical samples were carried out in accordance with approved guidelines and informed consent was obtained from all subjects prior to sample collection.

Voided urine specimens were collected from 10 men undergoing treatment for CRPCa and 5 healthy men with no clinical history of PCa. Urine specimens (30–50 ml) were centrifuged at 700 g for 10 min and urinary sediments were washed with ice-cold PBS buffer before being centrifuged again at the same conditions. 25 µl of lysis buffer (150 mM Tris-HCI pH 8.0, 4.5 M guanidium-HCI, 3% v/v Triton-X and 1.5 mM EDTA) was added to 50 µl of PBS- suspended urinary sediments with vigorous mixing to release total RNA. Next, total RNA was purifed from 25 µl of cell lysate using the Agencourt AMPure XP kit (Beckman Coulter, USA). Briefly, two volumes SPRI reagent was first incubated with the cell lysate for 10

Chapter 3.2 minutes. The RNA-bound magnetic beads were then separated from the lysate using a magnet and washed once with 100% isopropanol and twice more with 80% ethanol. Finally, total RNA was eluted in 25 µl of RNase-free water. cDNA synthesis and nucleic acid amplification For cell line experiments, the TwistAmp Basic RPA kit (Twist-DX, UK) was used with slight modifications to manufacturer’s instructions. Briefly, 1 µl of extracted RNA, 50 units of MMuLV reverse transcriptase (New England Biolabs, UK) and 500 nM of each primer (Table 1) were added to make a 12.5 µl reaction volume prior to incubation at 39°C for 20 min. For all other experiments, the TwistAmp Basic RT-RPA kit (Twist-DX, UK) was used with slight modifications to manufacturer’s instructions to simultaneously generate cDNA and rapidly amplify the cDNA templates in a single tube reaction. Briefly, 2 µl of extracted RNA and 500 nM of each primer (Table 1) were added to make a 12.5 µl reaction volume prior to incubation at 41°C for 30 min.

Table 1 Oligonucleotide sequences Oligonucletotides Sequence (5’ to 3’) RT-RPA Forward, RT-PCR Forward CCTGGAGCGCGGCAGGAAGCCTTATCAGTTG RT-RPA Reverse GCTAGGGTTACATTCCATTTTGATGGTGAC RT-PCR Reverse TCCTGCTGAGGGACGCGTGGGCTCATCTTG Oligonucleotides were purchased from Integrated DNA Technologies, USA

Bridging flocculation colorimetric readout RPA amplicons was similarly purified using SPRI as previously described but with only one ethanol wash step. After purification, 5 µl of purified RPA amplicons was incubated with 2 volumes of SPRI magnetic beads for 5 min before bead separation by magnet. Then, 20 µL of flocculation buffer (200 mM sodium acetate, pH 4.4) was added to the beads. After 1 min of incubation, the mixture was gently agitated using a magnet and observed for colour change.

RT-PCR RT-PCR was performed on extracted total RNA from clinical urine specimens to validate results from our assay. The KAPA SYBR FAST One-Step qRT-PCR kit (KAPA BIOSYSTEMS, USA) was used to set up a single reaction volume of 10 µl for each sample. Each reaction volume consist of 1X KAPA SYBR FAST qPCR Master Mix, 200 nM of each

Chapter 3.2 primer (Table 1), 1X KAPA RT Mix and 3 µl of extracted RNA. The tubes were incubated at 42°C for 10 min to synthesize cDNA, followed by 95°C for 5 min to deactivate RT before cycling 35 times (95°C for 30 s, 70°C for 30 s and 72°C for 1 min) and finished with 72°C for 10 min. Lastly, the RT-PCR products were visualized on 1.5% agarose gel to verify amplification.

RESULTS Assay principle Figure 1 depicts the working principle of our assay. This assay first employed magnetic beads to isolate and purify total RNA from urinary sediments by the Solid Phase Reversible Immobilization (SPRI) technique182. Next, isothermal reverse transcriptase-recombinase polymerase amplification83 (RT-RPA) was utilized to rapidly amplify TMPRSS2:ERG transcripts within 30 min. Lastly, amplified DNA (i.e. positive for TMPRSS2:ERG) were detected using a DNA-mediated bridging flocculation phenomenon whereby only amplicons of sufficient length (i.e., ≥200 bp)88 and amounts could initiate flocculation of magnetic beads to maintain a colourless solution.

Figure 1. Schematic representation of rapid and simple assay for TMPRSS2:ERG gene fusion detection in prostate cancer urine specimens. Total RNA which potentially include TMPRSS2:ERG transcripts is first isolated from the urine specimen of a screening candidate. Isothermal reverse transcriptase-recombinase polymerase amplification is used to generate

Chapter 3.2 cDNA amplicons exclusively in the presence of TMPRSS2:ERG fusion transcripts. This is followed by adding SPRI magnetic beads to spontaneously bind post-amplification sequences. Successful amplicons (i.e. TMPRSS2:ERG-positive) induces bridging flocculation of magnetic beads to produce a colourless solution. Specimens without TMPRSS2:ERG transcripts result in no amplification and beads bound with only primer sequences are of insufficient length to mediate flocculation, thereby producing a brown- coloured solution indicative of a negative TMPRSS2:ERG reading.

Assay specificity for TMPRSS2:ERG As a proof-of-concept, we applied our method to detect the most common TMPRSS2:ERG fusion transcript (TMPRSS2 exon1 to ERG exon 4) in cultured DuCap (TMPRSS2:ERG- positive) and LnCap (TMPRSS2:ERG-negative) human PCa cells4. The RT-RPA primers spanned the fusion junction of TMPRSS2 and ERG (Fig. 1), thus reducing the likelihood of amplifying wild type transcripts. As expected, DuCap cells were tested positive (colourless) for TMPRSS2:ERG while LnCap were negative (brown-coloured) (Fig. 2A). The RPA products were verified by agarose gel electrophoresis which only showed a 216 bp product band for the DuCap cell line but not for the LnCap cell line. When cell line RNA (NoT control) and reverse transcriptase (no RT control) were omitted from the RPA reaction, no products were detected to affirm that products were TMPRSS2:ERG RNA-dependent. To control for RNA loading, the housekeeping transcript, RN7SL1183, was used. We also confirmed the identity of the RPA amplicons through sequencing (Supplementary Fig. 1). The successful detection of TMPRSS2:ERG in DuCap cells demonstrated the specificity of our assay to detect TMPRSS2:ERG transcripts from total RNA. In addition, the colorimetric readout of the assay clearly showed the distinct colour changes between TMPRSS2:ERG states, thus potentially allowing a rapid and convenient binary evaluation of gene fusion status by the naked eye with minimal equipment.

Chapter 3.2

Figure 2. Detection of TMPRSS2:ERG transcripts in prostate cancer cell lines and clinical urine specimens. (a) TMPRSS2:ERG detection in DuCap and LnCap cell lines. (b) Titration assay of spiking different concentrations of synthetic TMPRSS2:ERG RNA into LnCap RNA background. (c) Two separate runs of TMPRSS2:ERG detection in urine specimens of 10 castrate-resistant prostate cancer patients and 5 healthy males with no prior prostate cancer history; 1st run (Left): PC1-PC10 & H1; 2nd run (Right): H2-H5. No RT and NoT refers to no-reverse transcriptase and no-template controls respectively. Top row: Images of flocculation assays for RT-RPA reactions. Bottom: Gel electrophoresis images corresponding to the RT-RPA reactions. Note: Gel electrophoresis experiments were all performed under similar experimental conditions (200V, 30 min) and images have been cropped for clarity of presentation, full-length blots/gels are presented in Supplementary Fig. 4.

Assay limit-of-detection We then tested the sensitivity of our assay by titrating known amounts of synthetic TMPRSS2:ERG RNA into a 10 ng of LnCap total RNA background. As little as 105 copies of TMPRSS2:ERG RNA (Fig. 2B) could be detected and this number of fusion transcripts was approximately equivalent to that of a single-cell184.

Urine TMPRSS2:ERG in PCa patients To demonstrate clinical utility, we then applied our assay to analyze TMPRSS2:ERG status in clinical urine specimens from 10 metastatic castration-resistant PCa (CRPCa) patients and 5 healthy control patients. Using RN7SL1 housekeeping transcripts as input control, we detected TMPRSS2:ERG transcripts from urine specimens in 7 out of 10 metastatic CRPC

Chapter 3.2 patients (70%) but not in the healthy patient and no-template (NoT) controls (Fig. 2C). These outcomes were further validated by the current gold standard RT-PCR assay (Supplementary Fig. 2) with a different pair of primers (Table 1). Both our assay and RT-PCR gave identical results, thus supporting our TMPRSS2:ERG assay as a feasible faster alternative to RT-PCR.

DISCUSSION The current use of PSA levels for PCa diagnosis has led to incidences of false- positive/negative results185. Gene fusions are one of the new generation of promising PCa biomarkers which could replace or supplement the PSA test as it is highly PCa-specific and absent in benign conditions6,168. Thus, we hypothesized that an assay which is able to reliably and rapidly detect gene fusion presence at a low-cost, could be a useful tool for diagnosing PCa in the point-of-care setting. Here we report a novel urine-based gene fusion assay by detecting the common TMPRSS2 (exon 1):ERG (exon 4)4 in PCa patients. This assay differs from previous reports of gene fusion detection methodologies in the literature as it is able to give a positive/negative flocculation-based outcome detectable by naked eye within 90 minutes with minimal equipment.

The bridging flocculation phenomenon in our assay refers to the ability of RT-RPA amplicons to initiate cross-linking of magnetic beads and flocculate out of an aqueous dispersion, thus rendering the solution colourless (Fig. 1). The length of the amplicons limits this flocculation mechanism as short-length primers (in the event of unsuccessful amplification) are unable to induce bridging of magnetic beads. Bridging flocculation of magnetic beads is particularly useful for visual detection of a variety of nucleic acid-based pathogens after successful amplification86,88. Whilst various lines of research have reported visual DNA detection through magnetic particles aggregation186-188, our assay is unique in requiring only a conventional magnetic plate (i.e. static magnetic field) after a rapid 30 min isothermal amplification.

Importantly, the sensitivity (105 TMPRSS2:ERG transcript copies) and specificity of this approach (Fig. 2A & B) could be useful for detecting low TMPRSS2:ERG amount in a complex clinical sample such as urine. Furthermore, the colorimetric readout of the assay clearly demonstrated the distinct colour changes between TMPRSS2:ERG states, thus potentially allowing a rapid and convenient binary evaluation of gene fusion status. Unlike

100

Chapter 3.2

RT-PCR/FISH assays which require time-consuming preparation, long assay times, and costly fluorescence readouts; our assay is able to avoid these complications. Firstly, the RT- RPA technique is readily-available commercially in pellet form and preparation steps are simple, quick and minimal. Secondly, the amplification process is isothermal and rapid to facilitate shorter assay turn-around-time. In contrast to the traditional fluorescence readouts of RT-PCR/FISH; our novel flocculation-based colorimetric readout is simpler and quicker to prepare, more cost-effective without need for specialized instruments, and free from conventional fluorescence bleaching issues.

To demonstrate potential clinical utility, we used our assay to investigate the TMPRSS2:ERG status in urine samples from 10 CRPCa patients and 5 healthy controls (Fig. 2C). The results were subsequently verified with gold standard RT-PCR with 100% concordance (Supplementary Fig. 2). Furthermore, our assay findings were also consistent with literature reports describing the presence of TMPRSS2:ERG in late-stage metastatic CRPCa patients189. It is also worth highlighting that the assay results were obtained in approximately 90 minutes, and we have also been able to detect TMPRSS2:ERG transcripts from whole urine specimens without any need for ultracentrifugation to concentrate urinary sediments (Supplementary Fig. 3). However, we found that detection from whole urine is only possible if the assay is performed immediately (within 2 hours) after specimen collection. We hypothesized that RNA from whole urine most likely degraded with time, therefore decreasing RT-RPA efficiency. In contrast, sediment RNA remained stable for extraction even after 12 hrs (Supplementary Fig. 3). While using whole urine may be convenient, due to logistic limitations and assay reproducibility, urinary sediments are the more viable RNA source for this study. Nonetheless, if the assay could be done immediately on-site, detecting TMPRSS2:ERG from whole urine may still be a possibility and thus further simplifying the method.

There are several aspects of our assay which could be further developed and improved upon in order to realize its full potential as a clinical tool. First, the flocculation colorimetric readout is not quantitative and only offers a discrete yes/no result in its current format. However, this could be overcome by designing softwares to measure the intensity and size of cross-linked magnetic beads. As more transcripts could lead to higher flocculation, the intensity and size of the flocculated magnetic beads could be quantitative for TMPRSS2:ERG

101

Chapter 3.2 transcripts. Furthermore, our assay could be modified as a chip-based assay in the future as isothermal amplification, magnetic beads, and visual readout are components which are ideal for chip-based assay development. By integrating our assay onto a chip, we could potentially increase assay throughput and multiplexibility, improve portability through assay miniaturization, and eventually automate the whole assay process. Nonetheless, the current assay is still an attractive low resource alternative as compared to RT-PCR or FISH.

In conclusion, we describe a novel coupling of isothermal RT-RPA with a flocculation readout to enable a rapid, convenient, inexpensive and non-invasive method for detecting PCa gene fusion RNA. To our knowledge, our colorimetric readout method is the first display of using the RT-RPA technique and bridging flocculation phenomena to evaluate urinary gene fusion transcripts such as TMPRSS2:ERG. This positive/negative readout is easily evaluated by the naked eye and also complements binary biomarkers such as gene fusions. We envisage that our assay may have wide application potential in detecting different gene fusion occurrences in other forms of cancers or diseases by simply tailoring primer sequences.

References 1. Schroder, F.H. et al. Prostate-Cancer Mortality at 11 Years of Follow-up. N. Eng. J. Med. 366, 981-990 (2012). 2. Prensner, J.R., Rubin, M.A., Wei, J.T. & Chinnaiyan, A.M. Beyond PSA: The next generation of prostate cancer biomarkers. Science Transl. Med. 4, 127rv3 (2012). 3. Cuzick, J. et al. Prevention and early detection of prostate cancer. Lancet Oncol. 15, E484-E492 (2014). 4. Andriole, G.L. et al. Prostate cancer screening in the randomized prostate, lung, colorectal, and ovarian cancer screening trial: Mortality results after 13 Years of follow-up. J. Natl. Cancer Inst. 104, 125-132 (2012). 5. Shoag, J. et al. Efficacy of prostate-specific antigen screening: Use of regression discontinuity in the PLCO cancer screening trial. JAMA Oncol. 1, (2015). 6. Tomlins, S.A. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science 310, 644-648 (2005). 7. Tomlins, S.A. et al. ETS gene fusions in prostate cancer: From discovery to daily clinical practice. Eur. Urol. 56, 275-286 (2009).

102

Chapter 3.2

8. Carver, B.S. et al. Aberrant ERG expression cooperates with loss of PTEN to promote cancer progression in the prostate. Nat. Genet. 41, 619-624 (2009). 9. Wang, J.H., Cai, Y., Ren, C.X. & Ittmann, M. Expression of variant TMPRSS2/ERG fusion messenger RNAs is associated with aggressive prostate cancer. Cancer Res. 66, 8347-8351 (2006). 10. Leyten, G. et al. Prospective multicentre evaluation of PCA3 and TMPRSS2-ERG gene fusions as diagnostic and prognostic urinary biomarkers for prostate cancer. Eur. Urol. 65, 534-542 (2014). 11. Vaananen, R.-M., Ochoa, N.T., Bostrom, P.J., Taimen, P. & Pettersson, K. Altered PCA3 and TMPRSS2-ERG expression in histologically benign regions of cancerous prostates: a systematic, quantitative mRNA analysis in five prostates. BMC Urol. 15, 88 (2015). 12. Rahim, S. & Uren, A. Emergence of ETS transcription factors as diagnostic tools and therapeutic targets in prostate cancer. Am. J. Transl. Res. 5, 254-268 (2013). 13. Rahim, S. et al. A small molecule inhibitor of ETV1, YK-4-279, prevents prostate cancer growth and metastasis in a mouse xenograft model. PLoS One 9, e114260 (2014). 14. Yoshimoto, M. et al. Three-color FISH analysis of TMPRSS2/ERG fusions in prostate cancer indicates that genomic microdeletion of chromosome 21 is associated with rearrangement. Neoplasia 8, 465-469 (2006). 15. Laxman, B. et al. Noninvasive detection of TMPRSS2 : ERG fusion transcripts in the urine of men with prostate cancer. Neoplasia 8, 885-888 (2006). 16. Hessels, D. et al. Detection of TMPRSS2-ERG fusion transcripts and prostate cancer antigen 3 in urinary sediments may improve diagnosis of prostate cancer. Clin. Cancer Res. 13, 5103-5108 (2007). 17. Laxman, B. et al. A first-generation multiplex biomarker analysis of urine for the early detection of prostate cancer. Cancer Res. 68, 645-649 (2008). 18. Tomlins, S.A. et al. Urine TMPRSS2:ERG fusion transcript stratifies prostate cancer risk in men with elevated serum PSA. Science Transl. Med. 3, 94ra72 (2011). 19. Rosi, N.L. & Mirkin, C.A. Nanostructures in biodiagnostics. Chem. Rev. 105, 1547- 1562 (2005).

103

Chapter 3.2

20. Wee, E.J.H., Lau, H.Y., Botella, J.R. & Trau, M. Re-purposing bridging flocculation for on-site, rapid, qualitative DNA detection in resource-poor settings. Chem. Commun. 51, 5828-5831 (2015). 21. Wee, E.J.H., Thu Ha, N. & Trau, M. A simple bridging flocculation assay for rapid, sensitive and stringent detection of gene specific DNA methylation. Sci. Rep. 5, 15028 (2015). 22. Ng, B.Y.C., Wee, E.J.H., West, N.P. & Trau, M. Rapid DNA detection of Mycobacterium tuberculosis-towards single cell sensitivity in point-of-care diagnosis. Sci. Rep. 5, 15027 (2015). 23. Huggins, M.L. Solutions of long chain compounds. J. Chem. Phys. 9, 440-440 (1941). 24. Flory, P.I. Thermodynamics of high polymer solutions. J. Chem. Phys. 10, 51-61 (1942). 25. Healy, T.W. & La Mer, V.K. The energetics of flocculation and redispersion by polymers. J. Colloid Sci. 19, 323-332 (1964). 26. Deangelis, M.M., Wang, D.G. & Hawkins, T.L. Solid-phase reversible immobilization for the isolation of PCR products. Nucleic Acids Res. 23, 4742-4743 (1995). 27. Piepenburg, O., Williams, C.H., Stemple, D.L. & Armes, N.A. DNA detection using recombination proteins. Plos Biol. 4, 1115-1121 (2006). 28. Galiveti, C.R., Rozhdestvensky, T.S., Brosius, J., Lehrach, H. & Konthur, Z. Application of housekeeping npcRNAs for quantitative expression analysis of human transcriptome by real-time PCR. RNA 16, 450-461 (2010). 29. Velculescu, V.E. et al. Analysis of human transcriptomes. Nat. Genet. 23, 387-388 (1999). 30. Thompson, I.M. et al. Prevalence of prostate cancer among men with a prostate- specific antigen level <= 4.0 ng per milliliter. N. Eng. J. Med. 350, 2239-2246 (2004). 31. Leslie, D.C. et al. New detection modality for label-free quantification of DNA in biological samples via superparamagnetic bead aggregation. J. Am. Chem. Soc. 134, 5689-5696 (2012). 32. Li, J. et al. Label-free DNA quantification via a 'pipette, aggregate and blot' (PAB) approach with magnetic silica particles on filter paper. Lab Chip 13, 955-961 (2013).

104

Chapter 3.2

33. Lin, C., Zhang, Y., Zhou, X., Yao, B. & Fang, Q. Naked-eye detection of nucleic acids through rolling circle amplification and magnetic particle mediated aggregation. Biosens. Bioelectron. 47, 515-519 (2013). 34. Cai, C.M., Wang, H.Y., Xu, Y.Y., Chen, S.Y. & Balk, S.P. Reactivation of androgen receptor-regulated TMPRSS2:ERG gene expression in castration-resistant prostate cancer. Cancer Res. 69, 6027-6032 (2009).

SUPPORTING INFORMATION

Supplementary Figure 1. DNA sequencing of DuCap RNA amplicons after RT-RPA. RT- RPA amplicons of extracted DuCap RNA was purified using SPRI magnetic beads and sequenced to verify amplification of target TMPRSS2:ERG region. *The ERG sequence bases are underlined to indicate the TMPRSS2:ERG fusion junction.

Supplementary Figure 2. RT-PCR of extracted RNA from patient urine specimens. Extracted RNA from urine specimens of 10 metastatic castrate-resistant prostate cancer patients and 5 healthy patients were amplified using RT-PCR for TMPRSS2:ERG detection. The RT-PCR amplicons were visualized on agarose gel and used to validate the screening results of our assay on the same group of patients (Figure 2C).

105

Chapter 3.2

Supplementary Figure 3. Comparison of RNA stability in whole urine and urinary sediments. RT-RPA of patient whole urine and urinary sediment RNA extracted at 0, 2 and 12 hrs after specimen collection.

106

Chapter 3.2

Supplementary Figure 4. Full-length gel images of Figure 2 results. Dotted outlines represent cropping lines of images shown in Figure 2A, B, and C.

107

Chapter 3.3

Colorimetric TMPRSS2-ERG Gene Fusion Detection in Prostate Cancer Urinary Samples via Recombinase Polymerase Amplification

Kevin M. Koo, Eugene J. H. Wee, Matt Trau

ABSTRACT TMPRSS2 (Exon 1)-ERG (Exon 4) is the most frequent gene fusion event in prostate cancer (PC), and is highly PC-specific unlike the current serum prostate specific antigen (PSA) biomarker. However, TMPRSS2-ERG levels are currently measured with quantitative reverse-transcription PCR (RT-qPCR) which is time-consuming and requires costly equipment, thus limiting its use in clinical diagnostics. Herein, we report a novel rapid, cost- efficient and minimal-equipment assay named “FusBLU” for detecting TMPRSS2-ERG gene fusions from urine. TMPRSS2-ERG mRNA was amplified by isothermal reverse transcription-recombinase polymerase amplification (RT-RPA), magnetically-isolated, and detected through horseradish peroxidase (HRP)-catalyzed colorimetric reaction. FusBLU was specific for TMPRSS2-ERG mRNA with a low visual detection limit of 105 copies. We also demonstrated assay readout versatility on 3 potentially useful platforms. The colorimetric readout was detectable by naked eye for a quick yes/no evaluation of gene fusion presence. On the other hand, a more quantitative TMPRSS2-ERG detection was achievable by absorbance/electrochemical measurements. FusBLU was successfully applied to 12 urinary samples and results were validated by gold-standard RT-qPCR. We also showed that sediment RNA was likely the main source of TMPRSS2-ERG mRNA in urinary samples. We believe that our assay is a potential clinical screening tool for PC and could also have wide applications for other disease-related fusion genes.

INTRODUCTION Prostate cancer (PC) is the most common form of cancer and the second leading cause of cancer death in males of developed countries 190. The customary screening approach for prostate biopsies is mainly based on high serum prostate specific antigen (PSA) levels and/or abnormal digital rectal examination (DRE). PSA is the primary PC biomarker used by

108

Chapter 3.3 clinical practitioners and has led to a significant increase in PC diagnosis since its introduction 191. However, due to its low PC-specificity, the majority of PSA-diagnosed cases are actually benign and do not require medical intervention. This has in turn led to redundant medical expenses as well as negative effects on patients’ overall well-being 192. Thus, there is a pressing need for better PC diagnostic biomarkers, particularly one with high PC-specificity for discriminating benign cases from aggressive PC cases.

In 2005, Tomlins and co-workers identified recurrent PC fusion genes between the promoter of transmembrane protease, serine 2 (TMPRSS2) and the coding sequence of erythroblastosis virus E26 (ETS) family members 4. These fusions were a result of chromosomal rearrangements-induced androgen-dependent overexpression of oncogenic ETS transcription factors 193. The fusion of TMPRSS2 exon 1 to ERG exon 4 (TMPRSS2-ERG) is the most frequent subtype of these gene fusions, appearing in about 50% of PC patients and 90% of all PC gene fusions 194. More importantly, TMPRSS2-ERG is highly PC-specific and absent in non-PC samples. Furthermore, TMPRSS2-ERG may correlate with PC aggression and metastatic potential 168,176,195; potentially making them more attractive as diagnostic and prognostic biomarkers than serum PSA.

TMPRSS2-ERG mRNA is commonly detected from the cell sediment fraction of urine 9,175. Recently, Nilsson and co-workers had reported TMPRSS2-ERG detection in urinary exosomes 196. Additionally, it is also highly plausible that TMPRSS2-ERG mRNA is freely circulating within the urine sample. Given that TMPRSS2-ERG is detectable from multiple sources within urine 197, it could be worth investigating the primary source (circulating free mRNA in whole urine, cellular RNA or exosomal RNA) of the fusion mRNA. This could potentially lead to a more in-depth understanding of the cancer biology and aid in future assay development.

Currently, quantitative reverse transcription PCR (RT-qPCR) is mainly used to measure TMPRSS2-ERG mRNA levels in urinary samples 4,6,175. However, this approach is too time- consuming and laborious for use in a clinical setting. Therefore, the development of newer, faster and convenient assays for quantifying TMPRSS2-ERG mRNA in clinical samples may aid in improved PC screening and treatment monitoring.

109

Chapter 3.3

Colorimetric assays have been developed for a wide range of target analytes and offer benefits such as low cost, short assay time, visual readout, and quantitative detection via absorbance measurements 198. In particular, the reaction between tetramethyl benzidine

(TMB) and horseradish peroxidase (HRP) in presence of hydrogen peroxide (H2O2) to produce a blue-colored product 87 has been employed in many colorimetric assays, with ELISA as a well-known example. Additionally, it has been demonstrated by previous studies 87,140,199 that TMB could serve as an electrochemical substrate. Given the excellent simplicity, sensitivity and speed of electrochemical biosensors 200-202, it may be ideal to adapt TMB- based colorimetric assays for alternative electrochemical readouts to exploit such advantages. Therefore, we hypothesized that a TMB-based assay which displays a quantitative colorimetric/electrochemical signal change in TMPRSS2-ERG presence could be potentially useful for diagnostic purposes.

Herein, we describe the FusBLU assay, a novel approach to detect and quantify TMPRSS2- ERG fusion mRNA. By the innovative merger of isothermal reverse-transcription recombinase polymerase reaction (RT-RPA) and HRP-catalyzed colorimetric readout, TMPRSS2-ERG mRNA in urine could be specifically amplified and detected by naked-eye or by quantitative absorbance measurements. The versatility of the assay readout was further shown by conveniently performing FusBLU on a portable potentiostat for electrochemical detection. The practical application of our assay was demonstrated using PC cell lines and patient urinary samples. We also applied our assay to detect TMPRSS2-ERG mRNA in whole urine, urinary sediments and exosomes of PC urinary specimens in order to investigate the primary source of the mRNA.

MATERIALS AND METHODS RNA extraction from cell lines DuCap and LnCap cells were kindly donated by Matthias Nees (VTT, Finland); Gregor Tevz (APCRC, Australia); and Michelle Hill (UQDI, Australia). DuCap and LnCap cells were cultured in RPMI-1640 growth media (Life Technologies, Australia) supplemented with 10% fetal bovine serum (Life Technologies, Australia) in a humidified incubator (Sanyo MCO- 19AIC, Japan) containing 5% CO2 at 37 ºC. Total RNA was extracted by lysing cells with Trizol® reagent (Life Technologies, Australia) before adding chloroform to separate RNA into a clear upper aqueous layer. Then, RNA was precipitated from the aqueous layer and

110

Chapter 3.3 resuspended in 50 µL of RNase-free water. RNA purity and integrity were checked using a NanoDrop Spectrophotometer (Thermo Fisher Scientific, Australia)

RNA extraction from clinical urinary samples Ethics approval was obtained from The University of Queensland Institutional Human Research Ethics Committee (Approval No. 2004000047). Informed consent was obtained from all subjects prior to sample collection and methods pertaining to clinical samples were carried out in accordance with approved guidelines.

For whole urine RNA extraction, 1 mL of fresh urine sample was briefly mixed with 500 µL of lysis buffer 86 (150 mM Tris-HCI pH 8.0, 4.5 M guanidium-HCI, 3% v/v Triton-X and 1.5 mM EDTA). Total RNA was purified from 25 µL of cell lysate using the Agencourt AMPure XP kit (Beckman Coulter, Australia). Briefly, two volumes of Solid Phase Reversible Immobilization (SPRI) reagent was incubated with the cell lysate for 10 minutes. The RNA- bound magnetic beads were then separated from the lysate using a magnet and washed twice with 80% ethanol. Finally, total RNA was eluted in 25 µL of RNase-free water.

For urinary cell sediment RNA extraction, urinary samples (30–50 mL) were centrifuged (Beckman Coulter AllegraTM X-22R, Australia) at 700 g, 4ºC for 10 min and urinary sediments (supernatant was kept for exosomal RNA extraction) were washed with ice-cold 10 mM PBS buffer before being centrifuged again at the same conditions. Then, 25 µL of lysis buffer was added to a 50 µL fraction of PBS-suspended urinary sediments with vigorous mixing to release total RNA. Next, total RNA was purified from 25 µL of cell lysate using the procedure as described for whole urine RNA extraction.

For urinary exosomal RNA extraction, microvesicles in the supernatant (after urinary sediment centrifugation) was firstly pelleted by ultracentrifugation (Beckman Coulter OptimaTM XL-100 K, Australia) at 100 000 g, 4ºC for 90 min and resuspended in 200 µL of 10 mM PBS buffer. Then, the Exo-spinTM kit (Cell Guidance Systems, USA) was used accordingly to manufacturer’s instructions to isolate exosomes from the pellet. To ensure that extracted RNA was of exosomal origin, the isolated urinary exosomes were treated with RNase A (New England Biolabs, Australia) to degrade all non-exosomal RNA before RNA extraction. Exosomal total RNA was extracted by Trizol® reagent (Life Technologies, Australia) using the procedure as described for RNA extraction from cells.

111

Chapter 3.3

Colorimetric FusBLU assay Synthetic oligonucleotide and primer sequences used in our experiments were obtained commercially (IDT, Singapore). RT-RPA primer sequences were designed based on the fusion junction of the most common TMPRSS2 (Exon 1):ERG (Exon 4) mRNA isoform reported in literature 4. The RT-qPCR primers were also designed to amplify the same target TMPRSS2:ERG isoform.

For cell line experiments, the TwistAmp Basic RPA kit (Twist-DX, UK) was used with slight modifications to manufacturer’s instructions. Briefly, 1 µL of extracted RNA, 50 units of MMuLV reverse transcriptase (New England Biolabs, Australia), 375 nM of each primer (Table 1) and 20 nM of biotinylated dUTPs (Thermo Fisher Scientific, Australia) were added to make a 12.5 µL reaction volume prior to incubation at 43°C for 20 min. For all other experiments, the TwistAmp Basic RT-RPA kit (Twist-DX, UK) with pre-included reverse transcriptase was used with the same modifications to manufacturer’s instructions. The 43ºC amplification temperature, which is above the manufacturer’s recommended optimum temperature, was required for higher amplification specificity.

Table 1. Oligonucleotide sequences used in experiments. Oligos 5'-Sequence-3'

RT-RPA TMPRSS2- ERG Fwd Primer CCTGGAGCGCGGCAGGAAGCCTTATCAGTTG Sequence

RT-RPA TMPRSS2-

ERG Rev Primer GCTAGGGTTACATTCCATTTTGATGGTGAC Sequence

RT-RPA Housekeeping Fwd GCTATGCCGATCGGGTGTCCGCACTAAGTT Primer Sequence

RT-RPA GACGGGGTCTCGCTATGTTGCCCAGGCTGG Housekeeping Rev

112

Chapter 3.3

Primer Sequence

RT-qPCR Fwd Primer CCTGGAGCGCGGCAGGAAGCCTTATCAGTTG Sequence RT-qPCR Rev Primer Sequence TCCTGCTGAGGGACGCGTGGGCTCATCTTG

For colorimetric detection of RT-RPA products, 2 µL of RT-RPA products was incubated with 1 µL of 1:1000 diluted streptavidin (SA)-HRP (BD Biosciences, Australia), 1 µL of SA- magnetic beads (New England Biolabs, Australia) and 10 µL of wash buffer (10 mM PBS buffer, 0.5% triton-X) for 5 min. Then, a magnet was used to separate the beads and beads were washed three times with wash buffer (10 mM PBS buffer, 0.1% Tween20). Lastly, 25 µL of 1-Step™ TMB substrate solution (Thermo Fisher Scientific, Australia) was added to the beads and the mixture was observed for color change after 5 min. For each sample, two separate RT-RPA assays were performed for independent detection of TMPRSS2-ERG and housekeeping (RN7SL1) RNA. For quantitative measurements of color change, absorbance readings were recorded at 650 nm with a NanoDrop Spectrophotometer (Thermo Fisher Scientific, Australia). For each sample, relative TMPRSS2-ERG level was calculated by normalizing the absorbance value of TMPRSS2-ERG to the respective housekeeping RN7SL1 as follows:

Relative TMPRSS2-ERG Level = XTmprss2-ERG / XRN7SL1 (1)

where XTmprss2:ERG and XRN7SL1 are the average absorbance readings at 650 nm for TMPRSS2-ERG and housekeeping RN7SL1 RNA respectively.

Alternative electrochemical readout For electrochemical detection of RT-RPA products, total RNA was isolated from patient urinary sediments and after which, the same steps as described previously for colorimetric

FusBLU assay were performed. At 5 min after TMB-H2O2 addition, 500 mM H2SO4 was added to stop the reaction and activate TMB for electrochemical detection. The electrodes used in our experiments were screen-printed electrodes (DRP110, Dropsens, Spain) with carbon as working and counter electrodes, and silver as reference electrode. Briefly, 45 µL of

the resulting mixture after H2SO4 addition was pipetted onto the electrode surface, and

113

Chapter 3.3 ensured that the solution was in contact with all three electrode types (i.e. working, counter, and reference). Amperometry measurements were carried out using a portable µSTAT 400 bipotentiostat/galvanostat (Dropsens, Spain) at 150 mV, 30 s. Each sample’s electrochemical measurement was scored for relative TMPRSS2-ERG level in the same manner as described previously for absorbance readings. All measurements were performed at room temperature.

RT-qPCR The KAPA SYBR® FAST One-Step qRT-PCR kit (KAPA Biosystems, Australia) was used to set up a single reaction volume of 10 µl for each sample. Each reaction volume consist of 1X KAPA SYBR® FAST qPCR Master Mix, 200 nM of each forward and reverse primer (Table 1), 1X KAPA RT Mix, 50 nM ROX dye and 30 ng of cell line total RNA template. RT-qPCR was performed using the Applied Biosystems® 7500 Real-Time PCR System (Thermo Fisher Scientific, Australia). The cycling protocol was: 42°C for 10 min to synthesize cDNA, followed by 95°C for 5 min to deactivate RT before cycling 35 times (95°C for 30 s, 50°C for 30 s and 72°C for 1 min) and finished with 72°C for 10 min.

RESULTS FusBLU assay for TMPRSS2-ERG detection As illustrated in Fig. 1, our colorimetric assay utilized a combination of isothermal RT-RPA and TMB-based colorimetric readout for detecting TMPRSS2-ERG mRNA. Total RNA was firstly isolated from various sources used in our study and specific primers were used to amplify the TMPRSS2-ERG mRNA region at a constant temperature of 43ºC. During the isothermal RT-RPA, biotinylated dUTPs were randomly incorporated into the newly- synthesized strands. Then, SA-magnetic beads and SA-HRP were added to select for and label biotinylated RT-RPA products through biotin-streptavidin interactions. Lastly, TMB-

H2O2 was added for color change reaction to test for the presence of HRP which in turn signifies the presence of RT-RPA products (i.e. TMPRSS2-ERG positive).

114

Chapter 3.3

Fig. 1. FusBLU assay for rapid TMPRSS2-ERG detection. Total RNA is isolated from urine and TMPRSS2-ERG mRNA is specifically amplified by RT-RPA isothermally. During strand polymerization, biotinylated dUTP bases are randomly incorporated and subsequently, SA- magnetic beads and SA-HRP are added to RT-RPA products for coupling through biotin-SA interactions. After magnetic isolation of RT-RPA products, TMB is added and a blue color change indicates the presence of TMPRSS2-ERG.

Assay specificity To demonstrate the specificity of FusBLU in detecting TMPRSS2-ERG, we applied our assay to total RNA from two different PC cell lines. DuCap and LnCap are well-studied cell lines with the presence and absence of TMPRSS2-ERG respectively 4. As shown in Fig. 2A, our assay was able to specifically amplify TMPRSS2-ERG mRNA from DuCap RNA and subsequently generated a blue color change. On the other hand, when the assay was performed on LnCap RNA, RT-RPA products were not generated at a detectable level colorimetrically (Fig. 2A). The specific colorimetric detection of TMPRSS2-ERG could be followed by naked-eye observation of color change or more quantitatively by absorbance measurements (Fig. 2B). Furthermore, gel electrophoresis was also used to verify that the RT-RPA primers used in our experiments were specific and generated 216 bp products only in TMPRSS2-ERG positive DuCap samples. RN7SL1 housekeeping RNA was used as a loading control to validate successful RNA isolation from the cell lines (Fig. 2A). In addition, control experiments without the reverse transcriptase did not generate any RPA products,

115

Chapter 3.3 thus demonstrating that amplification was dependent on TMPRSS2-ERG mRNA, not DNA targets (Fig. 2).

Fig. 2. Specificity of FusBLU assay. (A) HRP-catalyzed colorimetric results for DuCap (fusion positive) and LnCap (fusion negative) cell lines with corresponding gel electrophoresis images after RT-RPA. Top panel: TMPRSS2-ERG. Bottom panel: RN7SL1. No reverse transcriptase (No RT) and no-template control (NTC) were included. (B) TMPRSS2-ERG levels normalized to RN7SL1. Error bars represent standard deviation of three independent experiments.

Assay sensitivity To evaluate assay sensitivity, we prepared a titration of synthetic TMPRSS2-ERG mRNA (102-108 copies) into a background of LnCap total RNA to test the limit of detection (LOD). As expected, the color change intensity increased with increasing amounts of input targets (Fig. 3A). By visual detection, the assay LOD was 105 copies of TMPRSS2-ERG mRNA at which a blue color change in solution was still observable. The determination of assay sensitivity could also be followed quantitatively by absorbance measurements and assay LOD

116

Chapter 3.3 was further improved by 10-fold to 104 TMPRSS2-ERG mRNA copies with a linear range of 104-108 copies (Fig. 3B). However, it is worthy to note that the LOD of 105 copies by visual detection is approximately equivalent to single cell level of detection 184 and thus may potentially be adequate for TMPRSS2-ERG screening purpose in a clinical setting. FusBLU also showed good assay reproducibility with intra- and inter-assay variability of 12.5% and 10.5% respectively (n = 3).

Fig. 3. Sensitivity of FusBLU assay. (A) HRP-catalyzed colorimetric change generated from 102 - 108 copies of synthetic TMPRSS2-ERG mRNA, and corresponding gel electrophoresis images after RT-RPA. (B) Calibration plot of the average absorbance measurements at different initial RT-RPA input amounts. Inset shows the analogous linear calibration plot. (C)

117

Chapter 3.3

HRP-catalyzed current change generated from 103 - 107 copies of synthetic TMPRSS2-ERG mRNA. Inset shows the analogous linear calibration plot. Error bars represent standard deviation of three independent experiments.

Colorimetric assay performance on clinical specimens After establishing assay specificity and sensitivity, we challenged FusBLU with urine specimens to demonstrate potential clinical utility. To this end, we isolated sediment RNA from the urinary samples of 12 men and applied FusBLU for TMPRSS2-ERG detection. The magnetic SPRI method provided a minimal yield of 15 ng total RNA; an amount which was sufficient for our analysis using FusBLU. Of the 12 clinical urinary samples used in our experiments: 5 (P1-P5) were metastatic hormone-refractory PC patients undergoing treatment; 5 (P6-P10) were diagnosed with PC after biopsies; and the remaining 2 (H1 and H2) were healthy controls. As shown on Fig. 4A, we could visually detect TMPRSS2-ERG in 8 of the patient samples (P1, P3, P4, P5, P7, P8, P9, and P10), while the remaining samples (P2, P6, H1, and H2) showed no observable color change as similar to the no-target control (NTC). For quantitative detection, the relative absorbance measurements (Eqn. 1) provided additional information of TMPRSS2-ERG levels in different patients (Fig. 4B). In addition, we also performed RT-qPCR, the current gold standard gene fusion detection approach on the same 12 urinary samples as a benchmark comparison. The RT-qPCR threshold cycle (Ct) values (Fig. 4D) showed that higher Ct values (i.e. lower amount of targets in samples) were generated for samples with low FusBLU-detected TMPRSS2-ERG levels. We found that our FusBLU results were in excellent agreement with the RT-qPCR results, thus validating the TMPRSS2-ERG status of the patient samples.

118

Chapter 3.3

Fig. 4. FusBlu assay on clinical urinary samples. (A) HRP-catalyzed colorimetric results of 12 urinary samples over two different assay runs (1st run (Left): P1-P5; 2nd run (Right): P6- H2). Top panel: TMPRSS2-ERG. Bottom panel: RN7SL1. (B) TMPRSS2-ERG levels normalized to RN7SL1. (C) TMB-derived currents of 12 urinary samples compared to no template-control (NTC). (D) RT-qPCR threshold cycle (Ct) values for each sample as validation for FusBLU results. Error bars represent standard deviation of three independent experiments.

Alternative electrochemical readout Since TMB was electrochemically active and considering the multiple advantages of an electrochemical biosensor 201,203,204, we integrated the use of a portable potentiostat as an alternative FusBLU readout strategy to further demonstrate the potential for clinical screening applications. We firstly established the LOD of the electrochemical readout by a

119

Chapter 3.3 titration assay, and observed that 103 TMPRSS2-ERG copies were detectable with a linear range of 103-107 copies (Fig. 3C).

We next used the same collection of 12 urinary samples to demonstrate the electrochemical readout. Fig. 4C showed the TMPRSS2-ERG levels detected electrochemically were similar to that of spectrophotometry (Fig. 4B) and RT-qPCR (Fig. 4D). Taken together, these results indicated that FusBLU was a robust method for detecting TMPRSS2-ERG levels in urine and could be readily adapted for naked-eye, spectrophotometrical or electrochemical readout platforms.

Investigating TMPRSS2-ERG source in urine After demonstrating FusBLU’s potential for clinical applications, we used our assay to investigate the primary source of TMPRSS2-ERG mRNA in urine. The potential sources included cell-free circulating RNA, cell sediment RNA and exosomal RNA. To this end, 3 TMPRSS2-ERG-positive PC patients (P1, P3, and P4) were first selected (Fig. 4) and total RNA was then isolated from the three candidate sources and assayed with FusBLU. As shown on Fig. 5A, the sediment fraction gave the strongest color intensity change (i.e. highest amount of TMPRSS2-ERG mRNA), followed by circulating free fraction, while the exosomal fraction contain the least amount of TMPRSS2-ERG mRNA. In addition, quantitative absorbance measurements showed that TMPRSS2-ERG mRNA in the sediment fraction was 2-fold and 12-fold higher than in the circulating free and exosomal fractions respectively (Fig. 5B). Absorbance results were also validated by gel electrophoresis following RT-RPA of the isolated total RNA from the various candidate sources (Fig. 5A). Our results therefore suggested that the sediment fraction is likely the main source of TMPRSS2-ERG mRNA in urine.

120

Chapter 3.3

Fig. 5. Source of TMPRSS2-ERG mRNA in urine. Total RNA isolated from circulating free, sediment, and exosomal fractions of 3 patient urine samples were studied (A) HRP-catalyzed colorimetric changes and corresponding gel electrophoresis images after RT-RPA. Top panel: for TMPRSS2-ERG. Bottom panel: RN7SL1. (B) TMPRSS2-ERG levels normalized to RN7SL1. Error bars represent standard deviation of three independent experiments.

DISCUSSION Recurrent gene fusions between TMPRSS2 exon 1 and ERG exon 4 is one of the emerging PC biomarker alternative to serum PSA. TMPRSS2-ERG offers higher PC specificity, potential prognostic value and is detectable in urine as a non-invasive biomarker 9,175. However, the conventional RT-qPCR approach for measuring TMPRSS2-ERG levels is not ideal for routine

121

Chapter 3.3 clinical screening due to long assay time and the need for specialized instrument 4,6. Therefore, we have developed FusBLU, an assay (Fig. 1) which combined isothermal RT- RPA and HRP-based colorimetric readout for quick and quantitative TMPRSS2-ERG detection in 90 min without requiring expensive readout instruments. With FusBLU, we demonstrated (i) potential TMPRSS2-ERG detection in clinical urinary samples with a quick and convenient visual readout; (ii) quantitative TMPRSS2-ERG detection by alternative absorbance/electrochemical readout platforms; (iii) cell sediment RNA as potentially the main source of TMPRSS2-ERG mRNA in patient urinary samples.

In order for the assay to be used for TMPRSS2-ERG detection in real biological samples such as urine specimens, assay specificity and sensitivity are of paramount importance. High assay specificity is required to avoid false-positive/negative results due to detection of non-target mRNA and high assay sensitivity is required to detect trace amounts of TMPRSS2-ERG mRNA from a complex background of other biological molecules. FusBLU was highly specific for TMPRSS2-ERG mRNA in fusion-positive DuCap cells but not fusion-negative LnCap cells (Fig. 2). This high assay specificity was largely enabled by designing RT-RPA forward primer across the fusion junction to selectively amplify only gene fusion events. In future, this flexibility of primer design may also allow specific detection of other TMPRSS2- ERG isoforms for multiplexed FusBLU gene fusion detection. Our assay visual detection limit of 105 copies of TMPRSS2-ERG mRNA (Fig. 3A); which was approximately single-cell level 205, may be especially useful for non-invasively detecting the low levels of TMPRSS2- ERG in urine. Furthermore, this visual LOD could be further improved up to 100-fold with easily-adaptable absorbance/electrochemical readouts. In comparison to other standard urinary TMPRSS2-ERG detection methodologies such as RT-qPCR, FusBLU required lower sample input (10 ng total RNA vs. 50 ng total RNA for RT-qPCR in our experiments) and was able to produce results within a shorter timeframe (75 min vs. 150 min for RT-qPCR in our experiments).

To demonstrate the potential clinical utility of FusBLU, we applied our assay to 12 urinary samples (Fig. 4) and found excellent agreement between our assay results and current gold- standard RT-qPCR results (Fig. 4D). Importantly, the healthy and NTC samples generated negligible background signals, and varying TMPRSS2-ERG levels were detected for the other samples with good reproducibility over independent runs. These data therefore suggested the reliability and functional applicability of our assay for routine diagnostics. Furthermore, the

122

Chapter 3.3 presence of TMPRSS2-ERG in the urine samples of the late-stage metastatic, hormone refractory samples was consistent with previous reports in literature 206,207. Hence, our results suggested that FusBLU may be a useful screening tool for the TMPRSS2-ERG positive metastatic subtype of aggressive PC. Considering the good assay sensitivity as well as speed and simple naked-eye evaluation of color change detection; FusBLU may also be a rapid and convenient preliminary PC screening tool for guiding clinical decisions. In addition, quantitative absorbance measurements of the color intensity can also provide additional information that could be beneficial for tracking patient response to treatment or for relapse monitoring. Lastly, electrochemical approaches have high potential for rapid, sensitive, inexpensive and portable biosensing 130,208,209 and thus have attracted great interest for clinical diagnostic developments. To this end, we exploited TMB’s compatibility as an electrochemical substrate and adapted FusBLU into an electrochemical assay with a portable potentiostat to highlight the versatility of the method (Fig. 4C). Despite the use of a portable potentiostat, the FusBLU electrochemical readout provided the highest detection sensitivity (103 copies) out of the three different readout platforms. In short, FusBLU is potentially a flexible methodology that could be tailored to specific diagnostic needs.

While it is known that TMPRSS2-ERG mRNA is present in urine, it is not previously clear where the fusion mRNA originated from. Hence, identifying the primary source of TMPRSS2-ERG mRNA in urine might be useful for future assay development and may help better understand the cancer biology of TMPRSS2-ERG subtypes. Using FusBLU, we found that TMPRSS2-ERG mRNA was most abundant in the sediment fraction of urine (Fig. 5). TMPRSS2-ERG mRNA in circulating free fraction was the second most abundant source while minimal fusion mRNA was detected in the exosomal fraction. A possible explanation for our observation could be the shedding of tumor cells from the primary (and metastastic) tumors in the prostate into the urine through the prostatic urethra 210,211. It could also be from ingested tumor cells that are present in urine within white blood cells of the immune system 212. The presence of the TMPRSS2-ERG mRNA in circulating free fraction was likely from the spontaneous lysis of cells since minimal RNA was detected in the exosomal fraction. Nevertheless, the data is still preliminary and warrants further investigation. Notwithstanding, the data did suggested that although TMPRSS2-ERG was detectable in whole urine, the sediment fraction is most likely the best choice for sampling TMPRSS2-ERG mRNA due to ease of preparation and high abundance. Additionally, the use of our assay as a rapid and

123

Chapter 3.3 convenient research tool for biological studies was also demonstrated through this set of successful experiments to investigate the source of TMPRSS2-ERG mRNA in urine.

FusBLU shows promise as a faster, simpler, and more cost-effective solution for gene fusion biomarkers detection as compared to current techniques. Nevertheless, there remain some aspects of the assay which could be further investigated in future studies. Firstly, our main aim in this report is to demonstrate FusBLU as a successful proof-of-concept technology which could be applied to real patient samples. While FusBLU detected TMPRSS2-ERG levels in 12 clinical urinary samples as a demonstration, it would be ideal to apply the method to a larger pool of patients and also of samples at various stages of PC. This would allow a better evaluation of the clinical utility of the assay for PC diagnosis. Secondly, we envisaged an improved version of FusBLU assay with multiplexing capabilities to detection multiple biomarkers. It has been suggested that detecting a combination of gene fusion isoforms or other PC biomarkers could improve PC diagnostic and predictive accuracy 6. In this regard, we feel that an electrochemical readout would be better suited for this purpose as performing the assay on an array of miniaturized electrodes could greatly increase throughput.

In summary, we have developed FusBLU; a quantitative TMPRSS2-ERG assay with single- cell level detection sensitivity and highly specific for the most common gene fusion isoform in PC. This assay was achieved by the novel combination of isothermal RT-RPA and HRP- catalyzed colorimetric readout of RT-RPA products. Our assay was successfully applied to cell lines and patient urine samples with minimal equipment and at a low cost in approximately 75 minutes. Through the use of this assay, we also determined that total RNA isolated from urinary sediments as the main source for TMPRSS2-ERG mRNA detection from urinary samples. We believe that our colorimetric assay is potentially useful for POC detection of TMPRSS2-ERG, or other oncogenic fusion genes.

REFERENCES 1. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA A Cancer J Clinicians. 2012; 62: 10-29. 2. Welch HG, Albertsen PC. Prostate cancer diagnosis and treatment after the introduction of prostate-specific antigen screening: 1986-2005. J Natl Cancer Inst. 2009; 101: 1325-9. 3. Loeb S, Catalona WJ. Prostate-specific antigen in clinical practice. Cancer Lett. 2007; 249: 30-9.

124

Chapter 3.3

4. Tomlins SA, Rhodes DR, Perner S, Dhanasekaran SM, Mehra R, Sun XW, et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644-8. 5. White NM, Feng FY, Maher CA. Recurrent rearrangements in prostate cancer: Causes and therapeutic potential. Curr Drug Targets. 2013; 14: 450-9. 6. Prensner JR, Chinnaiyan AM. Oncogenic gene fusions in epithelial carcinomas. Curr Opin Genet Dev. 2009; 19: 82-91. 7. Tomlins SA, Aubin SMJ, Siddiqui J, Lonigro RJ, Sefton-Miller L, Miick S, et al. Urine TMPRSS2:ERG fusion transcript stratifies prostate cancer risk in men with elevated serum PSA. Science Transl Med. 2011; 3: 94ra72. 8. Tomlins SA, Bjartell A, Chinnaiyan AM, Jenster G, Nam RK, Rubin MA, et al. ETS gene fusions in prostate cancer: From discovery to daily clinical practice. Eur Urol. 2009; 56: 275-86. 9. Yoshimoto M, Joshua AM, Cunha IW, Coudry RA, Fonseca FP, Ludkovski O, et al. Absence of TMPRSS2:ERG fusions and PTEN losses in prostate cancer is associated with a favorable outcome. Mod Pathol. 2008; 21: 1451-60. 10. Hessels D, Smit FP, Verhaegh GW, Witjes JA, Cornel EB, Schalken JA. Detection of TMPRSS2-ERG fusion transcripts and prostate cancer antigen 3 in urinary sediments may improve diagnosis of prostate cancer. Clin Cancer Res. 2007; 13: 5103-8. 11. Laxman B, Tomlins SA, Mehra R, Morris DS, Wang L, Helgeson BE, et al. Noninvasive detection of TMPRSS2 : ERG fusion transcripts in the urine of men with prostate cancer. Neoplasia. 2006; 8: 885-8. 12. Nilsson J, Skog J, Nordstrand A, Baranov V, Mincheva-Nilsson L, Breakefield XO, et al. Prostate cancer-derived urine exosomes: a novel approach to biomarkers for prostate cancer. Br J Cancer. 2009; 100: 1603-7. 13. Dijkstra S, Mulders PFA, Schalken JA. Clinical use of novel urine and blood based prostate cancer biomarkers: A review. Clin Biochem. 2014; 47: 889-96. 14. Wang JH, Cai Y, Ren CX, Ittmann M. Expression of variant TMPRSS2/ERG fusion messenger RNAs is associated with aggressive prostate cancer. Cancer Res. 2006; 66: 8347-51. 15. Li HX, Rothberg L. Colorimetric detection of DNA sequences based on electrostatic interactions with unmodified gold nanoparticles. Proc Natl Acad Sci U S A. 2004; 101: 14036-9. 16. Wee EJH, Ngo TH, Trau M. Colorimetric detection of both total genomic and loci-specific DNA methylation from limited DNA inputs. Clin Epigenet. 2015; 7: 65. 17. Volpe G, Compagnone D, Draisci R, Palleschi G. 3,3 ',5,5 '- tetramethylbenzidine as electrochemical substrate for horseradish peroxidase based enzyme immunoassays. A comparative study. Analyst. 1998; 123: 1303-7. 18. Ng BYC, Wee EJH, West NP, Trau M. Naked-eye colorimetric and electrochemical detection of Mycobacterium tuberculosis–towards rapid screening for active case finding. ACS Sens. 2015; 1: 173-8. 19. Gasparac R, Taft BJ, Lapierre-Devlin MA, Lazareck AD, Xu JM, Kelley SO. Ultrasensitive electrocatalytic DNA detection at two- and three-dimensional nanoelectrodes. J Am Chem Soc. 2004; 126: 12270-1. 20. Ronkainen NJ, Halsall HB, Heineman WR. Electrochemical biosensors. Chem Soc Rev. 2010; 39: 1747-63. 21. Koo KM, Sina AAI, Carrascosa LG, Shiddiky MJA, Trau M. eMethylsorb: rapid quantification of DNA methylation in cancer cells on screen-printed gold electrodes. Analyst. 2014; 139: 6178-84.

125

Chapter 3.3

22. Wee EJH, Lau HY, Botella JR, Trau M. Re-purposing bridging flocculation for on-site, rapid, qualitative DNA detection in resource-poor settings. Chem Commun. 2015; 51: 5828-31. 23. Velculescu VE, Madden SL, Zhang L, Lash AE, Yu J, Rago C, et al. Analysis of human transcriptomes. Nat Genet. 1999; 23: 387-8. 24. Das J, Ivanov I, Montermini L, Rak J, Sargent EH, Kelley SO. An electrochemical clamp assay for direct, rapid analysis of circulating nucleic acids in serum. Nat Chem. 2015; 7: 569-75. 25. Sadik OA, Aluoch AO, Zhou A. Status of biomolecular recognition using electrochemical techniques. Biosens Bioelectron. 2009; 24: 2749-65. 26. Matthew T, Yang B, Jarrard DF. Toward the detection of prostate cancer in urine: A critical analysis. J Urol. 2013; 189: 422-9. 27. Mehra R, Tomlins SA, Yu J, Cao X, Wang L, Menon A, et al. Characterization of TMPRSS2-ETS gene aberrations in androgen-independent metastatic prostate cancer. Cancer Res. 2008; 68: 3584-90. 28. Attard G, Swermenhuis JF, Olmos D, Reid AHM, Vickers E, A'Hern R, et al. Characterization of ERG, AR and PTEN gene status in circulating tumor cells from patients with castration-resistant prostate cancer. Cancer Res. 2009; 69: 2912-8. 29. Nie Z, Nijhuis CA, Gong J, Chen X, Kumachev A, Martinez AW, et al. Electrochemical sensing in paper-based microfluidic devices. Lab Chip. 2010; 10: 477-83. 30. Hsieh K, Patterson AS, Ferguson BS, Plaxco KW, Soh HT. Rapid, sensitive, and quantitative detection of pathogenic DNA at the point of care through microfluidic electrochemical quantitative loop-mediated isothermal amplification. Angew Chem, Int Ed. 2012; 51: 4896-900. 31. Ferguson BS, Buchsbaum SF, Wu T-T, Hsieh K, Xiao Y, Sun R, et al. Genetic analysis of H1N1 influenza virus from throat swab samples in a microfluidic system for point-of-care diagnostics. J Am Chem Soc. 2011; 133: 9129-35. 32. Fujita K, Pavlovich CP, Netto GJ, Konishi Y, Isaacs WB, Ali S, et al. Specific detection of prostate cancer cells in urine by multiplex immunofluorescence cytology. Hum Pathol. 2009; 40: 924-33. 33. Hessels D, Gunnewiek J, van Oort I, Karthaus HFM, van Leenders GJL, van Balken B, et al. DD3(PCA3)-based molecular urine analysis for the diagnosis of prostate cancer. Eur Urol. 2003; 44: 8-15. 34. Kato T, Suzuki H, Komiya A, Imamoto T, Naya Y, Tobe T, et al. Clinical significance of urinary white blood cell count and serum C-reactive protein level for detection of non-palpable prostate cancer. Int J Urol. 2006; 13: 915-9.

126

Chapter 4

Label-Free Surface-Enhanced Raman Scattering Detection Systems for Clinical Biomarker Targets

Introduction Label-free surface-enhanced Raman Scattering (SERS) is a surface spectroscopic technique which directly detects adsorbed molecules on nanostructured metallic surfaces. Label-free SERS is an extremely attractive technique for simple and rapid detection of nucleic acid targets due to 1) direct target analysis without the need for detection readout labelling; 2) sensitive and rapid target detection due to SERS effect; 3) ability to provide chemical component and structure information of nucleic acid targets. Despite these benefits, label-free SERS has rarely, if any, been described for biomarker detection in real patient samples due to several severe limiting factors. In this chapter, prostate cancer RNA biomarker detection in patient urine samples was used as a model to showcase efforts in using isothermal reverse- transcription recombinase polymerase amplification (RT-RPA) to enable label-free SERS for clinical applications. Chapter 4.1 describes a new label-free SERS strategy for detecting gene fusion RNA in clinical samples via double-stranded-DNA/gold nanoparticles aggregation mechanism. Chapter 4.2 showcases a non-invasive prostate cancer subtyping technique based on multiplexed RT-RPA, label-free SERS via self-synthesized positively charged silver nanoparticles, and multivariate analysis to detect long nucleic acid biomarker amplicons.

Chapter 4 is based on two published articles in the Journal of Biomedical Nanotechnology (2016) [Chapter 4.1], and Nanoscale (2017) [Chapter 4.2].

127

Chapter 4.1

Rapid and Sensitive Fusion Gene Detection in Prostate Cancer Urinary Specimens by Label-free SERS

Kevin M. Koo, Benjamin McNamara, Eugene J. H. Wee, Yuling Wang, Matt Trau

ABSTRACT Recurrent chromosomal rearrangements such as fusion genes are associated with cancer initiation and progression. Prostate cancer (PCa) is a leading cause of cancer-related deaths in men and the TMPRSS2-ERG gene fusion is a recurrent biomarker in about 50% of all prostate cancers. However, current screening tools for TMPRSS2-ERG are generally confined to research settings and hence, the development of a rapid, sensitive and accurate assay for TMPRSS2-ERG detection may aid in clinical PCa diagnosis and treatment. Herein, we described a new strategy for non-invasive TMPRSS2-ERG detection in patient urinary samples by coupling the isothermal reverse transcription-recombinase polymerization amplification (RT-RPA) to amplify TMPRSS2-ERG transcripts and directly detect the amplicons using surface-enhanced Raman scattering (SERS). This novel coupling of both techniques allows rapid and quantitative TMPRSS2-ERG detection. Our assay can specifically detect as low as 103 copies input of TMPRSS2-ERG transcripts and was successfully applied to clinical PCa urinary samples. Hence, we believe our assay is a potential clinical screening tool for TMPRSS2-ERG in PCa and may have broad applications in detecting other gene fusion transcripts in other diseases.

INTRODUCTION In recent years, oncogenic gene fusion events have emerged as a promising new source of cancer biomarkers.1-3 Fusion genes have key roles in cancer progression and can serve as promising drug targets for cancer treatment. Of particular interest, the TMPRSS2-ERG gene fusion is the most common gene fusion occurrence amongst all solid tumor cancers and is found in approximately half of all prostate cancer (PCa) cases.4-6 In contrast to the traditional prostate-specific antigen (PSA), TMPRSS2-ERG is not only more specific for PCa but also associated with aggressive PCa and thus may be more informative as a biomarker.7, 8

128

Chapter 4.1

Furthermore, TMPRSS2-ERG transcripts are also present in patient urine samples,9, 10 thus offering its potential as an attractive non-invasive PCa biomarker. Classical methodologies, such as reverse transcription (RT)-PCR and fluorescence in situ hybridization (FISH), are commonly used to detect gene fusions including TMPRSS2-ERG.9, 11 Whilst these methods offer good detection sensitivity, they are laborious, time-consuming and prone to artifacts. Hence, the development of a simpler, rapid and accurate assay to detect the low TMPRSS2-ERG transcript copies in urine is very much sought after. An advancement in such a detection methodology could improve the clinical utility of the TMPRSS2-ERG biomarker for point-of-care (POC) applications.

Due to the relatively low TMPRSS2-ERG target copies in limited clinical biopsies, nucleic acid amplification methods are usually required to increase copy numbers to a sufficient level for detection. The recombinase polymerase amplification (RPA) is an isothermal amplification technique which offers advantages over conventional amplification methods such as PCR.12 These benefits include low constant operational temperature; therefore avoiding the need for expensive themocyclers and short amplification times (10-15 min) while achieving equivalent sensitivities to PCR. The applications of RPA has been demonstrated in the detection of a wide range of pathogens13, 14 and we hypothesized that RPA is an attractive option for POC gene fusion detection if readily coupled to a similarly convenient and sensitive readout platform.

Surface-enhanced Raman scattering (SERS) is an emerging analytical technique for highly sensitive (down to single molecule)15, 16 and specific detection of nucleic acids17, 18 based on electromagnetic field enhancement effect of molecules, e.g. DNA, near noble metal surfaces. With recent advancements in handheld portable SERS systems19, SERS-based assay may potentially be adopted for rapid, sensitive and portable POC applications. SERS-based assays typically use labels such as Raman reporter molecules-coated gold nanoparticles (AuNPs), to generate a distinctive Raman signal for detection, which requires additional modifications/hybridization on AuNPs. Nonetheless, label-free and direct SERS platform offers great opportunity for simple and rapid detection of nucleic acids by using AuNPs or silver nanoparticles (AgNPs) aggregates as an enhancement substrate.20, 21 However, since the duplex structure of double-stranded DNA (dsDNA), in conjunction with the strong electrostatic repulsion between negatively-charged dsDNA and AuNPs/AgNPs; prevent the bases from being exposed to the metal surface, direct SERS detection of dsDNA bases

129

Chapter 4.1 remains challenging.20 As such, most of these approaches typically require single-stranded DNA (ssDNA) detection by exploiting the strong affinity between ssDNA and AuNPs. While useful, ssDNA are inherently susceptible to rapid degradation and thus may limit its applications. To address this, Graham and co-workers recently reported the use of spermine- modified AgNPs to create stable dsDNA-AgNPs aggregates, thus allowing direct SERS detection of dsDNA bases.22 Nonetheless, we propose an alternate strategy whereby streptavidin-modified AuNPs (SA-AuNPs) may be used to bind biotinylated dsDNA analytes instead. This, therefore allows for better control over the dsDNA/AuNPs aggregation and in turn, enables a label-free and quantitative SERS assay for dsDNA detection.

Herein, we describe, to the best of our knowledge, the first assay to merge isothermal RT- RPA with dsDNA/AuNPs-mediated SERS to detect TMPRSS2-ERG transcripts in a fast, sensitive and specific fashion. Using this strategy, we were able to detect as little as 103 copies input of synthetic TMPRSS2-ERG transcripts and show specific TMPRSS2-ERG detection in different PCa cell lines. To demonstrate clinical feasibility, the method was also successfully extended to PCa urine samples where fusion gene status was determined accurately in 75 mins.

EXPERIMENTAL Reagents All reagents were purchased from Sigma Aldrich and used as is unless otherwise stated. DNA oligonucleotides (Integrated DNA Technologies, USA) were purchased and sequences are shown in Table 1.

Table 1. Oligonucleotide sequences used in experiments. Oligos 5'-Sequence-3'

GAGTAGGCGCGAGCTAAGCAGGAGGCGGAGGCGGAG GCGGAGGGCGAGGGGCGGGGAGCGCCGCCTGGAGCG CGGCAGGAAGCCTTATCAGTTGTGAGTGAGGACCAGT CGTTGTTTGAGTGTGCCTACGGAACGCCACACCTGGCT Synthetic TMPRSS2-ERG AAGACAGAGATGACCGCGTCCTCCTCCAGCGACTATG Sequence GACAGACTTCCAAGATGAGCCCACGCGTCCCTCAGCA GGATTGGCTGTCTCAACCCCCAGCCAGGGTCACCATC AAAATGGAATGTAACCCTAGCCAGGTGAATGGCTCAA G

RT-RPA TMPRSS2-ERG CCTGGAGCGCGGCAGGAAGCCTTATCAGTTG

130

Chapter 4.1

Fwd Primer Sequence

RT-RPA TMPRSS2-ERG Rev Primer Sequence GCTAGGGTTACATTCCATTTTGATGGTGAC

RT-RPA Housekeeping GCTATGCCGATCGGGTGTCCGCACTAAGTT Fwd Primer Sequence

RT-RPA Housekeeping GACGGGGTCTCGCTATGTTGCCCAGGCTGG Rev Primer Sequence

RT-PCR Fwd Primer CCTGGAGCGCGGCAGGAAGCCTTATCAGTTG Sequence

RT-PCR Rev Primer Sequence TCCTGCTGAGGGACGCGTGGGCTCATCTTG

Preparation of SA-AuNPs for SERS detection

AuNPs with ~60 nm diameter were synthesized by aqueous reduction of HAuCl4 with sodium citrate according to Frens’s method.23 SA-AuNPs were prepared as previously 24 reported. Briefly, 1 mL of 60 nm AuNPs was adjusted to pH 9.0-9.5 with 1 M NaHCO3 aqueous solution. Then, 25 µL of 1 mg/mL streptavidin was added to AuNPs under shaking for 1 hr at RT. Next, 10 µL of 2% PEG was added and incubated for 12 hrs at 4°C to stabilize the SA-AuNPs. Following that, the resultant mixture was centrifuged at 4°C for 15 min (7600 rpm) and the pellet was resuspended in 200 µL of 1M PBS for further use.

RNA extraction from cell lines DuCap and LnCap cells were kindly donated by Matthias Nees (VTT, Finland); Gregor Tevz (APCRC, Australia); and Michelle Hill (UQDI, Australia). The cells were cultured in RPMI- 1640 growth media (Life Technologies, Australia) supplemented with 10% fetal bovine serum (Life Technologies, Australia) in a humidified incubator containing 5% CO2 at 37 ºC. RNA was extracted using Trizol® reagent (Life Technologies, Australia). RNA purity and integrity were checked using a NanoDrop Spectrophotometer (Thermo Scientific, USA).

RNA extraction from clinical urinary samples Urinary samples (30–50 mL) were centrifuged at 700 g for 10 min and urinary sediments were washed with ice-cold PBS buffer before being centrifuged again at the same condition. 25 µL of lysis buffer (150 mM Tris-HCI pH 8.0, 4.5 M guanidium-HCI, 3% v/v Triton-X and 1.5 mM EDTA) was added to 50 µL of PBS-suspended urinary sediments with vigorous

131

Chapter 4.1 mixing to release total RNA. Next, total RNA was purified from 25 µL of cell lysate using the Agencourt AMPure XP SPRI kit (Beckman Coulter, USA). Briefly, two volumes SPRI reagent was first incubated with the cell lysate for 10 min. The RNA-bound magnetic beads were then separated from the lysate using a magnet and washed twice with 80% ethanol. Finally, total RNA was eluted in 25 µL of RNase-free water.

RT-RPA For cell line experiments, the TwistAmp Basic RPA kit (Twist-DX, UK) was used with slight modifications to manufacturer’s instructions. Briefly, 1 µL of extracted RNA, 50 units of MMuLV reverse transcriptase (New England Biolabs, UK), 250 nM of each primer (Table 1) and 20 nM of biotinylated dUTPs (ThermoFisher, Australia) were added to make a 12.5 µL reaction volume prior to incubation at 41°C for 20 min. For all other experiments, the TwistAmp Basic RT-RPA kit (Twist-DX, UK) was used with slight modifications to manufacturer’s instructions. Briefly, 1 µL of extracted RNA and 250 nM of each primer (Table 1) were added to make a 12.5 µL reaction volume prior to incubation at 41°C for 20 min. RT-RPA amplicons were visualized on 1.5% agarose gel to verify amplification and validate SERS results.

RT-PCR RT-PCR was performed to validate results of clinical urinary samples. The KAPA SYBR® FAST One-Step qRT-PCR kit (Kapa Biosystems, USA) was used to set up a single reaction volume of 10 µL for each sample. Each reaction volume consisted of 1X KAPA SYBR® FAST qPCR Master Mix, 200 nM of each primer (Table 1), 1X KAPA RT Mix and 3 µL of extracted RNA. The tubes were incubated at 42°C for 10 min to synthesize cDNA, followed by 95°C for 5 min to deactivate RT before cycling 35 times (95°C for 30 s, 69°C for 30 s and 72°C for 1 min) and finished with 72°C for 10 min. Lastly, the RT-PCR amplicons were visualized on 1.5% agarose gel to verify amplification.

SERS detection 1 uL of RT-RPA amplicons were incubated with 10 µL of synthesized SA-AuNPs for 10 min at room temperature to allow coating of biotin-dsDNA onto AuNPs based on biotin- streptavidin binding. SERS spectra were recorded with a portable IM-52 Raman microscope (Snowy Range Instruments, USA). The 785 nm laser was used for excitation of Raman

132

Chapter 4.1 scattering, and SERS spectra were obtained at 1 sec integration with laser power (70 mW) under a long distance objective (40 x, NA = 0.60).

For quantitative TMPRSS2-ERG measurements, the average Raman intensity at 795 cm-1 (wavenumber for uracil)25 on the SERS spectra of the sample of interest was normalized with the same point on the SERS spectra of the RN7SL1 housekeeping RNA. The normalization was calculated as follows:

Relative TMPRSS2:ERG Level = XSample / XHousekeeping

-1 where XSample and XHousekeeping are the average Raman intensity at 795 cm for sample of interest and housekeeping RNA respectively.

For convenient stratification of patients’ fusion gene status, a relative TMPRSS2:ERG value of 1.0 was established as the cut-off for differentiating patients who were positive or negative for TMPRSS2:ERG respectively. This value was obtained through DuCap and LnCap cell line studies (Fig. 3D) which showed that the relative TMPRSS2:ERG level for LnCap (negative for TMPRSS2:ERG) did not exceed the value of 1.0.

RESULTS AND DISCUSSION Principle of SERS assay for gene fusion detection Fig.1 depicts the schematic representation of our approach for gene fusion detection by RT- RPA and SERS. After total RNA purification from cells, a short region spanning across TMPRSS2-ERG fusion junction is amplified isothermally by RT-RPA. During the amplification process, biotinylated dUTPs are randomly incorporated in place of T into growing strands to generate biotin-dsDNA polymers. These biotin-dsDNA polymers then serve as binding substrates for SA-AuNPs and can result in a AuNPs-dsDNA network structure as shown in Fig. 1. Upon laser excitation, SERS signals from individual nucleic acid bases are generated. The SERS signal level is proportional to the amount of RT-RPA amplicons, which in turn, reflect the amount of TMPRSS2-ERG transcripts in the sample. In contrast, no biotin-dsDNA polymers are generated in the absence of fusion transcripts for SA-AuNPs binding and thus, no distinct SERS signal is observed.

133

Chapter 4.1

Fig. 1. Schematic representation of combining RT-RPA and SERS for rapid TMPRSS2-ERG detection. Total RNA is isolated from sample source and TMPRSS2-ERG transcripts are specifically amplified by RT-RPA isothermally. During strand polymerization, biotinylated dUTP bases are randomly incorporated and subsequently serve as binding sites for SA- AuNPs. Upon laser excitation, AuNPs enhance DNA-mediated SERS to signal positive amplification, which in turn, represents TMPRSS2-ERG presence. No amplification and therefore no SERS signal is generated if the fusion gene is not present.

Sensitive detection of TMPRSS2-ERG transcripts To evaluate the sensitivity of our assay, we titrated synthetic TMPRSS2-ERG RNA transcripts at known concentrations (0-108 input copies) and measured the resulting SERS signal. The SERS spectra shown in Fig. 2A clearly indicated the distinct signals of different bases in the RT-RPA amplicons. Signal peaks at 795 cm-1, 1240 cm-1 and 1385 cm-1 corresponded to uracil while the peak at 1560 cm-1 was that of guanine.25 As expected, the average SERS signal also increased proportionally with increasing input templates (Fig. 2A). The corresponding gel electrophoresis analysis of the RT-RPA amplicons, which showed increasing band intensities in an input-dependent manner, was also in good agreement with our SERS data (Fig. 2B). Although we observed low level of SERS signal peaks in the no- template control (NTC) (Fig. 2A), we attributed this to minimal amounts of non-specific amplification which could not be observed by gel electrophoresis (Fig. 2B) but detectable by

134

Chapter 4.1 more sensitive SERS technique. Using 795 cm-1 as the reference wavenumber for uracil bases, the detection limit of our assay was about 103 copies of TMPRSS2-ERG RNA input, at which the average SERS signal was approximately 1.5-fold higher (t-test P<0.05) above the background signal of the NTC (Fig. 2C). This low detection limit of our assay was approximately the amount of TMPRSS2-ERG transcripts present within a single cell.26 In addition, SERS signal from dsDNA (108 copies) only without AuNPs; AuNPs; and SA- AuNPs controls showed very weak intensity with no well-defined peaks presented, thus demonstrating that SERS signals in our assay were mainly generated from amplified nucleic acid bases using AuNPs for SERS enhancement. Considering the presented data, the low detection limit (103 input copies) and RSD (10.8%) demonstrated the high sensitivity and reproducibility of our TMPRSS2-ERG assay.

135

Chapter 4.1

Fig. 2. Sensitivity of assay. (A) Average SERS spectra generated using 0 - 108 copies of synthetic TMPRSS2-ERG transcripts as initial RT-RPA inputs. (B) Gel electrophoresis image generated using 0 - 108 copies of synthetic TMPRSS2-ERG transcripts as initial RT- RPA targets. (C) Average Raman intensity (at 795 cm-1) of different initial RT-RPA input amount. Error bars represent standard deviation of three independent experiment.

Specific TMPRSS2-ERG detection in DuCap cell line To test the specificity of our assay, we assayed the DuCap and LnCap model cell lines which were positive and negative for TMPRSS2-ERG transcripts respectively4. As expected, the

136

Chapter 4.1 average SERS signal generated from DuCap-derived RNA was distinctly higher than that of LnCap (Fig. 3A). In contrast, the average SERS signal of LnCap-derived RNA was low and close to the background signal of the no template (NTC). We also performed our assay on the RN7SL1 housekeeping gene27 as a loading control and observed consistent average SERS spectra (Fig. 3B) thus demonstrating that the difference in SERS signal was indeed due to the difference in TMPRSS2-ERG status. We further validated our results with gel electrophoresis (Fig. 3C) and similar to the SERS data, amplification was observed only with DuCap-derived RNA. By normalizing TMPRSS2-ERG signals to RN7SL1’s at 795 cm-1, the SERS signal for DuCapwas about 2.5 fold higher than that of LnCap (Fig. 3D). Furthermore, no amplification was detected when reverse transcriptase was removed from the RPA reaction, thus indicating that amplification was TMPRSS2-ERG transcript (RNA) dependent (Fig. 3E). We also observed that the average SERS signals for housekeeping RN7SL1-derived RNA were lower than that of DuCap (Fig. 3A & 3B); despite both having similar band intensities on gel (i.e. same amount of amplicons) (Fig. 3C). One possible explanation could be that the RN7SL1 amplicon sequence (45 adenine bases) contained eight lesser adenine bases and therefore lesser biotinylated dUTPs as compared to the TMPRSS2-ERG amplicons (53 adenine bases). This may, in turn, lead to lower amounts of SA-AuNPs binding and subsequently lower SERS signal for RN7SL1 amplicons.

137

Chapter 4.1

Fig. 3. Specificity of assay. (A) Average SERS spectra generated by our assay on cell lines which are positive (DuCap) and negative (LnCap) for TMPRSS2-ERG. (B) Corresponding average SERS spectra of housekeeping RN7SL1 RNA for DuCap and LnCap cell lines. (C) Gel electrophoresis image generated using RT-RPA on cell lines which are positive (DuCap) and negative (LnCap) for TMPRSS2-ERG. (D) Housekeeping RN7SL1-normalized relative TMPRSS2-ERG levels of DuCap, LnCap and no-template control (NTC). Error bars represent standard deviation of three independent experiments. (E) Gel electrophoresis image generated using RT-RPA on DuCap and LnCap RNA with/without reverse transcriptase.

Assay performance on clinical urinary specimens To demonstrate potential clinical utility, we applied our assay to urinary samples from 3 metastatic PCa patients (Fig 4). Each individual’s TMPRSS2-ERG- and RN7SL1-derived

138

Chapter 4.1

SERS spectra are shown in Fig. 4A and 4B respectively. To allow for convenient stratification of patients’ fusion gene status, a relative TMPRSS2:ERG level higher than the value of 1.0 was scored “TMPRSS2-ERG positive” as established from cell line experiments (Fig 3D). Using this approach, we found that Patient #1 was scored negative, while Patients #2 and #3 were positive for TMPRSS2-ERG. These SERS results were also in good agreement via standard gel electrophoresis analysis of the corresponding RT-RPA amplicons (Fig. 4D). An additional RT-PCR (Fig. 4E) using a different set of primers (Table 1) was also performed to validate our results.

Finally, considering all the data presented, our proposed SERS-based assay has demonstrated good sensitivity, specificity and reproducibility, as well as a practical application with clinical patient samples. Our entire assay can be completed within 75 mins and is, to the best of our knowledge, one of the fastest assays for the detection of TMPRSS2-ERG. The assay may be useful as a diagnostic screening tool and for monitoring changes in TMPRSS2-ERG transcript levels, for example, over the course of cancer treatment. The assay’s simplicity and rapid turnover can potentially enable physicians to make timely clinical decisions especially in aggressive PCa cases.

139

Chapter 4.1

Fig. 4. Assay performance on prostate cancer urinary samples. (A) Average TMPRSS2-ERG SERS spectra generated for 3 clinical urinary samples. (B) Average RN7SL1 SERS spectra for 3 clinical urinary samples. (C) Housekeeping RN7SL1-normalized relative TMPRSS2- ERG levels of 3 clinical urinary samples and no-template control (NTC) control. Error bars represent standard deviation of three independent experiments. (D) Gel electrophoresis image generated using RT-RPA to detect TMPRSS2-ERG transcripts in the 3 clinical urinary samples. (E) Gel electrophoresis image generated using RT-PCR on 3 PCa urinary samples.

CONCLUSIONS To conclude, we have developed a novel strategy for rapid TMPRSS2-ERG detection in PCa by combining RT-RPA with a direct and label-free SERS readout for the first time. Our assay is sensitive to 103 input copies of TMPRSS2-ERG RNA transcripts (approximately single-cell

140

Chapter 4.1 levels). Assay specificity was demonstrated by accurately differentiating between TMPRSS2- ERG positive and negative cells. Our method could also be applied to PCa clinical urinary samples as a potential non-invasive approach for TMPRSS2-ERG detection. Finally, this approach is easily adaptable for the detection of other fusion genes in cancer by simply redesigning the RT-RPA primers. We believe that our proposed assay has potential in both cancer diagnostics and research as a simple yet rapid method for evaluating gene fusion transcripts.

REFERENCES 1 A. Nussenzweig, and M. C. Nussenzweig. Origin of chromosomal translocations in lymphoid cancer. Cell. 141, 27-38 (2010) 2 T. Helleday, S. Eshtad, and S. Nik-Zainal. Mechanisms underlying mutational signatures in human cancers. Nat. Rev. Genet. 15, 585-598 (2014) 3 M. R. Stratton. Exploring the genomes of cancer cells: Progress and promise. Science. 331, 1553-1558 (2011) 4 S. A. Tomlins, D. R. Rhodes, S. Perner, S. M. Dhanasekaran, R. Mehra, X. W. Sun, S. Varambally, X. H. Cao, J. Tchinda, R. Kuefer, C. Lee, J. E. Montie, R. B. Shah, K. J. Pienta, M. A. Rubin, and A. M. Chinnaiyan. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 310, 644-648 (2005) 5 M. A. Rubin, C. A. Maher, and A. M. Chinnaiyan. Common gene rearrangements in prostate cancer. J. Clin. Oncol. 29, 3659-3668 (2011) 6 S. A. Tomlins, A. Bjartell, A. M. Chinnaiyan, G. Jenster, R. K. Nam, M. A. Rubin, and J. A. Schalken. ETS gene fusions in prostate cancer: From discovery to daily clinical practice. Eur. Urol. 56, 275-286 (2009) 7 S. A. Tomlins, S. M. J. Aubin, J. Siddiqui, R. J. Lonigro, L. Sefton-Miller, S. Miick, S. Williamsen, P. Hodge, J. Meinke, A. Blase, Y. Penabella, J. R. Day, R. Varambally, B. Han, D. Wood, L. Wang, M. G. Sanda, M. A. Rubin, D. R. Rhodes, B. Hollenbeck, K. Sakamoto, J. L. Silberstein, Y. Fradet, J. B. Amberson, S. Meyers, N. Palanisamy, H. Rittenhouse, J. T. Wei, J. Groskopf, and A. M. Chinnaiyan. Urine TMPRSS2:ERG fusion transcript stratifies prostate cancer risk in men with elevated serum PSA. Science Transl. Med. 3, 94ra72-94ra72 (2011)

141

Chapter 4.1

8 J. H. Wang, Y. Cai, C. X. Ren, and M. Ittmann. Expression of variant TMPRSS2/ERG fusion messenger RNAs is associated with aggressive prostate cancer. Cancer Res. 66, 8347-8351 (2006) 9 D. Hessels, F. P. Smit, G. W. Verhaegh, J. A. Witjes, E. B. Cornel, and J. A. Schalken. Detection of TMPRSS2-ERG fusion transcripts and prostate cancer antigen 3 in urinary sediments may improve diagnosis of prostate cancer. Clin. Cancer Res. 13, 5103- 5108 (2007) 10 B. Laxman, S. A. Tomlins, R. Mehra, D. S. Morris, L. Wang, B. E. Helgeson, R. B. Shah, M. A. Rubin, J. T. Wei, and A. M. Chinnaiyan. Noninvasive detection of TMPRSS2 : ERG fusion transcripts in the urine of men with prostate cancer. Neoplasia. 8, 885-888 (2006) 11 M. Yoshimoto, A. M. Joshua, S. Chilton-MacNeill, J. Bayani, S. Selvarajah, A. J. Evans, M. Zielenska, and J. A. Squire. Three-color FISH analysis of TMPRSS2/ERG fusions in prostate cancer indicates that genomic microdeletion of chromosome 21 is associated with rearrangement. Neoplasia. 8, 465-469 (2006) 12 O. Piepenburg, C. H. Williams, D. L. Stemple, and N. A. Armes. DNA detection using recombination proteins. Plos Biol. 4, 1115-1121 (2006) 13 M. Euler, Y. J. Wang, D. Heidenreich, P. Patel, O. Strohmeier, S. Hakenberg, M. Niedrig, F. T. Hufert, and M. Weidmann. Development of a panel of recombinase polymerase amplification assays for detection of biothreat agents. J. Clin. Microbiol. 51, 1110-1117 (2013) 14 E. J. H. Wee, H. Y. Lau, J. R. Botella, and M. Trau. Re-purposing bridging flocculation for on-site, rapid, qualitative DNA detection in resource-poor settings. Chem. Commun. 51, 5828-5831 (2015) 15 S. M. Nie, and S. R. Emery. Probing single molecules and single nanoparticles by surface-enhanced Raman scattering. Science. 275, 1102-1106 (1997) 16 K. Kneipp, Y. Wang, H. Kneipp, L. T. Perelman, I. Itzkan, R. Dasari, and M. S. Feld. Single molecule detection using surface-enhanced Raman scattering (SERS). Phys. Rev. Lett. 78, 1667-1670 (1997) 17 K. Faulds, F. McKenzie, W. E. Smith, and D. Graham. Quantitative simultaneous multianalyte detection of DNA by dual-wavelength surface-enhanced resonance Raman scattering. Angew. Chem., Int. Ed. 46, 1829-1831 (2007) 18 A. Barhoumi, D. Zhang, F. Tam, and N. J. Halas. Surface-enhanced Raman spectroscopy of DNA. J. Am. Chem. Soc. 130, 5523-5529 (2008)

142

Chapter 4.1

19 J. Zheng, S. Pang, T. P. Labuza, and L. He. Semi-quantification of surface-enhanced Raman scattering using a handheld Raman spectrometer: a feasibility study. Analyst. 138, 7075-7078 (2013) 20 E. Papadopoulou, and S. E. J. Bell. Label-free detection of nanomolar unmodified single- and double-stranded DNA by using surface-enhanced Raman spectroscopy on Ag and Au colloids. Chem. Eur. J. 18, 5394-5400 (2012) 21 E. Papadopoulou, and S. E. J. Bell. Label-free detection of single-base mismatches in DNA by surface-enhanced Raman spectroscopy. Angew. Chem., Int. Ed. 50, 9058-9061 (2011) 22 L. Guerrini, Z. Krpetic, D. van Lierop, R. A. Alvarez-Puebla, and D. Graham. Direct surface-enhanced Raman scattering analysis of DNA duplexes. Angew. Chem., Int. Ed. 54, 1144-1148 (2015) 23 G. Frens. Controlled nucleation for regulation of particle-size in monodisperse gold suspensions. Nat. Phys. Sci. 241, 20-22 (1973) 24 S. Yokota. Preparation of colloidal gold particles and conjugation to protein A, IgG, F(ab')(2), and streptavidin. Immunoelectron Microsc., Methods Protocols, Edited S. D. Schwartzbach and T. Osafune, (2010) 25 S. E. J. Bell, and N. M. S. Sirimuthu. Surface-enhanced Raman spectroscopy (SERS) for sub-micromolar detection of DNA/RNA mononucleotides. J. Am. Chem. Soc. 128, 15580-15581 (2006) 26 V. E. Velculescu, S. L. Madden, L. Zhang, A. E. Lash, J. Yu, C. Rago, A. Lal, C. J. Wang, G. A. Beaudry, K. M. Ciriello, B. P. Cook, M. R. Dufault, A. T. Ferguson, Y. H. Gao, T. C. He, H. Hermeking, S. K. Hiraldo, P. M. Hwang, M. A. Lopez, H. F. Luderer, B. Mathews, J. M. Petroziello, K. Polyak, L. Zawel, W. Zhang, X. M. Zhang, W. Zhou, F. G. Haluska, J. Jen, S. Sukumar, G. M. Landes, G. J. Riggins, B. Vogelstein, and K. W. Kinzler. Analysis of human transcriptomes. Nat. Genet. 23, 387-388 (1999) 27 C. R. Galiveti, T. S. Rozhdestvensky, J. Brosius, H. Lehrach, and Z. Konthur. Application of housekeeping npcRNAs for quantitative expression analysis of human transcriptome by real-time PCR. RNA. 16, 450-461 (2010)

143

Chapter 4.2

Nanoplasmonic Label-Free Surface-Enhanced Raman Scattering Strategy for Non-Invasive Cancer Genetic Subtyping in Patient Samples

Jing Wang, Kevin M. Koo, Eugene J. H. Wee, Yuling Wang & Matt Trau

ABSTRACT Simple nucleic acid detection methods could facilitate the progress of disease diagnostics for clinical uses. An attractive strategy is label-free surface-enhanced Raman scattering (SERS) due to its sensitivity and capability of providing structural fingerprinting of analytes that are close to or on nanomaterial surfaces. However, current label-free SERS approaches for DNA/RNA biomarker detection are limited to short and synthetic nucleic acid targets and have not been fully realized in clinical samples due to two possible reasons: i) low target copies in limited patient samples and ii) poor capability in identifying specific biomarkers from complex samples. To resolve these limitations and enable label-free SERS for clinical applications, we herein present a novel strategy based on a combination of multiplex reverse transcription-recombinase polymerase amplification (RT-RPA) to enrich for multiple RNA biomarkers, followed by label-free SERS aided by multivariate statistical analysis to directly detect, identify and distinguish between these long amplicons (~200 bp). As a proof-of- concept clinical demonstration, we employed the strategy for non-invasive subtyping of prostate cancer (PCa). In a training cohort of 43 patient urinary samples, we achieved 93.0% specificity, 95.3% sensitivity, and 94.2% accuracy. We believe that our proposed assay could pave the way for simple and direct label-free SERS detection of multiple long nucleic acid sequences in patient samples and thus facilitate rapid cancer molecular subtyping for personalized therapies.

INTRODUCTION The analysis of RNA biomarkers in cancer could inform on disease pathogenesis, medical diagnosis, disease staging, and therapeutic monitoring.1 For example, in prostate cancer (PCa), the molecular subtyping of patients based on their highly PCa-specific gene fusion mutation - TMPRSS2 exon 1 and ERG exon 4 (T1E4) - RNA status could potentially inform

144

Chapter 4.2 clinical decision-making and aid in directing precision therapies.2, 3 However, current commonly employed RNA detection strategies (e.g. reverse transcription-polymerase chain reaction (RT-PCR), next generation sequencing (NGS), NanoString, and Forster resonance energy transfer (FRET)-based biosensors, etc.) suffer from limitations such as costly fluorescence labelling and complex experimental procedures.1, 4-6 While other contemporary methods employing isothermal amplification have been proposed to shorten assay time and cost,7, 8 most still require some form of expensive reagents. Simpler and low-cost strategies are therefore urgently needed to facilitate the use of RNA biomarkers in clinic.

Label-free surface-enhanced Raman scattering (SERS), an approach to directly detect molecules absorbed onto nanostructure surfaces, is an attractive and powerful tool for nucleic acid detection. Compared to other technologies or SERS sensors that require auxiliary labelling modification/hybridization chemistry, label-free SERS sensors have the competitive characteristics in rapid, simple, cost-efficient and direct detection as well as providing the chemical component information for the target. However, most of current label-free SERS detection of nucleic acids are limited to short synthetic sequences, mainly focusing on characterizing chemical and structural differences within a single DNA/RNA sequence,9-14 and have rarely, if any, been translated into clinical applications. This could be due in part to the poor spectral reproducibility and/or limited sensitivity.10 Notwithstanding, applying label- free SERS technology to DNA/RNA biomarker detection in clinical samples has additional major challenges including: i) trace levels of target RNA biomarkers in limited, complex clinical samples; ii) poor capability to detect multiple DNA/RNA biomarkers simultaneously. Novel strategies are therefore needed to implement label-free SERS-based DNA/RNA bioassays for subtyping cancer.

To overcome limitations of label-free SERS for clinical applications, we herein present a novel DNA/RNA detection strategy consisting of multiplex reverse transcription- recombinase polymerase amplification (RT-RPA) to enrich for target sequences followed by sequence identification with label-free SERS. As previous works distinguish multiple sequences based on subtle SERS spectral differences,10, 12 they are unavoidably subject to subjectivity and poor spectral reproducibility. To address these issues, multivariate statistical analysis was used to distinguish between two > 200 bp long sequences and to predict for the presence of a particular sequence. As a proof-of-concept clinical demonstration, we applied the method in the risk stratification of PCa in 43 urinary samples based on a highly-promising

145

Chapter 4.2

PCa RNA biomarker - T1E4 fusion gene, as well as an endogenously-expressed housekeeping RNA (RN7SL1)15 that is used as a loading control to verify the presence of input RNA (i.e. rule out false-negative T1E4 results).15 With 93.0% specificity, 95.3% sensitivity, and 94.2% accuracy, we believe that the proposed assay could pave the way for simple and direct label-free SERS detection of long nucleic acid sequences in complex samples.

MATERIALS AND METHODS Materials All reagents were purchased from Sigma Aldrich, unless otherwise stated. Synthetic oligonucleotide and primer sequences used in our experiments were purchased from Integrated DNA Technologies (Singapore), and sequences are shown in Table 1.

Table 1. Oligonucleotide sequences Oligos 5'-Sequence-3' T1E4 RT-RPA Fwd Primer CCTGGAGCGCGGCAGGAAGCCTTATCAGTTG T1E4 RT-RPA Rev Primer GCTAGGGTTACATTCCATTTTGATGGTGAC RN7SL1 RT-RPA Fwd Primer GCTATGCCGATCGGGTGTCCGCACTAAGTT RN7SL1 RT-RPA Rev Primer GACGGGGTCTCGCTATGTTGCCCAGGCTGG CCTGGAGCGCGGCAGGAAGCCTTATCAGTTGTGAG TGAGGACCAGTCGTTGTTTGAGTGTGCCTACGGAA CGCCACACCTGGCTAAGACAGAGATGACCGCGTCC T1E4 RT-RPA amplicon TCCTCCAGCGACTATGGACAGACTTCCAAGATGAG CCCACGCGTCCCTCAGCAGGATTGGCTGTCTCAACC CCCAGCCAGGGTCACCATCAAAATGGAATGTAACC CTAGC GCTATGCCGATCGGGTGTCCGCACTAAGTTCGGCAT CAATATGGTGACCTCCCGGGAGCGGGGGACCACCA GGTTGCCTAAGGAGGGGTGAACCGGCCCAGGTCGG RN7SL1 RT-RPA amplicon AAACGGAGCAGGTCAAAACTCCCGTGCTGATCAGT AGTGGGATCGCGCCTGTGAATAGCCACTGCACTCC AGCCTGGGCAACATAGCGAGACCCCGTC

146

Chapter 4.2

RNA extraction from cell line DuCap cell line was kindly donated by Matthias Nees (VTT, Finland) and Gregor Tevz (APCRC, Australia). The cells were cultured in RPMI-1640 growth media (Life Technologies, Australia) supplemented with 10% fetal bovine serum (Life Technologies, Australia), 2 mM Glutamax (Gibco, Australia) and 1% PenStrep (Invitrogen, Australia) in a humidified incubator containing 5% CO2 at 37 ºC. RNA was extracted using Trizol® reagent (Life Technologies, Australia). RNA purity was checked using a NanoDrop 1000 Spectrophotometer (Thermo Scientific, USA).

RNA extraction from urinary samples For clinical samples, ethics approval was obtained from The University of Queensland Institutional Human Research Ethics Committee (Approval No. 2004000047) and informed consent was obtained from all subjects prior to sample collection. Methods pertaining to clinical samples were carried out in accordance with approved guidelines. Voided urinary samples were collected from male patients undergoing treatment for PCa and healthy men with no PCa history. Total RNA was extracted from urinary samples using the commercially- available ZR urine RNA isolation KitTM (Zymo Research, USA). For each sample, 30 mL of urine was passed through the supplied ZRC GFTM Filter to retain urinary cells. Then, 700 µL of supplied urine RNA buffer was passed through the filter and the filtered cells were collected in the flow-through. Next, as per manufacturer’s instructions, cells from the flow- through were lysed, washed and eluted in 10 µL of RNase-free water.

RT-RPA of RNA biomarkers The TwistAmp Basic RT kit (Twist-DX, UK) was used with slight modifications to manufacturer’s instructions to simultaneously generate complementary DNA (cDNA) and rapidly amplify cDNA templates in a single tube reaction. Briefly, 1 µl of extracted total RNA and 250 nM of each primer (Table 1) were added to the supplied reagents to make a 12.5 µl reaction volume prior to incubation at 41°C for 15 min. Amplicons were then purified using the Agencourt AMPure XP SPRI kit (Beckman Coulter, USA) and then eluted in RNase-free water for the following SERS measurement.

Preparation of AgNPs Silver nanoparticles of positive surface charges (AgNPs) were prepared as previously 16 reported. Briefly, 100 µL 0.5 M AgNO3 was added to 50 mL of distilled water, followed by

147

Chapter 4.2 the addition of 35 µL 0.1 M spermine. The mixture was under vigorous stirring for 20 min before the addition of 1.25 mL 0.01 M NaBH4. The reaction was kept under gentle stirring for 20 min until the colloid colour turned into grey. The ζ potential of the AgNPs (diluted in water) is +19.2 ± 0.4 mV, measured by Zetasizer NANO Series using standard settings (viscosity = 0.89, temperature = 25 ◦C). TEM (JEM-1010) and UV-vis spectra (NanoDrop 1000 spectrophotometer) detection were also performed to characterize AgNPs (Fig. S1). The maximum absorption peak for the AgNPs locates at 397 nm.

Label-free SERS measurements 1 µL of RT-RPA amplicons was mixed with 60 µl of AgNPs for 2 h. Next, SERS spectra in the range of 400-1800 cm-1 were collected from the sample mixture using a portable IM-52 Raman Microscope with a 785 nm wavelength laser as the excitation source. 70 mW of incident laser power in 1 s of illumination was used for the detection. Each sample was represented with the average spectrum calculated from 5 measurements.

Multivariate statistical analysis for biomarker prediction model The raw SERS spectra were firstly baseline-corrected to remove background noise using the Vancouver Raman Algorithm (a five-order polynomial fitting algorithm).17 In order to compensate for gross differences in the spectral response due to physical effects, area normalization of the spectroscopic data was performed by Origin 8 software (OriginLab Inc., USA).

A combination of principal component analysis (PCA) and linear discriminant analysis (LDA) - termed as PCA-LDA was applied in multivariate statistical analysis. As each SERS spectrum in the range of 400-1800 cm-1 possessed 1401 variables, PCA was therefore introduced to capture main variables for expression in different linear combinations of significant principal components (PCs). Spectrum i is in the form as:

where represents spectrum i and is a combination of parameters, values are PC scores and is the residual. The data is highly correlated in a multi-dimensional space.

Independent T-test of PC scores was used to examine ability in discriminating two data sets that were composed of separate T1E4 and RN7SL1 SERS spectra. Selected PC scores from

148

Chapter 4.2 significantly different PCs (p < 0.05) were used to build the LDA training model in an unbiased manner with the leave-one-out and cross-validation method. In this method, one sample (i.e., one spectrum) was left out from the data set, and the remaining samples were used for redeveloping the LDA model to classify the withheld sample. This process was repeated until all samples were classified. A receiver operating characteristic (ROC) curve which was generated by successively changing discrimination thresholds to determine discrimination sensitivity and specificity of the LDA model, was introduced to illustrate the discrimination ability of PCA-LDA model in predicting the different RNA biomarkers.

SPSS 19.0 software package (SPSS Inc., Chicago, Illinois) was used for Independent T-test, PCA, LDA, and ROC analysis.

RESULTS AND DISCUSSION Principle of target RNA biomarker detection using RT-RPA coupled with label-free SERS Fig. 1 depicts the RT-RPA coupled with label-free SERS strategy for cancer subtyping via RNA biomarkers in PCa patient samples. After total RNA extraction from urinary samples, target RNA biomarkers are amplified by isothermal RT-RPA and stabilized in the form of dsDNA amplicons. We believe this homogeneous mixture of highly-concentrated and stable DNA aids in achieving reproducible SERS signals. Following the amplification and purification, dsDNA amplicons are mixed with positively-charged AgNPs that function as SERS substrates. The AgNPs’ surfaces are covered with spermine to generate stable colloids in aqueous solution and enable formation of long-term (> 24 h) stable particle clusters after DNA addition.10, 16 When DNA adhesion reached its equilibrium (2 h), SERS spectra are collected in colloidal suspensions with a long working distance objective to represent a statistically-averaged result of a large ensemble of nanoparticle/DNA clusters in continuous Brownian motions within the scattering volume. This offers SERS spectra reproducibility that could represent molecular information of dsDNA amplicons such as base components and phosphate backbone structures. As such molecular information is distinct for different RNA biomarkers, it is therefore expected that the resulting SERS spectra will enable discrimination between different RNA biomarker sequences.

149

Chapter 4.2

Fig. 1. Schematic representation of label-free SERS detection for RNA biomarkers. The total RNA was firstly extracted from urinary samples. RT-RPA was then used to amplify the target RNA into dsDNA sequences. Purification of amplicons was performed prior to incubation with positively-charged AgNPs and subsequent SERS measurements.

Assay demonstration using DuCap cell line RNA To test the capability of our assay for clinical applications, we first used cultured RN7SL1- and T1E4-positive DuCap cell line as a preliminary model to individually detect T1E4 and RN7SL1RNA biomarkers. After performing the assay, we successfully acquired two unique SERS spectra for each target (Fig. 2A, curves a-b), which were also easily-discriminated from background signals of the RT-RPA no-template control (NTC) and AgNPs alone (Fig. 2A, curves c-d). The major vibrational peak assignments for SERS spectra of targets have been listed on Table 2. The peaks assigned to phosphate backbone (913 cm-1), purines (560, 742 and 1334 cm-1), and pyrimidines (1632 cm-1) could readily be observed on the SERS spectral profiles of both T1E4 and RN7SL1 target amplicons. In contrast, these peaks were absent on both SERS spectra of the no-template and AgNPs controls, and could therefore be used to specifically detect the presence of target amplicons.

150

Chapter 4.2

Fig. 2. (A) Typical average SERS spectra of (a) RN7SL1 and (b) T1E4 RNA from DuCap cell line sample, (c) no-template control (NTC), and (d) background signal of AgNPs. (B) Gel electrophoresis image of RT-RPA amplicons from DuCap cell line-derived total RNA.

Table 2. Typical vibrational assignments of major peaks in the observed SERS spectra of RNA transcripts Peak positions (cm-1) Major assignments 560 A18 742 Ring breathing of A,18, 19 T20 788 C, T21 913 Deoxyribose and backbone,20, 22 phosphate18, 1035 dT, dG, dC, deoxyribose,22 A,18 1180 T21 1247/1263 C, G21 1334 Ring breathing of A18, 19 1457 A,18 deoxyribose and backbone 20 1539 Ring stretching of dA and dG22 1632 C=C stretching of C23 Abbreviations: A - adenine; T - thymine; C - cytosine; G - guanine. dA - deoxyadenosine; dG - deoxyguanosine; dC - deoxycytidine; dT - deoxythymidine;

Gel electrophoresis analysis was used to validate the successful amplification and sizes of T1E4 and RN7SL1 RT-RPA amplicons from the DuCap cell line (Fig. 2B). As expected, T1E4 and RN7SL1 amplicons were 216 bp and 204 bp respectively, thus indicating specific RT-RPA amplification.

151

Chapter 4.2

Assay sensitivity Sensitive assays are very beneficial as availability of clinical specimens and target abundance are typically limited in quantity. To this end, we tested assay sensitivity by titrating synthetic T1E4 RNA (0-106 copies) in a background of total RNA isolated from a T1E4-negative cell line. In Fig. 3A, we observed that peak intensities at 742 and 1334 cm-1 (purines) increased with higher synthetic T1E4 input copies, and therefore the two peaks were utilized as candidates for assay sensitivity evaluation. As shown in Fig. 3B and 3C, the peaks at 742 and 1334 cm-1 for a low amount of synthetic T1E4 input copies (102) were significantly higher (p < 0.05) than no-input signals, and 102 T1E4 input copies was thus determined as our assay detection limit. RT-RPA amplification of target RNA biomarkers was fundamental in exponentially increasing low RNA target copies in the starting sample and stabilizing the amplicons in dsDNA form, thus leading to a sensitive and robust label-free SERS detection strategy for trace amounts of RNA biomarkers.

Gel electrophoresis analysis of the corresponding RT-RPA amplicons (for different T1E4 input copies) was further employed to verify RT-RPA products (Fig. 3D). As consistent with SERS, we observed a positive association between gel band intensities and T1E4 input copies. However, the detection sensitivity of gel electrophoresis was lower at 104 input copies, at which an amplicon band was still visible. It is worthy to note that recently described colorimetric RT-RPA assays for T1E4 detection also have poorer analytical sensitivity as compared to label-free SERS assay.24, 25 This 100-fold improvement in sensitivity further underscores a potential advantage of combining label-free SERS with RT- RPA. We attributed this increase in T1E4 detection sensitivity to the signal amplification conferred by SERS. Meanwhile, compared to other label-free SERS-based technologies,9, 11, 26 our methodology is unique in detecting longer sequences and at lower starting RNA concentration through the combination use of positively-charged AgNPs and RT-RPA.

152

Chapter 4.2

Fig. 3. Assay sensitivity evaluation. (A) Typical average SERS spectra after baseline- correction generated using various amounts of input synthetic T1E4 sequences. (B) Typical comparison of 742 and 1334 cm-1 peaks in the average SERS spectra of T1E4 after baseline- correction generated with no input and using 102 and 103 copies of input synthetic T1E4 sequences. (C) Average SERS intensities (at 742 and 1334 cm-1) using different amounts of input synthetic T1E4 sequences. Error bars represent standard deviations (SDs) of three independent experiments. (D) Gel electrophoresis image generated using various amounts of input synthetic T1E4 sequences.

Assay application to patient samples After establishing the feasibility of our assay for RNA biomarker detection in cultured cells, we proceeded to apply our method to clinical PCa samples. In this regard, we performed our assay for the detection of T1E4 and RN7SL1 RNA biomarkers from a T1E4-positive, a T1E4-negative, and a healthy control urinary samples. As shown in Fig. 4A, SERS spectra of RN7SL1 (curve a) and T1E4 (curve b) RNA biomarkers from a T1E4-positive urinary sample were similar to those of same biomarkers detected from DuCap cell line. In contrast, SERS spectra of T1E4-negative (curve c) and healthy (curve d) controls showed no peaks at positions for purines, pyrimidines and phosphate backbone (Table 2). Instead, SERS spectra from T1E4-negative and healthy samples resembled the background signal from AgNPs (Fig. 2A, curve d). As the presence of T1E4 biomarker in these three urinary samples was pre- verified via RT-RPA and gel electrophoresis analysis (Fig. 4A insert), we demonstrated the successful detection of PCa biomarkers in urinary samples using our assay combination of

153

Chapter 4.2

RT-RPA and AgNPs-based label-free SERS. Given that negligible signals were obtained for the control samples, our assay also exhibited accurate and selective target detection.

Most importantly, our assay offers several advantages over current RNA detection methodologies in clinical samples: i) relatively short assay time (~ 2 hr) and low cost; ii) no need for any labelling procedures for detection readout; iii) molecular information achieved from SERS spectra could be potentially useful for further cancer subtyping analysis.

We further investigated our assay reproducibility for detecting PCa urine-derived RNA biomarkers by including more T1E4-positive patient samples (n = 43). Fig. 4B showed that distinct T1E4 and RN7SL1 SERS spectra could be reproducibly detected in this larger cohort of patient samples. This data further suggested the clinical potential of our assay for non- invasive RNA biomarker detection in urinary samples.

Fig. 4. Typical average SERS spectra of RNA biomarkers from urinary samples. (A) Typical average SERS spectra of (a) internal control RN7SL1- and (b) T1E4-amplified from a T1E4- positive urinary sample, (c) a T1E4-negative urinary sample, and (d) a healthy control. The inset is the gel electrophoresis image of T1E4 and RN7SL1 RT-RPA amplicons from the T1E4-positive, T1E4-negative, healthy urinary samples. (B) Normalized average SERS spectra of RN7SL1 and T1E4 amplicons from 43 T1E4-positive urinary samples, with the shaded areas representing the SDs of the means.

Statistical analysis for prostate cancer subtyping We envisaged one potentially significant clinical application of our assay is for biomarker- based cancer subtyping. By sensitive and specific detection of cancer subtype-specific biomarkers, we could potentially classify patients for more precise and effective therapies with the proposed assay.

154

Chapter 4.2

In order to differentiate and predict T1E4-positive PCa with greater discrimination, we employed PCA-LDA-based multivariate statistical method to deconvolute the different SERS spectra. PCA identified key components (i.e. PCs) for differentiating SERS spectra of T1E4 and RN7SL1 amplicons. Using Independent T-test, we found that PC1, 3 and 4 have p values of < 0.001 but PC2 has a p value of > 0.05. This indicated that PC2 included less information in differentiating RN7SL1 and T1E4 and was therefore omitted. To further discuss the difference contributions of PC1, 3 & 4 towards discriminating between T1E4 and RN7SL1 RNA biomarkers, we used the individual PC correlation coefficient profiles (Fig. S2). We observed that PC1, 3 & 4 contributed peaks at 487 cm-1 (background AgNPs signal), 560 cm-1 (adenine), 742 cm-1 (adenine and thymine), 788 cm-1 (cytosine and thymine), 913-1035 cm-1 (adenine, deoxythymidine, deoxyguanosine, deoxycytidine, deoxyribose, backbone, and phosphate), 1180 cm-1 (thymine), 1247/1263 cm-1 (cytosine and guanine), 1457 cm-1 (deoxyribose, backbone, and adenine), 1539 cm-1 (adenine and guanine) and 1605-1684 cm-1 -1 18, 20-22 (cytosine and thymine) cm . Consequently, PC1, PC3 and PC4 scores were selected for discriminating SERS spectra of T1E4 and RN7SL1 amplicons (Figs. 5A-C) and used as axes in three-dimensional mapping of T1E4 and RN7SL1 SERS signals (Fig. 5D).

In Fig. 5D, T1E4 and RN7SL1 SERS signals (n = 43 for each) were clustered into two separate groups on the basis of their PC scores, thus indicating that the combination of (PC1, PC3, and PC4) was significant in differentiating T1E4 and RN7SL1 RNA biomarkers in PCa urinary samples. These three PCs were then incorporated to build a biomarker-discriminating training model and this training model is demonstrated to be able to predict SERS signals of T1E4 and RN7SL1 with the specificity of 93.0%, sensitivity of 95.3%, and accuracy of 94.2% (Fig. 5E).

In order to demonstrate our assay as a potential PCa subtype screening tool, we performed duplexed RT-RPA amplification and label-free SERS detection of both T1E4 and RN7SL1 biomarkers from each of 11 T1E4-positive urinary samples. As the presence of both biomarkers within a single sample after duplexed amplification led to different SERS spectral signatures from our well-differentiated individual T1E4 and RN7SL1 SERS signals, our developed PCA-LDA training model could accurately discriminate these 11 ‘mixed’ samples, as shown in Fig. 5E. The distribution of discriminant scores in Fig. 5E showed that the ‘mixed’ samples were located between individual T1E4 and RN7SL1 clusters with an area under ROC curve (AUC) of 0.978 (Fig. 5F). This indicated that our model was capable of

155

Chapter 4.2 identifying samples which were positive for both PCa biomarker T1E4 and internal control RN7SL1, thus suggesting a valuable first-line clinical screening tool for identifying T1E4- positive PCa patients (via non-invasive detection from urine) for effective subtype-specific treatment.

Fig. 5. Box plots of PC1 (A), PC3 (B), and PC4 (C) scores for RN7SL1, T1E4, and ‘mixed’ amplicons. The line within each notch box represents the median, and the lower and upper boundaries of the box indicate first and third quartiles respectively. Error bars (whiskers) represent the 1.5-fold interquartile range. Independent T-test yields *p < 0.001 for all three PCs, indicating these three PCs are diagnostically significant in differentiating RN7SL1 and T1E4. (D) A three-dimensional mapping of the PCA result for the RN7SL1 (black circle) and T1E4 (red triangle). (E) Box plots of the discriminant scores belonging to spectra of 43 RN7SL1, 43 T1E4 and 11 ‘mixed’ amplicons from urinary samples. (F) The receiver operating characteristic (ROC) curve for discriminating T1E4, RN7SL1 and ‘mixed’ amplicons from urinary samples.

The use of label-free SERS for multiplexed DNA/RNA detection is inherently difficult due to similar chemical structures of targets and poor spectral reproducibility.10, 12 To date, label- free SERS has been limited to classification of synthetic short sequences as longer nucleic acid sequences have more complicated interactions with SERS substrates.10, 12 To demonstrate our methodology is applicable for various RNA targets, we successfully showed detection of four other RNA biomarkers isolated from three patient urine samples (Fig. S3A). We performed Raman difference spectra analysis for all pairing combinations of RNA target signals (Figs. S3B-D) and the resultant spectral peaks display spectral differences between

156

Chapter 4.2 two different RNA targets. For example, in Fig. S3B, the analysis shows that T1E2 and RN7SL1 have spectral differences at 560 cm-1, 1035 cm-1, and 1334 cm-1, which could be assigned to adenine (560 cm-1, 1035 cm-1, and 1334 cm-1); and deoxythymidine, deoxyguanosine, deoxycytidine, and deoxyribose (1035 cm-1).

Taken together, while we have demonstrated a duplexed assay for two long sequences (i.e. RT-RPA amplicons of target RNA biomarkers) from clinical samples as well as the applicability of our method for detecting various RNA biomarker targets (Fig. S3), future work could improve on aspects such as nucleic acid-NPs interactions and data processing to increase multiplexing capabilities. Nonetheless, a duplexed system presented here is a step towards this direction and is potentially useful as a diagnostic tool for multiple RNA biomarkers detection.

CONCLUSIONS In this work, we presented a label-free SERS assay which incorporated RT-RPA and multivariate statistical analysis for detecting RNA biomarkers in clinical samples and further enabling PCa subtyping. Our assay overcame limitations with traditional RNA detection techniques and expanded label-free SERS detection to long-length nucleic acid sequences in clinical samples by: (1) enriching low amount of target RNA biomarkers in patient samples to enhance subsequent label-free SERS signals; and (2) introducing multivariate statistical analysis to differentiate SERS spectra of different RNA biomarkers.

Our assay was shown to be sensitive to 100 copies of input RNA, and multivariate statistical analysis was used to differentiate two different biomarkers with excellent discrimination accuracy. The detection of T1E4 PCa biomarker in PCa urinary samples demonstrated the clinical value of our assay for non-invasive molecular classification of a potentially aggressive PCa subtype to enable targeted treatments.

We envision that the novel combination of experimental and statistical techniques described in our proposed assay could be a viable way for simpler cancer subtyping (via nucleic acid biomarkers detection) in clinical patient samples.

157

Chapter 4.2

REFERENCES 1. I. Riedmaier and M. W. Pfaffl, Methods, 2013, 59, 3-9. 2. S. A. Tomlins, D. R. Rhodes, S. Perner, S. M. Dhanasekaran, R. Mehra, X. W. Sun, S. Varambally, X. H. Cao, J. Tchinda, R. Kuefer, C. Lee, J. E. Montie, R. B. Shah, K. J. Pienta, M. A. Rubin and A. M. Chinnaiyan, Science, 2005, 310, 644-648. 3. J. Wang, Y. Cai, C. Ren and M. Ittmann, Cancer Res., 2006, 66, 8347-8351. 4. P. J. Santangelo, B. Nix, A. Tsourkas and G. Bao, Nucleic Acids Res., 2004, 32, e57- e57. 5. C.-L. Shih, J.-D. Luo, J. W.-C. Chang, T.-L. Chen, Y.-T. Chien, C.-J. Yu and C.-C. Chiou, Cancer Genomics-Proteomics, 2015, 12, 223-230. 6. G. K. Geiss, R. E. Bumgarner, B. Birditt, T. Dahl, N. Dowidar, D. L. Dunaway, H. P. Fell, S. Ferree, R. D. George, T. Grogan, J. J. James, M. Maysuria, J. D. Mitton, P. Oliveri, J. L. Osborn, T. Peng, A. L. Ratcliffe, P. J. Webster, E. H. Davidson, L. Hood and K. Dimitrov, Nat. Biotechnol., 2008, 26, 317-325. 7. Y. Zhao, F. Chen, Q. Li, L. Wang and C. Fan, Chem. Rev., 2015, 115, 12491-12545. 8. K. M. Koo, E. J. Wee and M. Trau, Biosens. Bioelectron., 2017, 89, 715-720. 9. L.-J. Xu, Z.-C. Lei, J. Li, C. Zong, C. J. Yang and B. Ren, J. Am. Chem. Soc., 2015, 137, 5149-5154. 10. L. Guerrini, Ž. Krpetić, D. van Lierop, R. A. Alvarez‐Puebla and D. Graham, Angew. Chem. Int. Ed., 2015, 127, 1160-1164. 11. J. Morla-Folch, H. N. Xie, P. Gisbert-Quilis, S. Gomez-de Pedro, N. Pazos-Perez, R. A. Alvarez-Puebla and L. Guerrini, Angew. Chem. Int. Ed., 2015, 54, 13650-13654. 12. J. Morla-Folch, H. N. Xie, R. A. Alvarez-Puebla and L. Guerrini, ACS Nano, 2016, 10, 2834-2842. 13. J. L. Abell, J. M. Garren, J. D. Driskell, R. A. Tripp and Y. Zhao, J. Am. Chem. Soc., 2012, 134, 12889-12892. 14. E. Papadopoulou and S. E. Bell, Chemistry, 2012, 18, 5394-5400. 15. C. R. Galiveti, T. S. Rozhdestvensky, J. Brosius, H. Lehrach and Z. Konthur, RNA, 2010, 16, 450-461. 16. D. van Lierop, Ž. Krpetić, L. Guerrini, I. A. Larmour, J. A. Dougan, K. Faulds and D. Graham, Chem. Commun., 2012, 48, 8192-8194. 17. J. Zhao, H. Lui, D. I. McLean and H. Zeng, Appl. Spectrosc., 2007, 61, 1225-1232. 18. K. M. Koo, E. J. Wee, P. N. Mainwaring and M. Trau, Sci. Rep., 2016, 6, 30722.

158

Chapter 4.2

19. K. M. Koo, E. J. Wee and M. Trau, Theranostics, 2016, 6, 1415-1424. 20. K. M. Koo, B. McNamara, E. J. H. Wee, Y. Wang and M. Trau, J. Biomed. Nanotechnol., 2016, 12, 1798-1805. 21. M. Masetti, H. N. Xie, Z. Krpetic, M. Recanatini, R. A. Alvarez-Puebla and L. Guerrini, J. Am. Chem. Soc., 2015, 137, 469-476. 22. C. Muntean, N. Leopold, A. Halmagyi and S. Valimareanu, J. Raman Spectrosc., 2013, 44, 817-822. 23. S. E. J. Bell and N. M. S. Sirimuthu, J. Am. Chem. Soc., 2006, 128, 15580-15581. 24. W. Ke, D. Zhou, J. Wu and K. Ji, Appl. Spectrosc., 2005, 59, 418-423. 25. K. Kneipp, H. Kneipp, V. B. Kartha, R. Manoharan, G. Deinum, I. Itzkan, R. R. Dasari and M. S. Feld, Phys. Rev. E, 1998, 57, R6281-R6284. 26. E. Papadopoulou and S. E. Bell, Angew. Chem. Int. Ed., 2011, 50, 9058-9061.

SUPPLEMENTARY INFORMATION

Fig. S1. UV-Vis absorption spectrum of AgNPs. The maximum absorption locates at 397 nm. The insert is the corresponding TEM image.

159

Chapter 4.2

Fig. S2. Calculated correlation coefficient profiles of three diagnostically significant principal components (PCs) from a data set composed by 43 T1E4 and 43 RN7SL1 SERS spectra. Each PC revealed diagnostically significant spectral features (p < 0.001) for the discrimination between T1E4 and RN7SL1 RNA biomarkers.

Fig. S3. (A) SERS spectra of various target RNA biomarkers isolated from patient unary samples (n = 3), including ARV7, SChLAP1, fusion genes between TMPRSS2 exon 1 and ERG exon 5 (T1E5), TMPRSS2 exon 1 and ERG exon 2 (T1E2), TMPRSS2 exon 1 and ERG exon 4 (T1E4), and RN7SL1. Dotted lines indicate spectral positions for differentiating the distinct SERS signatures of each individual target. (B-D) Difference spectra obtained by digital subtraction between each combination pairs of target RNA signals.

160

Chapter 5

Amplification-Free Electrochemical RNA Biomarker Sensing

Introduction Chapters 3 and 4 have explored the use of isothermal RNA amplification for detection of urinary prostate cancer biomarkers. However, to avoid potential complications such as amplification bias/artifacts, primer design for the amplification process must be carefully considered and further optimized experimentally. Additionally, shorter-length RNA species such as microRNAs require customized amplification protocols due to their short primer- sized lengths. Thus, an amplification-free RNA biosensing system could be a viable alternative to detect different RNA biomarker species in their native forms directly. To this end, an electrochemical readout could provide the necessary analytical sensitivity for RNA detection without enzymatic amplification. In this chapter, a series of amplification-free electrochemical RNA nanobiosensors explored the novel concept of using affinity interactions between adenine bases and gold (Koo et al., Analytical Methods, 2015) to adsorb RNA targets onto electrode surfaces for amplification-free electrochemical detection. Chapter 5.1 and 5.2 demonstrate the amplification-free detection of messenger RNA and microRNA respectively, and Chapter 5.3 reports the combination of DNA-templated copper nanoparticle synthesis and adenine-gold adsorption for amplification-free detection of all common RNA biomarker species.

Chapter 5 is based on three published articles in Analytical Chemistry (2016) [Chapter 5.1], Analytical Chemistry (2016) [Chapter 5.2], and Nano Research (2017) [Chapter 5.3].

161

Chapter 5.1

Amplification-free Detection of Gene Fusions in Prostate Cancer Urinary Samples using mRNA-Gold Affinity Interactions

Kevin M. Koo, Laura G. Carrascosa, Muhammad J. A. Shiddiky, and Matt Trau

ABSTRACT A crucial issue in present-day prostate cancer (PCa) detection is the lack of specific biomarkers for accurately distinguishing between benign and malignant cancer forms. This is causing high degree of overdiagnosis and overtreatment of otherwise clinically insignificant cases. As around half of all malignant PCa cases display a gene fusion mutation between the TMPRSS2 promoter sequence and the ERG coding sequence (TMPRSS2:ERG) detectable in urine; non-invasive screening of TMPRSS2:ERG mRNA in patient urine samples could improve specificity of current PCa diagnosis. However, current gene fusion detection methodologies are largely dependent on RNA enzymatic amplification, which requires extensive sample manipulation, costly labels for detection, and it is prone to bias/artifacts. Herein we introduce the first successful amplification-free electrochemical assay for direct detection of TMPRSS2:ERG mRNA in PCa urinary samples by selectively isolating and adsorbing TMPRSS2:ERG mRNA onto bare gold electrodes without requiring any surface modification. We demonstrated excellent limit-of-detection (10 cells) and specificity using PCa cell line models; and showcased clinical utility by accurately detecting TMPRSS2:ERG in a collection of 17 urinary samples obtained from PCa patients. Furthermore, these results were validated with current gold standard reverse transcription (RT)-PCR approach with 100% concordance.

INTRODUCTION As cancer care is increasingly moving towards personalized treatment,1-4 the clinical applications of cancer detection assays are ultimately decided by novel biomarkers which could stratify cancer subtypes; for instance, to clearly distinguish malignant cancer forms from other benign conditions. This is particularly required for prostate cancer (PCa) diagnosis because the current use of prostate specific antigen (PSA) biomarker is associated with over diagnosis and overtreatment.5,6 Recurrent gene fusions are an emerging class of biomarkers

162

Chapter 5.1 which have been found to be widespread in different cancers such as prostate,7 breast,8 and lung9 cancers. These fusions result from chromosomal rearrangements events such as translocations, deletions, or inversions; and are associated with cancer initiation and progression by oncogene overexpression.10,11

Among all gene fusion cancer variants, the fusion between the exon 1 promoter sequence of TMPRSS2 gene and the exon 4 coding sequence of ERG gene (TMPRSS2:ERG) is especially common in PCa. In 2005, Tomlins et al. discovered TMPRSS2:ERG to be present in 50% of all malignant PCa cases and further work has related it to aggressive PCa subtypes.12-15 More importantly, TMPRSS2:ERG is absent in benign conditions which do not require treatment.7,12,16,17 Furthermore, it has been shown that TMPRSS2:ERG mRNA could be detected non-invasively through the urine samples of PCa patients, thereby providing a convenient way of diagnosing PCa without invasive tissue biopsy procedures.17,18 This has allowed TMPRSS2:ERG to be an exceptionally promising biomarker for predicting malignant PCa during early diagnosis and for stratifying PCa risk in individual patient to enable precision medicine.15,19-22

Current detection methodologies for TMPRSS2:ERG or any other gene fusion mRNA have largely been limited to amplification techniques such as reverse transcription-PCR (RT- PCR)18,19,23 or isothermal transcription mediation amplification (TMA).20,24,25 In particular, the detection of limited TMPRSS2:ERG mRNA copies from urinary samples requires enzymatic reverse transcription and amplification in order to achieve suitable detection sensitivity limit. Although RT-PCR and TMA are useful for sensitive and specific detection of TMPRSS2:ERG, they involve extensive sample manipulation during enzymatic RNA amplification. The use of enzymes will increase assay cost and require cold storage to prevent enzyme degradation. The amplification process might also cause amplification bias/artifacts to affect assay reliability. In addition, both these techniques require a level of technical expertise to perform and the subsequent gel electrophoresis or fluorescence readouts involves time-consuming preparation steps and costly reagents/probes. Hence, the development of an amplification-free technique with superior urinary TMPRSS2:ERG mRNA detection sensitivity could avoid all these issues and would be a major breakthrough towards better and more robust PCa diagnosis.

163

Chapter 5.1

In this report, we described a new approach for simple, sensitive, and amplification-free detection of TMPRSS2:ERG mRNA from cultured cells or clinical urinary samples. The underlying mechanism is based on poly adenine (polyA)-mediated mRNA-gold adsorption. Previous studies have shown that nucleic acid can spontaneously adsorb onto a bare gold surface by physisorption and chemisorption mechanisms.26 The adsorption affinity between nucleic acid bases and gold was found to be highly sequence-dependent and largely follows the trend A > C ≥ G > T27 with adenine bases displaying highest affinity interaction with gold over the other bases. This sequence affinity trend has been exploited in many applications26 such as self-attachment of DNA sequences on gold surfaces;28,29 electrochemical30,31 or optical32 detection of changes in DNA composition (e.g., base changes due to bisulfite-treated methylated DNA); and more recently, detection of miRNAs.33 However, to the best of our knowledge, the use of sequence-dependent adsorption has not been demonstrated for the detection of long-length RNA biomarkers such as TMPRSS2:ERG mRNA. Since all mRNA naturally comprise of 3’ polyA tail sequences, we exploited the strong affinity interaction between polyA-containing mRNA and gold for adsorbing TMPRSS2:ERG mRNA onto bare gold electrode surfaces with high efficiency. To this end, we performed magnetic isolation of target TMPRSS2:ERG mRNA using probes complementary to the gene fusion junction, followed by polyA-mediated TMPRSS2:ERG mRNA-gold adsorption onto a bare electrode surface for electrochemical detection. The adsorbed mRNA was then electrochemically 3-/4- quantified in the presence of the [Fe(CN)6] redox system. Conventionally, this redox 2+/3+ systems is coupled to the [Ru(NH3)6] to improve detection sensitivity (i.e., electrocatalytic enhancement of the signal) of mRNA detection.35-37 However, we26,30,31 and Zhang et al.34 have previously demonstrated that this system alone can be used for quantification of surface-bound oligonucleotides at bare gold electrodes. In this report, for the first time, we rationalized that a similar system could work for rapid electrochemical quantification of mRNA.

With our assay, the three main issues associated with conventional TMPRSS2:ERG detection techniques were resolved. Firstly, the direct adsorption of isolated TMPRSS2:ERG mRNA through their polyA tail maximized mRNA loading onto the electrode surface. This is crucial for highly-sensitive TMPRSS2:ERG detection without enzymatic reverse transcription and amplification steps. Furthermore, this adsorption mechanism is a rapid, simple, and cost- efficient process as compared to the conventional approach of using recognition and transduction layers in RNA biosensors; and does not require any tedious functionalization

164

Chapter 5.1 protocols underlying each step of the sensor fabrication or expensive reagents/labels. Secondly, the use of complementary capture probes to the gene fusion-junction ensured direct TMPRSS2:ERG isolation with high specificity, thus circumventing amplification bias/artefacts observed in conventional methods. Lastly, the use of magnetic beads for isolating captured TMPRSS2:ERG mRNA is an inexpensive, convenient and easily- automated procedure. Through this approach, we achieved TMPRSS2:ERG detection from as little as 10 PCa cells without any amplification or electrode surface modification. To demonstrate clinical utility, we have also successfully extended the application of our assay to urinary samples of PCa patients and our assay outcomes were consistent with the gold standard RT-PCR method.

EXPERIMENTAL SECTION Cell Culture DuCap and LnCap cells were kindly provided by Matthias Nees (VTT, Finland); Gregor Tevz (APCRC, Australia); and Michelle Hill (UQDI, Australia). The cell lines DuCap and LnCap were cultured in RPMI-1640 growth media (Life Technologies) supplemented with

10% fetal bovine serum (Life Technologies) in a humidified incubator containing 5% CO2 at 37ºC. Cell quantification was performed by first mixing an equal volume of 0.4% Trypan Blue (Sigma-Aldrich) to trypsinized cell suspension. Then, a sample mixture was loaded onto a haemocytometer for cell counting under an inverted phase contrast microscope. Cell number was diluted to the required amount for experiments accordingly and total RNA was extracted using Trizol® reagent (Life Technologies). RNA integrity and purity were checked using a NanoDrop Spectrophotometer (Thermo Scientific).

RNA Extraction from Clinical Urinary Specimens Ethics approval was obtained from The University of Queensland Institutional Human Research Ethics Committee (Approval No. 201400012) and informed consent was obtained from all subjects prior to sample collection. Methods pertaining to clinical samples were carried out in accordance with approved guidelines.

Voided urinary samples were collected from 17 men undergoing treatment for PCa and 4 healthy men with no clinical history of PCa at the Royal Brisbane and Women’s Hospital. Total RNA was extracted from the urine samples using the commercially-available ZR urine

165

Chapter 5.1

RNA isolation KitTM (Zymo Research). For each sample, 30 mL of urine was passed through the supplied ZRC GFTM Filter to retain urinary cells. Then, 700 µL of supplied urine RNA buffer was passed through the filter and the filtered cells were collected in the flow-through. Next, as per manufacturer’s instructions, the cells from the flow-through were lysed, washed and eluted in 10 µL of RNase-free water.

Probe Hybridization and Magnetic Isolation For probe hybridization, extracted total RNA from cell lines/ urinary samples were adjusted to required concentration (as required) in 5 µL of RNase-free water before mixing with 10 µL of 5x SSC buffer (pH 7) and 10 µL of 10 µM biotinylated capture probes (5’- CAACTGATAAGGCTTCCTGCCGCGCTCCAGG-Biotin -3’). *Underlined and bold are complementary to ERG and TMPRSS2 sequences respectively. The mixture was heated to 65 ºC for 2 min before being placed on a mixer for 60 min at room temperature to allow capture probe hybridization to TMPRSS2:ERG mRNA.

For magnetic isolation of TMPRSS2:ERG mRNA, 20 µL of streptavidin-labeled Dynabeads® MyOneTM Streptavidin C1 (Invitrogen) magnetic beads was firstly washed with 2x washing and binding (B&W) buffer (10 mM Tris-HCl, pH 7.5; 1 mM EDTA ; 2 M NaCl) and resuspended in 25 µL of 2x B&W buffer. Then, the resuspended magnetic beads were added to the prepared capture probes-TMPRSS2:ERG mRNA (as described above), mixed thoroughly, and incubated on a mixer for 30 min at room temperature to allow magnetic beads-labeling of capture probes. After which, the magnetic beads with attached capture probes-TMPRSS2:ERG mRNA were isolated with a magnet, washed twice with 2x B&W buffer and resuspended in 10 µL of RNase-free water.

Adsorption of Isolated mRNA on Gold Electrode Surface To release the captured TMPRSS2:ERG mRNA from the magnetic beads and capture probes, the magnetically-isolated TMPRSS2:ERG mRNA was heated for 2 min at 95°C, immediately placed on a magnet, and the supernatant containing the released TMPRSS2:ERG mRNA was collected. Then, 5 µL of this released mRNA was diluted to 30 µL in 5X SSC buffer (pH 7.0) (unless otherwise stated) and directly dropped onto the working electrode surface of a screen- printed gold electrode (Dropsens) and allowed to adsorb for 20 min (unless otherwise stated) with gentle shaking at room temperature. The electrodes were then washed with 10 mM

166

Chapter 5.1 phosphate buffer (137 mM sodium chloride, 2 mM potassium chloride, pH 7.4) before electrochemical measurements.

Electrochemical Detection All electrochemical measurements were performed on a CH1040C potentiostat (CH Instruments) with the three-electrode system (gold working and counter electrodes, silver reference electrode) on each screen-printed gold electrode. The electrolyte buffer consisted of 3- 4- 10 mM phosphate buffer solution (pH 7) containing 2.5 mM [Fe(CN)6] /[Fe(CN)6] (1:1) and 0.1 M KCl. Differential pulse voltammetry (DPV) signals were recorded from -0.1 V- 0.5 V with a pulse amplitude of 50 mV and a pulse width of 50 ms. The % current response change was calculated as follows:

% Current Response Change = [(iBare – iAdsorbed) / iBare] x 100 (1)

where iBare and iAdsorbed are current densities for bare electrode and electrode after sample adsorption respectively.

RT-PCR The KAPA SYBR® FAST One-Step qRT-PCR kit (KAPA BIOSYSTEMS) was used to set up a single reaction volume of 10 µl for each sample. Each reaction volume consist of 1X KAPA SYBR® FAST qPCR Master Mix, 200 nM of each forward (5’- CCTGGAGCGCGGCAGGAAGCCTTATCAGTTG-3’) and reverse (5’- GCTAGGGTTACATTCCATTTTGATGGTGAC-3’) primer, 1X KAPA RT Mix and 3 µl of input RNA target. The tubes were incubated at 42°C for 10 min to synthesize cDNA, followed by 95°C for 5 min to deactivate RT before cycling 35 times (95°C for 30 s, 69°C for 30 s and 72°C for 1 min) and finished with 72°C for 10 min. Lastly, the RT-PCR products were visualized on 1.5% agarose gel to verify amplification.

RESULTS AND DISCUSSION Principle of Amplification-free TMPRSS2:ERG Electrochemical assay The detection of the most common TMPRSS2:ERG fusion gene variant in approximately 50% of PCa cases,7,35 is rapidly gaining attention as a way to improve diagnosis and prognosis of early PCa.15,21,36 However, current methodologies such as amplification-based

167

Chapter 5.1 techniques are limited by the need for thermal cycling and enzymatic amplification to attain sufficient copy number level for detection.15,37 Therefore, by using the most frequent fusion between TMPRSS2 exon 1 and ERG exon 4 in PCa as a model to maximize clinical utility; we were motivated to design an amplification-free assay for detecting gene fusion mRNA from clinical urinary specimens with high detection sensitivity and specificity.

Our assay is made up of three crucial steps in order to realize electrochemical detection of TMPRSS2:ERG mRNA without enzymatic amplification: specific capture of TMPRSS2:ERG mRNA from sample using magnetic isolation followed by heat release of captured mRNA, and adsorption of isolated mRNA onto a bare gold electrode surface by mRNA-gold affinity interaction for electrochemical readout (Fig. 1). Firstly, we used biotinylated capture probes which are complementary to the unique TMPRSS2:ERG gene- fusion junction (Fig. 1a) for specific hybridization to TMPRSS2:ERG mRNA targets (Fig. 1b). Next, the hybridized mRNA targets are isolated from total RNA background and sequestered on the surface of streptavidin-coated magnetic beads through biotin-streptavidin interactions. Then, the TMPRSS2:ERG mRNA are magnetically purified and released from the capture probes through subsequent heating. Finally, the released TMPRSS2:ERG mRNA are allowed to adsorb onto the bare gold surface of screen-printed electrodes for 20 min (Fig. 1c). The adsorption process is facilitated by the high affinity between adenines of the TMPRSS2:ERG mRNA polyA tails and gold. For quantification of adsorbed TMMPRSS2:ERG mRNA, we optimized conditions for mRNA coverage on the electrode surface to be adequately low (i.e., partial blocking of electrode surface) for bulk ferricyanide ions to overcome coulombic repulsion (with negatively-charged mRNA); and approach the electrode surface to generate a detectable Faradaic current signal34. In the presence of adsorbed TMMPRSS2:ERG mRNA onto the electrode surface, coulombic repulsion of ferricyanide ions away from the surface resulted in a significantly lower signal as compared to a bare electrode30,31. This decrease in Faradaic current with respect to the baseline bare electrode current (% current response change) is also inversely proportional to the amount of adsorbed mRNA, thus allowing for quantitative TMPRSS2:ERG detection.

168

Chapter 5.1

Figure 1. Schematic representation of assay. (A) TMPRSS2:ERG mRNA resulted from the fusion between TMPRSS2 promoter sequence and ERG coding sequence. (B) Total RNA is firstly purified from urinary samples before specific capture of TMPRSS2:ERG mRNA transcripts by biotinylated capture probes and magnetic isolation with streptavidin-coated magnetic beads. (C) After magnetic purification, TMPRSS2:ERG mRNA are released from magnetic beads and allowed to adsorb onto the gold surface of screen-printed electrodes for electrochemical detection. The presence of adsorbed TMPRSS2:ERG mRNA will lead to higher coulombic repulsion of similarly-charged ferricyanide ions away from the electrode surface and lead to a decrease in electrochemical signal.

Assay Optimization The electrochemical signal level of our assay is dependent on the amount of mRNA being adsorbed on the electrode surface. The electrochemical signal of an electrode with low mRNA surface coverage is similar to that of a bare electrode as ferricyanide ions has easy access to the electrode surface (i.e., less coulombic repulsion) to generate a Faradaic current. Thus, to achieve excellent assay sensitivity and speed, it is ideal to maximize the amount of adsorbed mRNA on the electrode surface in the shortest timeframe possible. To this end, we optimized parameters such as adsorption time and pH, which are known to affect mRNA- gold adsorption kinetics.30-32 In these experiments, we used 3 ng (100 cells) of starting total RNA, an amount which we approximated to be within a urine sample. The starting total RNA

169

Chapter 5.1 was isolated from DuCap cells, a PCa cell line which carries the TMPRSS2:ERG fusion gene.

We firstly tested different adsorption times (5, 10, 15, 20, and 30 min) for the mRNA-gold adsorption process to take place before electrochemical readout. By using a clinically relevant amount of isolated total RNA from 100 cells, we found that a maximal current response change of approximately 50% was observed at 20 min and no significant increase (t- test P=0.098) in current response change was observed thereafter (Fig. 2a). This was probably due to maximal adsorption of TMPRSS2:ERG mRNA on the gold electrode surface at 20 min, thus leading to no signal improvement beyond this timepoint. Hence, we used a 20 min adsorption time for all further experiments.

Next, we optimized the adsorption buffer pH for the maximal current response change. Using 5 different buffer pH (pH 3, 5, 7, 9, and 11), we found that the optimal pH for generating maximal signal occurred at neutral pH 7 (Fig. 2b). The signal change level being generated improved with increasingly acidic pH up to pH 7 but subsequently decreased at even higher basic pH. We believe that the highest signal change was obtained at neutral pH because of electrode surface damage during electrochemical measurements at lower acidic pH, and on the other hand, mRNA is prone to degradation at higher basic pH.38 Therefore, the adsorption buffer pH was optimized at pH 7 for further experiments.

170

Chapter 5.1

Figure 2. Optimization of mRNA-gold adsorption. (A) Mean % current response change after 5, 10, 15, 20, 25, and 30 min of mRNA-gold adsorption time. (B) Mean % current response change using different buffer pH (3, 5, 7, 9, and 11) for mRNA-gold adsorption. Error bars represent standard deviation of three independent experiments.

Detection Sensitivity and Specificity To evaluate assay sensitivity, we performed detection on different amounts of total RNA isolated from known number of DuCap cells (0, 10, 100, 1000 and, 10 000 cells). Electrochemical measurements were taken following TMPRSS2:ERG mRNA capture, isolation, adsorption onto electrode surface. We observed that TMPRSS2:ERG mRNA (i.e. number of DuCap cells) was linearly correlated with % current response change (Fig. 3a) over the range of 0-10 000 cells, and the limit of detection (LOD) was established to be ~10 cells (300 pg of total RNA) (Fig. 3b). The observed electrochemical signal change using 10 DuCap cells is 15% and 25% higher (t-test P=0.030) than signals from no-template control (NoT) and bare electrode respectively. This LOD is achieved without amplification and

171

Chapter 5.1 highlights the high sensitivity of the electrochemical detection after rapid adsorption (20 min) of captured mRNA onto screen-printed gold electrode surface. Since tumor cells in the prostate tend to be shedded into the urine of PCa patients, this limit of detection of 10 cells (300 pg of total RNA) is adequate for the detection of TMPRSS2:ERG mRNA in patient urine samples. Our assay detection sensitivity is, to our knowledge, the highest for an electrochemical gene fusion assay. In literature, an electrochemical assay39 for TMPRSS2:ERG detection has previously demonstrated a detection sensitivity of 10 ng mRNA, a level which is readily available in tumor tissue but unlikely in urinary samples.

Figure 3. Sensitive of assay. (A) Differential pulse voltammograms corresponding to input total RNA from different numbers of DuCap cells. (B) Mean % current response change

172

Chapter 5.1 using input total RNA from different numbers of DuCap cells. Inset show the analogous linear calibration plot. Error bars represent standard deviation of three independent experiments.

To test for assay specificity, we targeted total RNA from LnCap, a negative control cell line which does not carry the TMPRSS2:ERG fusion gene. The resulting measurements showed a significant decrease in electrochemical response for RNA derived from 1000 DuCap cells, whilst RNA derived from a similar number of LnCap cells gave a level of electrochemical response similar to NoT (Fig. 4a). In quantitative terms, the electrochemical response change using DuCap RNA is 50% higher than that of NoT RNA (t-test P=0.001), but LnCap RNA is not significantly different from NoT background signals (t-test P=0.070) (Fig. 4b). This indicated that hybridization of capture probes to non-specific mRNA targets was negligible. Furthermore, we also verified the specific capturing and isolation of TMPRSS2:ERG mRNA by reverse-transcription amplification whereby specific TMPRSS2:ERG primers were used to amplify the isolated TMPRSS2:ERG mRNA after heat release from magnetic beads, and amplicons were later visualized by gel electrophoresis (Fig. 4c). The included NoT and no reverse transcriptase (No RT) controls verified that the amplification was dependent on RNA- based targets, whilst no capture probes and no heat release controls showed that the TMPRSS2:ERG mRNA targets were isolated by use of capture probes and released after heat treatment. This high specificity is largely due to the capture probe which was designed to hybridize across the fusion junction of TMPRSS2:ERG and helps minimize the likelihood of non-specific hybridization. The specific capture and release of mRNA targets is also essential for avoiding non-specific adsorption of non-target molecules (i.e. false-positive signals). Our assay clearly demonstrates specific TMPRSS2:ERG detection on cell lines, suggesting this assay could also be implemented in detection from complex biological specimens such as clinical urinary samples.

173

Chapter 5.1

Figure 4. Specificity of assay. (A) Differential pulse voltammograms corresponding to input total RNA from DuCap (TMPRSS2:ERG-positive) and LnCap (TMPRSS2:ERG-negative) cell lines. (B) Mean % current response change using input total RNA from DuCap and LnCap cell lines. Error bars represent standard deviation of three independent experiments. (C) RT-PCR of isolated TMPRSS2:ERG mRNA after heat release from beads with no- template (NoT), no reverse transcriptase (No RT), no capture probes, and no heat release controls.

174

Chapter 5.1

TMPRSS2:ERG Detection in Clinical Urinary Specimens To investigate the clinical application of our assay, we challenged our amplification-free electrochemical assay with the analysis of TMPRSS2:ERG levels in total RNA isolated from the clinical urinary samples of 12 metastatic castrate-resistant PCa patients (Patient 1-12), 5 PCa patients (Patient 13-17) with localized disease, and 4 healthy patient controls. Our assay found 12 of the samples to be TMPRSS2:ERG positive (Fig. 5a), and representative voltammograms of differing TMPRSS2:ERG levels from the clinical samples are shown in Fig. 5b. The 12 TMPRSS2:ERG-positive samples showed about 20-50% current response change as compared to bare electrodes, and the 5 TMPRSS2:ERG-negative samples (Patient 2, 11, 12, 15, 16 ) displayed <20% current response change (Fig. 5a).

Figure 5: Clinical applications of assay. (A) Mean % current responses corresponding to TMPRSS2:ERG detection in urinary samples (n = 21). Error bars represent standard deviation of three independent experiments. (B) Representative differential pulse voltammograms of TMPRSS2:ERG-positive and -negative patients, as well as healthy patient and NoT controls.

These results were also verified by current standard RT-PCR approach which showed the same 12 patients to be TMPRSS2:ERG-positive (Fig. 6). This is indicative that the electrochemical signals generated by our assay could discriminate TMPRSS2:ERG presence/absence in PCa urinary samples. With the healthy control samples, the electrochemical signals were equivalent to the background signal at <10% current response change, thus confirming assay specificity. The successful application of our assay to real clinical PCa samples (Fig. 5 and Fig. S1); and the subsequent validation of our assay results

175

Chapter 5.1 with RT-PCR (Fig. 6) as well as sequencing of RT-PCR amplicons (Fig. S2), suggested the viability of this technique for clinical use.

Figure 6. RT-PCR validation of TMPRSS2:ERG detection from urinary samples. RT-PCR validation of TMPRSS2:ERG status in urinary samples (n = 21). RT-PCR of housekeeping RN7SL1 transcripts were included as loading controls.

Nonetheless, we are aware that our current assay format is a promising technology demonstration, and application to a larger pool of patient samples is required to prove clinical utility. To our knowledge, our assay is the first amplification-free technique for non-invasive TMPRSS2:ERG detection. The clinical significance of our assay is broad as it is readily adaptable for detecting different gene fusion mRNA or any other RNA-based biomarkers by changing the capture probe sequences. As it has been indicated that a combination of TMPRSS2:ERG and other different PCa biomarkers could allow a diagnosis with better sensitivity and prognosis accuracy,12 the multiplexed detection of different PCa gene fusion biomarkers from an individual PCa patient could provide a more comprehensive diagnosis and personalized treatment plan.

Our developed assay is an inexpensive and amplification-free alternative to current amplification-based techniques for gene fusion detection (Fig. 1). Current RT-qPCR techniques mainly employ fluorescence readout for quantitative TMPRSS2:ERG detection.18,19 Fluorescence readout approaches are limited by higher assay cost, need for

176

Chapter 5.1 specialized instrument, and likelihood of high background fluorescence. Our data clearly demonstrates that our electrochemical readout could avoid these limitations as it provides rapid and quantitative electrochemical signals, and simply requires a potentiostat for measurements. Additionally, the mRNA-gold electrode surface adsorption mechanism prior to electrochemical detection made use of the natural strong affinity of adenine bases (i.e., polyA tails of captured mRNA) towards gold to bind isolated TMPRSS2:ERG mRNA onto the electrode surface. This simplistic approach is a convenient and rapid way for binding target mRNA sequences to gold electrode surface for electrochemical detection and has been previously demonstrated for DNA biosensing.26,30,31 Moreover, we performed our assay using screen-printed gold electrodes which are commercially-available and ensure highly reproducible results. On top of that, the low cost of these electrodes also enable one-time usage without need for time-consuming electrode cleaning. With this cost-effective electrode platform and without the need for tedious electrode surface modification and electrocatalytic labelling; our proposed approach is ideal for developing a clinically-friendly TMPRSS2:ERG assay.

CONCLUSIONS In conclusion, we have presented the first amplification-free electrochemical assay for gene fusion detection in PCa urinary samples. Our assay combines the use of magnetic TMPRSS2:ERG mRNA isolation, adsorption of isolated mRNA onto electrode surface by natural high binding affinity between adenine bases on mRNA polyA tails and gold, and rapid electrochemical detection of adsorbed mRNA using a ferricyanide system. Our proposed assay offers numerous advantages over existing methodologies. Most notably, it removes the need for enzymatic amplification, tedious experimental protocols and expensive fluorescence readout instruments; while also ensuring highly sensitive, specific, and non- invasive detection of TMPRSS2:ERG gene fusion, a more PCa-specific and potentially more superior alternative to current serum PSA biomarker. Furthermore, the clinical utility of this assay was demonstrated through successful TMPRSS2:ERG detection from PCa clinical urine samples. The analysis of urinary TMPRSS2:ERG in this assay offers a convenient and less invasive alternative to tissue biopsies, and could also be potentially used for monitoring drug responses and treatment effectiveness. Our assay is currently a proof-of-principle demonstration of TMPRSS2:ERG detection in PCa patients. For future work, we envisage it as a platform technology which, with slight modifications, could enable detection of other

177

Chapter 5.1 oncogenic gene fusion biomarkers. Furthermore, our assay could also be readily customised for detecting many other RNA-based cancer biomarkers such as microRNAs, splice variants, or non-coding RNAs detection. Finally, the simple workflow of our assay could allow for integration into many device-based biosensing applications such as microfluidic platforms. This may firstly reduce required sample volume, which could lead to faster hybridization kinetics (i.e., shorter assay time) within a smaller volume, could also enable high throughput and multiplexed detection of a selected panel of cancer biomarkers to characterize molecular subtypes.

REFERENCES (1) van 't Veer, L. J.; Bernards, R. Nature 2008, 452, 564-570. (2) Chin, L.; Andersen, J. N.; Futreal, P. A. Nat. Med. 2011, 17, 297-303. (3) Roychowdhury, S.; Iyer, M. K.; Robinson, D. R.; Lonigro, R. J.; Wu, Y.-M.; Cao, X.; Kalyana-Sundaram, S.; Sam, L.; Balbin, O. A.; Quist, M. J.; Barrette, T.; Everett, J.; Siddiqui, J.; Kunju, L. P.; Navone, N.; Araujo, J. C.; Troncoso, P.; Logothetis, C. J.; Innis, J. W.; Smith, D. C.; Lao, C. D.; Kim, S. Y.; Roberts, J. S.; Gruber, S. B.; Pienta, K. J.; Talpaz, M.; Chinnaiyan, A. M. Science Transl. Med. 2011, 3, 111ra121. (4) Soria, J. C.; Associate Editorial, T. Annals Oncol. 2014, 25, 5-6. (5) Schroeder, F. H.; Hugosson, J.; Roobol, M. J.; Tammela, T. L. J.; Ciatto, S.; Nelen, V.; Kwiatkowski, M.; Lujan, M.; Lilja, H.; Zappa, M.; Denis, L. J.; Recker, F.; Berenguer, A.; Maattanen, L.; Bangma, C. H.; Aus, G.; Villers, A.; Rebillard, X.; van der Kwast, T.; Blijenberg, B. G.; Moss, S. M.; de Koning, H. J.; Auvinen, A.; Investigators, E. N. Engl. J. Med. 2009, 360, 1320-1328. (6) Welch, H. G.; Albertsen, P. C. J. Natl. Cancer Inst. 2009, 101, 1325-1329. (7) Tomlins, S. A.; Rhodes, D. R.; Perner, S.; Dhanasekaran, S. M.; Mehra, R.; Sun, X. W.; Varambally, S.; Cao, X. H.; Tchinda, J.; Kuefer, R.; Lee, C.; Montie, J. E.; Shah, R. B.; Pienta, K. J.; Rubin, M. A.; Chinnaiyan, A. M. Science 2005, 310, 644-648. (8) Tognon, C.; Knezevich, S. R.; Huntsman, D.; Roskelley, C. D.; Melnyk, N.; Mathers, J. A.; Becker, L.; Carneiro, F.; MacPherson, N.; Horsman, D.; Poremba, C.; Sorensen, P. H. B. Cancer Cell 2002, 2, 367-376. (9) Soda, M.; Choi, Y. L.; Enomoto, M.; Takada, S.; Yamashita, Y.; Ishikawa, S.; Fujiwara, S.-i.; Watanabe, H.; Kurashina, K.; Hatanaka, H.; Bando, M.; Ohno, S.; Ishikawa, Y.; Aburatani, H.; Niki, T.; Sohara, Y.; Sugiyama, Y.; Mano, H. Nature 2007, 448, 561-U563.

178

Chapter 5.1

(10) Cerveira, N.; Ribeiro, F. R.; Peixoto, A.; Costa, V.; Henrique, R.; Jeronimo, C.; Teixeira, M. R. Neoplasia 2006, 8, 826-832. (11) Mitelman, F.; Johansson, B.; Mertens, F. Nat. Rev. Cancer 2007, 7, 233-245. (12) Wang, J. H.; Cai, Y.; Ren, C. X.; Ittmann, M. Cancer Res. 2006, 66, 8347-8351. (13) Demichelis, F.; Rubin, M. A. J. Clin. Pathol. 2007, 60, 1185-1186. (14) Attard, G.; Clark, J.; Ambroisine, L.; Fisher, G.; Kovacs, G.; Flohr, P.; Berney, D.; Foster, C. S.; Fletcher, A.; Gerald, W. L.; Moller, H.; Reuter, V.; De Bono, J. S.; Scardino, P.; Cuzick, J.; Cooper, C. S.; Transatlantic Prostate, G. Oncogene 2008, 27, 253-263. (15) Tomlins, S. A.; Day, J. R.; Lonigro, R. J.; Hovelson, D. H.; Siddiqui, J.; Kunju, L. P.; Dunn, R. L.; Meyer, S.; Hodge, P.; Groskopf, J.; Wei, J. T.; Chinnaiyan, A. M. Eur. Urol. 2015, doi:10.1016. (16) Soller, M. J.; Elfving, P.; Lundgren, R.; Panagopols, I. Genes Chromosomes Cancer 2006, 45, 717-719. (17) Laxman, B.; Tomlins, S. A.; Mehra, R.; Morris, D. S.; Wang, L.; Helgeson, B. E.; Shah, R. B.; Rubin, M. A.; Wei, J. T.; Chinnaiyan, A. M. Neoplasia 2006, 8, 885-888. (18) Laxman, B.; Morris, D. S.; Yu, J. J.; Siddiqui, J.; Cao, J.; Mehra, R.; Lonigro, R. J.; Tsodikov, A.; Wei, J. T.; Tomlins, S. A.; Chinnaiyan, A. M. Cancer Res. 2008, 68, 645-649. (19) Hessels, D.; Smit, F. P.; Verhaegh, G. W.; Witjes, J. A.; Cornel, E. B.; Schalken, J. A. Clin. Cancer Res. 2007, 13, 5103-5108. (20) Cornu, J.-N.; Cancel-Tassin, G.; Egrot, C.; Gaffory, C.; Haab, F.; Cussenot, O. Prostate 2013, 73, 242-249. (21) Tomlins, S. A. Eur. Urol. 2014, 65, 543-545. (22) Park, K.; Dalton, J. T.; Narayanan, R.; Barbieri, C. E.; Hancock, M. L.; Bostwick, D. G.; Steiner, M. S.; Rubin, M. A. J. Clin. Oncol. 2014, 32, 206-211. (23) Nilsson, J.; Skog, J.; Nordstrand, A.; Baranov, V.; Mincheva-Nilsson, L.; Breakefield, X. O.; Widmark, A. Br. J. Cancer 2009, 100, 1603-1607. (24) Tomlins, S. A.; Aubin, S. M. J.; Siddiqui, J.; Lonigro, R. J.; Sefton-Miller, L.; Miick, S.; Williamsen, S.; Hodge, P.; Meinke, J.; Blase, A.; Penabella, Y.; Day, J. R.; Varambally, R.; Han, B.; Wood, D.; Wang, L.; Sanda, M. G.; Rubin, M. A.; Rhodes, D. R.; Hollenbeck, B.; Sakamoto, K.; Silberstein, J. L.; Fradet, Y.; Amberson, J. B.; Meyers, S.; Palanisamy, N.; Rittenhouse, H.; Wei, J. T.; Groskopf, J.; Chinnaiyan, A. M. Science Transl. Med. 2011, 3, 94ra72.

179

Chapter 5.1

(25) Lin, D. W.; Newcomb, L. F.; Brown, E. C.; Brooks, J. D.; Carroll, P. R.; Feng, Z.; Gleave, M. E.; Lance, R. S.; Sanda, M. G.; Thompson, I. M.; Wei, J. T.; Nelson, P. S.; Canary Prostate Active, S. Clin. Cancer Res. 2013, 19, 2442-2450. (26) Koo, K. M.; Sina, A. A. I.; Carrascosa, L. G.; Shiddiky, M. J. A.; Trau, M. Anal. Methods 2015, 7, 7042-7054. (27) Kimura-Suda, H.; Petrovykh, D. Y.; Tarlov, M. J.; Whitman, L. J. J. Am. Chem. Soc. 2003, 125, 9014-9015. (28) Opdahl, A.; Petrovykh, D. Y.; Kimura-Suda, H.; Tarlov, M. J.; Whitman, L. J. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 9-14. (29) Schreiner, S. M.; Shudy, D. F.; Hatch, A. L.; Opdahl, A.; Whitman, L. J.; Petrovykh, D. Y. Anal. Chem 2010, 82, 2803-2810. (30) Sina, A. A. I.; Howell, S.; Carrascosa, L. G.; Rauf, S.; Shiddiky, M. J. A.; Trau, M. Chem. Commun. 2014, 50, 13153-13156. (31) Koo, K. M.; Sina, A. A. I.; Carrascosa, L. G.; Shiddiky, M. J. A.; Trau, M. Analyst 2014, 139, 6178-6184. (32) Sina, A. A. I.; Carrascosa, L. G.; Palanisamy, R.; Rauf, S.; Shiddiky, M. J. A.; Trau, M. Anal. Chem 2014, 86, 10179-10185. (33) Koo, K. M.; Carrascosa, L. G.; Shiddiky, M. J. A.; Trau, M. Anal. Chem. 2016, 88, 2000-2005. (34) Zhang, J.; Wang, L.; Pan, D.; Song, S.; Fan, C. Chem. Commun. 2007, 1154-1156. (35) Prensner, J. R.; Chinnaiyan, A. M. Curr. Opin. Genet. Dev. 2009, 19, 82-91. (36) Tomlins, S. A.; Bjartell, A.; Chinnaiyan, A. M.; Jenster, G.; Nam, R. K.; Rubin, M. A.; Schalken, J. A. Eur. Urol. 2009, 56, 275-286. (37) Leyten, G.; Hessels, D.; Jannink, S. A.; Smit, F. P.; de Jong, H.; Cornel, E. B.; de Reijke, T. M.; Vergunst, H.; Kil, P.; Knipscheer, B. C.; van Oort, I. M.; Mulders, P. F. A.; Hulsbergen-van de Kaa, C. A.; Schalken, J. A. Eur. Urol. 2014, 65, 534-542. (38) Li, Y. F.; Breaker, R. R. J. Am. Chem. Soc. 1999, 121, 5364-5372. (39) Fang, Z.; Soleymani, L.; Pampalakis, G.; Yoshimoto, M.; Squire, J. A.; Sargent, E. H.; Kelley, S. O. ACS Nano 2009, 3, 3207-3213.

180

Chapter 5.1

SUPPLEMENTARY INFORMATION

Figure S1. Representative Sanger sequencing of RT-PCR products from a TMPRSS2:ERG- positive urinary sample. *The TMPRSS2 sequence bases are underlined to indicate the TMPRSS2:ERG fusion junction.

181

Chapter 5.2

Poly(A) Extensions of miRNAs for Amplification-free Electrochemical Detection on Screen-Printed Gold Electrodes

Kevin M. Koo, Laura G. Carrascosa, Muhammad J. A. Shiddiky, and Matt Trau

ABSTRACT Current amplification-based microRNA (miRNA) detection approaches are limited by the small sizes of miRNAs as well as amplification bias/artifacts. Herein, we report on an amplification-free miRNA assay based on elevated affinity interaction between poly- adenylated miRNA and bare gold electrode. The poly(A) extension on 3’ ends of magnetically-isolated miRNA targets facilitated high adsorption efficiency onto gold electrode surfaces for electrochemical detection without any cumbersome electrode surface functionalization procedures. The assay showed excellent detection sensitivity (10 fM) and specificity, and was demonstrated for quantitative miR-107 detection in human cancer cell lines and clinical urine samples. We believe our assay could be useful as an amplification- free alternative for miRNA detection.

INTRODUCTION MicroRNAs (miRNAs) are short (~22 nucleotides) non-coding RNA molecules which are rapidly gaining interest as cancer biomarkers due to the crucial roles they play in cancer initiation and progression.1-3 In recent years, research has shown that miRNAs post- transcriptionally regulate gene expression mainly by binding to the 3’ untranslated region of targeted messenger RNAs (mRNAs) to inhibit translation, functioning as oncogenes and tumor suppressors.4 Besides showing promise as diagnostic and prognostic biomarkers in cancer, miRNAs are also attractive as non-invasive biomarkers with high stability in bodily fluids such as blood, urine, or saliva.5, 6 Thus, in order to implement miRNA biomarkers for use in clinical practice, it is essential to develop methodologies for quantitative miRNA detection with high sensitivity and specificity.

Quantitative reverse-transcription polymerase chain reaction (RT-qPCR) is the current gold standard for miRNA quantification.7, 8 The amplification-based RT-qPCR technique offers

182

Chapter 5.2 good sensitivity and specificity for miRNA detection. Yet, miRNAs still pose some challenging issues for amplification-based techniques due to their short lengths which is about the same size of conventional PCR primers. Furthermore, amplification-based methods have drawbacks such as errors due to amplification bias/artifacts; and the fluorescence readout also requires costly instruments, fluorescent labels, and is susceptible to background fluorescence interference. Therefore, an amplification-free miRNA detection methodology could represent an appealing alternative to alleviate these issues.

Electrochemical assays have been shown to offer excellent sensitivity and specificity for miRNA detection without any prior amplification process.9-12 Additionally, electrochemical assays provide rapid readout, and the required detection equipment and reagents are inexpensive. Typically, to enable electrochemical detection of miRNAs, gold electrode surfaces are functionalized with probes for specific hybridization with target miRNAs. To remove this tedious functionalizing process, different research groups have exploited the highly sequence-dependent nucleic acid-bare gold adsorption affinity trend of A > C ≥ G > T (i.e., adenine bases adsorb to gold most strongly and quickly)13, 14 for direct and efficient adsorption of poly-adenylated nucleic acid probes onto bare gold electrode surface.15 We therefore hypothesized that by modifying target miRNAs (i.e. nucleic acid sequences) with poly(A) sequence extensions, we could facilitate rapid and high miRNA-gold electrode surface adsorption for subsequent detection. This methodology would allow for electrochemical miRNA detection without any prior miRNA amplification or cumbersome electrode surface functionalization steps involved in fabricating recognition and transduction layers in conventional electrochemical miRNA biosensors. The adsorbed miRNAs could then 3-/4- 16, 17 be electrochemically detected with a [Fe(CN)6] redox system. Unlike conventional 3+ 3- 11, 18 redox system (e.g., [Ru(NH3)6] /[Fe(CN)6] ) , this system alone is sufficient for quantifying surface-bound nucleic acid sequences reliably.14, 16, 17, 19 As outlined by Zhang and co-workers19, it follows an electron transfer kinetics-based mechanism, where density of the adsorbed nucleic acid sequences at the electrode surface should be sufficiently low (i.e., partial blocking).

Herein we describe a novel approach for direct amplification-free detection of miRNA by firstly magnetically isolating target miRNAs, then conveniently extending target miRNAs with poly(A) sequences to enable rapid adsorption onto a gold electrode surface for electrochemical quantification. The practical application of the assay was showcased through

183

Chapter 5.2 the quantitative detection of a specific miRNA sequence in different breast and prostate cancer cell lines and assay results were validated with gold-standard RT-qPCR method. We also successfully detected target miRNAs in prostate cancer urine samples.

EXPERIMENTAL SECTION Isolation of Target miRNAs Total RNA was extracted from cancer cell lines (MDA-MB-231, MCF7, and 22Rv1 using Direct-zolTM kit (Zymo Research, USA), and target miRNAs were subsequently hybridized onto complementary capture probes-functionalized magnetic beads. Then, hybridized targets were washed, heat-released, and resuspended in RNA-free water. (See Supplementary Information for more details)

Electrochemical Detection of Target miRNAs Isolated target miRNAs were extended with poly(A) sequences separately within a tube before adsorption on screen-printed gold electrode surface. The adsorbed miRNAs were then detected by differential pulse voltammetry (DPV). (See Supplementary Information for more details)

RESULTS AND DISCUSSION Assay Principle Our novel assay for amplification-free electrochemical detection of miRNA is schematically depicted in Figure 1. miRNA targets are firstly specifically hybridized to complementary capture probes which have been functionalized onto the surfaces of streptavidin-coated magnetic beads through biotin-streptavidin interactions. Next, hybridized miRNA targets are magnetically purified and released from the capture probes through heating. The miRNA targets are then subjected to poly(A) extension on 3’ ends using polymerase A enzyme in order to increase miRNA-gold adsorption efficiency. After poly(A) extension, the miRNA targets are able to adsorb on the gold surface of screen-printed electrodes effectively due to high affinity between adenines of the polyA sequences and gold. Under optimized conditions, the coulombic repulsion between negatively-charged ferricyanide ions in the buffer and negatively-charged miRNA phosphate groups on the electrode surface will be sufficiently low for ferricyanide ions to overcome and diffuse on the electrode surface to generate a

184

Chapter 5.2 detectable Faradaic current signal which is significantly lower than the signal on a bare electrode.16, 17, 19

Figure 1. Schematic representation of assay. Magnetic beads-coupled capture probe sequences were firstly used for specific hybridization to miRNA targets. After hybridization, the miRNA targets were purified by magnetic separation before being released from capture probes through heating. Next, poly(A) extension was performed on the 3’ ends of released miRNA targets using polymerase A enzyme prior to adsorption onto the gold surface of screen-printed electrodes. The presence of adsorbed miRNAs leads to coulombic repulsion of ferricyanide ions away from the electrode surface, thus leading to a lower Faradaic current relative to a bare electrode. Quantitative miRNA detection is achieved as higher/lower amount of starting miRNA targets will lead to increased/decreased miRNA adsorption on electrode surface and in turn, lower/higher electrochemical signal.

Gold Surface Characterization Before and After miRNA Adsorption We used Fourier transform infrared spectroscopy (FTIR) to characterize the miRNA- and poly(A) extended miRNA-modified gold electrode surfaces. After miRNA adsorption, we observed FTIR spectra band changes generated from a bare gold surface (Figure S1). These strong bands (1418, 1655 cm-1)20 were characteristic signatures of C-H; C=O; C=N; N-H bonds in nucleic acids. In addition, poly(A) extended-miRNA adsorption resulted in enhanced absorbance at band positions (1553, 1603, 1639 cm-1)13 characteristic of adenine

185

Chapter 5.2 bases. The results showed the successful adsorption of both unmodified and poly(A) miRNAs on a bare gold surface.

Detection Sensitivity Before and After miRNA Poly(A) Extension To evaluate the effect of poly(A) extension on miRNA adsorption efficiency, we determined the detection sensitivity of our assay before and after poly(A) extension of isolated targeted miRNA. The sequence of miR-107 (Table S1) was used as a model target miRNA in our experiments. In previous studies, miR-107 overexpression has been associated with increased tumor growth and metastatic potential in cancer cells by regulating gene expression.21-23 Moreover, higher miR-107 expression level has also been detected in the urine and blood serum samples of cancer patients,24, 25 and has been suggested as a potential marker for cancer diagnosis and staging.

To test for assay detection sensitivity, a series of synthetic target miR-107 of different concentrations was detected. We observed an increasing relative DPV signal change (% i) with higher starting concentration of target miRNA (Figure 2) after 10 min adsorption time. This was attributed to the higher amount of target miRNA being isolated and subsequently adsorbed into the electrode surface. A higher amount of adsorbed miRNA caused higher 3- coulombic repulsion of [Fe(CN)6] ions from approaching the electrode surface, and resulted in a lower Faradaic current. Detection sensitivities were determined to be 1 pM (% i = 10%) (Figure 2A) and 10 fM (% i = 12%) (Figure 2B) before and after miRNA poly(A) extension respectively. The 100-fold improvement in detection sensitivity was attained due to poly(A) extension conferring higher adsorption affinity interaction between miRNA and gold surface as adenines have highest affinity interactions with gold amongst all nucleic acid bases.13 Thus, this resulted in an increased amount of adsorbed miRNA in the same amount of time as compared to miRNA without poly(A) extension. In addition, the negligible signal response from the no-template (NoT) control (% i = 3%) showed that electrochemical signals were mainly from adsorption of isolated miR-107 targets and unaffected by other molecules present in the adsorption mixture. Our detection limit of 10 fM with a linear range of 5 fM-5 pM is also superior or comparable with those of existing amplification-free miRNA electrochemical assays described in literature.11, 26-28 Most electrochemical miRNA sensing strategies29 developed in recent years achieved a fM detection sensitivity which is similar to our proposed assay. However, it is worthy to highlight the advantage of enhanced simplicity of our assay as compared to these assays. Our miRNA detection methodology does not

186

Chapter 5.2 require any tedious electrode surface functionalization; and utilized commercially-available screen printed electrodes which are inexpensive, disposable and provided good assay reproducibility with a RSD of 4.8% (n = 3).

Figure 2. Improved assay sensitivity after poly(A) extension. Differential pulse voltammetry (DPV) showing signal changes to different concentrations of starting miR-107 targets before (A1) and after (B1) poly(A) extension of isolated miRNA targets. % i (Relative DPV signal change) corresponding to different starting miR-107 target concentrations before (A2) and after (B2) poly(A) extension of isolated miRNA targets. Insets show the analogous linear calibration plots. Error bars represent standard deviation of three independent experiments.

Detection Specificity To evaluate the specificity of the capture probes in our assay for selectively isolating miR- 107 targets, we performed our assay on two unrelated and non-complementary miRNAs (miR-429 and miR-200c) and compared the obtained signals to that of miR-107 at the same

187

Chapter 5.2 assay starting miRNA concentration. We used an excessive starting miRNA concentration of 10 pM in order to test the capture probe specificity under saturating target condition. We found that electrochemical signals from experiments involving miR-429 (% i = 4.5%) and miR-200c (% i = 3.5%) were similar to the NoT control (Figure 3A). In contrast, electrochemical signal from the detection of miR-107 was about 10-fold higher (% i = 50%) than the signals of these non-complementary miRNA sequences (Figure 3B). These experiments demonstrated the high specificity of our capture probes in isolating miR-107 targets for miRNA-gold adsorption and subsequent electrochemical detection.

Figure 3. Assay specificity. (A) DPV showing signal changes after assay was performed on different starting miRNA sequences (miR-107; miR-429; miR-200c, 10 pM). (B) % i (Relative DPV signal change) corresponding to the different starting miRNA sequences. Error bars represent standard deviation of three independent experiments.

188

Chapter 5.2

Detection of miR-107 Levels in Cancer Cell Lines and Patient Samples After establishing the assay sensitivity and specificity towards miR-107 targets, we then challenged the assay with total RNA extracted from cultured human breast (MDA-MB-231 and MCF-7) and prostate cancer (22Rv1) cells to investigate assay performance with more complex biological samples. According to our assay results, miR-107 levels were expressed at varying levels in the different cell lines (Figure 4A). With the same amount of starting total RNA, our assay detected miR-107 to be lower-expressed in MCF7 cells, and 1.5-fold and 3.5-fold higher in MDA-MB-231 and 22Rv1 cells respectively (Figure 4B). Our assay results were in good agreement with previous findings in literature which also observed similar miR- 107 expression level patterns in these cell lines.21, 23 RT-qPCR was also performed as gold- standard assay validation and the miR-107 levels in all three cell lines were observed to be concordant with results of our assay (Figure 4C). In addition, we also quantitatively detected miR-107 levels in extracted total RNA from three clinical prostate cancer urine samples. As shown in Figures 4D and S2, the DPV signals from these patient samples were significantly higher than the NoT control sample. These results clearly indicated that our proposed assay could be applied for miR-107 detection in clinical samples. As miR-107 overexpression is linked to increased metastatic potential of cancer cells and tumor progression21, 23, our assay may aid in detecting miR-107 expression levels in cancer patients for diagnostic purpose or monitoring the treatment response of drugs targeting miR-107.

189

Chapter 5.2

Figure 4. Assay performance on human cancer cell lines and patient samples. (A) DPV showing signal changes for detection of different miR-107 expression levels in breast (MDA- MB-231; MCF7) and prostate (22Rv1) cancer cell lines. (B) % i (Relative DPV signal change) corresponding to miR-107 levels in the three different human cancer cell lines. (C) RT-qPCR validation of miR-107 expression levels in the three different human cancer cell lines. (D) % i (Relative DPV signal change) corresponding to miR-107 levels in three prostate cancer urinary samples. Error bars represent standard deviation of three independent experiments.

CONCLUSIONS In conclusion, the electrochemical assay described herein displayed excellent sensitivity and specificity for quantitative miRNA detection. We detected miR-107 levels in human cancer cells and urine samples without any amplification process, electrode surface functionalization, and costly specialized equipment. The high sensitivity (10 fM) was enabled by poly(A) extension of specifically-isolated miR-107 targets for better miRNA-gold adsorption efficiency. The electrochemical readout which was performed using screen- printed gold electrodes; paves the way for a rapid, quantitative, and inexpensive miR-107 detection that could be ideal for clinical diagnostics. We envisaged that the assay could also be readily extended to other miRNA markers through adaptation of capture probe sequences.

190

Chapter 5.2

REFERENCES (1) Garzon, R.; Calin, G. A.; Croce, C. M. Ann. Rev. Med. 2009, 60, 167-179. (2) Kong, Y. W.; Ferland-McCollough, D.; Jackson, T. J.; Bushell, M. Lancet Oncol. 2012, 13, E249-E258. (3) Hayes, J.; Peruzzi, P. P.; Lawler, S. Trends Mol. Med. 2014, 20, 460-469. (4) Krol, J.; Loedige, I.; Filipowicz, W. Nat. Rev. Genet. 2010, 11, 597-610. (5) Weber, J. A.; Baxter, D. H.; Zhang, S.; Huang, D. Y.; Huang, K. H.; Lee, M. J.; Galas, D. J.; Wang, K. Clin. Chem. 2010, 56, 1733-1741. (6) Mlcochova, H.; Hezova, R.; Stanik, M.; Slaby, O. Urol. Oncol.-Seminars and Original Invest. 2014, 32, 41.e1–41.e9. (7) Schmittgen, T. D.; Jiang, J. M.; Liu, Q.; Yang, L. Q. Nucleic Acids Res. 2004, 32, e43. (8) Chen, C. F.; Ridzon, D. A.; Broomer, A. J.; Zhou, Z. H.; Lee, D. H.; Nguyen, J. T.; Barbisin, M.; Xu, N. L.; Mahuvakar, V. R.; Andersen, M. R.; Lao, K. Q.; Livak, K. J.; Guegler, K. J. Nucleic Acids Res. 2005, 33, e179. (9) Bettazzi, F.; Hamid-Asl, E.; Esposito, C. L.; Quintavalle, C.; Formisano, N.; Laschi, S.; Catuogno, S.; Iaboni, M.; Marrazza, G.; Mascini, M.; Cerchia, L.; De Franciscis, V.; Condorelli, G.; Palchetti, I. Anal. Bioanal. Chem. 2013, 405, 1025-1034. (10) Yin, H.; Zhou, Y.; Zhang, H.; Meng, X.; Ai, S. Biosens. Bioelectron. 2012, 33, 247-253. (11) Yang, H.; Hui, A.; Pampalakis, G.; Soleymani, L.; Liu, F.-F.; Sargent, E. H.; Kelley, S. O. Angew. Chem.-Int. Edit. 2009, 48, 8461-8464. (12) Tavallaie, R.; De Almeida, S. R. M.; Gooding, J. J. Wiley Interdisciplinary Rev.- Nanomed. Nanobiotech. 2015, 7, 580-592. (13) Kimura-Suda, H.; Petrovykh, D. Y.; Tarlov, M. J.; Whitman, L. J. J. Am. Chem. Soc. 2003, 125, 9014-9015. (14) Koo, K. M.; Sina, A. A. I.; Carrascosa, L. G.; Shiddiky, M. J. A.; Trau, M. Anal. Methods 2015, 7, 7042-7054. (15) Li, H. X.; Rothberg, L. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 14036-14039. (16) Sina, A. A. I.; Howell, S.; Carrascosa, L. G.; Rauf, S.; Shiddiky, M. J. A.; Trau, M. Chem. Commun. 2014, 50, 13153-13156. (17) Koo, K. M.; Sina, A. A. I.; Carrascosa, L. G.; Shiddiky, M. J. A.; Trau, M. Analyst 2014, 139, 6178-6184. (18) Lapierre, M. A.; O'Keefe, M.; Taft, B. J.; Kelley, S. O. Anal. Chem. 2003, 75, 6327- 6333.

191

Chapter 5.2

(19) Zhang, J.; Wang, L.; Pan, D.; Song, S.; Fan, C. Chem. Commun. 2007, 1154-1156. (20) Dovbeshko, G. I.; Chegel, V. I.; Gridina, N. Y.; Repnytska, O. P.; Shirshov, Y. M.; Tryndiak, V. P.; Todor, I. M.; Solyanik, C. I. Biopolymers 2002, 67, 470-486. (21) Martello, G.; Rosato, A.; Ferrari, F.; Manfrin, A.; Cordenonsi, M.; Dupont, S.; Enzo, E.; Guzzardo, V.; Rondina, M.; Spruce, T.; Parenti, A. R.; Daidone, M. G.; Bicciato, S.; Piccolo, S. Cell 2010, 141, 1195-1207. (22) Wang, W.-X.; Kyprianou, N.; Wang, X.; Nelson, P. T. Cancer Res. 2010, 70, 9137- 9142. (23) Chen, P.-S.; Su, J.-L.; Cha, S.-T.; Tarn, W.-Y.; Wang, M.-Y.; Hsu, H.-C.; Lin, M.-T.; Chu, C.-Y.; Hua, K.-T.; Chen, C.-N.; Kuo, T.-C.; Chang, K.-J.; Hsiao, M.; Chang, Y.-W.; Chen, J.-S.; Yang, P.-C.; Kuo, M.-L. J. Clin. Invest. 2011, 121, 3442-3455. (24) Lodes, M. J.; Caraballo, M.; Suciu, D.; Munro, S.; Kumar, A.; Anderson, B. PLoS One 2009, 4, e6229. (25) Bryant, R. J.; Pawlowski, T.; Catto, J. W. F.; Marsden, G.; Vessella, R. L.; Rhees, B.; Kuslich, C.; Visakorpi, T.; Hamdy, F. C. Br. J. Cancer 2012, 106, 768-774. (26) Wu, X.; Chai, Y.; Yuan, R.; Zhuo, Y.; Chen, Y. Sens. Actuators B Chem. 2014, 203, 296-302. (27) Yin, H.; Zhou, Y.; Chen, C.; Zhu, L.; Ai, S. Analyst 2012, 137, 1389-1395. (28) Gao, Z.; Peng, Y. Biosens. Bioelectron. 2011, 26, 3768-3773. (29) Xia, N.; Zhang, L. P. Nanomaterials-Based Sensing Strategies for Electrochemical Detection of MicroRNAs. Materials 2014, 7, 5366-5384.

SUPPLEMENTARY INFORMATION

DETAILED EXPERIMENTAL SECTION Reagents and Materials All reagents were of analytical grade and purchased from Sigma Aldrich (Australia). UltraPureTM DNase/RNase-free distilled water (Invitrogen, Australia) was used throughout the experiments. Oligonucleotides were purchased from Integrated DNA Technologies (USA) and sequences are shown in Table S1. Screen-printed gold electrodes, DRP-C220BT (diameter ¼ 4mm), were acquired from Dropsens (Spain).

192

Chapter 5.2

Fourier Transform Infrared Spectroscopy (FTIR) for Surface Characterization FTIR experiments were performed on the Nicolet 5700 FTIR instrument (Thermo Fisher Scientific, USA). Spectra were collected from 50 scans in the 1400-1800 cm-1 region at 2 cm- 1 resolution. The measurements were done on a bare gold surface, and gold surfaces after adsorption of unmodified and poly(A)-extended synthetic miRNA (miR-107) sequences (Table S1).

RNA Extraction from Cell Lines MDA-MB-231 and MCF7 breast cancer cell lines were purchased from ATCC (USA), and 22Rv1 prostate cancer cell line was kindly donated by Colleen Nelson (APCRC, Australia). The cell lines were cultured in RPMI-1640 growth media (Life Technologies, Australia) supplemented with 10% fetal bovine serum (Life Technologies, Australia) in a humidified incubator containing 5% CO2 at 37ºC. RNA was extracted using Direct-zolTM kit (Zymo Research, USA) and RNA integrity and purity were checked using a NanoDrop Spectrophotometer (Thermo Scientific, Australia).

RNA Extraction from Clinical Urine Samples For clinical samples, ethics approval was obtained from The University of Queensland Institutional Human Research Ethics Committee (Approval No. 201400012) and informed consent was obtained from all subjects prior to sample collection. Methods pertaining to clinical samples were carried out in accordance with approved guidelines. Voided urinary samples were collected from 3 men patients undergoing treatment for prostate cancer. Total RNA was extracted from the urine samples using the commercially-available ZR urine RNA isolation KitTM (Zymo Research, USA). For each sample, 30 mL of urine was passed through the supplied ZRC GFTM Filter to retain urinary cells. Then, 700 µL of supplied urine RNA buffer was passed through the filter and the filtered cells were collected in the flow-through. Next, as per manufacturer’s instructions, the cells from the flow-through were lysed, washed and eluted in 10 µL of RNase-free water. miRNA Capture and Magnetic Isolation Firstly, 20 µL of streptavidin-labeled Dynabeads® MyOneTM Streptavidin C1 (Invitrogen, Australia) magnetic beads was washed with 2x washing and binding (B&W) buffer (10 mM Tris-HCl, pH 7.5; 1 mM EDTA ; 2 M NaCl) and resuspended in 25 µL of 2x B&W buffer. Next, an equal volume of 10 µM biotinylated capture probes (Table S1) was added and the

193

Chapter 5.2 resultant mixture was incubated on a mixer for 15 min at room temperature to generate capture probes-functionalized magnetic beads. The functionalized magnetic beads were washed thrice with 2x B&W buffer and resuspended in 10 µL of 5x SSC buffer (pH 7) before use.

For specific miRNA capture, 50 ng of extracted total RNA from cell lines or synthetic miR- 107 sequences (Table S1) of various concentrations were adjusted to 10 µL with RNase-free water before mixing with 10 µL of capture probes-functionalized magnetic beads. The mixture was placed on a mixer for 20 min at room temperature to allow capture probe hybridization to miRNA targets. After this step, the magnetic beads with attached miRNA targets were isolated with a magnet, washed twice with 2x B&W buffer and resuspended in 7.5 µL of RNase-free water.

Table S1. Oligonucleotide sequences used in experiments. Oligos 5'-Sequence-3'

Biotinylated miR-107 TGATAGCCCTGTACAATGCTGCT-Biotin Capture Probe

Synthetic miR-107 AGCAGCAUUGUACAGGGCUAUCA Sequence

RT-qRCR Forward AGCAGCATTGTACAG Primer

RT-qRCR Reverse TTTTTTTTTTTTTTTTTG Primer

Poly(A) Extension and Adsorption of Isolated miRNA on Gold Electrode Surface To release the captured miRNA targets from the magnetic beads and capture probes, the resuspended miRNA target mixture was heated for 2 min at 95°C, immediately placed on a

194

Chapter 5.2 magnet, and the supernatant containing the released miRNA targets was collected. Then, to extend poly(A) sequences on the released miRNA targets, 7.5 µL of released miRNA was added to 1 µL of 10x E. Coli Poly(A) polymerase reaction buffer (250 mM NaCl, 50 mM

Tris-HCl, 10 mM MgCl2, pH 7.9), 1 mM ATP, and 5 U E. Coli Poly(A) polymerase (New England BioLabs, Australia) to make up a 10 µL reaction mixture. The reaction mixture was incubated at 37°C for 15 min before being diluted to 30 µL in 5X SSC buffer (pH 7.0) and then directly dropped onto the working electrode surface of a screen-printed gold electrode (Dropsens, Spain) for 20 min of miRNA-gold adsorption with gentle shaking at room temperature. The electrodes were then washed with 10 mM phosphate buffer (137 mM sodium chloride, 2 mM potassium chloride, pH 7.4) before electrochemical measurements.

Electrochemical Detection All electrochemical measurements were performed on a CH1040C potentiostat (CH Instruments, USA) with the three-electrode system (gold working and counter electrodes, silver reference electrode) on each screen-printed gold electrode. The electrolyte buffer 3- consisted of 10 mM phosphate buffer solution (pH 7) containing 2.5 mM [Fe(CN)6] 4- /[Fe(CN)6] (1:1) and 0.1 M KCl. Differential pulse voltammetry (DPV) signals were recorded from -0.1 V- 0.5 V with a pulse amplitude of 50 mV and a pulse width of 50 ms. The % current response change was obtained by calculated as follows:

(% i) = [(iBare – iAdsorbed) / iBare] x 100

where iBare and iAdsorbed are current densities for bare electrode and electrode after sample adsorption respectively.

RT-qPCR The KAPA SYBR® FAST One-Step qRT-PCR kit (KAPA BIOSYSTEMS, Australia) was used to set up a single reaction volume of 10 µl for each sample. Each reaction volume consist of 1X KAPA SYBR® FAST qPCR Master Mix, 200 nM of each forward and reverse primer (Table S1), 1X KAPA RT Mix, 50 nM ROX dye and 30 ng of cell line total RNA template. RT-qPCR was performed using the Applied Biosystems® 7500 Real-Time PCR System (Thermo Scientific, Australia). The cycling protocol was: 42°C for 10 min to synthesize cDNA, followed by 95°C for 5 min to deactivate RT before cycling 35 times (95°C for 30 s, 50°C for 30 s and 72°C for 1 min) and finished with 72°C for 10 min.

195

Chapter 5.2

Figure S1. Representative FTIR spectra of bare gold surface (Black), and gold surfaces after adsorption of unmodified (Red) and poly(A)-extended (Blue) synthetic miRNA sequences. Band positions in grey and blue are chracteristic of nucleic acid bases (A, U, G, C)1 and adenine bases (A)2 respectively.

Figure S2. Representative DPV signals for detection of miR-107 levels in three prostate cancer urinary samples.

196

Chapter 5.2

REFERENCES (1) Dovbeshko, G. I.; Chegel, V. I.; Gridina, N. Y.; Repnytska, O. P.; Shirshov, Y. M.; Tryndiak, V. P.; Todor, I. M.; Solyanik, C. I. Biopolymers 2002, 67, 470-486. (2) Kimura-Suda, H.; Petrovykh, D. Y.; Tarlov, M. J.; Whitman, L. J. J. Am. Chem. Soc. 2003, 125, 9014-9015.

197

Chapter 5.3

DNA-Directed Assembly of Copper Nanoblocks with Inbuilt Fluorescent and Electrochemical Properties: Application in Simultaneous Amplification-Free Analysis of Multiple RNA Species

Kevin M. Koo, Laura G. Carrascosa, and Matt Trau

ABSTRACT The intrinsic affinity of DNA molecules towards metallic ions can drive specific formation of copper nanostructures within the nucleic acid helix structure in a sequence-dependent manner. The resultant nanostructures have interesting fluorescent and electrochemical properties which are attractive for novel biosensing applications. However, the potential of DNA-templated nanostructures for precision disease diagnosis remains unexplored. Particularly, DNA-templated nanostructures may have outstanding clinical diagnostics prospect in the universal amplification-free detection of different RNA biomarker species. Due to low cellular levels and differing species-dependent length and sequence features, simultaneous detection of different messenger RNAs (mRNAs), microRNAs (miRNAs) and long non-coding RNAs (lncRNAs) species with a single technique is a challenging task. Herein, we report a contemporary technique for facile in-situ assembly of DNA-templated copper nanoblocks (CuNBs) on various RNA species targets after hybridization-based magnetic isolation. Our approach circumvents typical drawbacks associated with amplification and labelling procedures of current RNA assays; and the synthesized CuNBs impart amplification-free fM-level detection limit with flexible fluorescence or electrochemical readouts. Furthermore, our nanosensing technique displays clinical application promise as demonstrated by non-invasive analysis of three diagnostic RNA biomarkers from a cohort of 10 prostate cancer patient urinary samples with 100%- concordance (qRT-PCR validation). We envision that the good analytical performance and versatility of our methodology could be of widespread use in both diagnostics and research fields.

198

Chapter 5.3

INTRODUCTION DNA molecules can exhibit intrinsic affinity towards a wide range of metallic ions and give rise to a variety of nanostrutures within the nucleic acid helix structure. In particular, silver (Ag) [1, 2] or copper (Cu) [3-5] nanostructures (CuNs) with interesting fluorescent and/or electrochemical properties can be created via DNA-directed assembly. CuNs are particularly attractive over other DNA-templated metal nanostructures due to a faster and more stable synthesis process as Cu2+ ions are less likely to form insoluble precipitate with common anions in biosensing environments. The synthesis process is also shown to be sequence- dependent and mostly favourable towards thymine (T)-rich sequences in either the single- stranded [6-8] or double-stranded [9-11] forms. This is a very attractive feature for nucleic acid detection as fluorescence signals can be effortlessly generated on specific DNA regions, thus avoiding complex target-labelling chemistry. The large Stokes shift difference between excitation and emission wavelength of fluorescent CuNs [9, 12] is also favourable for eliminating background signals in complex biological environments. Despite these appealing advantages of facile, rapid and low-cost DNA-directed CuNs synthesis; the biosensing potential of DNA-templated CuNs for precision diagnostic purposes has remained scarcely exploited. In this work we focused on developing a universal amplification-free nanosensor of multiple RNA species for cancer diagnosis using DNA-templated CuNs.

The detection of different RNA species targets such as messenger RNAs (mRNAs) [13], microRNAs (miRNAs) [14], and long non-coding RNAs (lncRNAs) [15, 16], is useful for developing cancer diagnostics and guiding treatment decisions [17, 18]. Taking prostate cancer (PCa) as an example, recent studies [19-22] have indicated that a suite of different RNA biomarker species could serve to improve current PCa diagnosis [23]. Explicitly, the gene fusion TMPRSS2:ERG (T1E4) mRNA biomarker has been found to be highly PCa- specific and rarely occurred in benign conditions [24, 25], whilst overexpression of miR-107 miRNA [26-28] and SChLAP1 lncRNA [29] biomarker is associated with increased risk for developing PCa metastasis and aggressive subtypes. These three different RNA biomarker species have also been readily detected in urine samples, and could allow early identification of malignant PCa cases without invasive biopsy procedures [30, 31]. However, the universal detection of multiple RNA biomarker species is complex with current tools.

199

Chapter 5.3

The universal analysis of different RNA species is challenging due to the widely-differing length or sequence (i.e. presence of poly-A tails) characteristics of individual RNA species. Presently, different variations of reverse transcription polymerase chain reaction coupled with fluorescence readout (qRT-PCR) are most commonly-used for quantitative RNA detection [32, 33]. Despite the good reliability of these qRT-PCR variations for sensitive and specific RNA detection, they are limited by the need for enzymatic amplification and use of fluorescent labels. Enzymatic amplification requires extensive sample manipulation and could lead to amplification artifacts/bias; and miRNA amplification is also especially challenging due to their short primer-sized lengths. Moreover, traditional qRT-PCR fluorescent organic dye labels are costly, complex and time-consuming to synthesize; unstable at ambient conditions; and requires cumbersome target-hybridization procedures.

Our nanosensing technique demonstrates, for the first time, DNA-directed assembly of Cu nanoblocks (CuNBs) on different RNA species for simultaneous amplification-free detection of multiple RNA biomarker species relevant to PCa detection. By enabling poly(A-T)- templated CuNBs synthesis on magnetically-purified RNA targets, our proposed technique exhibited high specificity, fM-level sensitivity, and flexible fluorescence/electrochemical RNA detection. We also showed the clinical translation promise by parallel analysis of various promising RNA biomarker (T1E4, miR-107, SChLAP1) species in PCa patient urinary samples.

EXPERIMENTAL Materials and Reagents All reagents were of analytical grade and purchased from Sigma Aldrich (Australia). UltraPureTM DNase/RNase-free distilled water (Invitrogen, Australia) was used throughout the experiments. Oligonucleotides were purchased from Integrated DNA Technologies (Singapore) and sequences are shown in Table 1.

RNA extraction from cultured PCa cells DuCap and 22Rv1 cells were cultured in RPMI-1640 growth media (Life Technologies, Australia) supplemented with 10% fetal bovine serum (Life Technologies, Australia) in a humidified incubator containing 5% CO2 at 37ºC. RNA was extracted using Direct-zolTM kit (Zymo Research, USA) according to the manufacturer’s instructions. Total RNA integrity

200

Chapter 5.3 and purity were checked using a NanoDrop Spectrophotometer (Thermo Scientific, Australia).

RNA extraction from PCa urinary samples Ethics approval was obtained from The University of Queensland Institutional Human Research Ethics Committee (Approval No. 2004000047) and informed consent was obtained from all subjects prior to sample collection. Methods pertaining to clinical samples were carried out in accordance with approved guidelines.

Voided urinary samples were collected from men undergoing treatment for PCa at the Royal Brisbane and Women’s Hospital. Total RNA was extracted from the urine samples using the commercially-available ZR urine RNA isolation KitTM (Zymo Research, USA). For each sample, 30 mL of urine was passed through the supplied ZRC GFTM Filter to retain urinary cells. Then, 700 µL of supplied urine RNA buffer was passed through the filter and the filtered cells were collected in the flow-through. Next, as per manufacturer’s instructions, the cells from the flow-through were lysed, washed and eluted in 10 µL of RNase-free water. Total RNA concentration for each sample was assayed using Qubit® RNA HS Assay Kit (Thermo Scientific, Australia) and total RNA input for each single assay run was standardized to 30 ng for non-biased comparisons of target biomarker expression levels.

Magnetic isolation of RNA targets Extracted total RNA from cell lines/urinary samples were adjusted to 30 ng in 5 µL of RNase-free water prior to mixing with 10 µL of 5x saline-sodium citrate (SSC) buffer (pH 7). RNA targets T1E4, miR-107 and SChLAP1 in samples were hybridized with 10 µL addition of 10 µM biotinylated complementary capture probes in individual tubes. For mRNA (T1E4) and lncRNA (SChLAP1) samples, the capture probes-targets mixture was heated to 90 ºC for 2 min before being incubated at 70 ºC for 60 min to facilitate hybridization. For miRNA (miR-107), the resultant mixture was incubated at 60 ºC for 30 min at room temperature.

For magnetic isolation of RNA targets, 20 µL of streptavidin-labelled Dynabeads® MyOneTM Streptavidin C1 (Invitrogen, Australia) magnetic beads was firstly washed with 2x washing and binding (B&W) buffer (10 mM Tris-HCl, pH 7.5; 1 mM EDTA ; 2 M NaCl) and resuspended in 25 µL of 2x B&W buffer. Then, the resuspended magnetic beads were added to each of the prepared capture probes-targets tubes, mixed thoroughly, and incubated on a

201

Chapter 5.3 mixer for 30 min at room temperature to allow magnetic beads binding to capture probes. After that, the magnetic beads with attached RNA targets were isolated with a magnet, washed twice with 2x B&W buffer and resuspended in 10 µL (for mRNA) or 7.5 µL (for miRNA and lncRNA) of RNase-free water.

PolyA tails synthesis on miRNA and lncRNA targets For miRNA and lncRNA targets, 3’-polyA tails were extended on the targets by mixing 7.5 µL of resuspended magnetic beads-miRNA/lncRNA targets- with 1 µL of 10x E. Coli

Poly(A) polymerase reaction buffer (250 mM NaCl, 50 mM Tris-HCl, 10 mM MgCl2, pH 7.9), 1 mM ATP, and 5 U E. Coli Poly(A) polymerase (New England BioLabs, Australia) to make up a 10 µL reaction mixture. Each reaction mixture was incubated at 37°C for 30 min (unless otherwise stated) and Poly(A) polymerase was inactivated by addition of 10 mM EDTA.

Next, 10 µL of 10 µM polyT sequences was added separately to each RNA target tube and incubated at ambient temperature for 20 min on a mixer to allow hybridization to polyA tails. After hybridization, RNA targets were magnetically purified as previously described and resuspended in 10 µL of RNase-free water. To release the captured RNA targets from the magnetic beads, each solution of resuspended RNA targets was heated for 2 min at 95°C, immediately placed on a magnet, and the supernatant containing the released RNA targets was collected.

DNA-templated CuNBs synthesis for fluorescence/electrochemical readout To synthesize CuNBs on polyA-T sequences, 10 µL released RNA targets was mixed with

MOPS buffer (10mM MOPS, 150 mM NaCl, pH 7.6), 200 µM CuSO4 (unless otherwise stated) and 2 mM sodium ascorbate (unless otherwise stated). The mixture was allowed to react for 10 min at ambient temperature for DNA-templated CuNBs synthesis.

For fluorescence readout, the incubated mixture was immediately subjected to fluorescence detection of RNA targets. The fluorescence measurements were performed on a Tecan Infinite® M200 PRO fluorescence plate reader (Tecan, Switzerland) with an excitation wavelength of 340 nm and an emission wavelength range of 500 nm-650 nm (max λem = 580 nm).

202

Chapter 5.3

For electrochemical readout, the incubated mixture was diluted to 30 µL in 5x SSC buffer (pH 7.0) and directly dropped onto the gold working electrode surface of a DRP-C223BT screen-printed electrode (Dropsens, Spain) and allowed to adsorb for 20 min with gentle agitation at ambient temperature. Subsequently, 1 mM 6 mercapto-1-hexanol in 50 mM PBS buffer (685 mM sodium chloride, 100 mM potassium chloride, 2.5% glycerol, pH 7.4) was added onto electrode surface for 15 min to block the remaining bare surface. The electrodes were then washed with 10 mM PBS buffer (pH 7.4) before amperometry measurements in the presence of 100 mM PBS buffer pH 7.4) containing 10 mM H2O2. The measurements were performed on a CH1040C potentiostat (CH Instruments, USA) at -0.3 V for 30 s. qRT-PCR validation The KAPA SYBR® FAST One-Step qRT-PCR kit (KAPA BIOSYSTEMS, Australia) was used to set up a single reaction volume of 10 µl for each sample. Each reaction volume consisted of 1X KAPA SYBR® FAST qPCR Master Mix, 200 nM of each forward and reverse primer (Table S1), 1X KAPA RT Mix, 50 nM ROX dye and 30 ng of sample total RNA template. RT-qPCR was performed using the Applied Biosystems® 7500 Real-Time PCR System (Thermo Fisher Scientific, Australia). The cycling protocol was: 42°C for 10 min to synthesize cDNA, followed by 95°C for 5 min to deactivate RT before cycling 35 times (95°C for 30 s, 50°C for 30 s and 72°C for 1 min) and finished with 72°C for 10 min.

The cycle threshold (Ct) value of each qPCR reaction was determined for comparison and validation with our assay results.

RESULTS AND DISCUSSION Mechanism of RNA detection using DNA-templated CuNBs The underlying mechanism of our proposed RNA detection method (Fig. 1) involves the coupling of parallel magnetic isolation of individual RNA targets and in-situ poly(A-T)- templated CuNBs synthesis for convenient fluorescence or electrochemical target detection. During initial magnetic target isolation, different RNA species (i.e. mRNA, miRNA, and lncRNA) are first complementarily bound to respective biotinylated capture probes in parallel reactions. Then, the captured targets are magnetically purified by streptavidin-magnetic beads, and 3’-polyA tails are enzymatically-synthesized [34] for miRNA and lncRNA targets (without natural polyA tails) exclusively. All targets are then hybridized to polyT sequences through their naturally-occurring (for mRNA targets)/enzymatically-synthesized polyA tails.

203

Chapter 5.3

After hybridization, RNA targets are magnetically purified again to remove unhybridized polyT sequences. The poly(A-T) sequences (on purified RNA targets) served as templates for fluorescent CuNBs synthesis and subsequent fluorescence measurements of corresponding RNA target levels. As fluorescence is spontaneously generated upon successful poly(A-T)- templated CuNBs synthesis, our approach does not require additional fluorescent dye labelling.

Figure 1. Schematic of assay using DNA-templated CuNBs synthesis for versatile detection of different RNA species. Total RNA is first isolated from sample before capture probes are used to capture RNA biomarker targets separately in individual assays for magnetic isolation. After magnetic enrichment of targets, poly A polymerase is used to enzymatically extend poly(A) tails on miRNA and lncRNA targets prior to target release from magnetic beads. Poly(T) sequences are then introduced for complementary hybridization to poly(A) tails of all RNA targets and resulting poly(A-T) served as templates for CuNBs synthesis on RNA targets. The presence of CuNBs allowed for RNA targets detection by either direct fluorescence readout or electrochemical readout after target adsorption onto gold electrode surface.

To improve detection sensitivity, we flexibly adapted the fluorescence readout for an alternative electrochemical readout (via CuNBs-induced reduction of hydrogen peroxide). We have previously shown that isolated RNA target sequences could be quickly adsorbed onto gold electrode surfaces based on nucleic acid bases-gold affinity interactions [34, 35]. Based on this mechanism, RNA targets-poly(A-T)-CuNBs (i.e. RNA targets after poly(A-T)-

204

Chapter 5.3 templated CuNBs synthesis) are conveniently dispensed onto a bare gold electrode surface for adsorption. The electrode surface-adsorbed CuNBs are then induced to catalyze hydrogen peroxide (H2O2) reduction, generating an electrochemical current response that is measurable by amperometry to directly reflect RNA target amount.

In this report, we used three different model PCa-related RNA targets to evaluate the versatility of detecting different RNA species for cancer diagnosis. The three targets comprised of mRNA (T1E4), miRNA (miR-107), and lncRNA (SChLAP1) biomarkers.

Synthesis of CuNBs on Poly(A-T) sequences We first evaluated the successful synthesis of fluorescent CuNBs on poly(A-T) sequences by performing our proposed detection methodology with miR-107 (1 pM) as a model target. As miR-107 targets do not naturally contain polyA tails, polyT complementary hybridization is only possible in the event of enzymatic 3’-polyA tails synthesis. Hence, miR-107 is an ideal target for showing that pre-assembly of poly(A-T) sequences on RNA targets was necessary for CuNBs synthesis upon CuSO4 and sodium ascorbate addition. Fig. 2 showed that fluorescence was emitted (Tube 1) under UV light source when all necessary procedures were performed. Control experiments with the omission of miR-107 target (i.e. no-template control (NTC)) (Tube 2); polyT sequences (Tube 3); enzymatic polyA tails extension (Tube

4); sodium ascorbate (Tube 5); and CuSO4 (Tube 6) demonstrated that fluorescence was only generated in the event of efficient CuNBs synthesis on poly(A-T) sequences (Fig. 2 inset). This is in good agreement with a previous report [9] which demonstrated that double- stranded poly(A-T) sequences could serve as templates for fluorescent CuNBs synthesis. TEM imaging also showed the successful synthesis of poly(A-T)-templated CuNBs with spherical shapes of 20 nm in average size (Fig. 2 inset).

205

Chapter 5.3

Figure 2. Synthesis of Poly(A-T)-templated CuNBs. Combined fluorescence spectra for (Tube 1); and control experiments with the omission of miR-107 target (i.e. no-template control (NTC) (Tube 2); polyT sequences (Tube 3); enzymatic polyA tails extension (Tube 4); sodium ascorbate (Tube 5); and CuSO4 (Tube 6). Inset shows the corresponding tubes under UV light source and TEM image of poly(A-T)-templated CuNBs (as indicated by arrows). Scale bar represents 100 nm.

Although single-stranded polyT sequences are also excellent templates for CuNBs synthesis [6], we did not observe any fluorescence with the addition of polyT sequences in absence of polyA tails (Tube 4). We reasoned that the second round of magnetic RNA target enrichment was effective in removing any unhybridized polyT sequences, thus resulting in the absence of background fluorescence signal. We also did not observe any fluorescence signal when CuNBs synthesis was attempted on only miR-107 templates with (Tube 3) or without (Tube 4) polyA tails, thus indicating highly-selective CuNBs synthesis on poly(A-T) templates solely. This observation of non-CuNBs growth on single-stranded RNA is consistent with previous findings that single-stranded polyA sequences [6] and random DNA [6, 9] are unable to support CuNBs synthesis. Thus, our results reinforced the concept that double- stranded nucleic acid templates are generally more stable and ideal for CuNBs synthesis in biosensing applications [36, 37]. Moreover, the use of poly(A-T)-templated CuNBs synthesis in our experiments provided advantages such as more effective CuNBs synthesis [9], lower required target concentration (i.e. no need for enzymatic amplification), as well as non-biased

206

Chapter 5.3 and reproducible fluorescence detection of different RNA targets since CuNBs synthesis was not sequence-dependent (i.e. solely poly(A-T)-dependent).

Optimization of Poly(A-T)-Templated CuNBs Synthesis Using miR-107 (1 pM) as a model target, we also optimized the polyA extension time as well as CuSO4 and sodium ascorbate concentrations to maximize the generated fluorescence signals (Fig. S1). First, we used various enzymatic polyA extension times (5, 15, 30, 45, 60 min) to test the minimal time for generating a significant fluorescence signal. We found that fluorescence signal saturated at 30 min with no significant increased beyond this timepoint (Fig. S1(a)). It is likely that the process of enzymatic polyA extension on miR-107 targets had reached its maximal at 30 min, and a longer extension time would not result in significantly longer polyA tails (i.e. longer poly(A-T)) for higher CuNBs-fluorescence. Hence, we used a polyA extension time of 30 min for all further experiments.

Next, we investigated the effect of different CuSO4 concentrations (50, 100, 200, 400, and 800 µM), followed by different sodium ascorbate concentrations (0.5, 1, 2, 4 and 8 mM) for

CuNBs synthesis. We observed signal saturation at 200 µM CuSO4 (Fig. S1(b)) and 2 mM sodium ascorbate (Fig. S1(c)), thereby indicating that synthesis of CuNBs was most favourable under these conditions to generate the highest fluorescence signal. Therefore,

CuNBs synthesis conditions were performed with 200 µM CuSO4 and 2 mM sodium ascorbate for further experiments.

Sensitivity for detecting different RNA target species To investigate the sensitivity of our proposed method in detecting different RNA target species, we prepared different concentrations of T1E4 (0-500 pM), miR-107 (0-100 pM), and SChLAP1 (0-500 pM) target sequences for detection. We observed increasing fluorescence intensities with higher concentrations of all RNA targets (Figs. 3(a), 3(d), and 3(g)), thus indicating increased synthesis of poly(A-T)-templated fluorescent CuNBs synthesis. Using the fluorescence readout, the limit-of-detection (LOD) was 500 fM (1.5 x 104 copies); 100 fM (3 x 105 copies); 500 fM (1.6 x 104 copies) for T1E4, miR-107, and SChLAP1 respectively (Figs. 3(b), 3(e), and 3(h)). By titrating different target concentrations (Fig. S2), we established working curves with good linear correlations between fluorescence intensities and target concentrations (Figs. 3(c), 3(f), and 3(i)). Amongst the three different RNA species, our data illustrated the highest assay sensitivity for miRNA detection. We conjectured that

207

Chapter 5.3 the shorter lengths of miRNAs allow for more efficient hybridization to similar-length capture probes and increased poly(A) extensions to result in higher signal generation.

Figure 3. Sensitivity of assay via fluorescence readout. Fluorescence signals generated using various concentrations of (a) T1E4 mRNA; (d) miR-107 miRNA; and (e) SChLAP1 lncRNA targets within the 500-640 nm wavelength range. The respective (580 nm) limit-of-detection (b, e, h) and linear calibration plot (c, f, g) for each of T1E4, miR-107, and SChLAP1 are shown. Error bars represent standard deviation of three independent experiments.

Next, we investigated if detection sensitivity could be further improved by using an alternative readout technique. As such, electrochemical readouts would be appealing because of their excellent sensitivity, simplicity, and inexpensive instrumentation. Dai et al. have previously showcased the electrocatalysis of DNA-RNA heteroduplex-templated Cu nanoclusters for high fM-target detection sensitivity [37]. Hence, we hypothesized that poly(A-T)-templated CuNBs might provide similar properties for enhanced electrochemical detection sensitivity. To enable electrochemical readout, we adsorbed RNA targets-poly(A- T)-CuNBs on a bare gold electrode surface [34], and utilized CuNBs-induced reduction of

H2O2 to generate an amperometric signal which directly correlated to biomarker quantity (Figs. 4(a), 4(d), and 4(g)). Using this alternative electrochemical readout, we achieved lower

208

Chapter 5.3 limits-of-detection as compared to the fluorescence readout (up to 10-fold improvement) (Figs. 4(b), 4(e), and 4(h)) : 100 fM (3 x 103 copies); 10 fM (3 x 104 copies); 100 fM (3.24 x 103 copies) for T1E4, miR-107, and SChLAP1 respectively. By titrating different target concentrations (Fig. S3), we established working curves with good linear relationships between current levels and target concentrations (Figs. 4(c), 4(f), and 4(i)) for all three RNA target species. The high sensitivity of electrochemical assays have been well-studied [38]; and the versatile adaptation of our assay for electrochemical readout provided additional benefits such as rapid analysis, ability for high-throughput parallel assays, and possibility for point-of-care testing with a portable potentiostat.

Figure 4. Sensitivity of assay via electrochemical readout. Electrochemical signals generated using various concentrations of (a) T1E4 mRNA; (d) miR-107 miRNA; and (e) SChLAP1 lncRNA targets. The respective limit-of-detection (b, e, h) and linear calibration plot (c, f, g) for each of T1E4, miR-107, and SChLAP1 are shown. Error bars represent standard deviation of three independent experiments.

As mRNAs, miRNAs, and lncRNAs are commonly used as disease diagnostic biomarkers in the clinic or for biological studies, the fM-level detection sensitivity of our methodology is capable of analyzing low amounts of different RNA biomarker species in clinical samples.

209

Chapter 5.3

This fM-level of sensitivity is also similar or superior to recently-developed detection techniques of a particular RNA species [37, 39-41]. Moreover, it is worthy to highlight that our developed methodology is novel in the detection of different RNA species via a single technique without any need for enzymatic amplification or conventional fluorescent labelling.

Selectivity of detecting RNA target To test the selectivity of the capture probes used in our proposed method for specific isolation and subsequent detection of RNA target, we utilized a single capture probe at a time in various presence of different targets (T1E4, miR-107, and SChLAP1). The capture probes used in our experiments were shown to be selective for respective T1E4 mRNA, miR-107 miRNA, and SChLAP1 lncRNA targets (Figs. 5(a), 5(b), and 5(c)), with approximately 8– fold decrease in fluorescence intensity (Fig. 5(d)) for non-specific targets. As the mixture of different RNA species is similar to the environment of isolated total RNA from cell, the high target selectivity of our methodology would be applicable for accurate RNA biomarker detection in the biologically-complex blood or urine samples.

Figure 5. Selectivity of assay. Fluorescence signals generated using different RNA targets and a single (a) T1E4 mRNA; (b) miR-107 miRNA; and (c) SChLAP1 lncRNA target capture probe within the 500-640 nm wavelength range. The fluorescence signal differences at 580

210

Chapter 5.3 nm (d) for specific detection of T1E4, miR-107, and SChLAP1 are shown. Error bars represent standard deviation of three independent experiments.

Analysis of RNA biomarker signatures using PCa cell line models After establishing the assay sensitivity and selectivity on synthetic RNA sequences, we continued the evaluation of our methodology on biologically-complex cell line models. To this end, we challenged our technique with total RNA extracts from PCa cell lines (DuCap and 22Rv1) to detect different RNA species (T1E4 mRNA, miR107 miRNA, and SChLAP1 lncRNA) by both fluorescence (Figs. 6(a) and 6(b)) and electrochemical readouts (Figs. 6(d) and 6(e)). It was observed that both fluorescence and electrochemical readouts produced similar cell line biomarker expression trends: DuCap showed T1E4 expression, whilst 22Rv1 overexpressed both miR-107 and SChLAP1 (Figs. 6(c) and 6(f)). These findings were in concordance with reports in literature [25, 26, 28, 29] which detected similar RNA biomarker expression trends in these cell lines. Furthermore, the successful selective RNA target detection in total RNA isolation from cultured cells also showed high resilience against interference from non-target molecules of our methodology; a parameter greatly required for clinical sample application. Although there have been developed assays for detecting RNA biomarkers from cultured cells, the simultaneous detection of different RNA species has rarely been shown. To the best of our knowledge, our assay is unique in the parallel detection of mRNA, miRNA, and lncRNA biomarkers from biological cells.

Figure 6. Detection of diverse RNA biomarker species using prostate cancer cell line models. Fluorescence signals generated for detection of T1E4, miR-107, and SChLAP1 in (a) DuCap

211

Chapter 5.3 and (b) 22Rv1 cells within the 500-640 nm wavelength range. The different biomarker signature of both cell lines via fluorescence readout at 580 nm (c) is shown. Electrochemical signals generated for detection of T1E4, miR-107, and SChLAP1 in (d) DuCap and (e) 22Rv1 cells. The different biomarker signature of both cell lines via electrochemical readout (f) is shown. Error bars represent standard deviation of three independent experiments.

Non-invasive detection of RNA biomarkers in patient urinary samples To investigate the clinical application of our methodology, we challenged it with the more sensitive electrochemical analysis of T1E4 (Fig. 7(a)), miR-107 (Fig. 7(b)), and SChLAP1 (Fig. 7(c)) RNA biomarkers from patient urinary samples (P1-P10 are from PCa patients, H1 and H2 are from healthy males). The use of urinary samples provided a form of non-invasive liquid biopsy as compared to painful, costly, and infection-risky tissue biopsies. Towards this end, we firstly extracted cellular total RNA from patient urine samples using a commercialized spin-column technique (see Experimental section for more details).

For the PCa patients, we generally observed higher expressions of the different target RNA biomarkers as compared to those of the healthy samples (Fig. 7). This is in good agreement with previous reports which stated T1E4 presence as well as miR-107 and SChLAP1 overexpression in malignant PCa cases [25, 26, 29]. To demonstrate diagnostic application, we established cutoff values (Fig. 7) to predict abnormal expression levels for each target RNA biomarker based on the mean signals of the healthy samples and compared our outcomes with conventional qRT-PCR. Each cutoff value was determined from the mean signals of the healthy samples plus three standard deviations. A generated signal from a patient sample exceeding the cutoff value would then suggest that the particular RNA biomarker is of an abnormal level. Using this analysis, we profiled T1E4, miR107, and SChLAP1 biomarker signatures for each patient and found 100% concordance with conventional qRT-PCR validation (Table S2). This successful detection of multiple clinically-promising RNA biomarkers species underscored the clinical application potential of our assay potential in obtaining biomarker signatures from biological samples. As mutations in PCa (and other cancer types) differ amongst different patients, a more effective diagnosis would utilize the combined detection of several different biomarkers to generate biomarker signatures (i.e. cancer molecular characterization) [42, 43]. Thus, the versatile nature of our proposed assay for different RNA biomarker species analysis, allows a potentially powerful diagnostic tool for generating informative biomarker signatures to enable potent personalized cancer therapies.

212

Chapter 5.3

Figure 7. Non-invasive RNA detection in clinical urinary samples. Electrochemical detection of (a) T1E4; (b) miR-107; and (c) SChLAP1 RNA biomarker species in urinary samples collected from twelve prostate cancer patients (P1-P10) and two healthy males (H1 and H2). Error bars represent standard deviation of three independent experiments.

CONCLUSIONS In conclusion, amplification-free detection of three different RNA species by poly(A-T)- templated CuNBs synthesis is reported for the first time. We have demonstrated a technique for versatile detection of different RNA biomarker species with high fM-level sensitivity and specificity via flexible fluorescence/electrochemical readout. The successful detection in urinary samples from PCa patients and subsequent result validation with conventional qRT- PCR; demonstrated the clinical potential of our assay for non-invasive detection of RNA biomarker signatures during PCa diagnosis.

Our proposed assay has demonstrated several benefits to mitigate conventional RNA detection challenges: (i) it does not require any enzymatic amplification to achieve a detectable readout signal, thus avoiding possible amplification bias/artifacts. (ii) it shows the versatile detection of different-length RNA species with a single technique (albeit a simple and convenient extension step for RNA species without natural polyA tails), and does not require any special assay designs such as stem-loop primers for detecting short-length miRNAs. (iii) it utilizes poly(A-T)-templated CuNBs for readouts in place of conventional fluorescence dye labels, therefore allowing for a more inexpensive and stable readout.

With future cancer diagnosis most likely to combine different biomarker species for comprehensive disease molecular characterization, we envisage that our described methodology could serve as an attractive amplification- and fluorescent dye-free alternative for different RNA species detection. Moreover, our assay could easily be adapted for RNA disease biomarkers in other diseases through rational capture probe design.

213

Chapter 5.3

REFERENCES [1] Petty, J. T.;Zheng, J.;Hud, N. V.; Dickson, R. M. DNA-templated Ag nanocluster formation. J. Am. Chem. Soc. 2004, 126, 5207-5212. [2] Richards, C. I.;Choi, S.;Hsiang, J. C.;Antoku, Y.;Vosch, T.;Bongiorno, A.;Tzeng, Y. L.; Dickson, R. M. Oligonucleotide-stabilized Ag nanocluster fluorophores. J. Am. Chem. Soc. 2008, 130, 5038-5039. [3] Monson, C. F.; Woolley, A. T. DNA-templated construction of copper nanowires. Nano Lett. 2003, 3, 359-363. [4] Rotaru, A.;Dutta, S.;Jentzsch, E.;Gothelf, K.; Mokhir, A. Selective dsDNA-templated formation of copper nanoparticles in solution. Angew. Chem., Int. Edit. 2010, 49, 5665-5667. [5] Jia, X. F.;Li, J.;Han, L.;Ren, J. T.;Yang, X.; Wang, E. K. DNA-hosted copper nanoclusters for fluorescent identification of single nucleotide polymorphisms. ACS Nano 2012, 6, 3311-3317. [6] Qing, Z. H.;He, X. X.;He, D. G.;Wang, K. M.;Xu, F. Z.;Qing, T. P.; Yang, X. Poly(thymine)-templated selective formation of fluorescent copper nanoparticles. Angew. Chem., Int. Edit. 2013, 52, 9719-9722. [7] Qing, Z. H.;He, X. X.;Qing, T. P.;Wang, K. M.;Shi, H.;He, D. G.;Zou, Z.;Yan, L.;Xu, F. Z.;Ye, X. S.; Mao, Z. G. Poly(thymine)-templated fluorescent copper nanoparticles for ultrasensitive label-free nuclease assay and its inhibitors screening. Anal. Chem. 2013, 85, 12138-12143. [8] Mao, Z. G.;Qing, Z. H.;Qing, T. P.;Xu, F. Z.;Wen, L.;He, X. X.;He, D. G.;Shi, H.; Wang, K. M. Poly(thymine)-templated copper nanoparticles as a fluorescent indicator for hydrogen peroxide and oxidase-based biosensing. Anal. Chem. 2015, 87, 7454- 7460. [9] Song, Q. W.;Shi, Y.;He, D. C.;Xu, S. H.; Ouyang, J. Sequence-dependent dsDNA- templated formation of fluorescent copper nanoparticles. Chem.-A Eur. J. 2015, 21, 2417-2422. [10] Chen, J. Y.;Ji, X. H.;Tinnefeld, P.; He, Z. K. Multifunctional dumbbell-shaped DNA- templated selective formation of fluorescent silver nanoclusters or copper nanoparticles for sensitive detection of biomolecules. ACS Applied Mat. Interfaces 2016, 8, 1786-1794.

214

Chapter 5.3

[11] Sha, L.;Zhang, X. J.; Wang, G. F. A label-free and enzyme-free ultra-sensitive transcription factors biosensor using DNA-templated copper nanoparticles as fluorescent indicator and hairpin DNA cascade reaction as signal amplifier. Biosens. Bioelectron. 2016, 82, 85-92. [12] Jia, X. F.;Yang, X. A.;Li, J.;Liab, D. Y.; Wang, E. K. Stable Cu nanoclusters: from an aggregation-induced emission mechanism to biosensing and catalytic applications. Chem. Comm. 2014, 50, 237-239. [13] Brinkman, B. M. N. Splice variants as cancer biomarkers. Clin. Biochem. 2004, 37, 584-594. [14] Van Roosbroeck, K.;Pollet, J.; Calin, G. A. miRNAs and long noncoding RNAs as biomarkers in human diseases. Expert Rev. Mol. Diagnostics 2013, 13, 183-204. [15] O'Leary, V. B.;Ovsepian, S. V.;Carrascosa, L. G.;Buske, F. A.;Radulovic, V.;Niyazi, M.;Moertl, S.;Trau, M.;Atkinson, M. J.; Anastasov, N. PARTICLE, a triplex-forming long ncRNA, regulates locus-specific methylation in response to low-dose irradiation. Cell Rep. 2015, 11, 474-485. [16] Mercer, T. R.;Dinger, M. E.; Mattick, J. S. Long non-coding RNAs: insights into functions. Nat. Rev. Genet. 2009, 10, 155-159. [17] Sharp, P. A. The centrality of RNA. Cell 2009, 136, 577-580. [18] Riedmaier, I.; Pfaffl, M. W. Transcriptional biomarkers - High throughput screening, quantitative verification, and bioinformatical validation methods. Methods 2013, 59, 3-9. [19] Prensner, J. R.;Rubin, M. A.;Wei, J. T.; Chinnaiyan, A. M. Beyond PSA: The next generation of prostate cancer biomarkers. Science Trans.l Med. 2012, 4, 127rv3. [20] Velonas, V. M.;Woo, H. H.;dos Remedios, C. G.; Assinder, S. J. Current status of biomarkers for prostate cancer. Int. J. Mol. Sci. 2013, 14, 11034-11060. [21] Fabris, L.;Ceder, Y.;Chinnaiyan, A. M.;Jenster, G. W.;Sorensen, K. D.;Tomlins, S.;Visakorpi, T.; Calin, G. A. The potential of microRNAs as prostate cancer biomarkers. Eur. Urol. 2016, 70, 312-322. [22] Ronnau, C. G. H.;Verhaegh, G. W.;Luna-Velez, M. V.; Schalken, J. A. Noncoding RNAs as novel biomarkers in prostate cancer. Biomed. Res. Int. 2014, 591703. [23] Pellegrini, K. L.;Sanda, M. G.; Moreno, C. S. RNA biomarkers to facilitate the identification of aggressive prostate cancer. Mol. Aspects Med. 2015, 45, 37-46.

215

Chapter 5.3

[24] Tomlins, S. A.;Bjartell, A.;Chinnaiyan, A. M.;Jenster, G.;Nam, R. K.;Rubin, M. A.; Schalken, J. A. ETS gene fusions in prostate cancer: From discovery to daily clinical practice. Eur. Urol. 2009, 56, 275-286. [25] Tomlins, S. A.;Rhodes, D. R.;Perner, S.;Dhanasekaran, S. M.;Mehra, R.;Sun, X. W.;Varambally, S.;Cao, X. H.;Tchinda, J.;Kuefer, R.;Lee, C.;Montie, J. E.;Shah, R. B.;Pienta, K. J.;Rubin, M. A.; Chinnaiyan, A. M. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science 2005, 310, 644-648. [26] Martello, G.;Rosato, A.;Ferrari, F.;Manfrin, A.;Cordenonsi, M.;Dupont, S.;Enzo, E.;Guzzardo, V.;Rondina, M.;Spruce, T.;Parenti, A. R.;Daidone, M. G.;Bicciato, S.; Piccolo, S. A microRNA targeting Dicer for metastasis control. Cell 2010, 141, 1195- 1207. [27] Wang, W.-X.;Kyprianou, N.;Wang, X.; Nelson, P. T. Dysregulation of the mitogen granulin in human cancer through the miR-15/107 microRNA gene group. Cancer Res. 2010, 70, 9137-9142. [28] Chen, P.-S.;Su, J.-L.;Cha, S.-T.;Tarn, W.-Y.;Wang, M.-Y.;Hsu, H.-C.;Lin, M.- T.;Chu, C.-Y.;Hua, K.-T.;Chen, C.-N.;Kuo, T.-C.;Chang, K.-J.;Hsiao, M.;Chang, Y.- W.;Chen, J.-S.;Yang, P.-C.; Kuo, M.-L. miR-107 promotes tumor progression by targeting the let-7 microRNA in mice and humans. J. Clin. Investig. 2011, 121, 3442- 3455. [29] Prensner, J. R.;Iyer, M. K.;Sahu, A.;Asangani, I. A.;Cao, Q.;Patel, L.;Vergara, I. A.;Davicioni, E.;Erho, N.;Ghadessi, M.;Jenkins, R. B.;Triche, T. J.;Malik, R.;Bedenis, R.;McGregor, N.;Ma, T.;Chen, W.;Han, S.;Jing, X.;Cao, X.;Wang, X.;Chandler, B.;Yan, W.;Siddiqui, J.;Kunju, L. P.;Dhanasekaran, S. M.;Pienta, K. J.;Feng, F. Y.; Chinnaiyan, A. M. The long noncoding RNA SChLAP1 promotes aggressive prostate cancer and antagonizes the SWI/SNF complex. Nature Genet. 2013, 45, 1392-1398. [30] Hessels, D.;Smit, F. P.;Verhaegh, G. W.;Witjes, J. A.;Cornel, E. B.; Schalken, J. A. Detection of TMPRSS2-ERG fusion transcripts and prostate cancer antigen 3 in urinary sediments may improve diagnosis of prostate cancer. Clin. Cancer Res. 2007, 13, 5103-5108. [31] Bryant, R. J.;Pawlowski, T.;Catto, J. W. F.;Marsden, G.;Vessella, R. L.;Rhees, B.;Kuslich, C.;Visakorpi, T.; Hamdy, F. C. Changes in circulating microRNA levels associated with prostate cancer. Br. J. Cancer 2012, 106, 768-774.

216

Chapter 5.3

[32] Bustin, S. A. Absolute quantification of mRNA using real-time reverse transcription polymerase chain reaction assays. J. Mol. Endocrinol. 2000, 25, 169-193. [33] Chen, C. F.;Ridzon, D. A.;Broomer, A. J.;Zhou, Z. H.;Lee, D. H.;Nguyen, J. T.;Barbisin, M.;Xu, N. L.;Mahuvakar, V. R.;Andersen, M. R.;Lao, K. Q.;Livak, K. J.; Guegler, K. J. Real-time quantification of microRNAs by stem-loop RT-PCR. Nucleic Acids Res. 2005, 33. [34] Koo, K. M.;Carrascosa, L. G.;Shiddiky, M. J. A.; Trau, M. Poly(A) extensions of miRNAs for amplification-free electrochemical detection on screen-printed gold electrodes. Anal. Chem. 2016, 88, 2000-2005. [35] Koo, K. M.;Carrascosa, L. G.;Shiddiky, M. J. A.; Trau, M. Amplification-free detection of gene fusions in prostate cancer urinary samples using mRNA-gold affinity interactions. Anal. Chem. 2016, 88, 6781-6788. [36] Xu, F. Z.;Shi, H.;He, X. X.;Wang, K. M.;He, D. G.;Guo, Q. P.;Qing, Z. H.;Yan, L. A.;Ye, X. S.;Li, D.; Tang, J. L. Concatemeric dsDNA-Templated copper nanoparticles strategy with improved sensitivity and stability based on rolling circle replication and its application in microRNA detection. Anal. Chem. 2014, 86, 6976- 6982. [37] Wang, Z. Y.;Si, L.;Bao, J. C.; Dai, Z. H. A reusable microRNA sensor based on the electrocatalytic property of heteroduplex-templated copper nanoclusters. Chem. Comm. 2015, 51, 6305-6307. [38] Bakker, E.; Qin, Y. Electrochemical sensors. Anal. Chem. 2006, 78, 3965-3983. [39] Das, J.;Ivanov, I.;Montermini, L.;Rak, J.;Sargent, E. H.; Kelley, S. O. An electrochemical clamp assay for direct, rapid analysis of circulating nucleic acids in serum. Nat. Chem. 2015, 7, 569-575. [40] Gliddon, H. D.;Howes, P. D.;Kaforou, M.;Levin, M.; Stevens, M. M. A nucleic acid strand displacement system for the multiplexed detection of tuberculosis-specific mRNA using quantum dots. Nanoscale 2016, 8, 10087-10095. [41] Zhang, P.;Zhang, J.;Wang, C.;Liu, C.;Wang, H.; Li, Z. Highly sensitive and specific multiplexed microRNA quantification using size-coded ligation chain reaction. Anal. Chem. 2014, 86, 1076-1082. [42] Kaffenberger, S. D.; Barbieri, C. E. Molecular subtyping of prostate cancer. Curr. Opinion. Urol. 2016, 26, 213-218.

217

Chapter 5.3

[42] Koo, K. M.;Wee, E. J.;Mainwaring, P. N.;Wang, Y.; Trau, M. Toward precision medicine: A cancer molecular subtyping nano‐strategy for RNA biomarkers in tumor and urine Small 2016, 12, 6233-6242.

SUPPLEMENTARY INFORMATION

Table S1. Oligonucleotide sequences used in experiments Oligos 5'-Sequence-3' CAACTGATAAGGCTTCCTGCCGCGCTCCAGG- T1E4 Capture Probe Biotin miR-107 Capture Probe TGATAGCCCTGTACAATGCTGCT-Biotin

SChLap1 Capture Probe TGTGTCCAGAACTGGTGGGTTCTTGGTCTC-Biotin

T1E4 RT-qRCR Forward ATTTAGGTACAACTCTTTCCCTCGTC Primer

T1E4 RT-qRCR Reverse Primer TGTATAGGAATCCCACTGAATTTTTC miR-107 RT-qRCR Forward AGCAGCATTGTACAG Primer miR-107 RT-qRCR Reverse TTTTTTTTTTTTTTTTTG Primer

SChLap1 RT-qRCR Forward GAGACCAAGAACCCACCAGTTCTGGACACA Primer

SChLap1 RT-qRCR Reverse CATCTGGCACTTCTTCCCCAGTCATTCCAT Primer

218

Chapter 5.3

Figure S1. Optimization of (a) polyA extension time; (b) CuSO4 concentration; and (c) sodium ascorbate concentration.

Figure S2. Fluorescence signals (580 nm) generated using various concentrations of (a) T1E4, (b) miR-107, and (c) SChLAP1. Error bars represent standard deviation of three independent experiments.

Figure S3. Electrochemical signals generated using various concentrations of (a) T1E4, (b) miR-107, and (c) SChLAP1. Error bars represent standard deviation of three independent experiments.

219

Chapter 5.3

Table S2. qRT-PCR validation of urinary sample results. Our Assay qRT-PCR

Sample T1E4 miR-107 SChLap1 T1E4 Ct miR-107 Ct SChLAP1 Ct

Status Status Status (Cycle) (Cycle) (Cycle)

P1 Normal Normal Abnormal 31.8 25.8 16.3

P2 Abnormal Abnormal Abnormal 15.3 16.9 16.7

P3 Abnormal Abnormal Normal 29.2 20.0 27.0

P4 Abnormal Abnormal Abnormal 14.7 15.7 18.7

P5 Normal Abnormal Abnormal 30.5 16.3 17.4

P6 Normal Abnormal Abnormal 31.4 16.9 17.4

P7 Abnormal Abnormal Abnormal 15.6 17.6 16.9

P8 Normal Abnormal Normal 32.4 16.4 27.3

P9 Abnormal Abnormal Abnormal 14.6 22.1 18.4

P10 Abnormal Normal Abnormal 19.1 25.8 21.2

H1 Normal Normal Normal 33.8 26.3 26.5

H2 Normal Normal Normal 31.2 25.9 26.3

220

Chapter 6

Simultaneous Analysis of Multiple Biomarkers via Multiplexing/High- Throughput Parallel Assays

Introduction As true for the majority of cancer types, prostate cancer is a highly heterogeneous disease which shows a large degree of inter- and intra-tumor molecular variability in patients. This makes the classification of prostate cancer into indolent and aggressive disease subtypes for precision therapy an extremely challenging task. Hence, to more accurately understand an individual cancer patient’s disease subtype, molecular biomarker diagnostic needs to detect multiple biomarkers (instead of a single biomarker) to provide a more complete screening outcome for clinical decision-making. Specifically for prostate cancer, the concurrent screening of multiple next-generation urinary RNA biomarkers from a single patient sample could improve clinical diagnosis, prognosis, and treatment prediction. To achieve this, multiple targets could be detected i) within a single tube by using multiple target-specific detection labels (i.e. multiplexed screening) or ii) using a multi-array approach whereby different biomarkers from a single sample are analyzed simultaneously (i.e. high-throughput parallel screening). Using these two concepts, this chapter describes the development of novel bioassays which enable simultaneous detection of multiple biomarkers for a more comprehensive prostate cancer diagnosis. Chapter 6.1 demonstrates the use of surface- enhanced Raman spectroscopy for rapid five-plexed detection of urinary biomarkers in prostate cancer subtyping. Chapter 6.2 showcases a novel technique which fuses multi-RNA target probe ligation and isothermal recombinase polymerase amplification for high-speed biosensing of different prostate cancer fusion gene variants. Chapter 6.3 describes a nanodevice that innovatively uses alternating current electrohydrodynamic (ac-EHD) fluidic manipulation to simultaneously profile the expression of three various types of next- generation urinary RNA biomarkers in a cohort of prostate cancer patients.

Chapter 6 is based on three published articles in Small (2016) [Chapter 6.1], Biosensors & Bioelectronics (2017) [Chapter 6.2], and Small (2018) [Chapter 6.3].

221

Chapter 6.1

Toward Precision Medicine: A Cancer Molecular Subtyping Nano-Strategy for RNA Biomarkers in Tumor and Urine

Kevin M. Koo, Eugene J. H. Wee, Paul N. Mainwaring, Yuling Wang, Matt Trau

ABSTRACT Cancer is a heterogeneous disease which manifests as different molecular subtypes. Due to the complex nature of tumor initiation, progression and metastatsis; cancer in different patients progresses at individual rates via various pathways. Currently, this cancer heterogeneity is largely unexploited, and the concept of precision medicine is to incorporate diagnostic technology to enable tailored treatments for different patients. To allow accurate cancer screening, early detection, therapy and monitoring; the characterization of multiple oncogenic biomarkers is required to characterize individualized cancer molecular subtypes. Despite the reliability of current multiplexed detection techniques; novel strategies are needed to resolve limitations such as long assay time, complex assay protocols, and difficulty in intepreting broad overlapping spectral peaks associated with conventional fluorescence readouts. Herein we present a rapid (80 min) multiplexed platform strategy for subtyping prostate cancer (PCa) tumor and urine samples based on their RNA biomarker profiles. This was achieved by combining rapid multiplexed isothermal reverse transcription-recombinase polymerase amplification (RT-RPA) of target RNA biomarkers with surface-enhanced Raman spectroscopy (SERS) nanotags for a “one-pot” assay. This is the first translational application of a RT-RPA/SERS-based platform for multiplexed cancer biomarker detection to address a clinical need. With excellent sensitivity of 200 zmol (100 copies) and specificity, we believe this platform methodology could be a useful tool for rapid multiplexed subtyping of cancers.

INTRODUCTION In recent years, there is overwhelming evidence that cancer is not just a single disease, but rather a variety of intricate subtypes which requires different treatment strategies.[1] For instance, prostate cancer (PCa) is a heterogeneous disease comprising of both clinically significant (aggressive and lethal) and insignificant (slow-growing and non-life threatening)

222

Chapter 6.1 subtypes. These different PCa subtypes could be characterized by their genetic (DNA, epigenetic and RNA expression) profiles,[2] and being able to stratify patients based on their profiles could potentially allow more effective personalized treatment decisions (i.e. precision medicine).[3] Recently, the potential of detecting multiple biomarkers for PCa subtyping and personalized treatment has been demonstrated in several reports.[4] The non-invasive use of urinary biomarkers is of particular interest due to the potential of eliminating financial, medical and psychological implications associated with biopsy procedures. Of note, novel urinary RNA biomarkers such as PCA3[5] and TMPRSS2:ERG fusion genes[6] were demonstrated to improve PCa diagnosis and risk stratification over the current use of a single prostate-specific antigen (PSA) biomarker. In these studies, the combination of both TMPRSS2:ERG and PCA3 was found to provide better risk prediction for high-grade PCa[4a, 7] Apart from diagnosis/prognosis, a splice variant of the androgen receptor (ARV7) has also been discovered recently to predict drug resistance response in advanced PCa patients,[8] and thus, the use of ARV7 for directing clinical decisions could potentially save valuable treatment time and cost. Thereby, it is of great clinical interest to develop a screening methodology for these promising PCa biomarkers simultaneously in a rapid manner. To meet this need, we herein describe a novel nano-subtyping platform for rapid multiplexed detection of PCa biomarkers, whilst addressing the limitations of methodologies used in current practice.

Existing technologies for multiplexed RNA biomarker detection and cancer molecular subtyping are generally dependent on established techniques such as quantitative real-time polymerase chain reaction (qRT-PCR);[4a, 4e] direct hybridization;[9] and next-generation sequencing (NGS).[3b] While effective for multiplexing, these techniques are limited by tedious protocols, slow turn-around time, high instrument cost, and requirement for high starting sample amount. Additionally, the multiplexed fluorescence-based readouts of traditional techniques are limited by large spectral overlaps and thus, convolute the interpretation of multiplexed detection data. To allow for fast, convenient, and sensitive amplification of trace RNA biomarker levels in biological samples, reverse transcription- recombinase polymerase amplification (RT-RPA) is a viable alternative to traditional RT- PCR techniques. RT-RPA is an isothermal amplification technique[10] which utilizes enzymatic strand separation during the amplification process instead of heat denaturation used in RT-PCR. This removal of thermal cycling steps has allowed rapid and cost-effective RT-RPA detection of a wide variety of human disease-related RNA targets in literature.[11]

223

Chapter 6.1

However, to the best of our knowledge, there still exists a lack of highly-multiplexed RT- RPA techniques which could be a potential solution for faster cancer molecular subtyping in patient samples. Surface-enhanced Raman spectroscopy (SERS) is an ideal candidate technique for detecting multiple target molecules (including RNA[12]) present in a single sample. SERS is a method which involves the adsorption of molecules (or labels) onto metal nanoparticles (eg. gold or silver) surfaces[13] to significantly amplify the Raman scattering effect through electromagnetic field amplification (via excitation of the localized surface plasmon of nanoparticles). Typically, target molecules are labelled with SERS labels (Raman probe molecules-coated metallic nanoparticles) for enhanced detection down to single-molecule level.[14] The unique fingerprint spectra of each specific label and the narrow well-separated SERS spectral peaks allow for the detection of multiple target molecules concurrently with easy interpretation unlike multiplexed fluorescence detection. The magnetic isolation of DNA and protein targets for labelled SERS detection has been investigated,[15] and shown to be highly sensitive as well as readily multiplexable. Yet, possibly due to the lack of a robust amplification techniques, similar applications for multiplexed RNA detection has rarely been demonstrated. Therefore, considering the advantages of SERS, an assay which integrates multiplexed RT-RPA of RNA biomarkers with “one-pot” SERS-based readout is a potentially attractive technique for PCa molecular subtyping.

Herein, for the first time, we described a five-plexed assay for simultaneously detecting promising next-generation biomarkers in PCa. Our proposed assay offered distinct benefits: (i) multiplexed biomarker detection could be achieved within a short time-frame with rapid isothermal RT-RPA and “one-pot” SERS readout; (ii) easy interpretation of detection outcomes from well-resolved SERS spectral peaks without the need for further data processing; (iii) applicable to clinical samples such as tissue biopsy specimens, and urine samples. With high specificity and zmol-level sensitivity, the combination of RT-RPA and SERS shows great potential as a cancer screening platform for multiplexed molecular subtyping.

224

Chapter 6.1

EXPERIMENTAL SECTION Reagents and Materials All reagents were of analytical grade and purchased commercially (Sigma Aldrich, Australia). UltraPureTM DNase/RNase-free distilled water (Invitrogen, Australia) was used throughout the experiments. Oligonucleotides were synthesized and purchased commercially (Integrated DNA Technologies, Singapore) and sequences are shown in Table S1.

Synthesis of SERS Nanotags: SERS nanotags were synthesized by the coating of Raman reporters and DNA probes on the AuNPs surface.[24] Briefly, AuNPs were synthesized by [25] citrate reduction of HAuCl4 and AuNPs (1 mL) was then mixed with TCEP-treated thiolated DNA oligonucleotides (10 μL, 50 μM) (IDT, Singapore) at room temperature (RT) for 12 hours. Then, Raman probe (100 μL, 1 mM): 4-mercaptobenzoic acid (MBA); 2,7- mercapto-4-methylcoumarin (MMC); 2-mercapto-4-methyl-5-thiazoleacetic acid (MMTAA); 4-mercapto-3-nitrobenzoic acid (MNBA); or 2,3,5,6-tetrafluoro-4-mercaptobenzoic acid (TFMBA) was added to the AuNPs for incubation at RT overnight. Afterwhich, NaCl (0.6 M) in PBS (1 mM) was used to age the SERS nanotags at RT for 12 hours before centrifugation and resuspension of SERS nanotags into PBS buffer (10 mM) prior to use. The final concentration (0.5 nM) of the SERS nanotags mix was determined by the UV extinction spectra of AuNPs.

Cell Culture and RNA Extraction DuCap, LnCap, and 22Rv1 cells were cultured in RPMI-1640 growth media (Life Technologies, Australia) supplemented with fetal bovine serum (10%) (Life Technologies,

Australia) in a humidified incubator containing CO2 (5%) at 37ºC. Total RNA was extracted using Direct-zolTM RNA MiniPrep kit (Zymo Research, USA) according to manufacturer’s instructions. Cultured cells were lysed using TRIzol® reagent (Life Technologies, Australia) before addition of ethanol. Then, the mixture was transferred into a supplied Zymo-SpinTM IIC Column in a collection tube for centrifugation. After several wash steps, the purified RNA was eluted with RNase-free water (50 µL). The RNA integrity and purity were checked using a NanoDrop Spectrophotometer (Thermo Fisher Scientific, Australia).

225

Chapter 6.1

RNA Extraction from Patient Tumor and Urine Samples Ethics approval was obtained from The University of Queensland Institutional Human Research Ethics Committee (Approval No. 201400012) and informed consent was obtained from all subjects prior to sample collection. Methods pertaining to clinical samples were carried out in accordance with approved guidelines. Voided urinary and biopsy samples were collected from men patients undergoing treatment for prostate cancer.

Total RNA was extracted from the urine samples using the commercially-available ZR Urine RNA Isolation KitTM (Zymo Research, USA). For each sample, urine (30 mL) was passed through the supplied ZRC GFTM Filter to retain urinary cells. Then, supplied urine RNA buffer (700 µL) was passed through the filter and the filtered cells were collected in the flow- through. Next, cells from the flow-through were lysed, and total RNA was washed and eluted in RNase-free water (10 µL).

Total RNA was extracted from biopsy samples using Qiagen All-Prep according to manufacturer’s instructions. Briefly, tissue samples were disrupted and homogenized before being transferred to an AllPrep DNA spin column for centrifugation. The flow-through was used for total RNA purification by first transferring into an RNeasy spin column. After several steps of washing and centrifugation, the purified RNA was eluted in RNase-free water (30 µL).

Multiplexed RT-RPA For multiplexed amplification of target biomarkers, the TwistAmp Basic RT kit (Twist-DX, UK) was used with slight modifications to manufacturer’s instructions. Each dried-freeze pellet was resuspended in buffer (29.5 µL) and DNase/RNase-free water (4.5 µL) before splitting into four equal aliqots. Then, each primer (375 nM) (Table S1) and purified RNA (1 µL, 50 ng) were added to make a reaction volume (12.5 µL) prior to incubation at 41°C for 20 min.

Hybridization to SERS Nanotags and Magnetic Isolation After amplification, RPA amplicons (10 µL) were incubated with MBA- (1 µL), MMC- (2 µL), MMTAA- (5 µL), MNBA- (5 µL), and TFMBA- (2 µL) SERS nanotags for 20 min at 37 °C. It is essential to highlight that due to the non-identical SERS signal levels in the stock solutions, the various SERS nanotags were added in different amounts such that a 1:1:1:1:1

226

Chapter 6.1

SERS signal was obtained for a 1:1:1:1:1 target mix. Before application to patient samples, we used cell line RNA to optimize the different amounts of added SERS nanotags (data not shown), and provide accurate quantification of target biomarkers. After incubation with SERS nanotags, streptavidin-magnetic beads (5 µL) (New England Biolabs, Australia) was added to the mixture for 10 min of incubation at room temperature. The amplicons-SERS nanotags complexes were then isolated on a magnetic plate with three consecutive washes of wash buffer (1 mM PBS buffer, 0.01% Tween20).

One-Pot SERS Detection of Biomarkers After the final wash, the pellet was then resuspended in PBS buffer (60 µL, 1 mM) and transferred to a quartz cuvette for SERS measurements on the IM-52 portable Raman microscope (Snowy Range Instruments, USA). An average SERS spectrum was obtained for each sample from five acquisitions of 2 sec integration by a 785 nm excitation laser at 70 mW. In all, this instantaneous SERS readout in combination with 20 min of RNA extraction; 20 min of multiplexed RT-RPA; 30 min of SERS naonotag hybridization and magnetic isolation; and 10 min of overall sample preparation time, gave a sample-to-answer assay time of 80 min.

Quantitative Comparison of Biomarker Levels To account for variations in starting sample RNA amounts, the Raman intensity of a target biomarker was normalized with respect to the stably-expressed housekeeping RN7SL1 RNA (i.e. loading control) of the same sample. The normalized Raman intensity which reflected the respective biomarker level (i.e. higher biomarker level led to higher Raman intensity), allowed for convenient biomarker level comparisons between samples:

Normalized Raman Intensity = (iBiomarker / iRN7SL1) (Equation 1)

where iBiomarker is the average Raman intensity for target RNA (T1E4, RN7SL1, T1E5,

ARV7, or PCA3), and iRN7SL1 is the average Raman intensity for RN7SL1 RNA loading control of the same sample. qRT-PCR Validation The KAPA SYBR® FAST One-Step qRT-PCR kit (KAPA BIOSYSTEMS, Australia) was used to set up a single reaction volume (10 µL) for each sample. Each reaction volume

227

Chapter 6.1 consist of KAPA SYBR® FAST qPCR Master Mix (1X), each forward and reverse primer (200 nM) (Table 1), KAPA RT Mix (1X), ROX dye (50 nM) and cell line total RNA template (30 ng). RT-qPCR was performed using the Applied Biosystems® 7500 Real-Time PCR System (Thermo Fisher Scientific, Australia). The cycling protocol was: 42°C for 10 min to synthesize cDNA, followed by 95°C for 5 min to deactivate RT before cycling 35 times (95°C for 30 s, 50°C for 30 s and 72°C for 1 min) and finished with 72°C for 10 min.

RESULTS AND DISCUSSION Working Principle The working principle of our five-plexed screening assay is illustrated in Figure 1. Firstly, the target RNA biomarkers in the sample are isothermally amplified concurrently by multiplexed RT-RPA using biomarker-specific primers. By use of modified primers, the generated amplicons are tagged with biotin molecules and target-specific 5’ overhang barcode sequences on either ends. Each 5’ barcode overhang sequence is achieved by inclusion of an internal carbon linker in the primer which prevents DNA extension on the complementary strand. Thus, this leaves the barcode sequence single-stranded to facilitate hybridization with complementary sequences on SERS nanotags. Next, all the amplicon- SERS nanotag complexes are attached to streptavidin (SA)-coated magnetic beads via biotin tags. These attachments facilitate slight AuNPs aggregation (Figure S1, Supporting Information) on the magnetic beads for higher SERS enhancement[16]. Finally, the magnetically enriched SERS-labelled amplicons are interrogated by Raman spectrometer to both identify (unique spectral peak) and quantify (peak intensity) the targets present in the sample.

In this study, we selected a panel of five RNA targets comprising of next-generation biomarkers which are exceptionally promising[8, 17] for PCa subtyping and risk stratification. The targets include two most common TMPRSS2:ERG gene fusion variants: TMPRSS2 exon 1-ERG exon 4 (T1E4) and TMPRSS2 exon 1-ERG exon 5 (T1E5); PCA3; ARV7; and a stably-expressed housekeeping RNA (RN7SL1)[18].

228

Chapter 6.1

Figure 1. Schematic representation of assay. Total RNA is firstly isolated from samples before target RNA biomarkers are simultaneously amplified using isothermal reverse transcription-recombinase polymerase amplification. During amplification, amplicons are tagged with biotin molecules and target-specific overhang hybridization sequences. The different biomarker-specific amplicons are then labelled with respective SERS nanotags through complementary sequence hybridization and magnetically-purified. Finally, the amplicons are detected by SERS concurrently and quantitative analysis of biomarker level is derived from spectral peak of each unique SERS nanotag.

Characterization of SERS Nanotags Five different Raman probe molecules, namely MBA; MMC; MMTAA; MNBA; and TFMBA (see experimental section for the full chemical names), were used to synthesize the SERS nanotags in our experiments. These Raman probes were assembled along with respective oligonucleotide sequences (Table S1, Supporting Information) on surfaces of gold nanoparticles (AuNPs) to generate different SERS nanotags. To ensure successful synthesis and function of the five different SERS nanotags, we first used transmission electron microscopy (TEM) to image the SERS nanotags (Figure 2A). The SERS nanotags were generally uniform in size at ~70 nm diameter. Next, we characterized the different SERS nanotags by UV/Vis absorption spectroscopy (Figure 2B). Here, as consistent with a refractive index change due to surface functionalization,[19] a spectral peak red shift for Raman probes/oligonucleotides-modified AuNPs was observed as compared to bare AuNPs, Finally, each SERS nanotag was analyzed both individually and as a mixture of all five nanotags. As shown in Figure 2C, the individual SERS signal for each nanotag was

229

Chapter 6.1 successfully detected and found to be different from each other. Distinct peak of each different Raman probe was observed at 1075 cm-1 (MBA), 1175 cm-1 (MMC), 1285 cm-1 (MMTAA), 1338 cm-1 (MNBA), 1380 cm-1 (TFMBA). These unique spectral peaks were distinct and easily-distinguished even during multiplexed detection of all five labels (Figure S2, Supporting Information). These characterization experiments thus demonstrated that our synthesized SERS nanotags were functional and able to generate distinctive SERS signals for multiplexed detection of target molecules.

Figure 2. Characterization of SERS nanotags. (A) Typical TEM image of SERS nanotags. (B) UV/Vis absorption spectra of bare AuNPs and different SERS nanotags. (C) Unique SERS spectral peaks of individual SERS nanotags (Left). Molecular structures of Raman probes used for synthesis of each SERS nanotag (Right).

230

Chapter 6.1

Assay Sensitivity A serial dilution of in vitro transcribed T1E4 RNA was first used to evaluate the sensitivity of our assay. SERS measurements were conducted following multiplexed RT-RPA amplification, SERS nanotag hybridization, and magnetic isolation. We observed that the Raman intensity associated with T1E4 (1075 cm-1) increased in a copy number-dependent manner, and the negative no-target control showed negligible background signal (Figure 3A). The limit of detection (LOD) of our assay was determined to be 100 copies (200 zmol) with a linear range of log(1-106) copies (Figure 3B). Moreover, the increase in Raman intensity with higher amount of input RNA also demonstrated the feasibility of the assay for quantitative biomarker detection. This could be useful for applications such as the monitoring of biomarker levels to assess drug response during therapy.

In addition, multiplexed detection sensitivity in a biologically-complex sample (e.g. cell line- derived RNA) was also evaluated. To this end, we titrated known amounts of in vitro transcribed ARV7 RNA into background of LnCAp cell line-isolated RNA. As ARV7 is absent in LnCap, we could readily observed Raman intensity changes due to different ARV7 levels. Figure 3C & D showed that the ARV7 Raman intensity (1338 cm-1) increased proportionally with increasing ARV7 copies whilst signals from background LnCap- endogenous RN7SL1 (1175 cm-1) and PCA3 (1375 cm-1) RNA remained constant. Similar to T1E4, the LOD for ARV7 in a complex system was maintained at approximately 100 copies with a linear range of log(1-106) copies (Figure S3, Supporting Information). Together, the data suggested a sensitive and robust approach for targeted RNA detection.

Our observed zmol-level LOD is generally less than the level of RNA material in a single cell,[20] and therefore adequate and suitable for detection in real clinical samples. This LOD is also superior to existing multiplexed SERS label-based assays[21] that reported higher LOD and/or required starting sample amount for disease detection. We reasoned that our higher assay sensitivity was due to dual signal amplification through isothermal amplification as well as SERS effect. It is also worthy to note that our assay could be performed in a much shorter assay timeframe (80 min) as compared to conventional techniques such as qRT-PCR, fluorescence in-situ hybridization (FISH), or NGS (at least 2.5 hours from sample-to-answer).

As a 70 mW laser was used to enable SERS in our assay, we investigated if assay performance such as sensitivity could be affected by sample degradation via laser excitation.

231

Chapter 6.1

To this end, samples were interrogated by laser and Raman intensities were recorded over multiple scans. Our findings indicated no significant sample degradation as the average Raman intensities remained constant with increasing number of signal acquisition (i.e. longer laser exposure) even at a low input amount of 100 ARV7 copies (Figure S4, Supporting Information). We reasoned that this was due to the small detection area that was being interrogated by a fixed laser beam in relation to the large body of sample volume; thus it was unlikely that freely-diffusing targets within the sample solution could be degraded by laser exposure during a single signal acquisition.

Figure 3. Assay sensitivity. (A) Average SERS spectra corresponding to different copy numbers of input T1E4 RNA. (B) Analogous linear calibration plot for T1E4 copy number titration. (C) Average SERS spectra corresponding to different copy numbers of input ARV7 RNA in background of LnCap RNA. (D) Corresponding normalized Raman intensities for ARV7 copy number titration. Error bars represent standard deviation of three independent experiments (n = 3).

Assay Specificity To evaluate assay specificity, we used our full set of RT-RPA primers and SERS nanotags to detect target biomarkers in three different PCa cell lines individually. Each of the three

232

Chapter 6.1 chosen cell lines carries different target biomarker(s) of interest: DuCap is positive for T1E4 and T1E5; LnCap is positive for PCA3; and 22Rv1 cell line is positive for ARV7 and PCA3. Additionally, all three cell lines were tested for RN7SL1 housekeeping RNA as a positive loading control. We assigned the five different SERS nanotags for the detection of each target: T1E4 (MBA, 1075 cm-1); RN7SL1 (MMC, 1175 cm-1); PCA3 (MMTAA, 1285 cm-1); T1E5 (MNBA, 1338 cm-1); ARV7 (TFMBA, 1380 cm-1). Figure 4 showed the specific SERS detection of biomarkers associated with each particular cell line. Typically, DuCap was only positive for T1E4, T1E5, and RN7SL1, whilst spectral peaks for ARV7 and PCA3 were noticeably absent (Figure 4A1 & A2); LnCap was only positive for PCA3 and RN7SL1 (Figure 4B1 & B2); and 22Rv1 was only positive for ARV7, PCA3, and RN7SL1 (Figure 4C1 & C2). These results coupled with the low background signals of the negative no-target controls (Figure 4) confirmed the specificity of the RT-RPA primers during selective amplification of targets, as well as the effect of stringent wash steps to remove non- specifically bound SERS nanotags on the magnetic bead surface. This high assay specificity was important for accurately discriminating target sequences in clinical blood or urine samples with large amounts of non-targets.

233

Chapter 6.1

Figure 4. Assay specificity. Average SERS spectra of multiplexed biomarker profiling in (A1) DuCap, (B1) LnCap, and (C1) 22Rv1 cell lines. Corresponding normalized Raman intensities are shown in (A2), (B2), and (C2) respectively. Error bars represent standard deviation of three independent experiments (n = 3).

Multiplexed Detection of PCa Biomarkers To demonstrate the multiplexing capability of our assay, we simultaneously detected the five PCa biomarkers by mixing purified total RNA from three different PCa cell lines. As shown

234

Chapter 6.1 in Figure 5A, all five targets could be detected and identified by their unique SERS spectral peaks from the mixture of different cell lines. Additionally, the normalized Raman intensities (Equation 1) of the respective targets reflected the different biomarker levels quantitatively (Figure 5B). The biomarker level trend determined by our assay outcome was subsequently validated by qRT-PCR with good agreement (Figure S5, Supporting Information). This experiment thus exhibited the multiplexing capability of our assay in biologically-complex cell line samples.

Figure 5. Multiplexed detection of biomarkers in prostate cancer cell lines. (A) Simultaneous SERS detection of five different biomarkers in a mixture of total RNA extracted from DuCap, LnCap, and 22Rv1 cells. (B) Corresponding normalized Raman intensities. Error bars represent standard deviation of three independent experiments (n = 3).

Multiplexed Biomarker Detection in Patient Tumor and Urine Samples The central aim of our effort was to allow simultaneous interrogation of PCa biomarkers in patient samples with a rapid multiplexed SERS assay. To demonstrate clinical feasibility, we challenged our assay by profiling biomarker levels in samples from PCa patients. We

235

Chapter 6.1 performed our five-plexed SERS assay on extracted total RNA from tumor (Figure 6A & B) and urine (Figure 6C-E) samples of different patients, and successfully measured different levels of four different PCa biomarkers (T1E4, T1E5, ARV7, and PCA3) in addition to the RN7SL1 loading controls for all patients. The normalized Raman intensities (Equation 1) of each biomarker allowed for easy quantitative comparisons of biomarker levels between samples. For instance, we observed that T1E4 level in U1 (Figure 6C2) was about 3-fold higher than in U2 (Figure 6C2) after normalizing T1E4 Raman intensities to the respective stably-expressed RN7SL1 housekeeping RNA Raman intensities (i.e. normalized biomarker levels to account for different starting sample RNA amounts).

We also validated our results with current standard qRT-PCR approach with 100% concordance (Figure S6-10, Supporting Information). In comparison to existing SERS techniques for RNA detection[12, 22], we found that our proposed assay: i) possessed superior detection sensitivity; ii) successfully showed multiplexed RNA detection for more than three targets for the first time; and most importantly iii) demonstrated clinical translation potential with real cancer patient samples.

The biomarker profiles obtained for the different PCa patients reflected the heterogeneity of the disease as each patient exhibited a different profile from each other. The availability of such PCa molecular profiling based on multiple biomarker levels could pave the way for widespread personalized cancer treatment in the near future. Given that numerous cancer molecular subtypes exist amongst different patients, the multiplexed profiling of relevant oncogenic biomarkers by our proposed strategy could aid in providing more timely and effective tailored-made therapy plans for each individual cancer patient and/or aid in biomarker level monitoring during drug treatment. Additionally, this would be a significant improvement over the single PSA biomarker test used in current PCa screening. The low specificity of PSA has been associated with high number of false-positive and -negative PCa diagnoses, leading to unnecessary medical procedures such as biopsies and prostatectomies. The ability of our assay to simultaneously detect multiple PCa biomarkers could provide a more informative and accurate PCa diagnosis to spare patients from redundant medical expenses. Lastly, our assay produced the same detection accuracy and reliability as qRT- PCR, making it attractive for facilitating rapid patient diagnosis in a clinical setting.

236

Chapter 6.1

Figure 6. Clinical applications of assay. Average SERS spectra of multiplexed biomarker profiling in patient biopsy tissue (A1 & B1), and urine (C1, D1, E1) samples. Corresponding normalized Raman intensities are shown in (A2), (B2), (C2), (D2), and (E2) respectively. Error bars represent standard deviation of three independent experiments (n = 3).

237

Chapter 6.1

CONCLUSION In summary, with an aim to better characterize PCa subtypes in a quick, sensitive, and quantitative fashion; we have developed a multiplexed assay that is capable of simultaneously detecting several biomarkers in patient samples quantitatively within a short timeframe of 80 min. Our proposed assay involved the use of multiplexed isothermal RT-RPA to amplify RNA targets in the sample before labelling the amplicons with different synthesized SERS nanotags for rapid one-pot SERS detection. We demonstrated that our proposed assay was capable of highly sensitive (200 zmol) and specific PCa molecular profiling in clinical urine and tissue specimens.

By combining the benefits of multiplexed isothermal amplification and one-pot SERS detection, our method firstly obviated the need for lengthy thermal cycling procedures associated with conventional amplification methods. Secondly, the multiplexed SERS readout offered speedy detection of target biomarkers with well-resolved spectral peaks to ensure easy interpretation of data. Thirdly, clinical application potential was demonstrated for use on real patient samples, with the detection of biomarkers in patient urine samples as an especially promising method for non-invasive PCa diagnostics.

We envisioned that our assay offers considerable potential for further improvement. First, the assay throughput could be scaled up by synthesizing more types of SERS nanotags with different encoded Raman probes. In literature, the successful fabrication of 31 different SERS-encoded nanoparticles with unique vibrational fingerprint has been reported.[23] Based on this, we could potentially create a similar number of SERS nanotags for detection of more cancer biomarkers. Second, besides the detection of RNA biomarkers, the proposed assay could be readily adopted for the detection of other oncogenic biomarkers such as single nucleotide variants (SNVs) or DNA methylation by changing RT-RPA primers. The analysis of different cancer mutations could allow a more explicit snapshot of tumor status. Furthermore, apart from cancer detection, our assay is applicable for biomarker detection in other human diseases as well. Third, with the emergence of portable handheld Raman spectrometers, the SERS readout could be made more convenient for on-site testing. Hence, we believe the present demonstration and promise of our proposed assay is a potential approach for PCa molecular subtyping and personalized therapy.

238

Chapter 6.1

REFERENCES [1] a) J. S. de Bono, A. Ashworth, Nature 2010, 467, 543; b) F. S. Collins, H. Varmus, N. Engl. J. Med. 2015, 372, 793. [2] a) C. S. Grasso, A. K. Cani, D. H. Hovelson, M. J. Quist, N. J. Douville, V. Yadati, A. M. Amin, P. S. Nelson, B. L. Betz, C. J. Liu, K. E. Knudsen, K. A. Cooney, F. Y. Feng, A. S. McDaniel, S. A. Tomlins, Annals Oncol. 2015, 26, 1110; b) S. A. Tomlins, M. Alshalalfa, E. Davicioni, N. Erho, K. Yousefi, S. Zhao, Z. Haddad, R. B. Den, A. P. Dicker, B. J. Trock, A. M. DeMarzo, A. E. Ross, E. M. Schaeffer, E. A. Klein, C. Magi-Galluzzi, R. J. Karnes, R. B. Jenkins, F. Y. Feng, Eur. Urol. 2015, 68, 555. [3] a) G. Attard, H. Beltran, Annals Oncol. 2015, 26, 1041; b) D. Robinson, E. M. Van Allen, Y. M. Wu, N. Schultz, R. J. Lonigro, J. M. Mosquera, B. Montgomery, M. E. Taplin, C. C. Pritchard, G. Attard, H. Beltran, W. Abida, R. K. Bradley, J. Vinson, X. H. Cao, P. Vats, L. P. Kunju, M. Hussain, F. Y. Feng, S. A. Tomlins, K. A. Cooney, D. C. Smith, C. Brennan, J. Siddiqui, R. Mehra, Y. Chen, D. E. Rathkopf, M. J. Morris, S. B. Solomon, J. C. Durack, V. E. Reuter, A. Gopalan, J. J. Gao, M. Loda, R. T. Lis, M. Bowden, S. P. Balk, G. Gaviola, C. Sougnez, M. Gupta, E. Y. Yu, E. A. Mostaghel, H. H. Cheng, H. Mulcahy, L. D. True, S. R. Plymate, H. Dvinge, R. Ferraldeschi, P. Flohr, S. Miranda, Z. Zafeiriou, N. Tunariu, J. Mateo, R. Perez- Lopez, F. Demichelis, B. D. Robinson, M. Schiffman, D. M. Nanus, S. T. Tagawa, A. Sigaras, K. W. Eng, O. Elemento, A. Sboner, E. I. Heath, H. I. Scher, K. J. Pienta, P. Kantoff, J. S. de Bono, M. A. Rubin, P. S. Nelson, L. A. Garraway, C. L. Sawyers, A. M. Chinnaiyan, Cell 2015, 161, 1215. [4] a) G. Leyten, D. Hessels, S. A. Jannink, F. P. Smit, H. de Jong, E. B. Cornel, T. M. de Reijke, H. Vergunst, P. Kil, B. C. Knipscheer, I. M. van Oort, P. F. A. Mulders, C. A. Hulsbergen-van de Kaa, J. A. Schalken, Eur. Urol. 2014, 65, 534; b) R. J. Bryant, D. D. Sjoberg, A. J. Vickers, M. C. Robinson, R. Kumar, L. Marsden, M. Davis, P. T. Scardino, J. Donovan, D. E. Neal, H. Lilja, F. C. Hamdy, JNCI-J. Natl. Cancer Inst. 2015, 107, djv095; c) H. Gronberg, J. Adolfsson, M. Aly, T. Nordstrom, P. Wiklund, Y. Brandberg, J. Thompson, F. Wiklund, J. Lindberg, M. Clements, L. Egevad, M. Eklund, Lancet Oncol. 2015, 16, 1667; d) G. Leyten, D. Hessels, F. P. Smit, S. A. Jannink, H. de Jong, W. J. G. Melchers, E. B. Cornel, T. M. de Reijke, H. Vergunst, P. Kil, B. C. Knipscheer, C. A. Hulsbergen-van de Kaa, P. F. A. Mulders, I. M. van

239

Chapter 6.1

Oort, J. A. Schalken, Clin. Cancer Res. 2015, 21, 3061; e) L. Mengual, J. J. Lozano, M. Ingelmo-Torres, L. Izquierdo, M. Musquera, M. J. Ribal, A. Alcaraz, BMC Cancer 2016, 16, 76. [5] M. J. G. Bussemakers, A. van Bokhoven, G. W. Verhaegh, F. P. Smit, H. F. M. Karthaus, J. A. Schalken, F. M. J. Debruyne, N. Ru, W. B. Issacs, Cancer Res. 1999, 59, 5975. [6] S. A. Tomlins, D. R. Rhodes, S. Perner, S. M. Dhanasekaran, R. Mehra, X. W. Sun, S. Varambally, X. H. Cao, J. Tchinda, R. Kuefer, C. Lee, J. E. Montie, R. B. Shah, K. J. Pienta, M. A. Rubin, A. M. Chinnaiyan, Science 2005, 310, 644. [7] a) S. A. Tomlins, S. M. J. Aubin, J. Siddiqui, R. J. Lonigro, L. Sefton-Miller, S. Miick, S. Williamsen, P. Hodge, J. Meinke, A. Blase, Y. Penabella, J. R. Day, R. Varambally, B. Han, D. Wood, L. Wang, M. G. Sanda, M. A. Rubin, D. R. Rhodes, B. Hollenbeck, K. Sakamoto, J. L. Silberstein, Y. Fradet, J. B. Amberson, S. Meyers, N. Palanisamy, H. Rittenhouse, J. T. Wei, J. Groskopf, A. M. Chinnaiyan, Science Transl. Med. 2011, 3, 94ra72; b) S. A. Tomlins, J. R. Day, R. J. Lonigro, D. H. Hovelson, J. Siddiqui, L. P. Kunju, R. L. Dunn, S. Meyer, P. Hodge, J. Groskopf, J. T. Wei, A. M. Chinnaiyan, Eur. Urol. 2015, doi:10.1016/j.eururo.2015.04.039. [8] E. S. Antonarakis, C. Lu, H. Wang, B. Luber, M. Nakazawa, J. C. Roeser, Y. Chen, T. A. Mohammad, Y. Chen, H. L. Fedor, T. L. Lotan, Q. Zheng, A. M. De Marzo, J. T. Isaacs, W. B. Isaacs, R. Nadal, C. J. Paller, S. R. Denmeade, M. A. Carducci, M. A. Eisenberger, J. Luo, N. Engl. J. Med. 2014, 371, 1028. [9] a) M. E. Lira, T. M. Kim, D. Huang, S. Deng, Y. Koh, B. Jang, H. Go, S.-H. Lee, D. H. Chung, W. H. Kim, E. F. P. M. Schoenmakers, Y.-L. Choi, K. Park, J. S. Ahn, J.- M. Sun, M.-J. Ahn, D.-W. Kim, M. Mao, J. Mol. Diagnostics 2013, 15, 51; b) J. I. Warrick, S. A. Tomlins, S. L. Carskadon, A. M. Young, J. Siddiqui, J. T. Wei, A. M. Chinnaiyan, L. P. Kunju, N. Palanisamy, Mod. Pathol. 2014, 27, 609. [10] O. Piepenburg, C. H. Williams, D. L. Stemple, N. A. Armes, Plos Biol. 2006, 4, 1115- 1121. [11] a) A. Abd El Wahed, P. Patel, D. Heidenreich, F. T. Hufert, M. Weidmann, PLOS Curr. 2013, 5, doi: 10.1371/currents.outbreaks.62df1c7c75ffc96cd59034531e2e8364; b) M. Euler, Y. J. Wang, D. Heidenreich, P. Patel, O. Strohmeier, S. Hakenberg, M. Niedrig, F. T. Hufert, M. Weidmann, J. Clin. Microbiol. 2013, 51, 1110; c) H. Zaghloul, M. El-Shahat, World J. Hepatol. 2014, 6, 916; d) B.-T. Teoh, S.-S. Sam, K.-K. Tan, M. B. Danlami, M.-H. Shu, J. Johari, P.-S. Hooi, D. Brooks, O.

240

Chapter 6.1

Piepenburg, O. Nentwich, A. Wilder-Smith, L. Franco, A. Tenorio, S. AbuBakar, J. Clin. Microbiol. 2015, 53, 830; e) E. J. H. Wee, M. Trau, ACS Sens. 2016, 1, 670. [12] a) Y. W. C. Cao, R. C. Jin, C. A. Mirkin, Science 2002, 297, 1536; b) L. Sun, J. Irudayaraj, J. Phys. Chem. B 2009, 113, 14021; c) E. Prado, N. Daugey, S. Plumet, L. Servant, S. Lecomte, Chem. Comm. 2011, 47, 7425. [13] a) M. Fleischmann, P. J. Hendra, McQuilla.Aj, Chem. Phys. Lett. 1974, 26, 163; b) D. L. Jeanmaire, R. P. Vanduyne, J. Electroanal. Chem. 1977, 84, 1; c) M. G. Albrecht, J. A. Creighton, J. Am. Chem. Soc. 1977, 99, 5215. [14] a) D. Graham, C. McLaughlin, G. McAnally, J. C. Jones, P. C. White, W. E. Smith, Chem. Comm. 1998, 1187; b) D. Graham, B. J. Mallinder, W. E. Smith, Angew. Chem., Int. Ed. 2000, 39, 1061; c) K. Faulds, F. McKenzie, W. E. Smith, D. Graham, Angew. Chem., Int. Ed. 2007, 46, 1829. [15] a) H. Chon, S. Lee, S. W. Son, C. H. Oh, J. Choo, Anal. Chem. 2009, 81, 3029-3034; b) J. Baniukevic, I. H. Boyaci, A. G. Bozkurt, U. Tamer, A. Ramanavicius, A. Ramanaviciene, Biosens. Bioelectron. 2013, 43, 281-288; c) J. Neng, M. H. Harpster, W. C. Wilson, P. A. Johnson, Biosens. Bioelectron. 2013, 41, 316-321; d) T. Donnelly, W. E. Smith, K. Faulds, D. Graham, Chem. Comm. 2014, 50, 12907-12910; e) H. T. Ngo, N. Gandra, A. M. Fales, S. M. Taylor, T. Vo-Dinh, Biosens. Bioelectron. 2016, 81, 8-14. [16] A. M. Schwartzberg, C. D. Grant, A. Wolcott, C. E. Talley, T. R. Huser, R. Bogomolni, J. Z. Zhang, J. Phys. Chem. B 2004, 108, 19191-19197. [17] J. R. Prensner, M. A. Rubin, J. T. Wei, A. M. Chinnaiyan, Science Transl. Med. 2012, 4, 127rv3. [18] C. R. Galiveti, T. S. Rozhdestvensky, J. Brosius, H. Lehrach, Z. Konthur, RNA 2010, 16, 450-461. [19] a) J. J. Storhoff, R. Elghanian, R. C. Mucic, C. A. Mirkin, R. L. Letsinger, J. Am. Chem. Soc. 1998, 120, 1959-1964; b) G. Raschke, S. Kowarik, T. Franzl, C. Sonnichsen, T. A. Klar, J. Feldmann, A. Nichtl, K. Kurzinger, Nano Lett. 2003, 3, 935-938. [20] V. E. Velculescu, S. L. Madden, L. Zhang, A. E. Lash, J. Yu, C. Rago, A. Lal, C. J. Wang, G. A. Beaudry, K. M. Ciriello, B. P. Cook, M. R. Dufault, A. T. Ferguson, Y. H. Gao, T. C. He, H. Hermeking, S. K. Hiraldo, P. M. Hwang, M. A. Lopez, H. F. Luderer, B. Mathews, J. M. Petroziello, K. Polyak, L. Zawel, W. Zhang, X. M.

241

Chapter 6.1

Zhang, W. Zhou, F. G. Haluska, J. Jen, S. Sukumar, G. M. Landes, G. J. Riggins, B. Vogelstein, K. W. Kinzler, Nat. Genet. 1999, 23, 387. [21] a) T. Kang, S. M. Yoo, I. Yoon, S. Y. Lee, B. Kim, Nano Lett. 2010, 10, 1189; b) K. K. Maiti, U. S. Dinish, A. Samanta, M. Vendrell, K.-S. Soh, S.-J. Park, M. Olivo, Y.- T. Chang, Nano Today 2012, 7, 85; c) K. Gracie, E. Correa, S. Mabbott, J. A. Dougan, D. Graham, R. Goodacre, K. Faulds, Chem. Sci. 2014, 5, 1030. [22] a) L. Sun, J. Irudayaraj, Biophys. J. 2009, 96, 4709-4716; b) J. D. Driskell, R. A. Tripp, Chem. Comm. 2010, 46, 3298-3300. [23] B. Mir-Simon, I. Reche-Perez, L. Guerrini, N. Pazos-Perez, R. A. Alvarez-Puebla, Chem. Mat. 2015, 27, 950. [24] Y. L. Wang, E. J. H. Wee, M. Trau, Chem. Commun. 2015, 51, 10953. [25] G. Frens, Nat. Phys. Sci. 1973, 241, 20.

SUPPORTING INFORMATION

Table S1. Oligonucleotide sequences used in experiments. Oligos 5'-Sequence-3'

TACACAGCAC(C3)CGCGGCAGGAAGCCTTAT T1E4 RT-RPA Forward Primer

Biotin-GTTACATTCCATTTTGAT T1E4 RT-RPA Reverse Primer

GCTACACGAT(C3)CCGATCGGGTGTCCGCAC RN7SL1 RT-RPA Forward Primer

Biotin-AGGCGCGATCCCACTACT RN7SL1 RT-RPA Reverse Primer

CAGATCGTCATGTTC(C3)AGCGCGGCAGGAACT T1E5 RT-RPA Forward Primer CTC

Biotin -CTGCTGGCACGATAACTC T1E5 RT-RPA Reverse Primer

ARV7 RT-RPA Forward Primer TCTGCACCAATGTAC(C3)CACTATTGATAAATTC CG

242

Chapter 6.1

ARV7 RT-RPA Reverse Primer Biotin-TGAGGCAAGTCAGCCTTT

ACGATGCATG(C3)CCTGATGATACAGAGGTGAG PCA3 RT-RPA Forward Primer

Biotin-GCACAGGGCGAGGCTCATCG PCA3 RT-RPA Reverse Primer

T1E4 qRT-PCR Forward Primer CCTGGAGCGCGGCAGGAAGCCTTATCAGTTG

T1E4 qRT-PCR Reverse Primer GCTAGGGTTACATTCCATTTTGATGGTGAC

RN7SL1 qRT-PCR Forward Primer CGCCGCCTGGAGCGCGGCAGGAACTCTCCT

RN7SL1 qRT-PCR Reverse Primer CTGCTGGCACGATAACTCTGCGCTCGTTCG

T1E5 qRT-PCR Forward Primer GCTATGCCGATCGGGTGTCCGCACTAAGTT

T1E5 qRT-PCR Reverse Primer GACGGGGTCTCGCTATGTTGCCCAGGCTGG

ARV7 qRT-PCR Forward Primer TGATTGCACTATTGATAAATTCCGAAGGAA ARV7 qRT-PCR Reverse Primer TTTGAATGAGGCAAGTCAGCCTTTCTTCAG PCA3 qRT-PCR Forward Primer CCTGATGATACAGAGGTGAG PCA3 qRT-PCR Reverse Primer GCACAGGGCGAGGCTCATCG *Underlined bases are complementary to sequences on SERS nanotags.

243

Chapter 6.1

Figure S1. Typical TEM image of the SERS nanotags on the magentic beads. Slight aggregation of SERS nanotags (indicated by arrows) was observed after hybridization to targets.

Figure S2. Combined SERS spectra of different SERS nanotags. Each SERS nanotag was synthesized from a different Raman probe molecule (Structure shown on right).

244

Chapter 6.1

Figure S3. Linear calibration plot for ARV7 copy number titration.

Figure S4. Average SERS spectra of increasing number (5; 10; and 20) of signal acquisitions (2 sec laser integration).

245

Chapter 6.1

Figure S5. Combined fluorescence spectra of qRT-PCR biomarker detection in mixture of isolated total RNA from 3 cell lines (DuCap, LnCap, 22Rv1).

Figure S6. Combined fluorescence spectra of qRT-PCR biomarker detection in isolated total RNA from tumor tissue sample A.

246

Chapter 6.1

Figure S7. Combined fluorescence spectra of qRT-PCR biomarker detection in isolated total RNA from tumor tissue sample B.

Figure S8. Combined fluorescence spectra of qRT-PCR biomarker detection in isolated total RNA from urine sample C.

247

Chapter 6.1

Figure S9. Combined fluorescence spectra of qRT-PCR biomarker detection in isolated total RNA from urine sample D.

Figure S10. Combined fluorescence spectra of qRT-PCR biomarker detection in isolated total RNA from urine sample E.

248

Chapter 6.2

High-Speed Biosensing Strategy for Non-Invasive Profiling of Multiple Cancer Fusion Genes in Urine

Kevin M. Koo, Eugene J. H. Wee, Matt Trau

ABSTRACT Aberrant chromosal rearrangements, such as the multiple variants of TMPRSS2:ERG fusion gene mutations in prostate cancer (PCa), are promising diagnostic and prognostic biomarkers due to their specific expression in cancerous tissue only. Additionally, TMPRSS2:ERG variants are detectable in urine to provide non-invasive PCa diagnostic sampling as an attractive surrogate for needle biopsies. Therefore, rapid and simplistic assays for identifying multiple urinary TMPRSS2:ERG variants are potentially useful to aid in early cancer detection, immediate patient risk stratification, and prompt personalized treatment. However, current strategies for simultaneous detection of multiple gene fusions are limited by tedious and prolonged experimental protocols, thus limiting their use as rapid clinical screening tools. Herein, we report a simple and rapid gene fusion strategy which expliots the specificity of DNA ligase and the speed of isothermal amplification to simultaneously detect multiple fusion gene RNAs within a short sample-to-answer timeframe of 60 min. The method has a low detection limit of 2 amol (1000 copies), and was successfully applied for non-invasive fusion gene profiling in patient urine samples with subsequent validation by a PCR-based gold standard approach.

INTRODUCTION Fusion genes are a result of chromosomal rearrangements such as translocations, deletions, or inversions (Rabbitts 2009). This can result in abnormal gene expression and in turn, leads to tumour initiation and development (Edwards 2010; Mitelman et al. 2007; Prensner and Chinnaiyan 2009). With regards to cancer diagnosis and prognosis, fusion genes are emerging as biomarkers of immense value due to their specific presence in tumor tissue only. Additionally, it is now known that recurrent gene fusions are useful for characterizing specific subtypes of prostate, breast, and lung cancers for targeted treatments (Kumar-Sinha et al. 2015; Kumar-Sinha et al.

249

Chapter 6.2

2008). The most prevalent group of gene fusions affecting approximately 50% of all prostate cancer (PCa) cases, is the fusion between TMPRSS2 and ERG genes (Soller et al. 2006; Tomlins et al. 2005). There are about 20 TMPRSS2:ERG variants with different fusion junctions (Clark et al. 2007; St. John et al. 2012), and the presence of certain variants in tumour cells has been associated with cancer aggressiveness and metastatic potential (Lapointe et al. 2007; Wang et al. 2006). Additionally, TMPRSS2:ERG mRNAs are also detectable in urine (Hessels et al. 2007), thus presenting as a convenient and non-invasive mean to detect and stratify PCa as compared to a tissue biopsy which is both painful and expensive. Therefore, TMPRSS2:ERG fusion genes are of increasing interest as diagnostic and prognostic biomarkers for early PCa detection, immediate patient risk stratification, and prompt personalized treatment (Tomlins et al. 2009). Herein, we describe a novel technique for rapid and non-invasive profiling of multiple TMPRSS2:ERG variants concurrently while achieving results within a considerably shorter timeframe as compared to current methodologies.

Ligase-based strategies (Schouten et al. 2002) are currently used to detect multiple gene rearrangements (Cui et al. 2012; Lee et al. 2010) by enzymatically ligating adjacent barcoded probes that flank fusion junctions to effectively generate a library of PCR-amplifiable and identifiable fusion events for further downstream analysis. Despite being an effective research tool, this approach has drawbacks which we aimed to address in this report. Firstly, the use of DNA ligases (typically T4 DNA ligase) is inefficient and slow on RNA templates; hence requiring long ligation protocols, alternative costly ligases or high starting sample amount to achieve detectable signals (Bullard and Bowater 2006; Lohman et al. 2014; Nilsson et al. 2001; Wee and Trau 2016). Next, the subsequent PCR step (Kvastad et al. 2015; Roy et al. 2015; Zhang et al. 2014) is also usually time-consuming (typically 120 min); and requires expensive and bulky instrumentation such as capillary electrophoresis systems (Lee et al. 2010; Schouten et al. 2002; Zhang et al. 2014) to identify and quantify targets in the sample. These drawbacks limit the use of ligase-based assay for quick and simple clinical screening of fusion gene RNAs.

The use of a quicker isothermal amplification method such as Recombinase Polymerase Amplification (RPA) (Piepenburg et al. 2006) after probe ligation may

250

Chapter 6.2 allow for faster and simpler assays to be performed on cheaper and simpler equipment. Additionally, the high sensitivity of RPA (Euler et al. 2013; Kersting et al. 2014) could be useful in amplifying lower amount of ligation products (i.e. shorter ligation time). This could resolve the issue of lengthy ligation time associated with conventional ligation-based strategies (Lee et al. 2010; Roy et al. 2015; Zhang et al. 2014), and also potentially improve ligation specificity (Kuhn and Frank-Kamenetskii 2005) for better signal-to-noise ratio. Currently, to the best of our knowledge, a ligation-based RPA assay for detecting cancer fusion gene RNAs such as TMPRSS2:ERG variants in PCa remains unexplored.

Herein we describe a rapid assay which first involved rapid T4 DNA ligase-based ligation of probes specific to the fusion junctions of different TMPRSS2:ERG mRNA variants in a single reaction. This was then followed by isothermal RPA in individual RPA assays for real-time fluorescence identification (via convenient DNA- intercalating dye) of fusion variants present in the sample. The assay was readily completed within 60 min, possessed amol-level detection sensitivity and could successfully profile the status of two different TMPRSS2:ERG variants and a RNA loading control in PCa cell lines and patient urine samples.

EXPERIMENTAL Reagents and materials All reagents were of analytical grade and purchased from Sigma Aldrich (Australia). UltraPureTM DNase/RNase-free distilled water (Invitrogen, Australia) was used throughout the experiments. Oligonucleotides were purchased from Integrated DNA Technologies (USA) and sequences are shown (Supporting Information, Table S1).

RNA extraction from cell lines DuCap and LnCap cells were kindly donated by Matthias Nees (Vtt, Finland); Gregor Tevz (APCRC, Australia); and Michelle Hill (UQDI, Australia). The cells were cultured in RPMI- 1640 growth media (Life Technologies, Australia) supplemented with 10% fetal bovine serum (Life Technologies, Australia) in a humidified incubator containing 5% CO2 at 37 ºC. RNA was extracted using Trizol® reagent (Life Technologies, Australia). RNA purity and integrity were checked using a NanoDrop Spectrophotometer (Thermo Scientific, USA).

251

Chapter 6.2

RNA extraction from patient urine samples Ethics approval was obtained from The University of Queensland Institutional Human Research Ethics Committee (Approval No. 201400012) and informed consent was obtained from all subjects prior to sample collection. Methods pertaining to clinical samples were carried out in accordance with approved guidelines.

Total RNA was extracted from urine samples using the commercially-available ZR urine RNA isolation KitTM (Zymo Research, USA). For each sample, 30 mL of urine was passed through the supplied ZRC GFTM Filter to retain urinary cells. Then, 700 µL of supplied urine RNA buffer was passed through the filter and the filtered cells were collected in the flow- through. Next, as per manufacturer’s instructions, the cells from the flow-through were lysed, washed and eluted in 10 µL of RNase-free water. RNA purity and integrity were checked using a NanoDrop Spectrophotometer (Thermo Scientific, USA).

Multiplexed hybridization and ligation of probes A 10 µL hybrization mixture containing 100 nM of each ligation probe and 30 ng of total RNA extract or various concentrations of synthetic target sequences was made. The hybridization mixture was heated to 85°C for 2 min and then cooled on ice to facilitate hybridization. Next, 1 µL of the hybridization mixture was added to a 10 µL ligation mixture containing 40 U T4 DNA Ligase (New England Biolabs, Australia) and 1x T4 DNA Ligase Buffer. The ligation mixture was heated at 37°C (unless otherwise stated) for 5 min to ligate the hybridized probes before 2 min at 95°C to stop enzyme activity.

Quantitative RPA The TwistAmp Basic RPA kit (Twist-DX, UK) was used with slight modifications to manufacturer’s instructions. Each isothermal amplification reaction was carried out in individual tubes. In each reaction, 7.38 µL of RPA rehydration buffer, 250 nM of each primer (Supporting Information, Table S1), 2 nM of SYTO® 9 (SYTO9) fluorescent dye (Thermo Fisher, Australia), and 1 µL of ligation mixture (containing hybridized and ligated probe templates) were added to a final 12.5 µl reaction volume. RT-RPA was performed on the Applied Biosystems® 7500 Real-Time PCR System (Thermo Scientific, Australia) at 37°C for 15 min where fluorescence measurements were acquired every 30 seconds. The relative fluorescence intensity at assay endpoint was calculated as follows:

252

Chapter 6.2

Relative Fluorescence Intensity = (iSample – iRPA Control) / iRPA Control (1)

where iSample and iRPA Control are fluorescence intensities for target sample and RPA NoT control at the 15 min timepoint respectively.

Quantitative PCR (qPCR) validation For validation analysis of each target (T1E4, T1E5, or RN7SL1) in clinical urine samples, the KAPA SYBR® FAST One-Step qRT-PCR kit (KAPA BIOSYSTEMS, Australia) was used to set up a single reaction volume of 10 µl. Each reaction volume consist of 1X KAPA SYBR® FAST qPCR Master Mix, 200 nM of each forward and reverse primer (Table 1), 1X KAPA RT Mix, 50 nM ROX dye and 30 ng of total RNA template. qPCR was performed using the Applied Biosystems® 7500 Real-Time PCR System (Thermo Fisher Scientific, Australia). The cycling protocol was: 42°C for 10 min to synthesize cDNA, followed by 95°C for 5 min to deactivate RT before cycling 35 times (95°C for 30 s, 50°C for 30 s and 72°C for 1 min) and finished with 72°C for 10 min. The cycle threshold (Ct) value (i.e. initial target amount) was determined for each qPCR reaction in order to cross-validate the data of our proposed assay.

RESULTS AND DISCUSSION Assay principle The principle of our assay for detecting multiple TMPRSS2:ERG variants is shown in Fig. 1. Three pairs of ligation probes (Probes 1A/1B/2A/2B/3A/3B) are designed to contain complementary sequences (in blue/orange/green) to each target sequence. Each ligation probe pair also contains a unique primer recognition sequence (in black/grey/brown) for RPA amplification. All three ligation probe pairs are added to a single starting sample for hybridization to target sequences of interest within 5 min, and successfully hybridized probes are subsequently ligated with T4 DNA ligase. The T4 DNA ligase catalyzed a phosphodiester bond formation between the 5’ phosphate and 3’ hydroxyl blunt ends of each hybridized ligation probe pair; and this ligation process could discriminate single base mismatches which is useful for differentiating highly-similar TMPRSS2:ERG fusion variant junctions (Tang et al. 2008). Moreover, since the ligation probes were designed for specific hybridization to gene fusion mRNA targets, the possibility of non-specific ligation probe hybridization to

253

Chapter 6.2 genomic DNA (with non-complementary sequences) and subsequent T4 DNA ligase- mediated ligation is therefore eliminated. Following ligation, individual ligated probe pairings for each target sequence are amplified isothermally by RPA (via. unique primer sequences) in individual tubes within 15 min. As the unique primer set in each RPA reaction is specific for each target sequence (T1E4, T1E5 or RN7SL1), the generation of double- stranded amplicons in each target-specific RPA reaction could be converted into a fluorescence signal using a double-stranded DNA-intercalating dye. The incorporation of SYTO9 fluorescent dye in each RPA reaction allowed for convenient and sensitive real-time detection of successful amplification in presence of a target sequence. SYTO9 is a double- stranded DNA intercalating dye which has been shown to possess better dye stability and fluorescence reproducibility as compared to traditional SYBR Green dyes (Monis et al. 2005). In all, the whole sample-to-answer time for detection of three targets concurrently is 60 min.

In our experiments, we used three different RNA sequences as model targets to evaluate our proposed assay for PCa screening. The three target sequences comprised of two closely-related TMPRSS2:ERG variants of different fusion junctions: TMPRSS2 exon 1 with ERG exon 4 (T1E4) and TMPRSS2 exon 1 with ERG exon 5 (T1E5), and an endogenously-expressed mRNA housekeeping sequence (RN7SL1).

254

Chapter 6.2

Fig. 1. Schematic representation of ligation-based recombinase polymerase amplification (RPA) assay. Ligation probe pairs (Probes 1A/1B/1C) are designed to contain complementary sequences (in blue/orange/green) to each target sequence, and unique RPA primer recognition sequences (in black/grey/brown). Ligation probe pairs hybridize to multiple target sequences in total RNA starting sample and are subsequently enzymatically ligated. The ligated probes are then amplified using target-specific RPA primers in separate reactions. The inclusion of intercalating fluorescence dye allowed the amplification to be monitored in real-time.

Optimization of Ligation Temperature The ligation temperature plays an important role in assay specificity as the optimal ligation temperature could allow probes to hybridize specifically to target sequences

255

Chapter 6.2 and minimize false-positive signals from non-specific ligation and subsequent amplification (Zhang et al. 2014). To this end, we used ligation temperatures of 35°C, 37°C, and 39°C to investigate the optimal temperature for using T1E4 probes to discriminate between closely-related T1E4 and T1E5 target sequences. An optimal temperature of 37°C was found to achieve the largest difference in relative fluorescence intensity and thus, maximized discrimination between T1E4 and T1E5 (Supporting Information, Fig. S1). Hence, 37°C was selected as the ligation temperature for all further experiments.

Efficiency of downstream RPA To investigate the efficiency of using downstream isothermal RPA instead of traditional PCR in our ligase-based assay, we firstly tested the selectivity of RPA primers for ligated probes amplification. We omitted two out of three of our ligation probe pairs at a time and repeated our assay on a 3-target mixture (T1E4, T1E5, and RN7SL1) with the remaining ligation probe pair. We found that the absence of particular ligation probe pairs led to negligible fluorescence signal for the corresponding target sequences (Fig. 2 and Supporting Information Fig. S2). Thus, this demonstrated the efficiency of each RPA primer pairs for successful amplification of respective ligated probe pairs to generate detectable fluorescence signals. In addition, the selectivity of each RPA primer set for exclusive amplification of its assigned ligated probe pairs was also tested (Supporting Information, Fig. S3). It was found that no RPA amplicons were generated for non-specific targets (eg. T1E4 RPA primers generated amplicons only in presence of T1E4 mRNA targets, and no amplification were observed with a mix of both T1E5 and RN7SL1 mRNA targets). This high primer selectivity will be essential for accurately discriminating target sequences in biologically-complex samples such as blood or urine samples.

Moreover, the five min ligation step in our assay was significantly faster as compared to the typical T4 ligation protocol (Nilsson et al. 2001) which required at least two hours. Although a smaller amount of ligation products is expected as a consequence of reduced ligation time, this highlighted the high sensitivity and efficiency of the downstream RPA step in amplifying low quantities of ligated products to detectable levels. It is worthy to note that the low quantity of ligated product for each RPA reaction may allow for a single patient sample to be split into several parallel RPA analyses of multiple fusion gene biomarkers. In comparison to traditional ligation-

256

Chapter 6.2 based assays (with lengthy PCR amplification procedures (Kvastad et al. 2015; Roy et al. 2015; Zhang et al. 2014) which required about five hours, our approach (60 min) showed a marked five-fold improvement in total assay time. We envisage that the rapid nature of our proposed assay will be greatly beneficial for parallel-processing of patient samples, thus enabling a more efficient workflow to process high amount of clinical samples as well as quicker cancer diagnosis/prognosis for each patient.

Fig. 2. Evaluation of downstream RPA efficiency. Relative fluorescence intensities after target-specific ligation and RPA of T1E4, T1E5, or RN7SL1 target sequences. Error bars represent standard deviation of three independent experiments.

Assay sensitivity To evaluate detection sensitivity, we analyzed a serial dilution of synthetic T1E4 RNA transcripts using all three ligation probe pairs in our assay. The resulting positive relationship between fluorescence signal and different T1E4 concentrations are shown in Fig. 3. The assay detection limit was determined to be 1000 copies (2 amol) with a linear range between 103-107 copies (Supporting Information, Fig. S4). This detection limit is superior to existing ligation-based assays (Arefian et al. 2011; Roy et al. 2015) as well as existing multiplexed gene fusion detection methodologies which required a higher amount of starting material (Frampton et al. 2013; Lee et al. 2010; Lira et al. 2013; Peter et al. 2001). Furthermore, the low amol detection limit in our proposed

257

Chapter 6.2 assay was achieved with the simplicity of DNA-intercalating fluorescent dye readout in individual reactions. This enabled quick and simultaneous detection of multiple fusion gene in real-time without any prior need for complicated sequence-specific labelling procedure, thus offering a powerful tool for sensitive, low-cost and accurate PCa screening. Moreover, the quantitative nature of our assay in measuring gene fusion mRNA levels may be helpful for quantifying tumor load during diagnosis or monitoring treatment response during therapy.

Fig. 3. Assay sensitivity. Combined fluorescence spectra generated from 103 – 107 copies of synthetic T1E4 target sequences.

Detection of multiple RNA targets To demonstrate the capability of our assay for detecting multiple targets simultaneously, we challenged the assay to detect the three target sequences simultaneously from a single starting sample. It was observed that each target-specific downstream RPA reaction generated a positive fluorescence signal (Fig. 4), thus demonstrating successful multiplexed detection of all three different sequences. The low background signals in both ligation and RPA no-target (NoT) controls showed that resulting fluorescence signals were attributed to the presence of target sequences.

258

Chapter 6.2

Fig. 4. Combined fluorescence spectra for simultaneous detection of T1E4, T1E5, and RN7SL1 synthetic target sequences. Ligation and RPA no-target (NoT) controls were included.

Specific detection of fusion genes in PCa cell lines After establishing the assay’s capability using synthetic sequences, we sought to investigate the assay performance on biologically-complex samples. To this end, we first challenged our assay with total RNA extracts from PCa cell lines to detect T1E4 and T1E5 fusion mRNA variants. T1E4 is the predominant fusion variant in PCa, and accounts for ~90% of TMPRSS2:ERG variants identified in tumor samples (Tu et al. 2007; Wang et al. 2006). T1E5 is another common TMPRSS2:ERG variant which has been found to co-exist with T1E4 (Clark et al. 2007; Tu et al. 2007). Importantly, both variants are highly PCa-specific biomarkers which are absent in clinically benign cases (Tomlins et al. 2005). Therefore, it is of interest to detect these different fusion variants to aid in diagnosing PCa and predicting disease development. As an initial test, two PCa cell lines were tested with our assay: DuCap (T1E4 and T1E5 positive) and LnCap (T1E4 and T1E5 negative). The resulting fluorescence signals are shown (Supporting Information, Fig. S5); and as shown in Fig. 5, T1E4 and T1E5 were successfully detected in DuCap whilst negligible fluorescence signals were generated for the detection of either fusion variants in LnCap. To control for sample loading, the detection of housekeeping RN7SL1 mRNA was also included in our assay. Additionally, the sole detection of the RN7SL1 target in LnCap further underscored the assay specificity

259

Chapter 6.2 during multiplexed probe ligation and parallel RPA reactions. This high detection specificity is most likely conferred by the dual target recognition events (first by ligation probes, followed by RPA primers) within our assay. These detection outcomes for presence/absence of TMPRSS2:ERG variants in PCa are also in agreement with previous studies (Koo et al. 2016; Tomlins et al. 2005; Yoshimoto et al. 2006), thus validating the accuracy of our assay. Lastly, the assay also showed good reproducibility with a CV of 8.9% (n = 3).

Fig. 5. Profiling of multiple gene fusion biomarkers in cell lines and clinical urine samples. Two prostate cancer cell lines (DuCap and LnCap) and 11 clinical urine samples (P1- P9, H1 and H2) were tested for T1E4 (yellow), T1E5 (green), and RN7SL1 (grey) presence. Shaded bar plots indicates positive presence of target sequence in sample. Error bars represent standard deviation of three independent experiments.

Gene fusion profiling in clinical patient samples The eventual aim of our assay was to show clinical utility for rapid profiling fusion genes in patient urine samples. The non-invasive detection of fusion genes from urine offers evident benefits such as being pain-free, inexpensive, and without the risk of infections as compared to surgical biopsy sampling. After preliminary work on cell lines, we proceeded to test our assay on nine urine samples obtained from PCa patients as well as two urine samples from healthy individuals (Fig. 5 and Supporting Information, Fig. S6). T1E4 and T1E5 cutoff values were determined based on the mean signal of healthy controls plus three standard deviations. We conveniently

260

Chapter 6.2 scored patient samples to be positive/negative for each target biomarker by noting if fluorescence signals are higher/lower than the cutoff values. Using or assay, we found that three PCa samples (P1, P5, and P8) were positive for both T1E4 and T1E5, whilst five samples (P2, P3, P4, P7 and P9) were positive for T1E4 only. For P6 and healthy controls (H1 and H2), only loading control RN7SL1 signals were observed to indicate the presence of input RNA and absence of both T1E4 and T1E5.

From our assay results, the variable fusion gene expressions of different PCa patients highlighted the promise of using fusion gene profiling to inform clinical diagnosis. Though large-scale population-based clinical studies are still ongoing, previous pilot studies have shown TMPRSS2:ERG fusion variants to be present in about half of all PCa cases (Soller et al. 2006; Tomlins et al. 2005) and linked to lethal PCa forms (Wang et al. 2006). Hence, the successful demonstration of our assay for TMPRSS2:ERG profiling in clinical samples may provide quick and accurate diagnosis for clinically-significant PCa and enable targeted therapies for TMPRSS2:ERG inhibition. Nonetheless, we are aware that our current assay format is a promising diagnostic demonstration, and application to a larger pool of patient samples is required to prove clinical utility. Our assay results were validated with qPCR (Supporting Information, Table S2) with excellent agreement, thus demonstrating analytical accuracy in clinical samples. In each assay run involving cell line and clinical patient samples, the total assay time was ~60 min and required only 30 ng of total RNA starting sample. The successful detection of multiple TMPRSS2:ERG variants in urine samples suggested a possibility for future non- invasive PCa screening as an alternative to current surgical tissue biopsies. Due to the ease of urine sampling and the release of PCa cancer cells into urine, urine is a promising substrate for developing non-invasive PCa diagnostics. Additionally, the readout of our assay only required a conventional fluorescence detector, and we envisaged that the use of portable fluorometers (Boyle et al. 2013; Clancy et al. 2015; Li et al. 2010) could allow our assay to be increasingly cost-efficient and amenable for clinical application in the future.

261

Chapter 6.2

CONCLUSIONS In conclusion, we have developed a rapid ligation-based RPA assay for simultaneously detecting different TMPRSS2:ERG fusion variants in PCa patient samples with high amol-sensitivity and specificity. Our assay results on patient urine samples were in excellent agreement with the gold standard PCR-based approach, thereby underscoring the promise of an accurate and reliable clinical screening technique for non-invasive PCa diagnosis. Most crucially, our 60 min sample-to-answer assay time is a remarkable improvement on the current tedious and time-consuming detection of multiple fusion genes. For future work, the multiplexing capability of our proposed assay could be scaled up by the rational design and addition of more ligation probe pairs. This could be useful for the detection of other TMPRSS2:ERG variants during PCa screening to aid in better diagnosis and therapy. Moreover, by altering probe design, our assay could be modified for multiple gene fusion detection in other cancers or diseases.

REFERENCES Arefian, E., Kiani, J., Soleimani, M., Shariati, S.A.M., Aghaee-Bakhtiari, S.H., Atashi, A., Gheisari, Y., Ahmadbeigi, N., Banaei-Moghaddam, A.M., Naderi, M., Namvarasl, N., Good, L., Faridani, O.R., 2011. Nucleic Acids Res. 39(12), e80. Boyle, D.S., Lehman, D.A., Lillis, L., Peterson, D., Singhal, M., Armes, N., Parker, M., Piepenburg, O., Overbaugh, J., 2013. mBio 4(2), e00135. Bullard, D.R., Bowater, R.P., 2006. Biochem. J. 398(1), 135-144. Clancy, E., Higgins, O., Forrest, M.S., Boo, T.W., Cormican, M., Barry, T., Piepenburg, O., Smith, T.J., 2015. BMC Infect. Dis. 15, 481. Clark, J., Merson, S., Jhavar, S., Flohr, P., Edwards, S., Foster, C.S., Eeles, R., Martin, F.L., Phillips, D.H., Crundwell, M., Christmas, T., Thompson, A., Fisher, C., Kovacs, G., Cooper, C.S., 2007. Oncogene 26(18), 2667- 2673. Cui, J., Azimi, M., Adekile, A.D., Al Awadhi, H., Hoppe, C.C., 2012. Hemoglobin 36(3), 276-282. Edwards, P.A.W., 2010. J. Pathol. 220(2), 244-254. Euler, M., Wang, Y.J., Heidenreich, D., Patel, P., Strohmeier, O., Hakenberg, S., Niedrig, M., Hufert, F.T., Weidmann, M., 2013. J. Clin. Microbiol. 51(4), 1110-1117.

262

Chapter 6.2

Frampton, G.M., Fichtenholtz, A., Otto, G.A., Wang, K., Downing, S.R., He, J., Schnall- Levin, M., White, J., Sanford, E.M., An, P., Sun, J., Juhn, F., Brennan, K., Iwanik, K., Maillet, A., Buell, J., White, E., Zhao, M., Balasubramanian, S., Terzic, S., Richards, T., Banning, V., Garcia, L., Mahoney, K., Zwirko, Z., Donahue, A., Beltran, H., Mosquera, J.M., Rubin, M.A., Dogan, S., Hedvat, C.V., Berger, M.F., Pusztai, L., Lechner, M., Boshoff, C., Jarosz, M., Vietz, C., Parker, A., Miller, V.A., Ross, J.S., Curran, J., Cronin, M.T., Stephens, P.J., Lipson, D., Yelensky, R., 2013. Nat. Biotechnol. 31(11), 1023-1031. Hessels, D., Smit, F.P., Verhaegh, G.W., Witjes, J.A., Cornel, E.B., Schalken, J.A., 2007. Clin. Cancer Res. 13(17), 5103-5108. Kersting, S., Rausch, V., Bier, F.F., von Nickisch-Rosenegk, M., 2014. Malaria J. 13, e99. Koo, K.M., Wee, E.J.H., Mainwaring, P.N., Wang, Y., Trau, M., 2016. Small. DOI:10.1002/smll.201602161 Kuhn, H., Frank-Kamenetskii, M.D., 2005. FEBS J. 272(23), 5991-6000. Kumar-Sinha, C., Kalyana-Sundaram, S., Chinnaiyan, A.M., 2015. Genome Med. 7, 129. Kumar-Sinha, C., Tomlins, S.A., Chinnaiyan, A.M., 2008. Nat. Rev. Cancer 8(7), 497-511. Kvastad, L., Solnestam, B.W., Johansson, E., Nygren, A.O., Laddach, N., Sahlen, P., Vickovic, S., Bendigtsen, S.C., Aaserud, M., Floer, L., Borgen, E., Schwind, C., Himmelreich, R., Latta, D., Lundeberg, J., 2015. Sci. Rep. 5, e16519. Lapointe, J., Kim, Y.H., Miller, M.A., Li, C., Kaygusuz, G., van de Rijn, M., Huntsman, D.G., Brooks, J.D., Pollack, J.R., 2007. Mod. Pathol. 20(4), 467-473. Lee, S.-T., Yoo, E.-H., Kim, J.-Y., Kim, J.-W., Ki, C.-S., 2010. Bri. J. Haematol. 148(1), 154-160. Li, Z.H., Wang, Y., Wang, J., Tang, Z.W., Pounds, J.G., Lin, Y.H., 2010. Anal. Chem. 82(16), 7008-7014. Lira, M.E., Kim, T.M., Huang, D., Deng, S., Koh, Y., Jang, B., Go, H., Lee, S.-H., Chung, D.H., Kim, W.H., Schoenmakers, E.F.P.M., Choi, Y.-L., Park, K., Ahn, J.S., Sun, J.-M., Ahn, M.-J., Kim, D.-W., Mao, M., 2013. J. Mol. Diagnostics 15(1), 51- 61. Lohman, G.J., Zhang, Y., Zhelkovsky, A.M., Cantor, E.J., Evans, T.C., Jr., 2014. Nucleic Acids Res. 42(18), 11846-11846. Mitelman, F., Johansson, B., Mertens, F., 2007. Nat. Rev. Cancer 7(4), 233-245. Monis, P.T., Giglio, S., Saint, C.P., 2005. Anal. Biochem. 340(1), 24-34.

263

Chapter 6.2

Nilsson, M., Antson, D.O., Barbany, G., Landegren, U., 2001. Nucleic Acids Res. 29(2), 578- 581. Peter, M., Gilbert, E., Delattre, O., 2001. Lab. Investig. 81(6), 905-912. Piepenburg, O., Williams, C.H., Stemple, D.L., Armes, N.A., 2006. Plos Biol. 4(7), 1115- 1121. Prensner, J.R., Chinnaiyan, A.M., 2009. Curr. Opin. Genet. Dev. 19(1), 82-91. Rabbitts, T.H., 2009. Cell 137(3), 391-395. Roy, C.K., Olson, S., Graveley, B.R., Zamore, P.D., Moore, M.J., 2015. eLife 4, e03700. Schouten, J.P., McElgunn, C.J., Waaijer, R., Zwijnenburg, D., Diepvens, F., Pals, G., 2002. Nucleic Acids Res. 30(12), e57. Soller, M.J., Elfving, P., Lundgren, R., Panagopols, I., 2006. Genes Chromosomes Cancer 45(7), 717- 719. St. John, J., Powell, K., Conley-LaComb, M.K., Chinni, S.R., 2012. J. Cancer Sci. Ther. 4(4), 94-101. Tang, H., Yang, X., Wang, K., Tan, W., Li, H., He, L., Liu, B., 2008. Talanta 75(5), 1388- 1393. Tomlins, S.A., Bjartell, A., Chinnaiyan, A.M., Jenster, G., Nam, R.K., Rubin, M.A., Schalken, J.A., 2009. Eur. Urol. 56(2), 275-286. Tomlins, S.A., Rhodes, D.R., Perner, S., Dhanasekaran, S.M., Mehra, R., Sun, X.W., Varambally, S., Cao, X.H., Tchinda, J., Kuefer, R., Lee, C., Montie, J.E., Shah, R.B., Pienta, K.J., Rubin, M.A., Chinnaiyan, A.M., 2005. Science 310(5748), 644-648. Tu, J.J., Rohan, S., Kao, J., Kitabayashi, N., Mathew, S., Chen, Y.-T., 2007. Mod. Pathol. 20(9), 921-928. Wang, J.H., Cai, Y., Ren, C.X., Ittmann, M., 2006. Cancer Res. 66(17), 8347-8351. Wee, E.J.H., Trau, M., 2016. ACS Sens. 1(6), 670–675. Yoshimoto, M., Joshua, A.M., Chilton-MacNeill, S., Bayani, J., Selvarajah, S., Evans, A.J., Zielenska, M., Squire, J.A., 2006. Neoplasia 8(6), 465-469. Zhang, P., Zhang, J., Wang, C., Liu, C., Wang, H., Li, Z., 2014. Anal. Chem. 86(2), 1076- 1082.

264

Chapter 6.2

SUPPORTING INFORMATION

Table S1. Oligonucleotide sequences used in experiments. *Underlined and bold represent RPA primer sequences on ligation probes. Oligos 5'-Sequence-3'

T1E4 Probe 1A -PO-GCTTCCTGCCGACGAGGGAAAGAGTTGTACCTAAAT

T1E4 Probe 1B TGTATAGGAATCCCACTGAATTTTTCCTCACAACTGATAAG

T1E5 Probe 1A -PO-AGTTCCTGCCGATAAGGGAGATAATAAGAGTTGGGT

T1E5 Probe 1B TAACTCATATTGTAGAAGAGTAGAAGCATTCATCAGGAG

-PO- RN7SL1 Probe 1A TAGTGCGGACACACAGGGAGCCGAAAGACGAAAGGCC

RN7SL1 Probe 1B GGCTGTATACATTTCCCTCAGGACAGTGATGCCGAACT

T1E4 RPA Fwd ATTTAGGTACAACTCTTTCCCTCGTC Primer

T1E4 RPA Rev TGTATAGGAATCCCACTGAATTTTTC Primer

T1E5 RPA Fwd ACCCAACTCTTATTATCTCCCTTATC Primer

T1E5 RPA Rev TAACTCATATTGTAGAAGAGTAGAAG Primer

RN7SL1 RPA Fwd GGCCTTTCGTCTTTCGGCTCCCTGTG Primer

RN7SL1 RPA Rev GGCTGTATACATTTCCCTCAGGACAG Primer

T1E4 RT-PCR Fwd CCTGGAGCGCGGCAGGAAGCCTTATCAGTTG Primer

T1E4 RT-PCR Rev GCTAGGGTTACATTCCATTTTGATGGTGAC Primer

265

Chapter 6.2

T1E5 RT-PCR Fwd GCTATGCCGATCGGGTGTCCGCACTAAGTT Primer

T1E5 RT-PCR Rev GACGGGGTCTCGCTATGTTGCCCAGGCTGG Primer

RN7SL1 RT-PCR CGCCGCCTGGAGCGCGGCAGGAACTCTCCT Fwd Primer

RN7SL1 RT-PCR CTGCTGGCACGATAACTCTGCGCTCGTTCG Rev Primer

Fig. S1. Effect of ligation temperature on assay specificity.

Fig. S2. Combined fluorescence spectra for the detection of T1E4, T1E5, and RN7SL1 target sequences using only (A) T1E4 probes, (B) RN7SL1 probes, (C) T1E5 probes.

266

Chapter 6.2

Fig. S3. Gel electrophoresis of RPA reactions with different primers and target combinations.

Fig. S4. Linear calibration plot of the average florescence measurements at different initial T1E4 input copies. Error bars represent standard deviation of three independent experiments.

Fig. S5. Combined fluorescence spectra for the detection of T1E4, T1E5, and RN7SL1 in (A) DuCap, and (B) LnCap cell lines.

267

Chapter 6.2

Fig. S6. Combined fluorescence spectra for the detection of T1E4, T1E5, and RN7SL1 in patient urine samples.

268

Chapter 6.2

Table S2. qPCR validation of urine sample results. qPCR cycle threshold (Ct) values for the detection of T1E4, T1E5, and RN7SL1 in patient urine samples. Ct values in bold are targets which were tested negative using our assay. qPCR Our Assay

Sample T1E4 T1E5 RN7SL1 T1E4 T1E5 RN7SL1 P1 20.6 27.2 13.3 + + + P2 26.8 33.8 12.9 + - +

P3 29.1 32.7 14.7 + - + P4 27.7 32.9 10.2 + - + P5 15.1 17.2 13.1 + + +

P6 30.4 31.5 15.9 - - + P7 19.3 31.7 14.2 + - + P8 21.2 23.3 10.4 + + +

P9 23.4 30.8 14.4 + - + H1 33.1 33.8 15.2 - - + H2 32.5 33.9 14.4 - - +

269

Chapter 6.3

Amplification-Free Multi-RNA Type Profiling for Cancer Risk Stratification via Alternating Current Electrohydrodynamic Nanomixing

Kevin M. Koo, Shuvashis Dey, and Matt Trau

ABSTRACT Simultaneous analysis of messenger RNA (mRNA), microRNA (miRNA), and long non- coding RNA (lncRNA) - multi-RNA type profiling - is increasingly crucial in cancer diagnostics. Yet, rapid multi-RNA type profiling is challenging due to enzymatic amplification reliance and RNA type-dependent characteristics. Here we report a nanodevice that uniquely uses alterable alternating current electrohydrodynamic (ac-EHD) forces to enhance probe-target hybridization prior to direct native RNA target detection, without target amplification or surface functionalization. To exemplify clinical applicability, we performed non-invasive screening of next-generation prostate cancer (PCa) RNA biomarkers (of different types) in patient urine samples. A strong correlation between multi-RNA type expression and aggressive PCa was found, and the nanodevice performance is statistically evaluated. We believe our miniaturized system exhibits potential for cancer risk stratification via multi-RNA type profiling.

INTRODUCTION The combinational analysis of different RNA cancer biomarker types (multi-RNA type profiling) is becoming increasingly useful in personalized patient screening and monitoring. Taking prostate cancer (PCa) for example, multi-RNA type profiling could help identify aggressive PCa for personalized patient treatment, whilst sparing indolent cases from unnecessary treatments (i.e. risk stratification).[1] Lately, interest in non-invasive “liquid biopsies”[2] has uncovered various types of next-generation PCa RNA biomarkers[3] in urine. Such next-generation biomarkers include: highly PCa-specific gene fusion messenger RNA (mRNA) between TMPRSS2 and ERG genes (T2:ERG)[4] which rarely occurred in benign cases, and overexpression of miR-107[5] microRNA (miRNA) and SChLAP1[6] long non-coding RNA (lncRNA) that are strongly related to aggressive PCa. T2:ERG is present in about half of all clinically-detected PCa, and increased T2:ERG levels are linked to

270

Chapter 6.3 higher tumor volume and worse clinical prognosis. miR-107 has been found to be a potentially useful circulating biomarker with significantly higher levels in PCa patients as compared to healthy controls. SChLAP1 was discovered as one of the top-ranked prognostic PCa urine biomarker candidates associated with metastatic PCa progression in over 1000 patients. Hence, from a clinical standpoint, simultaneous profiling of these different types of urinary RNA biomarkers could be a valuable non-invasive diagnostic tool for differentiating aggressive forms of PCa to enable personalized treatment.

Presently, the simultaneous detection of multiple RNA biomarkers, especially so for different RNA types, is a challenging task. Rapid multi-RNA type profiling from patient samples is generally performed by quantitative polymerase chain reaction (qPCR) which may cause amplification artifacts/bias. Moreover, qPCR has to be modified according to detect different RNA target types due to type-dependent characteristics.[7] Thus, we have recently developed amplification-free RNA assays that eliminated cumbersome surface functionalization, and used adenine-driven target adsorption to allow direct detection of RNA biomarkers without enzymatic modification of native RNA copies.[8] Yet, similar to existing amplification-free RNA detection systems like the prominent NanoString Technologies,[9] our methodology is impeded by (i) hours-long probe-target hybridization due to slow-diffusion kinetics, (ii) large sample amount requirement which is impractical for multiple target detection.

To circumvent these issues, we hereby report a nanodevice which demonstrates adjustable alternating current electrohydrodynamic (ac-EHD) nanoscaled fluidic mixing (i.e. nano- mixing) to enhance probe-target hybridization kinetics. This nano-mixing phenomenon arises from the application of an ac potential across a pair of asymmetrical microelectrode. This gives rise to fluidic movement within nanometres of the microelectrode surfaces to maneuver molecules in solution, driven primarily by a body force on the free charges within the two non-uniformly charged double layers on the microelectrode surfaces. Unlike most microfluidic manipulation with limited control, ac-EHD nano-mixing can be uniquely manipulated and “tuned” through deft selection of the ac frequency to alter the nanoscaled fluidic force strength.[10] Herein, we designed a new paired circular electrode design to generate an optimal ac-EHD nano-mixing effect. For the first time, we applied ac-EHD nanofluidics for fast concentration of probes and RNA targets from bulk solution, as well as enhanced hybridization to an extent whereby conventional molecular amplification (eg. PCR) is unnecessary and native RNA targets could be directly detected.

271

Chapter 6.3

EXPERIMENTAL SECTION Materials All reagents (Sigma Aldrich) were of analytical grade and used without further purification unless otherwise stated. UltraPureTM DNase/RNase-free distilled water (Invitrogen) was used throughout the experiments. Oligonucleotides (Integrated DNA Technologies) sequences used in this work are shown in Table S1.

Nanodevice design and fabrication Design and fabrication of the nanodevice were carried out at Australian National Fabrication Facility-Queensland Node. Nanodevice design was prepared using L-edit (MEMS-pro, v15) and then inscribed on to a 5-inches chrome mask using direct laser writer. Exposed masks were then developed using standard mask development procedure that included sequential treatment with Az726 solution (1 min), then chrome etching (1.5 min) followed by acetone cleaning and finally drying by nitrogen gas. This design contains three sets of two separate microwells, termed as “Mixing” and “Detection” microwells. Each “Mixing” and “Detection” microwell contains a circular working microelectrode (1000 µm and 800 µm in diameter respectively), that is surrounded by a 50 µm-thick ring microelectrode. Both circular and ring microelectrodes are connected separately to a pair of common connection pads for energizing the chip with external ac power source, and the inner circular and ring microelectrode are separated by 50 µm. To pattern the design on substrate, silicone wafers were water-cleaned and dried (5 min). After cleaning, a 200 nm-thick layer of AzNolF 2070 negative photoresist (Microchem) was coated on the clean wafer by spin coating (3000 rpm, 30 s). Following this, the polymer-coated wafers were then UV exposed to transfer the design. UV exposed wafers were then developed using PGMEA (propylene glycol methyl ether acetate) and dried with liquid nitrogen, and were transferred for gold deposition. This step is followed by overnight lift-off step to reveal gold structures on the nanodevice.

To fabricate microwells, PDMS (polydimethylsiloxane) master mix was poured onto the flat silicon surface, degassed and cured at 65˚C for 1 h. After curing step, two holes were punched and the PDMS layer was plasma-bonded to the prepared nanodevice.

272

Chapter 6.3 ac-EHD nano-mixing of probes and targets Total RNA extract from cell lines/urinary samples were adjusted (20 ng, 2.5 µL) using phosphate buffer saline (PBS) buffer (137 mM sodium chloride, 2 mM potassium chloride, pH 7.4), and added to each of the three “Mixing” microwell (one microwell for each RNA target type) on the nanodevice. Briefly, biotinylated capture probes (2.5 µL, 10 µM) for different RNA targets (T2:ERG, miR-107 and SChLAP1) was also added to the respective “Mixing” microwell. Then, ac-EHD nano-mixing of targets and probes was induced in each

“Mixing” microwell through application of an ac field strength of f = 500 Hz and Vpp = 800 mV for 10 min. The ac field parameters were optimized in this study for maximal enhancement of probe-target collisions and complementary hybridization.

Magnetic isolation of hybridized targets For magnetic isolation of biotinylated probe-target molecules after ac-EHD nano-mixing, streptavidin-labeled Dynabeads® MyOneTM Streptavidin C1 (Invitrogen) magnetic beads (5 µL) was added to each microwell, mixed thoroughly, and incubated for 15 min. Afterwhich, a permanent magnet was placed under the nanodevice, and the magnetic beads with bound RNA targets were separated from solution, washed twice with washing buffer (10 mM Tris- HCl, pH 7.5; 1 mM EDTA ; 2 M NaCl) within the microwells, and resuspended in RNase- free water (10 µL for T2:ERG mRNA or 7.5 µL for miR-107 miRNA and SChLAP1 lncRNA targets).

Target adsorption via adenine-gold affinity interactions To promote the adsorption of RNA targets onto bare gold microelectrode surface for detection, 3’-polyA tails were extended on miRNA and lncRNA targets (without natural polyA tails) by mixing resuspended magnetic beads-miRNA/lncRNA targets (7.5 µL) with E.

Coli Poly(A) polymerase reaction buffer (250 mM NaCl, 50 mM Tris-HCl, 10 mM MgCl2, pH 7.9), ATP (1 mM), and E. Coli Poly(A) polymerase (5 U) (New England BioLabs) to make up a reaction mixture (10 µL). Each reaction mixture was incubated at 37°C for 10 min by placing the nanodevice on a heating plate before Poly(A) polymerase was inactivated by addition of EDTA (10 mM). To remove leftover reagents, the magnetic beads with bound polyA-extended targets were washed twice as previously described, and resuspended in RNase-free water (10 µL).

273

Chapter 6.3

Next, all targets were heat-released from probes by heating the solution in each microwell at 95°C for 2 min. A magnet was immediately placed under the nanodevice, and the supernatant of each microwell containing different released RNA targets was collected, diluted in 5x saline-sodium citrate (SSC) buffer (20 µL, pH 7.0), and introduced into respective downstream “Detection” microwells. In the “Detection” microwells, the targets were allowed to adsorb onto the surface of the gold working microelectrode for 10 min through adenine- gold affinity interactions. The microelectrodes were then washed with PBS (10 mM) before electrochemical measurements.

Detection of adsorbed targets All electrochemical measurements were performed on a CH1040C potentiostat (CH Instruments). The electrolyte buffer consisted of PBS solution (10 mM, pH 7) containing 1:1 3- 4- [Fe(CN)6] /[Fe(CN)6] (2.5 mM) and KCl (0.1 M). Differential pulse voltammetry (DPV) signals were recorded from -0.2 V- 0.15 V with a pulse amplitude of 50 mV and a pulse width of 50 ms. The % current response change (% i) was obtained by calculated as follows:

(% i) = [(iBare – iAdsorbed) / iBare] x 100

where iBare and iAdsorbed are current densities for bare microelectrode and microelectrode after sample adsorption respectively.

Statistical analysis The Passing and Bablok regression analysis was used to compare between the patients’ RNA biomarker expression levels measured on our nanodevice, and the data generated by a reference qPCR technique. The analyses were performed using the XLSTAT software (Addinsoft).

Cell culture PCa cells were cultured in RPMI-1640 growth media (Life Technologies) supplemented with fetal bovine serum (10%) (Life Technologies) in a humidified incubator containing CO2 (5%) at 37ºC.

274

Chapter 6.3

Preparation of patient urine samples Ethics approval was obtained from The University of Queensland Institutional Human Research Ethics Committee (Approval No. 2004000047) and informed consent was obtained from all subjects prior to sample collection. Methods pertaining to clinical samples were carried out in accordance with approved guidelines. Voided urinary samples were collected from men undergoing treatment for PCa at the Royal Brisbane and Women’s Hospital.

Total RNA was extracted from the urine samples using the commercially-available ZR urine RNA isolation KitTM (Zymo Research). Each urine sample (30 mL) was passed through the supplied ZRC GFTM Filter to retain urinary cells. Then, supplied urine RNA buffer (700 µL) was passed through the filter and the filtered cells were collected in the flow-through. Next, as per manufacturer’s instructions, the cells from the flow-through were lysed, washed and eluted in RNase-free water (10 µL). Total RNA concentration for each sample was assayed using Qubit® RNA HS Assay Kit (Thermo Scientific) and total RNA input for each single assay run was standardized (20 ng) for non-biased comparisons of target biomarker expression levels.

RESULTS AND DISCUSSION Nanodevice for Amplification-free Multi-RNA Type Profiling Our fabricated nanodevice enables parallel analysis of three RNA biomarkers, and consists of three sets of “Mixing” and “Detection” microwells. Each microwell contains asymmetric circular and ring microelectrode pairs (Figure 1a) for ac-EHD nano-mixing or adsorption- based target detection.

Upon ac potential difference application across the microelectrodes to generate a non- uniform field (E), the asymmetrical microelectrode geometry engenders a lateral variation in number of accumulated double-layer charges with spatial charge distribution on the microelectrode surfaces (Figure 1b). These charges experience a force F (F = ρEt, where ρ = charge density and Et = tangential component of E). Due to the lateral force on the larger ring microelectrode (FL) being stronger than that of the smaller circular microelectrode (Fs), the resultant net fluid flows towards the larger microelectrode (i.e. FL > Fs). Consequently, this unidirectional fluid flow causes molecules in solution to be dragged by the fluidic flow[11] and induces the necessary nano-mixing effect to enhance probe-target hybridization kinetics. To

275

Chapter 6.3 visualize ac-EHD nano-mixing, we have utilized fluorescent beads to showcase the nano- mixing motion (Video S1, Supporting Information). Notably, these resultant fluidic forces enable intensified concentration and collision frequency of probes and targets in bulk solution at nm-length from the microelectrode surface, thus speeding up hybridization in minuscule fluid volume within the “Mixing” microwells.

Figure 1c depicts the assay workflow with total RNA extract from samples being firstly added to “Mixing” microwells on the device, with each containing a different target-specific probe. After enhanced probe-target hybridization via ac-EHD nano-mixing, streptavidin- magnetic beads are added to magnetically-enrich captured targets. Prior to electrochemical target detection by adenine- gold adsorption,[8] enzymatic 3’ polyA extensions are briefly performed on miRNA and lncRNA targets without endongenous polyA tails. RNA targets are subsequently heat-released from probes and transferred to “Detection” microwells for rapid adsorption onto gold microelectrode surfaces through their endogenous (for mRNA targets)/enzymatically-synthesized polyA tails. This transfer step to an unmodified bare gold electrode is to ensure minimal adsorption of non-target molecules and prevent false positive detection. Figure 1d shows the atomic force microscopy (AFM) characterization of the gold “Detection” microelectrode surface after rapid target adsorption. Lastly, to quantify the 3-/4- amount of adsorbed RNA biomarker targets, a [Fe(CN)6] redox system is used for generating a detectable Faradaic current signal.

276

Chapter 6.3

Figure 1. a) Brightfield microscopic images showing the asymmetric microelectrode dimensions in the “Mixing” (left) and “Detection” (right) microwells. b) Mechanism of ac- EHD nano-mixing effect. Under ac field (E) application, the asymmetrical microelectrodes result in a lateral quantity difference of double-layer charges nm above the microelectrode surface. This leads to a stronger lateral force on the larger microelectrode and a resultant net fluid flow towards it. c) Schematic of the analysis workflow. (i) ac-EHD nano-mixing to speed up probe-target hybridization; (ii) Addition of streptavidin-magnetic beads to bind biotinylated probe-target molecules for magnetic target purification, and subsequent polyA extensions of purified targets; (iii) Heat release of targets from capture probes; (iv) PolyA sequences facilitated rapid target adsorption onto bare gold microelectrode surface for electrochemical detection. d) Atomic force microscopy (AFM) images showing successful RNA target adsorption on the gold microelectrode surface. ac-EHD Nano-mixing for Enhanced Probe-Target Hybridization The main bottleneck of all hybridization-type assays is the long hybridization time required for freely-diffusing probes and targets to interact in solution. Microfluidic mixing can enhance the hybridization performance but mostly with little control over the mixing forces.[12] We have thus employed ac-EHD nano-mixing to resolve the dilemma of sluggish hybridization kinetics. A unique feature of ac-EHD nano-mixing effect is that the frequency and amplitude of the applied ac-field could be flexibly tuned for absolute control over the generated fluidic forces. This thus allows for optimal enrichment and probe hybridization of RNA targets.

277

Chapter 6.3

We first tested a frequency (f) range of 250-1000 Hz at constant amplitude (Vpp) of 800 mV to determine the optimal applied field strength for specific target detection in minimal time (Figure S1, Supporting Information). We found that hybridization at lower field strength of 500 Hz is more effective than at higher field strengths at which the fluidic mixing is possibly too vigorous for efficient hybridization. Furthermore, as compared to a 100 min incubation control experiment which solely relies on Brownian motion diffusion for probe-target hybridization, a similar level of detectable signal was achieved in 10 min (10-fold decrease in assay time) via ac-EHD nano-mixing (500 Hz, 800 mV) (Figure S2, Supporting Information). This demonstrates that our optimized ac-EHD nano-mixing can effectively augment probe- target hybridization, and significantly reduce total sample-to-answer duration to within 60 min.

Detection Limit for RNA Targets A high level of detection sensitivity is required to detect low levels of RNA targets present in clinical samples. To evaluate the detection limit of our nanodevice, we titrated different concentration of T2:ERG (500 aM - 500 fM) (Figure 2a), miR-107 (50 aM - 50fM) (Figure 2b), and SChLAP1 (500 aM - 500 fM) (Figure 2c) target sequences and monitored the concentration-dependent changes in electrochemical signals during detection. We found that our biosensing platform generated signals that increased linearly with increasing target concentrations, and has a limit-of-detection of 500 aM (1.5 x 104 copies); 50 aM (3 x 105 copies); 500 aM (1.6 x 104 copies) for T2:ERG (Figure 2d), miR-107 (Figure 2e), and SChLAP1 (Figure 2f) respectively. With aM-level detection sensitivity without any prior enzymatic amplification step and a total assay time of 60 min, our nanodevice outperforms current amplification-free RNA detection methods[8, 13] in sensitivity and speed. Particularly, with regards to short-length miRNA target detection which benefits the most from amplification-free biosensing strategies, we achieved an excellent miRNA detection limit similar to recent hybridization-based miRNA detection approaches[14] but without any need for intricate nanoparticle synthesis and labeling. This series of titration experiments show that the nanodevice is viable for quantifying different types of RNA targets over a range of clinically-relevant levels in patient samples,[15] and also avoids the caveats (eg, artifacts, bias) associated with enzymatic nucleic acid amplification.

278

Chapter 6.3

Figure 2. Detection sensitivity of a) T2:ERG mRNA; b) miR-107 miRNA; and c) SChLAP1 lncRNA targets. The average concentration-dependent % current response change and corresponding linear calibration plot (inset) for d) T2:ERG; e) miR-107; and f) SChLAP1 respectively. NTC represents no-target controls. Error bars represent standard deviations of triplicate independent measurements.

Specificity of Target Capture The specific capturing of intended targets is fundamental for developing accurate and reliable clinical assays. To ensure a high level of capture probe specificity is achieved for intended RNA targets, we investigated whether the capture probes used in our experiments are able to detect their respective target sequences from a heterogeneous mixture of non-target sequences. Within individual microwells on our miniaturized device, we challenged each of our designed capture probes with a mixture of different T2:ERG, miR-107, and SChLAP1synthetic RNA sequences and used the resultant electrochemical responses as a measure of successful probe-target hybridization and subsequent detection. Each of our three capture probe sequences showed high specificity towards its respective complementary target, and generated a measured signal only in the event of matched probe-target pairings (Figure S3a-c, Supporting Information). In quantitative terms, the electrochemical response change between capture probes and specific targets is about 7- to 9-fold higher than that of non-specific targets (Figure S3d, Supporting Information), and generated signals from non- specific targets are not significantly different from no-target control background signals (t- test P > 0.05).

279

Chapter 6.3

Furthermore, the hybridization between probes and targets occurs in solution without immobilized surface-bound probes and thus removes the issue of non-specific surface fouling interactions. For eventual applications in patient sample testing, the high specificity of our proposed nanodevice is essential for discriminating low levels of RNA targets from a complex background of non-target biomolecules.

Figure 3. Screening of T2:ERG, miR-107, and SChLAP1 biomarkers in a) DuCap, b) 22Rv1, and c) LnCap prostate cancer cell lines. The average % current response change and heat map representation d) of the multi-RNA type profiling data. NTC represents no-target controls Error bars represent standard deviations of triplicate independent measurements.

RNA Expression Profiling in Cultured PCa Cells To further investigate capture probe specificity and test the feasibility of our nanodevice for use in biological samples, we challenged our nanodevice with the simultaneous detection of T2:ERG, miR-107, and SChLAP1 by titrating DuCap (Figure 3a), 22Rv1 (Figure 3b), and LnCap (Figure 3c) PCa cells into healthy donor urine. It was found that the three cell lines displayed different expression profiles for the three RNA biomarkers: DuCap showed high T2:ERG with low miR-107 and SChLAP1 expressions; 22Rv1 contrastingly showed low T2:ERG with high miR-107 and SChLAP1 expressions; whilst LnCap showed high SChLAP1 with low T2:ERG and miR-107 expressions (Figure 3d). These findings were successfully validated with qPCR, and reports in literature.[4a, 5a, 6a] These results exhibited the robustness

280

Chapter 6.3 and practical utility of our nanodevice in detecting potential diagnostic PCa biomarkers from actual biological fluid samples, and underlined the unique ability of our nanodevice in detecting different RNA types simultaneously.

Non-invasive Multi-RNA Type Profiling in Patient Urine Samples To demonstrate clinical feasibility, we attempted the non-invasive profiling of T2:ERG, miR- 107, and SChLAP1 biomarkers from PCa patient urine samples (Figure 4). In total, we tested pre-treatment urine samples from 18 PCa patients and three healthy controls. Overall, we found that the PCa samples displayed more frequent T2:ERG presence as well as higher miR- 107 and/or SChLAP1 expression as compared to healthy controls (Figure 4a). Amongst the PCa patients, eight samples (P2, P3, P5, P9, P11, P14, P15, and P17) which showed higher expression for all T2:ERG, miR-107 and SChLAP1 biomarkers, were highly correlated with poorer biopsy outcomes (Table S2, Supporting Information) which indicated aggressive PCa forms. It is notable that the combined use of all three biomarkers outperformed a single biomarker’s ability in aggressive PCa discrimination. These results showcase the ability of our nanodevice in obtaining multi-RNA type biomarker signatures, and the possibility of using the resultant signature for personalized PCa risk stratification.

281

Chapter 6.3

Figure 4. a) Multi-RNA type profiling of 18 prostate cancer urine samples and three healthy controls. (*) indicates patients with poor biopsy outcomes indicative of aggressive cancer. b) Monitoring of multi-RNA type expression changes in eight patient urine samples before and after treatment. Pre- (black bars) and post- (grey bars) treatment RNA biomarker levels are directly reflected by % current response changes, with corresponding heat map diagrams. Error bars represent standard deviations of duplicate independent measurements.

282

Chapter 6.3

We performed qPCR to validate the patient sample results of our nanodevice (Table S3, Supporting Information), and used Passing and Boblok regression analysis[16] for robust evaluation of the linear relationship between two analytical methods.[17] Through this statistical test, we obtained linear regression plots which shows the degree of agreement between the two methods for T2:ERG, miR-107 and SChLAP1 detection (Figure S4, Supporting Information). The y-intercepts and slope values of the fitted regression line (in red) of our nanodevice data are close to “0” and “1” respectively within 95% confidence intervals (CI) of the reference line (in black), thus showing good agreement between both methods. Furthermore, Cusum linearity tests showed that both methods share a linear relationship with no significant differences (p > 0.05) for the detection of all three RNA target types (Figure S4, Supporting Information), and residual plots also presented narrow distribution of standard deviations around the fitted regression lines (Figure S5, Supporting Information). Therefore, we found that the qPCR data and our nanodevice data are consistent and highly correlated, and our nanodevice platform could be an ideal alternative to qPCR.

We then examined our nanodevice’s potential for monitoring a patient’s RNA biomarker expression profile in post-treatment urine samples (Figure 4b). From the eight patients with aggressive PCa who subsequently underwent radical prostatectomy treatment, the post- treatment urine samples of five patients (P2, P3, P9, P15, and P17) were detected with lower T2:ERG, miR-107 and SChLAP1 expression. These outcomes were correlated with no PCa recurrence for these patients whilst the three remaining patients (P5, P11, P14) who later developed post-treatment PCa recurrence were detected with no significant decrease in biomarker expression between their pre- and post-treatment urine samples. Besides displaying the potential of our nanodevice as a tool for PCa treatment monitoring, these results also underscore the usefulness of combining different RNA types for recurrence prediction in future clinical studies.

CONCLUSION In summary, we reported the use of nanodevice-based ac-EHD nano-mixing for amplification-free multi-RNA type profiling, with enhancements in analytical speed and sensitivity over existing assays. Using our nanodevice on non-invasive PCa urine samples as surrogates for tumor biopsies, we displayed the potential of combining our nanodevice with multi-RNA type profiling for disease diagnosis, prognosis, and monitoring. Despite our multi-step assay protocol, it is noteworthy to highlight that the current short total assay

283

Chapter 6.3 timeframe of 60 min may be further reduced through feasible automation of the heating and fluidic transfer steps. Moreover, our platform is amenable for further higher-throughput RNA analysis to provide a more comprehensive disease snapshot. We envision that our nanodevice could be a possible tool for early personalized risk stratification in individual cancer patients, thereby potentially allowing prompt therapeutic opportunities.

REFERENCES [1] a) S. A. Tomlins, J. R. Day, R. J. Lonigro, D. H. Hovelson, J. Siddiqui, L. P. Kunju, R. L. Dunn, S. Meyer, P. Hodge, J. Groskopf, J. T. Wei, A. M. Chinnaiyan, Eur. Urol. 2016, 70, 45; b) L. Fabris, Y. Ceder, A. M. Chinnaiyan, G. W. Jenster, K. D. Sorensen, S. Tomlins, T. Visakorpi, G. A. Calin, Eur. Urol. 2016, 70, 312; c) G. Leyten, D. Hessels, F. P. Smit, S. A. Jannink, H. de Jong, W. J. G. Melchers, E. B. Cornel, T. M. de Reijke, H. Vergunst, P. Kil, B. C. Knipscheer, C. A. Hulsbergen-van de Kaa, P. F. A. Mulders, I. M. van Oort, J. A. Schalken, Clin. Cancer Res. 2015, 21, 3061; d) L. Van Neste, R. J. Hendriks, S. Dijkstra, G. Trooskens, E. B. Cornel, S. A. Jannink, H. de Jong, D. Hessels, F. P. Smit, W. J. G. Melchers, G. Leyten, T. M. de Reijke, H. Vergunst, P. Kil, B. C. Knipscheer, C. A. Hulsbergen-van de Kaa, P. F. A. Mulders, I. M. van Oort, W. Van Criekinge, J. A. Schalken, Eur. Urol. 2016, 70, 740- 748; e) C. G. H. Ronnau, G. W. Verhaegh, M. V. Luna-Velez, J. A. Schalken, Biomed. Res. Int. 2014, 2014, 5917033. [2] J. C. M. Wan, C. Massie, J. Garcia-Corbacho, F. Mouliere, J. D. Brenton, C. Caldas, S. Pacey, R. Baird, N. Rosenfeld, Nat. Rev. Cancer 2017, 17, 223. [3] a) J. R. Prensner, M. A. Rubin, J. T. Wei, A. M. Chinnaiyan, Science Transl. Med. 2012, 4, 127rv3; b) S. Dijkstra, P. F. A. Mulders, J. A. Schalken, Clin. Biochem. 2014, 47, 889. [4] a) S. A. Tomlins, D. R. Rhodes, S. Perner, S. M. Dhanasekaran, R. Mehra, X. W. Sun, S. Varambally, X. H. Cao, J. Tchinda, R. Kuefer, C. Lee, J. E. Montie, R. B. Shah, K. J. Pienta, M. A. Rubin, A. M. Chinnaiyan, Science 2005, 310, 644; b) S. A. Tomlins, S. M. J. Aubin, J. Siddiqui, R. J. Lonigro, L. Sefton-Miller, S. Miick, S. Williamsen, P. Hodge, J. Meinke, A. Blase, Y. Penabella, J. R. Day, R. Varambally, B. Han, D. Wood, L. Wang, M. G. Sanda, M. A. Rubin, D. R. Rhodes, B. Hollenbeck, K. Sakamoto, J. L. Silberstein, Y. Fradet, J. B. Amberson, S. Meyers, N. Palanisamy, H. Rittenhouse, J. T. Wei, J. Groskopf, A. M. Chinnaiyan, Science Transl. Med. 2011, 3, 94ra72.

284

Chapter 6.3

[5] a) G. Martello, A. Rosato, F. Ferrari, A. Manfrin, M. Cordenonsi, S. Dupont, E. Enzo, V. Guzzardo, M. Rondina, T. Spruce, A. R. Parenti, M. G. Daidone, S. Bicciato, S. Piccolo, Cell 2010, 141, 1195; b) R. J. Bryant, T. Pawlowski, J. W. F. Catto, G. Marsden, R. L. Vessella, B. Rhees, C. Kuslich, T. Visakorpi, F. C. Hamdy, British J. Cancer 2012, 106, 768. [6] a) J. R. Prensner, M. K. Iyer, A. Sahu, I. A. Asangani, Q. Cao, L. Patel, I. A. Vergara, E. Davicioni, N. Erho, M. Ghadessi, R. B. Jenkins, T. J. Triche, R. Malik, R. Bedenis, N. McGregor, T. Ma, W. Chen, S. Han, X. Jing, X. Cao, X. Wang, B. Chandler, W. Yan, J. Siddiqui, L. P. Kunju, S. M. Dhanasekaran, K. J. Pienta, F. Y. Feng, A. M. Chinnaiyan, Nature Genet. 2013, 45, 1392; b) J. R. Prensner, S. Zhao, N. Erho, M. Schipper, M. K. Iyer, S. M. Dhanasekaran, C. Magi-Galluzzi, R. Mehra, A. Sahu, J. Siddiqui, E. Davicioni, R. B. Den, A. P. Dicker, R. J. Karnes, J. T. Wei, E. A. Klein, R. B. Jenkins, A. M. Chinnaiyan, F. Y. Feng, The Lancet Oncol. 2014, 15, 1469. [7] C. F. Chen, D. A. Ridzon, A. J. Broomer, Z. H. Zhou, D. H. Lee, J. T. Nguyen, M. Barbisin, N. L. Xu, V. R. Mahuvakar, M. R. Andersen, K. Q. Lao, K. J. Livak, K. J. Guegler, Nucleic Acids Res. 2005, 33, e179. [8] a) K. M. Koo, L. G. Carrascosa, M. Trau, Nano. Res. 2018, 11, 940; b) K. M. Koo, L. G. Carrascosa, M. J. A. Shiddiky, M. Trau, Anal. Chem. 2016, 88, 2000; c) K. M. Koo, L. G. Carrascosa, M. J. A. Shiddiky, M. Trau, Anal. Chem. 2016, 88, 6781; d) K. M. Koo, A. A. I. Sina, L. G. Carrascosa, M. J. A. Shiddiky, M. Trau, Anal. Methods 2015, 7, 7042. [9] G. K. Geiss, R. E. Bumgarner, B. Birditt, T. Dahl, N. Dowidar, D. L. Dunaway, H. P. Fell, S. Ferree, R. D. George, T. Grogan, J. J. James, M. Maysuria, J. D. Mitton, P. Oliveri, J. L. Osborn, T. Peng, A. L. Ratcliffe, P. J. Webster, E. H. Davidson, L. Hood, Nature Biotechnol. 2008, 26, 317. [10] a) M. Trau, D. A. Saville, I. A. Aksay, Science 1996, 272, 706; b) R. Vaidyanathan, S. Rauf, E. Dray, M. J. A. Shiddiky, M. Trau, Chemistry-A Eur. J. 2014, 20, 3724; c) R. Vaidyanathan, M. J. A. Shiddiky, S. Raul, E. Dray, Z. Tay, M. Trau, Anal. Chem. 2014, 86, 2042; d) M. J. A. Shiddiky, R. Vaidyanathan, S. Rauf, Z. Tay, M. Trau, Sci. Rep. 2014, 4, 3716. [11] S. Rauf, M. J. A. Shiddiky, M. Trau, Chem. Commun. 2014, 50, 4813. [12] a) T. J. Johnson, D. Ross, L. E. Locascio, Anal. Chem. 2002, 74, 45; b) C. Y. Lee, C. L. Chang, Y. N. Wang, L. M. Fu, Int. J. Mol. Sci. 2011, 12, 3263.

285

Chapter 6.3

[13] a) Z. Fang, L. Soleymani, G. Pampalakis, M. Yoshimoto, J. A. Squire, E. H. Sargent, S. O. Kelley, ACS Nano 2009, 3, 3207; b) M. N. Islam, V. Gopalan, M. H. Haque, Mostafa K. Masud, M. S. A. Hossain, Y. Yamauchi, N.-T. Nguyen, A. K.-Y. Lam, M. J. A. Shiddiky, Biosens. Bioelectron. 2017, 98, 227; c) M. H. Lin, Y. L. Wen, L. Y. Li, H. Pei, G. Liu, H. Y. Song, X. L. Zuo, C. H. Fan, Q. Huang, Anal. Chem. 2014, 86, 2285; d) K. M. Koo, L. G. Carrascosa, M. Trau, Nano Res. 2017, doi: 10.1007/s12274-017-1706-0. [14] a) Li, L. G. Xu, W. Ma, X. L. Wu, M. Z. Sun, H. Kuang, L. B. Wang, N. A. Kotov, C. L. Xu, J. Am. Chem. Soc. 2016, 138, 306; b) W. Ma, P. Fu, M. Z. Sun, L. G. Xu, H. Kuang, C. L. Xu, J. Am. Chem. Soc. 2017, 139, 11752. [15] S. O. Kelley, ACS Sens. 2017, 2, 193. [16] H. Passing, W. Bablok, J. Clin. Chem. Clin. Biochem. 1988, 21, 709. [17] W. Bablok, H. Passing, R. Bender, B. Schneider, J. Clin. Chem. Clin. Biochem. 1988, 26, 783.

SUPPORTING INFORMATION

Table S1. Oligonucleotide sequences used in experiments Oligos 5'-Sequence-3' CAACTGATAAGGCTTCCTGCCGCGCTCCAGG- TMPRSS2:ERG Capture Probe Biotin miR-107 Capture Probe TGATAGCCCTGTACAATGCTGCT-Biotin

SChLAP1 Capture Probe TGTGTCCAGAACTGGTGGGTTCTTGGTCTC-Biotin

TMPRSS2:ERG qRCR Forward ATTTAGGTACAACTCTTTCCCTCGTC Primer TMPRSS2:ERG qRCR Reverse TGTATAGGAATCCCACTGAATTTTTC Primer miR-107 qRCR Forward Primer AGCAGCATTGTACAG miR-107 qRCR Reverse Primer TTTTTTTTTTTTTTTTTG

286

Chapter 6.3

SChLAP1 qRCR Forward Primer GAGACCAAGAACCCACCAGTTCTGGACACA

SChLAP1 qRCR Reverse Primer CATCTGGCACTTCTTCCCCAGTCATTCCAT

Optimization of ac-EHD nano-mixing field strength

Figure S1. Optimization of frequency (f) (250-1000 Hz) at constant amplitude (Vpp) of 800 mV to determine the optimal applied field strength for ac-EHD nano-mixing.

Enhanced probe-target hybridization

Figure S2. ac-EHD nano-mixing to enhance probe-target collision frequency. Average % current response change at different time points over 100 min after using ac-EHD nano-

287

Chapter 6.3 mixing (black) or static incubation (red) for probe-target hybridization. Error bars represent standard deviations of triplicate independent measurements.

Specificity of RNA target capture

Figure S3. Electrochemical signals using a mixture of different RNA targets and a single a) T2:ERG mRNA; b) miR-107 miRNA; and c) SChLAP1 lncRNA target capture probe. The average % current response change d) for specific detection of T2:ERG, miR-107, and SChLAP1. NTC represents no-target controls Error bars represent standard deviations of triplicate independent measurements.

Table S2. Tumor Gleason scores for patient biopsy samples. Highlighted in bold are patients with a Gleason score of >7 that is indicative of aggressive cancer. Sample no. Gleason Score Sample no. Gleason Score P1 7 P10 7 P2 8 P11 8 P3 9 P12 6 P4 6 P13 6 P5 8 P14 9

288

Chapter 6.3

P6 6 P15 9 P7 7 P16 6 P8 7 P17 8 P9 9 P18 7

Table S3. qPCR validation of urinary sample results. (*) indicates patients with poor biopsy outcomes indicative of aggressive cancer. qPCR Ct (Cycle) Our Device Data (% i) Sample TMPRSS2:ERG miR1- SChLAP1 TMPRSS2:ERG miR1- SChLAP1 no. 07 07 P1 33.2 25.1 15.9 0 30 90 P2* 17.5 17.7 16.5 80 80 85 P3* 20.5 19.3 17.6 65 70 80 P4 24.5 23.1 25.6 40 50 30 P5* 17.7 15.4 17.9 80 95 80 P6 28.3 22.9 24.8 20 60 40 P7 30.3 23.6 25.5 10 45 30 P8 29.8 25.5 23.6 10 30 45 P9* 23.0 17.4 16.0 60 80 90 P10 28.5 25.3 17.7 20 50 80 P11* 16.3 17.3 20.1 90 80 70 P12 25.8 22.2 17.0 30 60 75 P13 29.7 24.1 24.8 10 40 40 P14* 19.6 16.3 16.6 70 90 90 P15* 17.1 17.3 17.0 80 80 80 P16 25.4 24.7 28.2 30 40 20 P17* 22.4 19.1 17.8 60 70 80 P18 25.1 25.0 25.4 30 30 30 H1 30.1 24.5 28.2 10 40 20 H2 32.7 25.3 28.4 0 30 20 H3 33.0 28.0 25.6 0 20 30

289

Chapter 6.3

Regression plots of Passing-Boblok analyses

Figure S4. Passing−Boblok regression analyses (n = 21) of our nanodevice-based assay and qPCR for the detection of a) T2:ERG, regression line equation: y = 0.89 x + 5.96 (95% CI, 0.84-0.97 for slope and 3.23-8.84 for intercept); b) miR-107, regression line equation: y = 0.98 x + 2.65 (95% CI, 0.88-1.07 for slope and -1.84-2.65 for intercept); and c) SChLAP1, regression line equation: y = 0.96 x + 3.02 (95% CI, 0.89-1.01 for slope and 0.40-8.86 for intercept) biomarkers.

7. Residual plots of Passing-Boblok analyses

Figure S5. Passing−Boblok residual plots showing standard deviation distribution around the fitted regression lines of our device-based assay and qPCR for the detection of a) T2:ERG; b) miR-107; c) SChLAP1 biomarkers.

290

Chapter 7

Clinical Evaluation of Non-Invasive Nanodiagnostics for Prostate Cancer Risk Stratification

Introduction To facilitate the translation of novel nanodiagnostic approaches from research bench to the clinic, it is an essential requirement to perform a comprehensive clinical evaluation of the developed nanotechnologies with well-annotated patient sample cohorts. Explicitly, en-route to the eventual translation of academic nanotechnological innovations into clinical tools for improving patient care; clinical metrics such as clinical sensitivity and specificity, positive and negative predictive values, and the area-under-curve value of the receiver operator characteristic curve have to be assessed in order to demonstrate nanodiagnostic performance and utility in real human patients. This chapter discusses the clinical evaluation of a nanodiagnostic assay that has been previously introduced for non-invasive prostate cancer early detection and risk stratification in patient urine samples. Moreover, during the assay clinical evaluation process, the clinical value of a novel target biomarker panel for improving prostate cancer screening, is also being simultaneously investigated. Therefore, pilot clinical studies permitted combined clinical evaluation of both nanodiagnostic technologies and urinary RNA biomarkers. Chapter 7 reports the clinical performance of a flocculation-based TMPRSS2:ERG triage assay for non-invasive visual identification of significant prostate cancer in pre- and post-prostatectomy prostate cancer urine samples.

Chapter 7 is based on one submitted research manuscript.

291

Chapter 7

TMPRSS2:ERG Status in Pre- And Post-Prostatectomy Urine: A Novel First-Line Triage Nanoassay for Non-Invasive Visual Identification of Significant Prostate Cancer

Kevin M. Koo, Eugene J. H. Wee, Renée S. Richards, Martin F. Lavin, Robert A. Gardiner, Paul N. Mainwaring, Matt Trau

ABSTRACT Background: Novel screening assays for identification of significant (Gleason score ≥7) prostate cancer (PCa) are needed to address the negative overtreatment impact on patients with insignificant PCa. The nanoscaled Single Drop Genomics (SDG) assay has been demonstrated for visual detection of urinary TMPRSS2:ERG (T2:ERG) transcripts with single cell-level sensitivity. By relating urinary T2:ERG status to pathology outcomes, we herein investigated SDG’s clinical potential as a first-line triage assay to impact clinical decision making. Methods: Pre- and post-prostatectomy urine samples were collected and urinary sediment RNA was extracted. T2:ERG status was determined using SDG which combines isothermal RNA amplification and visual evaluation. SDG/T2:ERG outcomes were correlated to clinical and pathological data, and compared against serum prostate-specific antigen (PSA) and Prostate Cancer Prevention Trial (PCPT) calculator risk prediction tools. SDG clinical evaluation was performed by receiver operating characteristic curve analysis. Results: In this cohort of 46 paired pre- and post-prostatectomy and 15 healthy volunteers/ PCa biopsy-negative control urine samples, positive and negative T2:ERG status by SDG were significantly correlated with significant and insignificant PCa respectively. The observed positive and negative predictive value for detecting significant PCa is 90.8%, and 72.3% respectively. In our patient cohort, the area under curve (AUC) of SDG/T2:ERG is 0.75 as compared to 0.54 by use of serum PSA test. Conclusions: In perspective, the T2:ERG detection outcomes of SDG are promising for significant PCa identification during early screening.

292

Chapter 7

INTRODUCTION Despite being one of the most common solid tumor malignancy and leading cause of cancer- related deaths in men, the current prostate-specific antigen (PSA)-based screening of prostate cancer (PCa) remains an imperfect test which results in significant over-diagnosis and - treatment. The general conclusion of the recent findings of two major randomized PSA-based PCa screening trials [1, 2] has suggest that the net benefit of PSA testing is insufficient to risk the proven harms of unnecessary radical prostatectomy/radiation therapy [3]. Hence, there is a critical need for better screening strategies to accurately identify patients (Gleason score ≥7) who actually require and can benefit from PCa treatments.

We envisage that an improved PCa screening test in the clinic would encompass both the use of a more PC-specific biomarker, and an accompanying screening technology which could perform biomarker detection in a fast, affordable, and accurate fashion. T2:ERG (T2:ERG) gene fusions are the most common chromosomal rearrangement mutations in PCa, and have been shown to be highly specific for PCa [4], and only detectable in high-grade prostatic intraepithelial neoplasia (PIN) lesions in close proximity to T2:ERG-positive tumor foci [5]. There have been also several reports of an association between T2:ERG presence and clinically-significant aggressive PCa [6-8]. Additionally, T2:ERG RNA are detectable from patient urine [9], thus allowing for convenient non-invasive testing.

Recently, we have demonstrated a novel miniaturized Single Drop Genomics (SDG) technique for visual detection of urine-isolated T2:ERG RNA within a minuscule fluid droplet [10]. In principle, SDG employed magnetic beads for DNA-mediated bridging flocculation to initiate a solution color change in T2:ERG presence. SDG displayed single- cell detection sensitivity, and exploited the binary (presence/absence) nature of T2:ERG in PC for rapid naked-eye evaluation with minimal equipment. Importantly, the isothermal target amplification time of the SDG is adjustable to measure a certain threshold of T2:ERG level. Therefore, SDG could be an ideal triage assay for risk stratifying PCa non-invasively using urine samples.

In this paper, we investigated the use of our developed technique as a first-line T2:ERG triage assay for non-invasive visual identification of significant prostate cancer. In a cohort of 46 paired pre- and post-prostatectomy urine samples, we correlated SDG-detected T2:ERG

293

Chapter 7 status to pathology outcomes, and investigated the potential of our SDG-T2:ERG approach in comparison to other current PCa predictive tools.

METHODS Study Population This study consists of men who underwent radical prostatectomy treatment, due to suspicious DRE and/or PSA screening results during initial screening. Men were enrolled at the Royal Brisbane and Women’s Hospital. Ethics approval was obtained from The University of Queensland Institutional Human Research Ethics Committee (Approval No. 2004000047). All subjects provided informed consent prior to sample collection and methods pertaining to clinical samples were carried out in accordance with approved guidelines.

Sample collection and processing Voided urine specimens were collected from men and assigned a study ID as per standard procedure. Samples were stored at -80 °C until sample processing. Total RNA was extracted from 1.5 mL of urine sample using the ZR urine RNA isolation KitTM (Zymo Research, USA) according to manufacturer’s instructions, and eluted in 15 µl of RNase-free water.

SDG assay The SDG assay is performed in the following steps: isothermal biomarker amplification, magnetic amplicon purification, visual detection of biomarker presence. For isothermal amplification, the TwistAmp Basic RT-RPA kit (Twist-DX, UK) was used with slight modifications to manufacturer’s instructions. Briefly, 2 µl of extracted RNA and 500 nM of each primer (Table S1) were added to make a 12.5 µl reaction volume prior to incubation at 41°C for 30 min. Next, amplicons were purifed using the Agencourt AMPure XP kit (Beckman Coulter, USA). Two volumes of magnetic beads were first incubated with one volume of amplicons for 10 minutes. The amplicon-bound magnetic beads were then separated using a magnet and washed once with 100% isopropanol and once with 80% ethanol. Finally, total RNA was eluted in 25 µl of RNase-free water. After purification, 5 µl of purified RPA amplicons was incubated with two volumes of magnetic beads for 5 min before bead separation by magnet. Then, 20 µL of flocculation buffer (200 mM acetate buffer, pH 4.4) was added to the beads. After 1 min of incubation, the mixture was gently agitated using a magnet and observed for colour change.

294

Chapter 7

Statistical analysis All data processing and ROC curve generation were performed using the Origin software (OriginLab, USA).

RESULTS Patient characteristics Non-DRE urine samples were collected from 61 men. Of the 61 men, 15 were healthy volunteers/PCa biopsy-negative controls. For the remaining 46 men, pre- and post- prostatectomy urine samples were collected for T2:ERG expression analysis. For each sample, RNA was extracted and presence/absence of T2:ERG and RN7SL1 internal control transcripts were determined using the nanotechnological-based SDG assay.

The median age of the men is 62.5 years and median pre-prostatectomy serum PSA is 7.5 ng/mL (Table 1). All 61 men were of Cucasian descent, and the tumor-positive rate was 75.4% with 97.8% having a Gleason score (GS) ≥7 (69.9% GS 3+4, 45.7% GS 4+3)

SDG screening outcomes for PCa prediction For each sample, positive T2:ERG status was visually evaluated by a change in solution color [10]. An evaluation for RN7SL1 internal control was also being performed to ensure successful RNA extraction from each urine sample. In our cohort of pre- urine samples, SDG exhibited good clinical performance in the specific detection of PCa. As compared to the use of PSA, the SDG-detected T2:ERG status demonstrated greater positive predictive value (PPV) (90.8% vs 60.2% ), negative predictive value (NPV) (72.3% vs 59.4%), and AUC (0.90 vs 0.54) for tumor-positive samples (Figure 1a). The SDG assay generated a positive T2:ERG status in all tumor-positive samples and displayed negative outcomes in all the control samples.

295

Chapter 7

Table 1. Patient clinical characteristics and SDG-T2:ERG status of pre-prostatectomy urine samples. Pre-Op Pre-Op PSA (>4 T2:ERG (+) T2:ERG (-) ng/mL) 46 Patients, n (%) (100) 25 (54.3) 21 (45.7) 34 (76.1 %) Control cohort 15 Age (years) 62.5 Median serum PSA level (ng/mL) 7.5

Gleason Score GS<7 1 (2.2) 0 (0.0) 1 (4.8) 3 (2.1) 17 GS (3+4) (36.9) 1 (4.0) 16 (76.2) 13 (37.1) 21 GS (4+3) (45.7) 18 (72.0) 3 (14.2) 10 (28.6) 7 GS>7 (15.2) 6 (24.0) 1 (4.8) 9 (25.7)

Pathologic Stage 27 T1, T2 (58.7) 19 >T2 (41.3)

296

Chapter 7

Figure 1. ROC curves of SDG-T2:ERG and PSA for tumor-positive diagnoses (a), and SDG- T2:ERG and PCPT for high-grade PCA detection (b).

Detection of high grade prostate cancer The Gleason scoring of prostate tissue is presently the most powerful predictor of PCa prognosis. The predictive ability of the SDG-T2:ERG assay was further evaluated in the prediction of pathological GS associated with significant PCa (GS ≥7 and pathologic tumor >2). From pre-prostatectomy samples, positive SDG-T2:ERG assay outcomes were associated with significant PCa (AUC = 0.75). As compared to Prostate Cancer Prevention Trial (PCPT) risk calculator (AUC = 0.58) that incorporates PSA and other clinical factors in an attempt to provide an individual risk estimate, the SDG-T2:ERG assay showed better predictive ability for significant PCa (Figure 1b).

As studies have shown GS≥4+3 to have a worse prognosis as compared to GS 3+4 PCa [11, 12], we investigated SDG-T2:ERG’s accuracy in improving prognostic prediction for pathologic GS 4+3/3+4 status. Among 45 samples (97.8%) with GS ≥7, the SDG-T2:ERG assay correctly discriminated 72% GS 4+3/>7 vs 4% GS 3+4 samples (AUC = 0.88) (Table 1).

297

Chapter 7

Monitoring of biochemical recurrence Using post-prostatectomy urine samples, we also investigated the SDG-T2:ERG assay’s utility as a PCa monitoring tool for predicting biochemical recurrence. Amongst the 42 post- prostatectomy samples, 58.7% developed biochemical recurrence as indicated by a raising PSA level. The SDG assay detected positive T2:ERG status in 79.2% of these samples, thus suggesting that SDG-T2:ERG is a viable alternative post-treatment urinary assay for monitoring of PCa biochemical recurrence (Table 2).

Table 2. Patient biochemical recurrences and SDG-T2:ERG status of post-prostatectomy urine samples. Post-Op Fusion Post-Op Fusion (+) (-) Patients, n (%) 46 (100) 24 (34.8) 22 (65.2)

Biochemical Recurrence 27 (58.7) 19 (79.2) 8 (36.4) No Biochemical Recurrence 19 (41.3) 5 (20.8) 14 63.6)

DISCUSSION We have described a novel nanotechnological SDG assay which evaluates urinary T2:ERG status for diagnosing men with high-grade (GS≥7) PCa at early screening. The T2:ERG biomarker is one of the most PCa-specific diagnostic biomarkers, and Laxman et al. first reported detection of T2:ERG in patient urine samples as an attractive and non-invasive assay for PCa early detection [9]. Notably, it was reported that the quantity of urinary T2:ERG transcripts strongly correlated with tumour burden [7, 8]. Furthermore, several research groups have also demonstrated the potential prognostic utility of urinary T1E4 detection through association with aggressive PCa [6, 13]. While T2:ERG has routinely been detected in urine samples by conventional research-based RT-PCR techniques [13, 14], we identified that the use of a nanotechnological-based assay may enable a faster T2:ERG screening approach with a simplified readout that is amenable for use as a clinical PCa triage assay. SDG has previously been shown to possess single-cell detection sensitivity [10], and could potentially be useful for non-invasive T2:ERG detection in urine without prior digital rectal exam (DRE) as demonstrated in this work.

298

Chapter 7

In our cohort of samples, SDG-T2:ERG specifically detects PCa (AUC = 0.90, NPV = 72.3%) (Figure 1a) and also accurately predicts significant (GS≥7) PCa (AUC = 0.75) (Figure 1b). Furthermore, among the high-grade PCa, SDG-T2:ERG is able to accurately discriminate a larger portion of higher-grade GS 4+3 tumors (AUC = 0.88) (Table 1). These results highlighted the PCa-specificity of the T2:ERG biomarker over serum PSA and its association with clinically significant PCa. It also demonstrated the robustness of the SDG assay in T2:ERG detection from urine samples. Our use of T2:ERG to detect significant PCa is consistent with a clinical-grade, transcription-mediated amplification assay which found urine T2:ERG expression to be associated with clinically significant cancer at biopsy and prostatectomy [8], thereby supporting a role for T2:ERG as a potential PCa detection and risk stratification biomarker. Additionally, we also utilized the SDG-T2:ERG assay on post- prostatectomy urine samples to predict for biochemical recurrence and the obtained results reflected that our assay may be of use for disease recurrence monitoring purpose after radical prostatectomy treatment (Table 2).

Despite the highly promising initial results, there are some limitations to this study. Firstly, the statistical power of this study is limited by a relatively small cohort. However, it is noteworthy that our cohort is well-represented with men of GS≥7 (including both 4+3 and 3+4) pathology findings in order to study the high grade disease discriminating ability of our SDG-T2:ERG assay. Our initial data herein is encouraging, and we are progressing with expansion of our clinical sample cohort by collecting and longitudinal monitoring of clinical data and samples. Another point of concern is that our SDG assay is unable to quantify T2:ERG levels. However, we will like to highlight that an optimal SDG cut-off point for T2:ERG detection could be easily set by adjusting the isothermal target amplification time.

CONCLUSION This study highlights the use of SDG, a novel nanotechnological assay which targets the urinary T2:ERG biomarker for discriminating significant PCa in men presented for early screening.

REFERENCES 1. Andriole, G.L., et al., Mortality results from a randomized prostate-cancer screening trial. New England Journal of Medicine, 2009. 360(13): p. 1310-1319.

299

Chapter 7

2. Schroeder, F.H., et al., Screening and prostate-cancer mortality in a randomized european study. New England Journal of Medicine, 2009. 360(13): p. 1320-1328. 3. Pinsky, P.F., P.C. Prorok, and B.S. Kramer, Prostate cancer screening - a perspective on the current state of the evidence. New England Journal of Medicine, 2017. 376(13): p. 1285-1289. 4. Tomlins, S.A., et al., Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science, 2005. 310(5748): p. 644-648. 5. Mosquera, J.M., et al., Characterization of TMPRSS2-ERG fusion high-grade prostatic intraepithelial neoplasia and potential clinical implications. Clinical Cancer Research, 2008. 14(11): p. 3380-3385. 6. Wang, J.H., et al., Expression of variant TMPRSS2/ERG fusion messenger RNAs is associated with aggressive prostate cancer. Cancer Research, 2006. 66(17): p. 8347- 8351. 7. Young, A., et al., Correlation of urine TMPRSS2:ERG and PCA3 to ERG plus and total prostate cancer burden. American Journal of Clinical Pathology, 2012. 138(5): p. 685-696. 8. Tomlins, S.A., et al., Urine TMPRSS2:ERG fusion transcript stratifies prostate cancer risk in men with elevated serum PSA. Science Translational Medicine, 2011. 3(94): 94ra72. 9. Laxman, B., et al., Noninvasive detection of TMPRSS2 : ERG fusion transcripts in the urine of men with prostate cancer. Neoplasia, 2006. 8(10): p. 885-888. 10. Koo, K.M., et al., A simple, rapid, low-cost technique for naked-eye detection of urine-isolated TMPRSS2: ERG gene fusion RNA. Scientific Reports, 2016. 6: 30722. 11. Stark, J.R., et al., Gleason score and lethal prostate cancer: does 3+4=4+3? Journal of Clinical Oncology, 2009. 27(21): p. 3459-3464. 12. Sinnott, J.A., et al., Prognostic utility of a new mRNA expression signature of Gleason score. Clinical Cancer Research, 2017. 23(1): p. 81-87. 13. Leyten, G., et al., Prospective multicentre evaluation of PCA3 and TMPRSS2-ERG gene fusions as diagnostic and prognostic urinary biomarkers for prostate cancer. European Urolology, 2014. 65(3): p. 534-542. 14. Laxman, B., et al., A first-generation multiplex biomarker analysis of urine for the early detection of prostate cancer. Cancer Research, 2008. 68(3): p. 645-649.

300

Chapter 7

SUPPORTING INFORMATION

Table S1 Primer sequences Primer Sequence (5’ to 3’)

RT-RPA Forward CCTGGAGCGCGGCAGGAAGCCTTATCAGTTG

RT-RPA Reverse GCTAGGGTTACATTCCATTTTGATGGTGAC

Oligonucleotides were purchased from Integrated DNA Technologies, USA

301

Chapter 8

Conclusions & Future Work

8.1 Conclusions In conclusion, with the goal to improve PCa early detection for better disease risk stratification in the clinic, this thesis has advanced both the fields of clinical PCa screening methodologies and accurate non-invasive identification of aggressive PCa subtypes from urine samples. The core research work being showcased throughout this dissertation on the design of novel nanodiagnostics for next-generation PCa biomarker detection has successfully met the stated thesis objectives under Section 1.2.

The first research objective is the development of gene fusion (TMPRSS2:ERG) diagnostics towards non-invasive clinical use. This thesis has described various TMPRSS2:ERG nanobiosensing methods which innovatively unified isothermal recombinase polymerase amplification with a variety of detection readouts to enable different applications. These include a bridging flocculation-based assay for simple naked-eye detection (Chapter 3.2); an enzyme-catalyzed assay with flexible visual and quantitative detection (Chapter 3.3); two different label-free approaches with direct SERS detection (Chapter 4). Depending on the need for visual, quantitative, or molecular analysis of TMPRSS2:ERG status in PCa urinary specimens, each of these nanodiagnostic tools could be employed suitably for the intended purpose.

The second research objective is the design of amplification-free nanobiosensors for universal detection of various RNA biomarker species. By using nucleic acid target-magnetic enrichment and facile adsorption-based electrochemical target detection (Chapter 5), this thesis has reported a suite of enzymatic amplification-free techniques for all major species of next-generation PCa urinary RNA biomarkers. These techniques have allowed direct RNA detection without the related shortcomings of nucleic acid amplification, cumbersome sensing surface modification, and detection label hybridization chemistry.

302

Chapter 8

The third research objective is the expansion into simultaneous detection of multiple biomarkers. This thesis has achieved this objective through two different strategies, namely by multiplexed screening or high-throughput parallel screening. For multiplexed screening, the first-ever “one-pot” five-plexed urinary RNA biosensor by use of various self-synthesized SERS nanotags (Chapter 6.1) is shown. For high-throughput parallel screening, rapid real- time fluorescence detection of different PCa gene fusion variants is demonstrated through a novel merger of multiple concurrent DNA probe-RNA target ligation and rapid real-time recombinase polymerase amplification (Chapter 6.2); and a nanofluidic manipulation technique for rapid amplification-free multi-RNA type profiling in pre- and post-treatment patient urinary samples (Chapter 6.3). These assays have enabled quick simultaneous detection of different next-generation PCa biomarkers for potentially more specific and informative screening.

The fourth and fifth research objectives are the clinical evaluation of developed nanotechnological assays, and the investigation of biomarker value via pilot clinical studies. These two objectives are intricately linked because the clinical performance of both the nanodiagnostic assay and target biomarker panel were jointly analyzed when being applied to a broad cohort of patient samples. Chapter 7 presents the clinical evaluation of a nanoscaled assay developed during the course of this thesis for TMPRSS2:ERG detection in pre- and post-prostatectomy urine samples. By relating urinary TMPRSS2:ERG status to pathology outcomes, we investigated the combination of nanotechnological assay and TMPRSS2:ERG biomarker as a first-line triage assay to impact clinical decision making. The “blueprint” for clinical assay evaluation and biomarker value investigation in this chapter could subsequently be adapted for other nanotechnologies to translate academic-laboratory diagnostic innovations for patient care efficiently.

8.2 Future work As concluded above, the work described in this thesis has significantly contributed towards advancements in a new era of more accurate and personalized PCa screening. Consequently, the research outcomes which were established within this thesis will serve as a basis for future studies.

From an analytical standpoint, the target detection limit for most techniques may still be further improved upon through nanoscaled innovations. There are two approaches to increase

303

Chapter 8 the detection sensitivity, namely, amplification of target copies and/or readout signals. For target amplification, it is worth investigating the utility of other isothermal nucleic acid amplification techniques besides recombinase polymerase amplification. This would allow side-by-side evaluations of the benefits and shortcomings of integrating different isothermal amplification techniques into the various nanodiagnostic platforms being described in this thesis. For readout signal amplification, a promising direction would research into using more sophisticated surface-enhanced Raman scattering readout labels. These readout labels may possess distinctive morphologies or surface nanostructures which aid in producing greater Raman enhancement effects. Thus, it would be interesting to pursue the incorporation of these advanced labels into our multiplexed biosensing technique (Chapter 6.1) to potentially achieve greater detection sensitivity.

For all nanodiagnostic approaches showcased throughout this thesis, the sample preparation step to extract cellular nucleic acid has primarily been achieved through commercial magnetic bead purification or spin-column kits. Whilst efficient and straightforward to perform, these methodologies are currently the most time-consuming step in the entire sample-to-answer workflow. A more ideal preparation protocol would be a direct sampling methodology whereby tumor cells are lyzed in its native biological medium (i.e. blood or urine), and the released nucleic acid targets could then be promptly used for enzymatic amplification or capture probe hybridization via the developed nanotechnological techniques of this thesis work. Hence, further work is to be done with regards to rapid and simplified target extraction from clinical samples. Specifically, this requires research into ways to keep cellular nucleic acids stable and protected from degradation once being released, as well as methodologies to collect the released nucleic acid targets for transfer to downstream processing.

From a clinical point of view, extensive clinical evaluation of newly-developed nanodiagnostic technologies and novel biomarker panel is essential to ensure adequate performance for patient use. In this regard, the described work in Chapter 7 has attempted to address this need for clinical evaluation, and has resulted in promising initial outcomes. However, these pilot clinical studies are still preliminary; and future work in testing with a higher number of patient samples to further evaluate clinical metrics, blinded studies, cross- institutional technological validation, will be necessary to demonstrate improved PCa screening benefits. Furthermore, during these further studies, discoveries of more next-

304

Chapter 8 generation PCa biomarkers may be uncovered. These new promising biomarkers may be readily incorporated for assessment of significant gain in area-under-curve value of the receiver operator characteristic curve. Finally, it would be worthy to utilize the clinical studies to answer some current pressing biological issues about urinary PCa screening. For instance, the feasibility of using urine as an ideal surrogate for blood/biopsy tissue could be thoroughly examined by using nanodiagnostics to test the biomarker correlation in the urine, blood, and biopsy tissue specimens of the same patient. Another potential longitudinal study could also be performed by using nanodiagnostics to investigate urinary biomarker expression (i.e. tumorigenesis) changes at different time points during PCa progression.

This thesis has yielded several nanodiagnostic platforms for RNA biosensing. While PCa cancer gene fusion detection is primarily used as the biological/clinical model in this thesis, further extension of these platform applications for gene fusion screening in other cancers (eg. BCR-ABL1in leukemia; BRAF fusions in melanoma; EML4-ALK in lung cancer) as well as RNA biomarker detection in other diseases (eg. oncogenic, pathogenic or viral) would be worth pursuing.

305