Point-Of-Care Diagnostics for Lung Transplantation

Point-of-Care Diagnostics for Lung Transplantation

Andrew Thomas Sage

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Pharmaceutical Sciences University of Toronto

Point-of-Care Diagnostics for Lung Transplantation

Andrew Thomas Sage

Doctor of Philosophy

Graduate Department of Pharmaceutical Sciences University of Toronto

2015 Abstract

Over the last few decades, lung transplantation (LTx) has become a well-established therapy for patients suffering from end-stage lung disease. Despite the many innovations that have occurred in recent years, the outcomes following LTx still lag behind those of other organ transplant procedures due to a lack of technologies that could monitor predictive biomarkers. Recently, several genomic and proteomic analytes have been identified that are prognostic of LTx outcome; however, traditional analytical methods are impractical for use in the transplant setting. Rapid analysis of molecular biomarkers reporting on the initial and long-term status of a donated organ would improve utilization rates and transplant outcomes.

The work herein focuses on developing a novel class of chip-based sensors that would enable direct analysis of mRNAs and proteins that are correlated with the suitability of a lung for transplant. The biosensing approach utilized in these studies was performed using a simple and elegant electrochemical reporter system with metal electrodes of varying morphologies and sizes.

By fabricating fractal circuit sensors (FraCS), we demonstrated the technological and biological validation required for a point-of-care (POC) diagnostic device to be used for gene quantitation in the transplant setting. This approach was applied to confirm the clinical validity of previously reported biomarkers (IL-6, IL-10, EGLN1, and ATP11B) and, importantly, was used to develop a FraCS prediction model (FPM), which was highly predictive (AUC 0.82) of primary graft dysfunction (PGD) in donor lungs.

Using nanostructured microelectrodes (NMEs), a sensing platform was developed to detect two proteins implicated in chronic lung rejection: TGFβ1 and ET-1 – biomarkers involved in restrictive allograft syndrome (RAS) and bronchiolitis obliterans syndrome

(BOS). We highlight novel strategies used for the electrochemical detection of small (<

50-kDa) proteins and demonstrate that this approach is compatible for use with lung perfusate – a biopsy-free approach to donor lung outcome prediction.

This work reports the rapid detection of lung transplant specific biomarkers for the first time. It achieved this advance with the use of novel electrochemical biosensors. These diagnostic tools represent a significant development in the field of transplantation. In summary, this work further strengthens the role of POC devices in the transplant setting and will have broad implications for the transplant community.

iii

Acknowledgments

This work is the culmination of the efforts of many people, whose contributions are truly appreciated. First and foremost, I would like to thank my Ph.D. supervisor, Dr. Shana Kelley. I am extremely grateful for the opportunities I have been given in your research lab and the independence in which you let me pursue this project with. Your guidance and support throughout this process has helped me become a better scientist.

To my committee members, Dr. Shaf Keshavjee, Dr. Mingyao Liu, and Dr. Edward Sargent: I am so very thankful for your efforts in this project. Together, you helped push this work into something I am extremely proud of. Your advice and suggestions have taught me so much and has given me such a strong foundation in this field.

There are so many people in both the Kelley Laboratory and Latner Thoracic Surgery Research Group that have had a hand in this work. For all of your contributions, I am sincerely grateful. Projects such as these would not be possible without the hard work and dedication of so many.

I would like to thank my family for always being so supportive through every step of this process – your personal triumph in the face of adversity continues to inspire me. Finally, I would like to thank Holly for being there for me throughout this work. You have been there for the ups and downs of research and your love and support mean the world to me. Thank you for being the sounding board to my ideas, the voice of reason, and my best friend.

Table of Contents

Abstract ...... ii

Acknowledgements ...... iv

List of Tables ...... viii

List of Figures ...... ix

List of Abbreviates ...... xi

CHAPTER 1 - Introduction

1.0 ...... 1

1.1 The Challenges Facing Lung Transplantation ...... 2

1.2 Origins and Rationale for Lung Transplantation ...... 4

1.3 Lung Transplantation – A Donor Lung Shortage ...... 5

1.4 Novel Strategies for Organ Preservation ...... 6

1.5 Patient Outcomes Following Lung Transplantation ...... 8

1.6 Prognostic Tools in Lung Transplantation - Biomarkers ...... 10

1.7 The Challenges Surrounding Current Lung Transplantation Diagnostics ..... 17

1.8 The Current State of Diagnostics ...... 18

1.9 Assays for Protein Detection ...... 19

1.10 Assays for Nucleic Acids Detection ...... 22

1.11 The Emergence of Electrochemical Biosensors ...... 23

1.12 Meeting the Challenges of RNA Extraction ...... 25

1.13 Putting it all Together – Integrated Diagnostic Devices ...... 26

1.14 The Future of Electrochemical Sensors–Nanostructured Microelectrodes . 27

1.15 Using the Nanostructured Microelectrode Platform for Superior Clinical Performance ...... 33 v

1.16 Meeting the Challenges of Lung Transplantation Diagnostics–Summary .. 36

1.17 Project Hypothesis and Goals ...... 37

1.18 Overview of Thesis Chapters ...... 39

CHAPTER 2 – The Development of a Quantitative Biosensing Platform

2.0 ...... 41

2.1 Abstract ...... 42

2.2 Introduction ...... 42

2.3 Results ...... 44

2.4 Discussion ...... 59

2.5 Conclusions ...... 65

2.6 Materials and Methods ...... 66

CHAPTER 3 – Using the FraCS Platform to

Predict Lung Transplantation Outcomes

3.0 ...... 71

3.1 Abstract ...... 72

3.2 Introduction ...... 72

3.3 Results ...... 75

3.4 Discussion ...... 84

3.5 Conclusions ...... 87

3.6 Materials and Methods ...... 87

CHAPTER 4 – The Development of an Electrochemical Assay

for Small Protein Targets Using TGFβ1

4.0 ...... 92

4.1 Abstract ...... 93

4.2 Introduction ...... 93

4.3 Results ...... 96

4.4 Discussion ...... 101

4.5 Conclusions ...... 103

4.6 Materials and Methods ...... 104

CHAPTER 5 – Detection of Small Peptides –

An Electrochemical Assay for Endothelin-1 Analysis

5.0 ...... 107

5.1 Abstract ...... 108

5.2 Introduction ...... 108

5.3 Results ...... 111

5.4 Discussion ...... 118

5.5 Conclusions ...... 119

5.6 Materials and Methods ...... 120

CHAPTER 6 – Conclusions and Future Outlook

6.0 ...... 123

6.1 Ph.D. Summary and Perspective ...... 124

6.2 Concluding Remarks ...... 127

CHAPTER 7 – References

7.0 ...... 129

7.1 List of Contributed Works ...... 130

7.2 References ...... 131

vii

List of Tables

Table 1. Lung transplant recipients ...... 4

Table 2. Summary of nucleic acid and protein biomarkers implicated in LTx. .... 16

Table 3. Point-of-care diagnostic capabilities ...... 17

Table 4. Biosensor hybridization signals and aperture parameters ...... 47

Table 5. FraCS signal reproducibility statistics ...... 53

Table 6. PNA probe sequences for LTx assay ...... 55

Table 7. Bland-Altman results of FraCS vs. qPCR ...... 79

Table 8. PGD-predictive value of LTx biomarkers ...... 82

Table 9. The FraCS Prediction Model (FPM) ...... 83

Table 10. Diagnostic characteristics of the FPM ...... 84

Table 11. Estimated endogenous ET-1 concentrations ...... 117

Table 12. Solid organ transplantation in Canada and the United States ...... 126

viii

List of Figures

Figure 1. Normothermic ex vivo lung perfusion ...... 6

Figure 2. Inflammatory pathways involved in LTx outcomes ...... 13

Figure 3. Detection platforms for nucleic acid detection ...... 23

Figure 4. The nanostructured microelectrode assay ...... 28

3+ 3- Figure 5. [Ru(NH3)6] /[Fe(CN)6] electrochemical reporter system ...... 30

Figure 6. Voltammetric techniques ...... 32

Figure 7. Biosensor morphology ...... 45

Figure 8. Biosensor morphology and hybridization currents ...... 46

Figure 9. Electrochemical performance of circular and linear apertures ...... 47

Figure 10. Nucleic acid quantification using linear apertures ...... 49

Figure 11. Sensor surface density and limit of detection ...... 50

Figure 12. Rapid analyte detection using FraCS ...... 51

Figure 13. Biosensor response in unpurified lysate ...... 52

Figure 14. FraCS signal reproducibility ...... 53

Figure 15. LTx assay probe validation ...... 54

Figure 16. Validation of LTx probes with DNA targets ...... 56

Figure 17. Quantification of RNA markers predictive of lung transplant outcomes ...... 57

Figure 18. GAPDH validation ...... 58

Figure 19. Donor lung profiling using LTx biosensors ...... 59

Figure 20. Lung transplant assessment assay ...... 75

Figure 21. Fractal Circuit Sensors (FraCS) ...... 76

Figure 22. FraCS vs. Gold Standard, qPCR ...... 77

Figure 23. qPCR validation ...... 78 ix

Figure 24. Bland-Altman analysis of FraCS vs. qPCR ...... 79

Figure 25. Relative expression of LTx biomarkers ...... 80

Figure 26. EGLN1 expression in donor lungs ...... 81

Figure 27. FraCS Prediction Model (FPM) ...... 83

Figure 28. Protein detection using ferricyanide/ferrocyanide-blocking ...... 96

Figure 29. Electrochemical scanning using the ferricyanide/ferrocyanide-blocking assay ...... 97

Figure 30. Electrochemical LAP detection ...... 97

Figure 31. LAP limit of detection and comparison to ELISA ...... 98

Figure 32. Timing improvements of LAP detection using the ferricyanide/ferrocyanide- blocking assay ...... 99

Figure 33. Validation of LAP detection in lung perfusate (STEEN) solution .... 100

Figure 34. ET-1 detection using ferricyanide/ferrocyanide blocking assay ...... 113

Figure 35. Validation of ET-1 detection scheme ...... 114

Figure 36. ET-1 biosensor characteristics ...... 114

Figure 37. Comparison of ET-1 assay to ELISA ...... 115

Figure 38. Effect of sensor size on ET-1 LOD ...... 115

Figure 39. Endogenous ET-1 detection ...... 117

List of Abbreviations

ANOVA – analysis of variance

Akt – protein kinase B

AR – acute rejection

ATP11B – ATPase, class VI, type 11B

AUC – area under the curve

BAL – bronchoalveolar lavage

BOS – bronchiolitis obliterans syndrome

BSA – bovine serum albumin

CF – cystic fibrosis

CLAD – chronic lung allograft dysfunction

COPD – chronic obstructive pulmonary disease

CV – cyclic voltammetry

DBD – donation following brain death

DCD – donation following cardiac death

DNA – deoxyribonucleic acid

DPV – differential pulse voltammetry

ECE – endothelin converting enzyme

EDTA – ethylenediaminetetraacetic acid

EGLN1 – Egl nine homolog 1

ELISA – enzyme-linked immuno assay

ET-1 – endothelin-1

EVLP – normothermic ex vivo lung perfusion

3- Ferricyanide - [Fe(CN)6]

4- Ferrocyanide - [Fe(CN)6]

FEV1 – volume exhaled at end of first second of forced expiration

FiO2 – fraction of inspired oxygen

FPM – FraCS prediction model

FraCS – fractal circuit sensors

GAPDH – glyceraldehyde 3-phosphate dehydrogenase

HIV – human immunodeficiency virus

ICU – intensive care unit

IL-10 – interleukin 10

IL-6 – interleukin 6

IL-8 – interleukin 8

IPAH – idiopathic pulmonary arterial hypertension

IPF – idiopathic pulmonary fibrosis

JAK – Janus kinase

LAP – latency associated peptide

LLC – large latent complex

xii

LOD – limit of detection

LTx – lung transplantation

MAP – mitogen-activated protein

MCH – 6-mercapto-1-hexanol mRNA – messenger ribonucleic acid miRNA – micro ribonucleic acid

MSC – mesenchymal stem cell

NDA – neutralizer displacement assay

NME – nanostructured microelectrode

NO – nitric oxide

NPV – negative predictive value

OD – optical density

PAH – pulmonary arterial hypertension

PaO2 – partial pressure of oxygen in arterial blood

PBS – phosphate buffered saline

(q)PCR – (quantitative) polymerase chain reaction

PGD – primary graft dysfunction

PKC – protein kinase c

PNA – peptide nucleic acid

POC – point-of-care

xiii

PPV – positive predictive value

RAS – restrictive allograft syndrome

RNA – ribonucleic acid

ROC – receiver operating characteristic

SAM – self-assembled monolayer

SEM – scanning electron microscope

SPR – surface plasmon resonance

SSC – saline-sodium citrate

TGFβ1 – transforming growth factor beta 1

TH17 – T helper 17 cell

TMB – 3,3’,5,5’-tetramethylbenzidine

xiv 1

Chapter 1 Introduction

Sections of this chapter are represented in:

Sage, A.T., Besant, J.D., Lam, B., Sargent, E.H., and Kelley, S.O. Ultrasensitive Electrochemical Biomolecular Detection Using Nanostructured Microelectrodes. Accounts of Chemical Research. 47(8): 2417-25 (2014). and

Sage, A.T., Besant, J.D., Mahmoudian, L., Poudineh, M., Bai, X., Zamel, R., Hsin, M., Sargent, E.H., Cypel, M., Liu, M., Keshavjee, S., and Kelley, S.O. Fractal circuit sensors enable rapid quantification of biomarkers for donor lung assessment for transplantation. Science Advances, in press (2015).

1 Introduction

1.1 The Challenges Facing Lung Transplantation

Each year, diseases such as COPD and pulmonary fibrosis cause irreparable damage to the lungs of thousands of individuals in North America. In order to save these patients, the only treatment option is to physically remove the lungs and replace them with that of a deceased donor. Only in the last three decades have surgeons been able to perform such a complicated surgery and have now completed thousands worldwide. Yet, due to the complexity and invasiveness of the procedure, surgeons must be extremely cautious when selecting donor lungs. In doing so, transplant teams reject about 85% of donor lungs that could potentially be available for transplant. In fact, many of these organs that are turned down could be transplanted safely. Furthermore, of the transplanted lungs, nearly a third of the recipients will develop a severe form of acute rejection and half may suffer from chronic rejection. Thus, the patient in need of a lung transplant is burdened with numerous challenges including long waitlist times and persistent rejection risks post-transplant.

Confidence in donor lung selection and subsequent patient outcomes would be greatly enhanced if metrics existed that would predict both the long- and short-term health of the recipient. Recently, a series of lung biomarkers have been uncovered that do indeed correlate with how well a donor lung will perform in a patient. These discoveries represent a valuable tool for surgeons, but simultaneously create a new challenge. Current technologies to analyze these biomarkers are too slow – biomarker data must be available to a surgeon within a two-hour window and current techniques can take upwards of four hours.

Recently, the development of electrochemical biosensors has been shown to represent a novel class of rapid diagnostics. This technology is often based on miniaturized electrodes that convert the physiologic levels of disease-specific biomarkers into an electrical current coupled with a simple user-readout. Thus, using an electrochemical approach could allow for the development of novel sensing platforms that could make rapid biomarker measurements in donor lungs. The value of this technology is expanded even further as it could be integrated into portable, hand-held devices that can be used directly in an operating room.

The deployment of rapid, electrochemical sensors in transplantation centers around the world would allow surgeons to predict transplant outcomes and better utilize the available donor lungs. Importantly, these sensors could provide valuable insight into donor lungs by allowing surgeons to know what damage a lung may have with the time to fix it. Thus, one can conceive of a future where a damaged, rejected lung is diagnosed using electrochemical biosensors then repaired and transplanted as a known, healthy lung – precision medicine for lung transplants.

This chapter first provides a background on lung transplantation – successes, challenges, and the search for predictive biomarkers. We then describe the current state of rapid diagnostics and their subsequent use in medical practice. Next, we focus on the development of a novel class of electrochemical biosensors, nanostructured microelectrodes, designed by the Kelley laboratory and show how this platform has been used to meet numerous clinical challenges. We then conclude how a nanostructured microelectrode-based assay could be configured to meet the demands of a transplant-specific diagnostic tool set.

1.2 Origins and Rationale for Lung Transplantation

The history of lung transplantation (LTx) originates in the 1950s where the first reports of experimental lung transplants were documented in the Soviet Union (1). The first human lung transplantation was performed successfully by Dr. James Hardy in the United States of America in 1963; however, another twenty years passed before long- term survival of the procedure was achieved by the Toronto Lung Transplant Group (1). Following the first successful single-lung transplant in 1983, the Toronto Group then performed the first successful double-lung transplant three years later on a patient who survived fifteen years post-transplant. The outcomes for lung transplantation have seen continual improvement since that time due to innovations in organ preservation techniques, patient care, and therapeutic developments with the five-year survival rate reaching approximately 66% (2). As a result of these advances, lung transplantation has now become a well-established therapy for patients suffering end-stage lung disease, when all other medications and treatments have failed. In Canada and the United States, approximately 73% and 94% of the patients requiring a lung transplant have been diagnosed with idiopathic pulmonary fibrosis (IPF), chronic obstructive pulmonary disease (COPD), or cystic fibrosis (CF), respectively (Table 1) (2, 3).

Table 1. Lung Transplant Recipients

Canada (%) United States (%) Recipient Diagnoses 2003 - 2012 2002 - 2012

Obstructive lung disease (COPD/emphysema) 22.4 36.8

Pulmonary vascular disease (IPAH) 4.4 5.2

Cystic fibrosis/immunodeficiency 31.4 14.6

Restrictive lung disease (IPF) 24.2 42.3

Other 17.7 1.2

The etiology of each transplant-evoking disease may be different, but the ultimate physiological impairment of lung function is consistent. In obstructive lung disease (COPD and emphysema), the damage to the walls of the lung is often associated with cigarette smoking. COPD is the third-leading cause of death in United States (4) where an estimated 20% of the adult population smoked cigarettes in 2012 (5). For patients suffering from cystic fibrosis (inability to clear airways) and α-1-antitrypsin deficiency (tissue degradation), the origin of lung damage is genetic in nature. In contrast, the origins of lung disease (airway narrowing) are poorly understood in both idiopathic pulmonary arterial hypertension (IPAH) and fibrosis (IPF).

1.3 Lung Transplantation – A Donor Lung Shortage

In North America, the number of patients on the waiting list to receive a donor lung has risen dramatically over the past decade (~100%) (2, 3). In this time, the national donation rates have only modestly increased (~60%) and transplant teams have performed an average of 180 and 1700 transplants per annum in Canada and the United States respectively (2, 3). Together, the growing waitlist combined with an unmatched increase of available donor lungs underscores the high waitlist mortality (~20%) associated with lung transplantation (2). Perhaps the greatest challenge surrounding the donor/recipient gap in lung transplantation is organ utilization. Compared to liver and kidney transplantation where utilization rates approach 75% of donor organs, lung transplantation uses less than 25% of donated organs (3). In current practice, donor organs go through a rigorous evaluation procedure; however, the metrics that are traditionally used to assess donor lung suitability are primarily qualitative in nature. Selection criteria such as age, X-ray results, lung function

(PaO2/FiO2), smoking and cancer history, and overall transplant time can vary across transplant centers and do not accurately predict the suitable organs, chronic, or acute lung injury. Due to the complexity of the lung transplant procedure and a lack of predictive metrics reporting on both the short and long-term survival of a recipient, many

transplant teams employ very conservative practices of organ selection thus resulting in the extremely low utilization rate. Several approaches have been suggested to improve the availability of donor lungs including expansion of the donor pool to include cardiac death donors (6), ex-vivo lung perfusion (7), and greater inclusion criteria in the selection process (it is estimated that ~40% of rejected donor lungs may be suitable for transplant) (8).

1.4 Novel Strategies for Organ Preservation

A significant change to the transplantation process has recently been proposed, tested, and clinically validated (7). Normothermic ex vivo lung perfusion (EVLP) (Figure 1) represents a promising technique that may allow transplant teams to utilize a higher

Gas (Deoxygenation)

Ventilator

xVIVO Chamber

From Portal Vein

To Portal Artery Perfusate Reservoir Leukocyte Filter

Heat & Gas Exchanger

Figure 1. Normothermic ex vivo lung perfusion. Schematic of the EVLP system – perfusate solution is deoxygenated and heated to 37°C then passed through a leukocyte filter. This solution is then returned to the lung through the portal artery where physiological gas exchange can occur. Oxygenated perfusate solution is subsequently returned to the perfusate reservoir through the portal vein. Lungs are kept on a ventilator during this process.

percentage of the available donor lungs, thus addressing the disparity between donors and recipients. Traditionally, static hypothermia represents the method of choice for organ preservation during transplantation (15, 16). However, the hypothermic process represents a significant challenge in donor lung assessment. By slowing down the metabolic processes (intended to inhibit damage-inducing pathways), cold lung preservation makes both reassessment and repair of the donor lung challenging (7). In contrast, during EVLP, donor lungs are maintained at normothermic conditions in an acellular blood-free perfusate solution. In doing so, the process of EVLP mimics physiological conditions and, as a result, provides transplant teams an opportunity to further assess and treat the donor organ. Studies have shown that the EVLP technique can increase the window of time in which a donor lung can be maintained outside of the human body (as long as 12 hours) (7). By increasing the assessment time, transplant surgeons were able to closely examine the marginal donor lungs over a greater period of time and demonstrated that, with an extended observation time, these lungs perform equally well as lungs that were initially accepted (7). The incidence of primary graft dysfunction (PGD) and other measures of outcome were comparable for lungs transplanted using conventional methodology and EVLP thus validating the safety of this technique (7).

The extended opportunity for assessment using EVLP provides two essential donor expansion capabilities: (i) expansion of donor pool to included cardiac-death patients and (ii) treatment and inclusion of additional lungs in the existing donor pool. Currently, the most common approach to obtain donor lungs is through the use of donation following brain death (DBD) (9). Of note, donation after cardiac death (DCD) was the only source of organ donation during the early stages of lung transplantation; however, that source of donor organs was quickly supplanted by donation after brain death donors (9). Through the use of EVLP, normothermic ischemia, and improved preservation and surgical techniques, outcomes following the use of donors of cardiac

death approach those of brain death (7, 9-12); however, adoption into the clinical setting remains slow and DBD donor organs can outnumber DCD donors by approximately 80% (13). An important characteristic of the EVLP system is the ability to reassess and treat donor lungs prior to transplantation. Of note, the use of a blood-free, acellular perfusate allows for the delivery of targeted therapies including gene-delivery systems. Importantly, the use of the blood-free perfusate can overcome many of the challenges associated with the host-immune response. Recently, the EVLP system has been used to effectively repair human donor lungs through treatment with a protective cytokine in an adenovirus-mediated gene-transfer system (14). Together, these results demonstrate the power of the EVLP system to allow better utilization of donor lungs and subsequently improve patient outcomes.

1.5 Patient Outcomes Following Lung Transplantation

Following lung transplantation, primary graft dysfunction (PGD) is a major cause of early death for lung transplant patients. PGD occurs within the first 72 hours post-transplant and is characterized by nonspecific diffuse alveolar damage, lung edema, and severe hypoxia (15-18). The gradations of PGD span from 0 (PaO2/FiO2 greater than 300 mmHg) to III (PaO2/FiO2 less than 200 mmHg with pulmonary edema) (19). The development of PGD has a poor indication for transplant recipients as PGD accounts for approximately one-third of all of the 30-day mortality cases following lung transplantation (20). Typically, PGD manifests as progressive hypoxemia and radiographic pulmonary infiltrates with a lack of causality (18, 19, 21, 22). PGD shares many similar predisposing characteristics with Acute Respiratory Distress Syndrome (ARDS) including trauma and sepsis (23). As such, efforts to preventatively treat PGD through improved organ selection, preservation, and reperfusion techniques in addition to therapeutic interventions have been met with varying degrees of success (24-26). It is thought that PGD develops sequentially, with the donor lung providing the initial source of damage-inducing factors followed by a second wave arising from those

present in the recipient (27). Critical to the pathogenesis of PGD is the generation of reactive oxygen species (ROS) and the activation of the damage-amplifying inflammatory cascade (16). There have been a number of attempts to identify metrics that would accurately predict the prevalence of PGD during transplantation. Studies aimed at identifying clinical risk factors of PGD have been limited by trial size and design, data quality, and variability within the PGD definition (28). Recent efforts in the transplantation field have sought to uncover the molecular basis behind PGD and identify biomarkers that were prognostic of PGD in a donor lung. Although there are now many promising PGD biomarkers, the successful integration of rapid molecular testing programs that clinically predict the development, diagnosis, or prognosis of PGD remains elusive.

Long-term success of LTx is hindered by the development of chronic lung allograft dysfunction (CLAD). CLAD is characterized by persistent graft dysfunction and, within 5 years of transplant, it is estimated that about half of LTx patients will experience some form of CLAD (29). Recipients who display symptoms of CLAD post-transplant are burdened with a 5-year survival rate of approximately 50% (29, 30). CLAD presents as several different sub-phenotypes including: bronchiolitis obliterans syndrome (BOS) and restrictive allograft syndrome (RAS). BOS is characterized by obstructive lung physiology and graded from 0 to III based on the FEV1 decline in a patient (31). In contrast, the RAS phenotype presents with restrictive pulmonary physiology in addition to a decline in FEV1 and has a prognosis poorer than that of BOS (32, 33). Over the past few decades, there have been extensive studies aimed at identifying possible CLAD biomarkers in an effort to both predict and understand the mechanisms of chronic lung rejection. To this end, transplant teams have looked to bronchoalveolar lavage (BAL) fluid as a possible source for biomarker discovery. The BAL procedure introduces then collects a small amount of fluid from a specified lung region using a bronchoscope. A number of studies have identified several broad biomarker categories

that were highly associated with BOS including, but not limited to: cellular composition (neutrophils) (34) and cytokines/chemokines (IL-6, IL-8, TGFβ1, ET-1) (35-37). Together, the discovery and monitoring of prognostic biomarkers that specifically identify CLAD and subsequent phenotypes will help transplant physicians better treat and manage their patients.

1.6 Prognostic Tools in Lung Transplantation - Biomarkers

Shortly after the first successful lung transplant was performed, a search began to identify common characteristics of transplants with good and poor outcomes. The identification of potential lung transplant biomarkers has developed extensively over the past few decades and has grown to include physiological, metabolomic, genomic, proteomic, and whole cell prognostic markers. These markers have been demonstrated in a variety of sources, typically found in patient plasma/serum, bronchoalveolar lavage (BAL) fluid, and tissue biopsies.

In 1994, it was reported that a differential expression of B and T immune cells in peripheral blood lymphocytes was associated with the development of BOS in transplant patients (38). Furthermore, peripheral blood provided another source of prognostic markers as natural killer (NK) cells were found to be associated with chronic rejection in the transplant patient (39, 40). Cell-specific labeling of MIB-1 and HLA class I antigens in transbronchial biopsies (TBB) identified proliferating cells as additional cellular markers for BOS prediction (41, 42). In addition, the proliferative phenotype of cells involved in the development of BOS was confirmed in BAL fluid samples and expanded to include the ratio of FoxP3 to CD4+ (43) and activated mesenchymal cells (44). It has also been proposed that acute lung rejection could also be diagnosed using T cell expression profiles found in BAL fluid samples (45). Together, these findings highlight the predictive value of whole cell expression patterns in lung transplantation;

however, the methodologies used (cell staining and flow cytometry) in these studies are slow and labor intensive thus precluding their practical integration as prognostic tools.

Physiological function measurements of donor lungs are the current standard for lung assessment and have been used as a non-invasive source of biomarkers for acute and chronic lung rejection. The incidence of primary graft dysfunction has been shown to correlate with physiological metrics in a large, multicenter study (28). These variables were easily measured/assessed and included: donor smoking habits, FiO2, weight, and arterial pressures (28). Simple measurements of nitric oxide (NO) and carbon monoxide (CO) levels, and the helium slope in exhaled breath samples, have also been investigated and shown to be potential markers of early BOS onset (46). Although the development of physiological risk factors in lung transplantation is promising, challenges surrounding biomarker validation, standardization, and integration into clinical practice persist. The validation of these and other potential physiological biomarkers represent exciting possibilities for outcome predictions as they can be easily and constantly measured without undue stress on a patient.

Similar to physiological biomarkers, metabolomic biomarkers are becoming increasing prevalent in transplant monitoring. One such application for metabolomic markers is in the diagnosis of invasive aspergillosis in transplant recipients. Aspergillus infection occurs frequently transplant patients (estimated 30% of cases (47)) and is associated with poor prognosis if left untreated. Thus, it is crucial that physicians are able to diagnose an aspergillus infection rapidly and accurately. To this end, the metabolomic marker galactomannan has been studied and shown to be an accurate diagnostic biomarker found in serum and BAL fluid (48, 49). Although promising, galactomannan levels are determined by an ELISA sandwich assay-technique, which hinders the rapid detection of aspergillus infection in the clinic.

One of the most common sources of prognostic biomarkers in lung transplantation has been at the proteomic level. A number of studies have identified various proteins and the predictive ability of these markers in most facets of transplantation. Small peptides have been found in BAL fluid that are predictive of BOS (50) and others, such as procalcitonin, were used to differentiate bacterial infections from allograft rejection (51, 52). As with whole cell analysis, serum and BAL fluid have been important sources of proteomic biomarkers that are predictive of chronic rejection and BOS in transplant patients. These matrices are less invasive than a transbronchial biopsy and are relatively easy to obtain. Proteomic markers of BOS in serum and BAL fluid have included: neutrophil peptides (HNP) (53), IgGs (54), soluble CD14 (55), chemokines (IL- 8, RANTES, MIP3α/β) (56, 57), glycoproteins (YKL-40) (58), surfactant proteins (SP-A) (59, 60), and inflammatory markers (IL-15, IL-17, TNFα, IL-6) (61, 62). Interestingly, inflammatory pathways and cytokines have been shown to play an important role in the prognosis of a donor lung (Figure 2).

IL-6 IL-10

Pro-Inflammatory Anti-Inflammatory Pathway Activation Pathway Activation IL-6R IL-10R

NF-κB JAK1/STAT3 Positive Feedback

Negative Feedback Implicated in: Acute Rejection Chronic Rejection Inflammatory cytokines, chemokines Bacterial Infection

SOCS3

Figure 2. Inflammatory pathways involved in LTx outcomes. Signaling from IL-6 (red) activates the IL-6 receptor (IL-6R), which then activates NF-κB thereby promoting the increased expression of additional inflammatory cytokines and genes (pink pathway). In contrast, IL-10 (green) activates the JAK/STAT pathway (green pathway) causing the activation of SOCS3 – an inhibitor of NF- κB, which negatively regulates the inflammatory effects of IL-6 (orange pathway).

In addition to procalcitonin, several other studies have identified novel protein markers for the diagnosis of community-acquired respiratory virus (CARV) (63) and cytomegalovirus (CMV) (64) infections in BAL fluid and serum. Although a number of proteomic biomarkers have been shown to be prognostic for long-term transplant survival, it has been of great importance to identify makers of acute lung rejection as

this can play a dramatic role in immediate patient outcomes. As such, a number of clinical studies have sought to validate proteins that were predictive of primary graft dysfunction and short-term transplant outcomes. A porcine-based study in 2005 demonstrated that increased IL-1β protein levels were associated with graft function (65). Additional trials have demonstrate a role for: soluble CD30 (66), ICAM-1 (67, 68), protein C (67, 68), PAI-1 (67, 68), sRAGE (68), SP-D (68), and IL-8 (69) protein levels in determining the incidence of PGD in patients using plasma and serum samples. Taken together, these studies demonstrate the plethora of available protein biomarkers that could be used to guide transplant teams to better predict short- and long-term donor organ outcomes as well as discriminate viral/bacterial infections from rejection episodes. A common theme throughout all of these studies is the method in which the biomarkers were identified. Almost exclusively, protein levels were measured using a variation of an ELISA-based protocol. As such, these studies are mostly retrospective in that the protein measurements were conducted sometime after the transplant procedure. This finding is due largely in part to the lengthy procedure of the ELISA test and the associated challenges in dealing with complex sample matrices that involve initial handling/processing steps (ex. serum/plasma). The advent of rapid protein-based diagnostics with high performance standards in complex samples will allow for integration of many of these protein biomarkers into transplant practice.

The last category of LTx biomarkers are one of the most challenging to measure as nearly all of the available analytical techniques require purification of the prospective samples prior to assessment. Nucleic acid biomarkers are contained within cells alongside many other enzymes and proteins that can degrade the analytes and interfere with the downstream measurement assays. As such, sample purification is required in most nucleic acid tests, which is in contrast to protein measurements that often do not require a purification step. Nevertheless, a number of studies have sought to identify mRNA markers that are prognostic for lung transplant outcomes using microarrays and

quantitative PCR. Animal studies using rat and swine models have determined a predictive role for FoxP3 (70), Tim-3 (71), and IL-2 (72) mRNA in acute lung dysfunction with the role of FoxP3 being confirmed in human studies (73, 74). Changes in overall gene expression patterns (75), an upregulation of IFNγ and ICAM-1 (76, 77), or increase in IL-15 and IL-17 mRNA (78, 79) were shown to be associated with acute cellular rejection and subsequent development of BOS in cells obtained from BAL samples. Chronic rejection has also been correlated with inflammatory cytokine mRNA expression levels found in BAL fluid (including ET-1, IL-8, IL-17, and IL-23) (80, 81). By sampling tissue, prognostic nucleic acids found within the cells of the lung provide valuable information about the genetic-makeup of the donor organ. A porcine study conducted using lung tissue identified matrix metalloproteinase (MMP) mRNA transcripts that were predictive of lung inflammation and fibroblast proliferation (82). Additional human studies in alveolar and bronchiolar tissue further confirmed the presence of MMP biomarkers as well as additional mRNA markers including TGFβ, RANTES, IL-8, and CD80/86 in the development of BOS (35, 83-85). Furthermore, donor lung biopsies have been used to identify key, predictive mRNA molecules that preoperatively predict PGD (15, 86) or CLAD (60, 62) in donor lungs.

Notably, predisposition to inflammatory pathway activation observed as a high ratio of IL-6 to IL-10 mRNA has been reported to impart a twenty-times higher relative risk of recipient death due to PGD (86). In a follow-up study of 107 donor lungs, PGD- predictive mRNA markers included ATP11B, FGFR2, EGLN1, and MCPH1 (15). Although encouraging, the purification steps and lengthy assay times of PCR and microarray protocols used in the gene-based biomarker studies represent the major limiting factors for the integration of nucleic acid testing in the transplant setting.

A summary of nucleic acid and protein biomarkers for LTx is found in Table 2. These studies demonstrate the numerous biomarkers that have been found to be associated with transplantation outcome; however, the clinical validation of many of these markers is lacking and, currently, there are no clinically approved prognostic LTx biomarkers.

Table 2. Summary of nucleic acid and protein biomarkers implicated in LTx.

Biomarker Analyte Biomarker(s) (Reference) Source

Nucleic IL-6:IL-10 (86), ATP11B (15), EGLN1 (15), FGFR2 (15), MCPH1 Acid (15), MMP-2/9* (82), IL-8 (35), CD80/86 (85), SP-A (60), IL-6 (62) Lung Tissue

Protein HO-1* (87) , NAC* (88), IL-10 (89), SDF-1 (90)

Nucleic IFNγ (76, 77), ICAM-1 (76), IL-15 (78), IL-17 (79), FoxP3 (73, 74), Acid ET-1 (80), IL-23:IL-17 (81) Bronchoalveolar Lavage (BAL) IgG (54), IL-1β* (65), Small peptides (50), sCD14 (55), IL-8 (56), Protein RANTES (56), SP-A (59), MIP3α/β (57), HNP (53)

Nucleic FoxP3* (70) Acid Blood/Plasma sCD30 (66), ICAM-1 (67, 68), Protein C (67, 68), PAI-1 (67, 68), Protein sRAGE (68), SP-D (68), YKL-40 (58), sCD14 (55)

* Indicates an animal model.

1.7 The Challenges Surrounding Current Lung Transplantation Diagnostics

The use of point-of-care diagnostics has expanded greatly in the infectious disease field and is growing rapidly in oncology; however, use as personalized diagnostic tools in lung transplantation remains very limited. One of the primary reasons for this is the narrow time window for assessment in which transplant teams have to work within – a timeframe that is outside of capabilities of current molecular diagnostics (Summarized in Table 3 (Adapted from (91)).

Table 3. Point-of-care diagnostic capabilities

Diagnostic Platform Signal-to-Answer Time

Microscopy Morphology: minutes, Staining: hours

Cell Culture Days to weeks

Antibody/Antigen (ELISA) Hours

Real-time PCR Hours*

Mass Spectrometry Seconds to minutes*

Flow Cytometry/FACS Hours*

Gene Sequencing Hours*

Microarray Hours*

*Does not account for sample purification and preparation processes prior to the execution of the assay.

Microarray and polymerase chain reaction (PCR) technologies are powerful tools for the discovery (92) and validation (15, 73, 74, 78, 79, 86) of nucleic acid biomarkers for lung transplant outcomes. The identification of novel biomarkers typically begins by using

genome-wide or pathway-targeted panels in a microarray (93) followed by PCR validation of prospective gene candidates. Once verified by PCR, the genes may go through several more rounds of validation using a new set of candidate lungs and if the association holds true, the identified gene is then classified as a predictive biomarker. The integration of the identified gene targets into the transplant setting is often delayed at this step as both microarrays and PCR are labour intensive, cost prohibitive, and slow to provide transplant teams with the appropriate analysis/information. Additionally, microarrays and PCR tests only provide information at the genomic level and do not reflect the proteomic status of the donor lung. The gold standard in protein detection is the enzyme-linked immuno assay (ELISA). Similar to microarrays and PCR, ELISA testing requires significant operator handling/manipulation (through various wash steps) and upwards of several hours to complete. Together, all of these technologies have been instrumental in the identification and validation of potential lung transplant biomarkers. Moving forward, the development of testing methods that are rapid, easily integrated, and hands-free represent a significant advance in the field of the transplantation. The incorporation of genomic and proteomic testing in the transplant setting will allow for better utilization and quality of donor lungs, fewer time spent on the waiting list and in the ICU, and improved detection and treatment of potential graft dysfunction. The development of these types of test will greatly reduce health care costs associated with transplants in North America and, most importantly, significantly improve the quality of life for transplant patients.

1.8 The Current State of Diagnostics

Point-of-care (POC) diagnostics have become an important component of modern-day medicine. With an increased understanding of the biological pathways and processes of disease, predictive biomarkers have become increasingly prevalent. The use of POC diagnostics has allowed for the practical testing of these biomarkers in a clinical setting, such as at home or the doctor’s office, and will continue to transform medicine towards

a personalized medicine approach – tailoring the right treatment for the right patient at the right time.

The basis for POC diagnostics has been developed, with varying degrees of success for the detection of nucleic acids, proteins, metabolites, mammalian cells, bacteria, and viruses. As seen in Table 3, there are a number of different strategies for a diagnostic platform; however, a majority of POC platforms utilize an optical detection method to quantify the presence of the desired analyte. By coupling molecules to fluorescent dyes or enzymatically activating an optical reporter, optical-based devices require expensive and sophistical equipment capable of measuring and quantifying these signals (~400- 700 nm wavelengths). This strategy is used in enzyme-linked immunosorbent assay

(ELISA) (3,3’,5,5’-Tetramethylbenzidine (TMB)/H2SO4: 450 nm), quantitative polymerase chain reaction (qPCR) (SYBR Green: 520 nm), and reporter conjugate (microarrays, flow cytometry: dye wavelength dependent) based POC platforms. In contrast, novel strategies that include optical changes that are binary (signal on/off, ex. dipstick technologies) or possess non-optical readouts (ex. electrochemical methods) will produce highly desirable POC devices.

1.9 Assays for Protein Detection

Rapid and sensitive protein biosensors have developed extensively over the last few decades. There are now numerous platforms that can achieve the desired characteristics of speed, sensitivity and cost for a variety of applications; however, there are several limitations of each device that would preclude use as a quantitative lung transplantation diagnostic.

Silver nanoparticles have received a great deal of attention as protein diagnostic platforms because of the rapid sample-to-answer times they have achieved. These

assays are based on the aggregation of the silver nanoparticles and are monitored by a spectrophotometer. Using this technique, a neuronal peptide, neurogenin 1, was detected within 10 minutes (94). The drawback of this approach is that was only validated in a purified, buffered solution which does not demonstrate adequate sensitivity in complex matrices such as blood or tissue. In addition, the assay required use of UV-Vis spectrophotometer, which limits the ability of this detection platform to be easily integrated into a point-of-care device. Optical protein sensors have been developed that rapidly (< 20 minutes) monitor protein levels in a variety of samples such as serum (95) and livestock samples (96); however, like silver nanoparticle detectors, these assays require the use of wavelength specific detectors and lasers which significantly increases the cost and reduces the portability of these platforms. Similarly, a fluorescence-based protein biobarcode approach has been successfully demonstrated in blood using an inexpensive and integrated microfluidic system (97). In contrast, lateral flow protein assays have become a successful approach to rapid, portable, and sensitive protein detection. These devices have the unique characteristic of providing a visual readout that is observable by the naked eye. A wide range of analytes have been successfully testing using lateral flow including: IgGs (98), C- reactive protein (99), and HIV (100). The visual readout obtained by these sensors was rapid (10 – 20 minutes) (98-100) and shown to be compatible with extremely complex, whole blood (100). Initial assessment would suggest that this approach is ideal for the use as a transplant diagnostic, but transplant diagnostics have an additional requirement of being quantitative. The detection of HIV using a lateral flow device is binary (i.e. infection is present or not), thus the relative strength of the signal (readout) is not important. However, for a transplant protein device, the expression of analytes is variable, and it is the variability amongst patients that is predictive. Therefore, the use of lateral flow assays, which are not highly quantitative, is impractical. Interestingly, the lateral flow approach could be implemented for the detection of transplant-specific infections such as aspergillus, CMV, and CARV.

Although optical methods of protein detection predominate and have displayed record- breaking sensitivities (101), electrochemical platforms have been used to achieve similar results (102). For example, using a label-free approach with semiconducting nanowires, Stern et al. demonstrated real-time, 100 fM detection of cellular antibodies (103). A biobarcode approach (101), where a specific DNA sequence acts as a reporter for protein detection, has been used with a high degree of success to detect HIV (104) and prostate cancer antigens (105). Many protein biosensors are based on the use of gold/silver nanoparticles and electrochemical measurements. Biochemiresistors are sensors that pair functionalized gold nanoparticles with electrochemical resistance measurements to successfully monitor antibody-antigen displacement reactions of small molecules (106). Other studies have demonstrated the use of functionalized electrodes for the detection of lectins (107), viral particles (108), and fungal infections (109) using electrochemical reporters. Even though these tests demonstrated good performance in terms of sensitivity and feasibility (use with human serum), they failed to meet the time sensitive nature of rapid diagnostics with assay times significantly greater than one hour (107-109). In contrast, silicon nanowires have been utilized for rapid (< 5 minutes) detection of predictive myocardial infarction proteins, but this system was only validated using buffered solutions (110). Gold nanoparticles have been used in place of the traditional ELISA reagents for electrochemical sandwich assays. These devices have been validated with such analytes as thrombin (111) and lung cancer biomarkers (112), but the ELISA-like sequential addition of assay regents and wash steps produced assay times that approached two hours.

A significant challenge that exists in protein detection, irrespective of the detection platform is sensitivity (113). Two critical parameters affecting protein detection sensitivity is the availability of high-affinity capture molecules (antibodies, aptamers) and the potentially large quantities of sample required to achieve detectable limits of antigens. As such, advances in protein diagnostic technologies are aimed at improving sensitivity through various nanomaterials (114) and sampling volumes using microfluidic approaches (115, 116).

1.10 Assays for Nucleic Acids Detection

Significant progress has been made towards meeting the challenges of integrating rapid nucleic acid detection systems into the clinical setting. Traditional PCR and microarray techniques require amplification and several labeling steps, which contribute to a multi- hour signal-to-answer timeframe. Thus, the development of novel nucleic acid sensors that are rapid and do not require an amplification step are highly desirable for many applications. As with protein biosensors, the use of gold nanoparticles has seen widespread use as nucleic acid sensors. The sensing readout of gold nanoparticles can be visual, electrochemical, scanometric, chemiluminescent, surface plasmon resonance, or quartz crystal microbalance-based (117). The vast majority of these techniques require hybridization times of at least 60 minutes thus limiting their use as rapid biosensors (117). Quartz crystal microbalance readouts combined with nanoamplicons have been shown to rapidly (< 10 minutes) detect ultralow concentrations of DNA (0.17 aM) in buffered solutions without target amplification (118). To demonstrate the clinical utility of this platform, further studies using clinical samples that achieve the same speed and sensitivity are required. A gold nanoparticle platform has also been shown to have clinical utility in the detection MDR1 polymorphisms in human peripheral blood mononuclear cells (119), viral H5N1 RNA (120), and genomic KRAS detection (121); yet in order to achieve sufficient detection levels, these assays were performed over the course of several hours (119-121). Optical approaches to rapid, nucleic acid detection have shown clinical promise. These platforms have been developed to monitor circulating DNA in lung cancer patient serum (122), yeast/pathogenic detection (123), and E. coli 16S rRNA (124). In order to achieve a high degree of specificity within the platform, these optical platforms employ microfluidics (122), magnetic microbeads (123), or isolation of RNA (124) to separate clinical samples. Alongside the need for expensive and impractical optical equipment, these tests can provide a sample-to-answer no faster than one hour. Transmission electron microscopy (TEM) has been used to for the identification and characterization

of modifications within human genomic DNA (125). This approach was successfully employed for the detection of tandem repeats found throughout the genome and the analysis of genetic polymorphisms (125). The drawback of this type of platform is the need for an expensive TEM and assay times of about two hours. Similar to TEM platforms, surface-enhanced Raman spectroscopy platforms have been developed to detect the breast cancer biomarker, BRCA1, in cultured cells without the use of PCR- amplification (126), but require extended operational times (> 12 hours) (126).

1.11 The Emergence of Electrochemical Biosensors

Nucleic acid electrochemical biosensors have been shown to detect a wide range of analytes with varying sensitivities and speeds (127). In comparison to other detection methods, electrochemical platforms provide fast sample-to-answer times (Figure 3).

qPCR Microarray Sequencing Electrochemical

> 4 hours > 5 hours 10 hours < 30 minutes

Figure 3. Detection platforms for nucleic acid detection. Various assays that are capable of performing assays for nucleic acid monitoring including: qPCR, microarrays, gene sequencing, and electrochemistry. The approximate signal-to-answer time is shown below each test.

Electrical conductance measurements have been used in capillary based assays for the overnight detection of DNA oligomers (128). The clinical utility of miRNA detection has spurred the development of several companion diagnostic platforms (129). For example, a label-free and PCR-free sensor was developed for the profiling of miRNAs by coupling quantum dots (QD) with electrochemical sensors (130). Electrochemical platforms have also been coupled with enzymatic reactions such as HRP (131) and glucose oxidase (Gox) (132), to achieve sensitive and specific miRNA detection in total RNA purified from tissues within several hours (131, 132). Interestingly, a label-free approach that uses electrical traps with Raman hot spots was demonstrated to be effective with small sample sizes (1 µL) and femtomolar detection limits; however, further studies involving clinical samples is required (133). Important in the development of transplant-specific sensors are those nucleic acid platforms capable of making rapid mRNA measurements in complex samples. Chen et al. developed a novel electrical sensor array based on nanogaps and conductance that was able to achieve 0.1 fM levels of detection of the purified mRNA analogue, cDNA, within an hour (134). Furthermore, an electrochemical, branched-DNA biosensor was shown to detect BCR- ABL mRNA in purified nucleic acids from cell cultures overnight (135). These tests represent promising sensing platforms; however, improvements to hybridization time and validation using unpurified sample lysates are essential.

The commercial and clinical success of electrochemical glucose sensors has prompted the pursuit of similar analogous systems for the detection of additional biomolecular targets (136). Whereas the glucose biosensor was based upon the reaction with glucose oxidase, a naturally occurring redox enzyme, novel strategies for the detection of other metabolomic, genomic, and proteomic targets requires a distinct redox coupling strategy. To this end, the glucose-sensing platform has been shown to be amenable to modifications using enzymatic labels to detect low levels of other metabolites such as

cocaine (137). Novel electrochemical platforms and approaches are currently being studied to detect nucleic acids and these strategies have included: biobarcodes (138), non-covalent (139) and structural (140) reporters, DNA nanostructures (141), biodiscs (142), and sandwich assays (143, 144). Sample preparation and amplification steps are key to many electrochemical platforms and this can negatively impact the usability of these platforms. To date, there exists only one electrochemical analysis system for clinical use and it requires both sample purification and amplification prior to use (145). Electrochemical platforms that do not require prior processing or purification steps and inherently amplify signals are primed for success in the diagnostic field.

1.12 Meeting the Challenges of RNA Extraction

Several challenges arise when using unpurified samples in nucleic acid analysis. First, the presence of off-target sequences, proteins, enzymes, and metabolites can significantly reduce the sensitivity of the assay. Secondly, the presence of these additional molecules can inhibit and interfere with downstream analytical steps. Finally, for mRNA detection, the integrity of the mRNA is often lost in unpurified samples over time as numerous enzymes and autolytic processes rapidly degrade RNA. Thus, a great deal of research has been explored to identify techniques and processes that can rapidly extract RNA from cells and tissue while maintaining the integrity of the analyte. The goal of such processes is to engineer a system that rapidly extracts nucleic acid from complex samples including blood and tissue with little to no user manipulation. This feat has been achieved with varying degrees of success. Using programmable robotic handlers, Griesemer et al. demonstrated rapid RNA extraction from respiratory swabs of clinical influenza samples (24 samples in 33 minutes) (146). However, a bulky, automated machine was required for this process in addition to off-board lysis and manual manipulation steps (146). Similarly, automation of the standard Qiagen nucleic acid extraction protocol has been validated and has demonstrated rapid RNA purification (147). Novel purification methods of RNA, such as the use of magnetic

beads, have been developed into an automated approach for mRNA analysis of cytokines and chemokines; yet slow processing (1.5 hours for 32 samples), reverse- transcription (~1 hour), and bulky equipment were used (148). Although these devices have been shown to be efficient and speedy, the challenge persists to develop purification methods that are readily integratable and available at the point-of-care. Significant strides have been made towards miniaturization of purification techniques. Nucleic acids from spiked bacterial/viral cultures were rapidly purified within 10 minutes using a microfluidic, lab-on-a-chip, device (149). The purified DNA and RNA were shown to be free of contaminants and thus compatible with downstream applications (149). Further validation with clinical samples is needed to fully demonstrate the potential success of this method.

1.13 Putting it all Together – Integrated Diagnostic Devices

The integration of rapid extraction units with detection methods is the gold standard for point-of-care nucleic acid platforms. To date, these devices have focused on simple and quick extraction methods followed by fast-PCR. In order to meet these characteristics, analytes for validation have been strategically chosen. Predominately, integrated devices have been focused on the detection of foreign pathogens in human samples. Influenza (150) and HIV diagnoses (151) have been achieved within several hours using patient samples with little-to-no prior sample handling on devices that are now commercially available (151). Furthermore, detection of foot-and-mouth disease was demonstrated in cultured livestock samples using this approach (152).

The development of integrated nucleic acid testing platforms has achieved success in both oncology and respiratory fields. Mycobacterium tuberculosis (TB) infects approximately 9 million people annually and is responsible for nearly 2 million deaths worldwide (153). Using the rpoB gene as a reporter for mycobacterium tuberculosis, an

integrated processing device has been shown to detect the presence of TB with good sensitivity and specificity (69% and 98.4% respectively (154)) in clinical samples that included pleural fluid and tissue biopsies (154, 155). Furthermore, the identification of human gene mutations has seen realization in clinical testing units. Platforms have been shown to detect oncology-specific mutations, such as BRAF V600, from paraffin- embedded tissue samples in approximately 90 minutes (156). It is believed that platforms such as these and others will become a mainstay in the clinic in the near future. By selecting analytes with binary outcomes (presence of TB infection or BRAF mutation), these devices are able to actualize clinical utility at the expense of quantitation. To a clinician, the diagnosis of TB or the identification of treatment-guiding BRAF mutations take precedent over the exact expression levels of the target gene. As such, these devices are able to make rapid diagnoses in clinical samples, yet would not be able to distinguish between two positive samples with a high degree of accuracy. To meet this challenge, devices will need to be reconfigured, not only to detect analytes with a high degree of sensitivity and specificity, but also to accurately and precisely identify expression levels within the sample. In doing so a plethora of additional companion diagnostic tests, including those for transplantation, can be realized.

1.14 The Future of Electrochemical Sensors – Nanostructured Microelectrodes

Electrochemical detection platforms represent a means by which the diagnostic needs of transplant teams can be met. The ability to receive genomic and proteomic information in a matter of minutes, not hours, could be made possible using redox- active reporter molecules and metal sensing electrodes. Devices based on electrochemical platforms have the distinct advantage of being portable, rapid, sensitive, and sample processing-free.

The general principle for all electrochemical detection methods is the conversion of a biological event into a change in the flow of electrons, current, at or near an electrode. Changes in currents can then be measured using relatively simple equipment providing a rapid sample-to-answer turnaround. One such platform, designed and developed in the Kelley laboratories, is based on the use of nanostructured microelectrodes (NMEs) (Figure 4) (157, 158).

Figure 4. The nanostructured microelectrode (NME) assay. Images depicting the glass-substrate NME microchip (left panel) with an SEM image of an NME sensor following electrochemical deposition (central panel) and a schematic of sensor functionalization (orange) and target (blue) hybridization (right panel).

Classical metal electrodes used in biosensing are gold, disk electrodes. They are characterized by large (1.6 mm diameter), planar gold surfaces. Similar to bulk disk electrodes, NMEs are metal sensors with large surface areas; however, NMEs are non- planar. Metal electrodeposition (159) of pure noble metals or metal alloys (ex. gold, silver, palladium) can be performed under a variety of conditions to generate NMEs of different sizes and morphologies (158). The electrodeposition process used for NME growth produces electrodes that extend three-dimensionally into solution. In addition, the metal surface of NMEs is not smooth; rather it can be highly textured at nanometer length scales (1 – 100 nm). The process of nanostructuring microelectrodes has produced sensors that have achieved attomolar sensitivities (158). NMEs can be grown on a variety of substrates using a simple microfabrication process that exposes gold

apertures within a glass-based microchip (119, 157, 160-162). As such, numerous configurations or alignments of NME-based sensors/microchips can be tailored to specific applications.

It is well established that several metals (including palladium (163) and gold (164)) have strong thiol-binding affinities. This unique bonding is crucial for electrochemical biosensors as it allows for the functionalization of metal electrodes using thiolated biological reporter probes. The complex structures and surfaces of NMEs provide an additional benefit over traditional planar electrodes in terms of probe functionalization. The surface roughness of NMEs allows for a more diverse range of probe deflection angles compared to planar electrodes (160). This property of NMEs reduces the steric- interference of probe/probe and probe/target interactions allowing for more efficient analyte capture.

Having an effective set of electrodes is only one component of a good electrochemical- sensing platform. A redox-active reporter system that is fast (rapid diffusion rates), sensitive, and specific is also crucial to the success of an assay. One such electrochemical reporter that has been shown to specifically report the presence of nucleic acids are complexes of the transition metal ruthenium. Ruthenium complexes have gained a great deal of attention in recent years as they have been shown to bind electrostatically to the anionic DNA backbone (Figure 4) and can be used to quantify immobilized single-stranded and double-stranded DNA with a high degree of accuracy (165-167). Thus, using a ruthenium-based approach, the NME platform monitors changes in bioreactions that occur on the electrode surface as a result of probe and target hybridization (166, 168).

Under standard hybridization conditions, the ruthenium reporter system has limited sensitivity. Only a single electron is transferred per reporter molecule in the electrochemical reaction (Figure 5 #1), which, at low analyte concentrations, can result in undetectable sensitivities (169). Yet, nucleic acid biosensors have a strict requirement of being extremely sensitive. In order to maximize the current generated by a single ruthenium molecule, a number of amplification strategies have been investigated including redox cycling. The addition of an iron-based secondary redox probe was shown to promote the electronic amplification of the original ruthenium signal (Figure 5). Importantly, the secondary redox complex was selected based on the overall negative charge of the complex, which prevents the complex from reaching the electrode surface and contributing to a non-specific signal (169, 170). Thus, a 3- 2+ secondary electron acceptor [Fe(CN)6] could rapidly oxidize [Ru(NH3)6] (Figure 5 #2) 3+ and regenerate the primary electron acceptor [Ru(NH3)6] (Figure 5 #3) allowing multiple electron transfers, amplified current, from a single ruthenium molecule. To this end, the inclusion of a secondary electron acceptor was proposed and validated using

3+ 3- Figure 5. [Ru(NH3)6] /[Fe(CN)6] electrochemical reporter system. Schematic 3+ 3- depicting the [Ru(NH3)6] /[Fe(CN)6] electrochemical assay readout (PNA probe (black) and resulting DPV signal (right panel, shown in dark blue), target mRNA (light blue) hybridization and resulting DPV signal (right panel, shown in red)).

the NME platform (169). Several studies have demonstrated that a ruthenium and iron- based electrochemical redox system provided a significant enhancement over ruthenium-derived currents alone and have achieved 10-50 fold signal amplifications (158, 169). Compared to the various other techniques for analyte analysis (optical, PCR, etc.), the ruthenium/iron electrochemical detection method is label-free and does not require any sample processing steps. As such, this unique property has significant implications on the ability of this platform to be easily integrated into a rapid, POC device.

The final component of an electrochemical biosensing platform is the means by which an electrochemical current is measured. The flow of electrons between a metal sensor and a redox reporter complex at the electrode surface can be measured using a variety of voltammetry-based techniques. Cyclic voltammetry (CV) is widely used in electrochemical measurements as a qualitative assessment of the electrode surface and redox reporter system. The potential waveform for a CV scan is comprised of repeating units of linearly increasing and decreasing potentials (Figure 6a). The output of the CV scanning technique (Figure 6b) qualitatively reflects the electrode surface and can contain significant interference from background currents. In contrast, differential pulse voltammetry (DPV) scanning produces a peaked current-potential graph, which can easily provide quantitative information regarding an electrochemical process. The DPV method is similar to other voltammetric methods; however, DPV scans are comprised of a series of step-wise potential increments overlaid on a linear sweep (Figure 6c). This step-wise approach translates into two sampling times using the DPV technique – one before and after a step-pulse (Figure 6c “Sample Period”). By reporting the difference between the currents measured at the two sampling times in a DPV sweep, background signals (capacitive or non-Faradaic currents) are normalized and thus the DPV-output signal is due solely on the redox reporter (Faradaic current). The output of a DPV scan is a defined peak (Figure 6d) that can be quantified in terms

of peak amplitude or area. This level of analysis is easily programmable using simple computer software programs and can be measured using equipment that is readily available, cost-effective, and easily miniaturized (127). As such, the DPV technique represents a sensitive and quantitative scanning method for electrochemical biosensors.

a b Current (I) Potential

time Potential (V)

Sample Period c d Step E

Pulse Amplitude Current (I)

Potential Sample Period

Pulse Width

Pulse Period time Potential (V)

Figure 6. Voltammetric Techniques. The potential waveform (a) and voltammogram (b) of the cyclical voltammetry (CV) scanning technique. The potential waveform (c) and voltammogram (d) of the differential pulse voltammetry (DPV) scanning technique (adapted from www.basinc.com).

1.15 Using the NME Platform for Superior Clinical Performance

The combined integration of NME-based sensors, the ruthenium/iron redox reporter system, and a DPV scanning technique has helped to create the foundation of a successful electrochemical biosensing platform.

The NME platform has been shown to be capable of detecting a diverse range of analytes at clinically relevant concentrations. The primary targets of interest of most biosensors are nucleic acid reporters as they are specific markers of disease. Using the NME platform, the detection of nucleic acids was demonstrated at unprecedented sensitivities (10 aM) (158). Importantly, this detection strategy was found to be effective in analyzing RNA from prostate cancer biopsies and leukemia cells (157, 161). In addition to RNA detection, miRNAs are important biomarkers for the diagnosis and classification of various diseases including cancer (171). The versatility of the NME electrochemical detection platform demonstrated that the detection of low-abundance miRNAs was possible (119, 172).

The speed of a nucleic acid sensor is highly dependent on the concentration and subsequent kinetics of the target analyte. Traditional validation of electrochemical biosensors uses short sequenced oligonucleotides (10-100 base pairs (bp)) as proof-of- concept; however, analyzing gene expression in clinical samples involves measuring large mRNA molecules (> 1000 bp). For example, a 4000 bp mRNA target specific to bacterial detection would take 24 days to accumulate on a 100 nm sensor – well outside an acceptable signal-to-answer timeframe (173). The discovery that the NME-based sensors could be fine-tuned to various sizes and structures led to significant improvements ultra-rapid detection. For the same 4000 bp target, using the NME

platform with sensors of 100 um in size, the time to accumulate a single target dropped to only 12 minutes, which was well within a clinically-relevant time frame (173).

The use of the NME platform has been further extended for the analysis of proteins and small molecules. However, these molecules are challenging to detect because many proteins and small molecules have varying degrees of charge or a lack of charge. Thus, the conventional ruthenium-based approach that exploits the negative charge of a nucleic acid backbone would not be feasible. A novel adaptation of the NME assay was developed for protein detection that employed steric hindrance for the inhibition of electron transfer after the target protein binds the sensor (174). This strategy was successfully employed for the detection of a large, ovarian cancer marker (0.1 units/mL) and enumeration of circulating cancer cells (175). Although adaptations of the NME assay were successful for the detection of large proteins and whole cells, small molecules and proteins remained undetected until the advent of the neutralizer displacement assay (NDA). Through the addition of a positively charged neutralizing strand, semi-complementary to a probe sequence on the surface of NMEs, the NDA was able to indirectly measure the presence of target analyte by monitoring the dissociation of the neutralizing strand. Because the probe and neutralizing strands were comprised of nucleic acids and derivatives, the classical ruthenium-based redox reporter system could be used with this approach. The NDA method was validated using several small molecules and protein targets with specific aptamer probes (176). Having met the challenges of metabolomic and proteomic detection using variations of the NME platform, one of the last remaining analytes of interest for biosensor was antibody detection. The production of antibodies against specific pathogens is an important diagnostic tool in HIV screening. To this end, a completely novel strategy of detection using the NME platform was developed. By confining the bioreactions to locations proximal to an NME sensor, the production of an electrochemical signal resulting from the hybridization of a conjugated secondary antibody was accurately

measured (177). Together with nucleic acid detection, the NME platform has the demonstrated capability of rapid and sensitive detection of various important biomolecules thus providing valuable diagnostic and prognostic information to health care professionals.

Lastly, in order for the NME platform to fulfill the characteristics of a point-of-care diagnostic tool, testing surrounding integration in addition to multiplexing needed to be addressed. Using an electrical lysis method, a sample-processing unit was coupled to the NME platform and successfully detected E. coli or S. saprophyticus samples at clinically relevant concentrations (a single bacterial cell within thirty minutes) (127, 178). Other limitations that exist for electrochemical biosensors involve the number of electrical contacts that are required for high degrees of multiplexing. Because each biosensing electrode requires an individually addressable contact, strategies to expand beyond the typical microchip format without a concomitant increase in microchip size have been investigated. Through the creation of perpendicular liquid channels on glass microchips that contained several working or reference/counter electrodes in series, a solution-based circuit could be created to address an individual electrode via the activation of a specific channel combination (179). In doing so, arrays of up to 100 NMEs could be profiled on a single microchip with only 20 electrical connections. This strategy was shown to be very effective in monitoring and determining the origins of various combinations of urinary tract infections within two minutes (179).

Although the NME platform has been shown to exhibit the necessary characteristics of a point-of-care diagnostic tool, use as a gene-profiling tool has been limited. Novel strategies that move beyond detection of infectious analytes are required for such applications. Critically, developing NME-based sensors that are capable of quantifying small differences in analytes expressed at high levels in human tissue are needed. In

addition, the starting substrate for gene profiling can often take the form of a solid piece of tissue (biopsy), which presents significant challenges with respect to specificity, as there are many off-target analytes present as well as potentially large pieces of debris from the lysis process. Together, an NME senor to be used for mammalian gene profiling must be developed and display the key characteristics of being highly quantitative and specific.

1.16 Meeting the Challenges of Lung Transplantation Diagnostics – Summary

The goal of the work herein is to develop novel classes of point-of-care diagnostics that could be used to transform the lung transplant community. By developing electrochemical biosensors based on the established success of the nanostructured microelectrode platform, this work will enable access to a plethora of knowledge and insights that were previously unavailable using existing technologies. Sensing platforms designed to detect relevant nucleic acid and protein targets that identify episodes of acute and chronic lung rejection would be truly impactful in the decision-making process of transplant surgeons. The successful demonstration of this technology would have far reaching implications beyond that of lung transplantation. Immediately, the work could be generalizable to other solid organ transplantation procedures such as the heart, liver, and kidneys as rapid access to proteomic and genomic information is also extremely valuable for these procedures. Beyond the field of transplantation, there are a number of companion-like diagnostic tests that could be spurred from this work. For decades there have been studies that identify metabolites, genes, and proteins that correlate with various human maladies; however, as in transplantation, many of these prognostic markers are the result of changes in physiological expression levels for which lengthy testing procedures occur in centralized laboratories within the health care system. Therefore, the development of rapid biosensors that provide quantitative information to physicians could be instrumental in guiding diagnoses and therapies for these cases.

Together, this work supports the key role that point-of-care diagnostics will play in the transplantation process – one that could change how donor lungs are selected, managed and treated, ultimately improving the quality of life for patients.

1.17 Project Hypothesis and Goals

The nanostructured microelectrode platform can be used to develop rapid and highly sensitive diagnostic tools that establish biomarker expression profiles for the transplantation process.

Two main classifications of devices will be investigated in this work:

(I): A rapid mRNA diagnostic tool that predicts the incidence of primary graft dysfunction in donor lungs.

A predictive nucleic acid test that specifically reports on the mRNA levels of transplant- specific biomarkers will be designed. First, technical validation of the proposed platform will be undertaken. This will require the development of a new, quantitative microchip platform. Subsequent work will demonstrate the key parameters of this new platform including: sensitivity, speed, and feasibility for use in the transplant setting. Importantly, the reproducibility of the new platform will be determined to complete the technical validation aspect of this work. Continuing with the development of nucleic acid biosensors, a series of studies for biological validation will be conducted. Probes that specifically report on the physiological levels of PGD-predictive genes in addition to internal assay controls will be designed and developed. The complexity of the assay will be tested and built up from a simple buffered solution to unpurified human lung biopsies. In order to confirm the validity of the test, a head-to-head comparison with the industry standard for nucleic acid assays, qPCR, will be conducted. Finally, a series of

clinical validation studies will confirm the prognostic validity of the mRNA biomarkers used in this work. With this information, an algorithm that combines the results of multiple mRNA biomarkers will be designed and validated for clinical utility.

(II): The foundation for a fast protein detection device will provide valuable insight into the short- and long-term success of the lung transplant procedure.

Complementary to the nucleic acid assay, the groundwork of a transplant-relevant protein assays will be investigated. Initially, an electrochemical detection assay will be sought that specially reports on the levels of small (< 50-kDa) proteins. As a model analyte involved in the development of chronic lung allograft dysfunction, TGFβ1 represents the basis for the proof-of-concept studies. The technical validation of this assay requires the evaluation of the speed and feasibility of electrochemical TGFβ1 detection and demonstration of the capabilities of the assay in transplant-relevant media (lung perfusate solution) will constitute biological validation. Building from the success of the TGFβ1 platform, a novel sensor for the detection of the ET-1 peptide will be conducted. Peptide (~20 amino acids) detection remains a significant challenge for electrochemical biosensors based on the steric blocking. As such, the ET-1 assay will represent a new schematic for electrochemical peptide detection and provide an important biological test that could predict acute or chronic rejection (primary graft dysfunction and bronchiolitis obliterans syndrome) in lung transplant patients. Work with this assay will require technical and biological validation consisting of proof-of- concept and lung perfusate studies respectively.

1.18 An Overview of Thesis Chapters

In this work we describe our efforts to develop a new class of quantitative electrochemical biosensors. We utilize these sensors to accurately predict lung transplant outcomes by assessing the risk for primary graft dysfunction and chronic lung allograft dysfunction.

Chapter 2 looks at the development of a novel class of sensors, fractal circuit sensors (FraCS), derived from the NME platform. We demonstrate the ability of these new sensors to provide rapid and, importantly, quantitative gene-expression information. Technical validation of this platform is described in Chapter 2 where the sensors are validated with a variety of analytes and we show the development of a lysis technique in which crude lung biopsy samples can be used as the starting substrate for FraCS analysis. Lastly, we biologically validate the ability of FraCS to provide gene-based profiles of donor lungs with varying outcomes.

Clinical validation of the FraCS approach is shown in Chapter 3. We describe our efforts to validate the approach using qPCR as the gold standard and go on to show the clinical utility of the biomarkers used in the FraCS system. Importantly, we use the FraCS assay to develop and cross-validate a prediction model that accurately predicts the incidence of primary graft dysfunction in donor lungs with a high degree of sensitivity and specificity.

Chapter 4 looks at the development of an electrochemical platform that can serve as the foundation for small protein detection assays. We use the TGFβ1 surrogate, LAP, as a

proof-of-concept maker and demonstrate that this platform is amenable to the transplant process with timely performance in transplant media.

In Chapter 5, we describe the development of a novel assay for the detection of endothelin-1 (ET-1), a biomarker specific to primary graft dysfunction and bronchiolitis obliterans syndrome (BOS). By using an approach analogous to that of a competitive ELISA, we show the detection of ET-1 using an electrochemical approach. We describe the current advantages of this testing and show the possibility of this assay to be fine- tuned to a wide variety of target concentrations.

Lastly, Chapter 6 provides a summary of the work contained herein and gives a general outlook towards the future of the personalized medicine approach and its use within the transplant setting.

Chapter 2 The Development of a Quantitative Biosensing Platform

Chapter 2 contains materials from the manuscript:

2 The Development of a Quantitative Biosensing Platform

2.1 Abstract

Previous work by our group has focused on the development of rapid and highly sensitive electrochemical biosensors for the detection of rare analytes (127, 157, 161, 172, 179). These sensors were designed to display maximal signal under the conditions of limited analyte binding; however, for lung transplant gene-profiling techniques, quantification of small differences in large analyte expression levels is crucial. Herein, we sought to develop a novel class of chip-based sensors that would enable rapid and quantitative analysis of mRNA in tissue. Through the exploration of various sensors, we developed fractal circuit sensors (FraCS), three-dimensional metal structures with large surface areas that displayed signals with maximal concentration dependence. This platform was validated using a logical buildup of analyte complexity and ultimately tested using crude, unpurified tissue biopsies. Using FraCS, we were able to rapidly (< 20 minutes) and reproducibly quantify small differences in the expression of transplant-relevant biomarkers. This work provides an important step towards bringing rapid diagnostic mRNA profiling to clinical application in lung transplantation.

2.2 Introduction

To bring molecular diagnostics to the bedside in lung transplantation (LTx), there exists the critical challenge of delivering reliable and reproducible results to the clinician in less than thirty minutes of sample receipt. The provision of molecular diagnostic results based on predictive biomarkers in this time frame could be truly impactful on real-time decision-making in clinical lung transplantation. Existing molecular analysis

technologies, including PCR, require highly trained personnel, contamination-free laboratory facilities, and extensive sample purification. As a result, they require in the range of 6-12 hours to deliver results in a hospital workflow (113, 180). Delays encountered when methods requiring lab facilities like PCR are used become even more significant when one accounts for the fact that lung transplantation most often occurs outside normal lab operating hours. In the limited time window of lung transplantation, tissue must be taken from its crude state and processed into a form that can be analyzed for gene expression. As such, there exist no prior reports of sample-to- answer processing and analysis of lung tissue biomarkers for LTx suitability using techniques requiring fewer than 12 hours.

There has been significant progress in the area of new molecular analysis strategies, with chip-based sensors achieving clinically relevant sensitivity and specificity on samples subjected to minimal cleanup (119, 127, 157, 161, 176). The Kelley laboratory has developed a class of sensors designed to capture and analyze mRNA transcripts. These devices have been built to be highly sensitive for rapid, qualitative mRNA detection for diagnosis of infectious disease and other rare analytes. However, these sensors were designed to achieve saturation of the sensor surface and electrical signal at very low levels of mRNA bound to the surface. Thus, these sensors do not allow for the quantification of mammalian genes expressed at high levels.

Herein we report a new class of nanoscale chip-based sensors, termed fractal circuit sensors (FraCS), which enable quantitative analysis of multiple biomarkers present at relevant concentrations in human lung tissue. The LTx-assessment chip enabled by FraCS rapidly assesses predictive mRNA biomarkers in donor lungs from a lung tissue biopsy performed by the surgeons, relying on a simplified sample preparation step to release biomarkers into solution, thus essentially requiring no sample purification.

This assay represents important proof-of-principle for the deployment of a quantitative gene-profiling point-of-care diagnostic tool for lung transplantation. It is our hope that in the future, the evaluation of donor lungs could be made more precisely based on molecular-level information that the FraCS approach provides.

2.3 Results

Development of quantitative electrochemical sensors

In this work, we endeavored to create a new type of sensor with maximal surface area. Sensors made from electrodeposited gold were templated in apertures created in a SU- 8 layer patterned over gold contacts on a glass chip. The electrochemical currents collected at these sensors and used for nucleic acid quantitation are generated by a 3+ 3- label-free, redox-active, [Ru(NH3)6] /[Fe(CN)6] reporter system (Figure 5). Briefly, the 3+ [Ru(NH3)6] electron acceptor complex is positively charged and binds to the sensors at levels that correspond to the amount of negatively charged nucleic acid that binds to the 3- sensor. The [Fe(CN)6] electron acceptor complex is negatively charged, and so does 3+ not bind to the sensor, but remains in solution to accept the electrons from [Ru(NH3)6] to allow multiple turnovers for this ion.

First we investigated the effects of aperture shape on sensor morphology (Figure 7) and hybridization efficiency (Figure 8 and Table 4). For sensors plated in circular, square, and line-based apertures with approximately the same aperture area, we observed that the circular and square apertures produced sensors with greater electrode growth at the edge of the sensor and a recessed interior (Figure 7, bottom row). For sensors deposited in line-based apertures, electrode growth was restricted to only the observed ‘edge-effects’ (Figure 7f). When we compared the changes in hybridization currents against assay background for complementary DNA targets for the various aperture footprints. We determined that the line apertures produced the highest current changes (364%) followed by square apertures (193%) and then circular apertures (119%) (Figure 8). Consistent with the observation of the effects of sensor shape and edge growth, we

a b c

d e f

19.0 µm 19.0 µm 50.0 µm

Figure 7. Biosensor morphology. Graphic representation of biosensing apertures created with (a) circular, (b) square, or (c) linear footprints. Images collected using

scanning electron microscopy for sensors electroplated in 50 mM HAuCl4 for 30 s using (d) circular, (e) square, or (f) linear apertures. Scale bars are shown on each image.

then demonstrated that a change in hybridization signals (currents) followed an increasing pattern that was consistent with the % increase in calculated perimeter for each sensor (circle to square: +125% perimeter, +162% current; circle to line: +267% perimeter, +306% current), thus confirming a role for edge growth on hybridization currents (Table 4). Lastly, in an independent set of experiments we confirmed that that the catalytic currents (20 nA vs 5 nA) and subsequent current density (0.25 mA/cm2 vs. 0.20 mA/cm2) was indeed higher in the line-based apertures (Figure 9) leading us to hypothesize that these sensors, termed Fractal Circuit Sensors (FraCS), would allow for an enhanced ability to quantitate gene targets.

500 (-) DNA 400 (+) DNA

300

200

Delta I (%) 100

-100

Line Circle Square

Figure 8. Biosensor morphology and hybridization currents. Currents generated from biosensors templated with circular, square, and linear apertures for (+) complementary (black bars) and (-) non-complementary (white bars) DNA targets at 100 nM. Data represented are of n=10 different sensors. Columns represent mean and error-bars indicate s.e.m.

Table 4. Biosensor hybridization signals and aperture parameters. Dimensions / Area / Perimeter / Hybridization Signal / Shape µm µm2 µm %

Circle r = 12.5 491 78.5 119

Square 25 x 25 625 100 193

Line 5 x 100 500 210 364

a ) b -2 catalyticcurrent (nA) catalytic current density (mA cm catalyticcurrent density (mA electrodeposition time (s) electrodeposition time (s)

Figure 9. Electrochemical performance of circular and linear apertures. Comparison of (a) catalytic current and (b) catalytic current density for biosensors electrodeposited in circular (black circles) and linear (black squares) apertures for various plating times. Data represented are of n=15 different sensors. Each point represent mean and error-bars indicate s.e.m.

Figure 9 is adapted with permission from reference (181): Zhou YG, Wan Y, Sage AT, Poudineh M, and Kelley SO. “Effect of microelectrode structure on electrocatalysis at nucleic acid-modified sensors.” Langmuir. 2; 30 (47): 14322-8 (2014).

Rapid and quantitative gene analysis with FraCS

3+ In an effort to maximize the amount of [Ru(NH3)6] that could access the surface of a sensor, we moved towards a FraCS approach and away from the sensors described in our previous work, which involved using smaller round templates. This would theoretically allow larger electrochemical currents to be generated and the signals would then exhibit maximal concentration dependence. A mathematical model was 3+ constructed that compared the total amount of [Ru(NH3)6] (current) on sensors made with round templates to FraCS by taking into account the electrode size, hybridization 3+ efficiency, and diffusional properties of [Ru(NH3)6] (Figure 10a). By modeling the amount of ruthenium associated with analyte on the surface of FraCS, we observed that FraCS possessed a two-fold advantage over the nanostructured microelectrodes. First, the dynamic range for currents produced by FraCS could be greatly enhanced (Figure 10a) and secondly, FraCS could reach a saturation current that was significantly higher than the circular microelectrodes (3.67 nA vs. 0.67 nA Figure 10a). We then experimentally tested whether FraCS were superior for nucleic acid quantification compared to our previously developed electrochemical sensors. Both sets of electrodes achieved signal-to-noise ratios greater than 2.0 (Figure 10b/c); however, as quantitative sensors, FraCS had an improved ability to achieve larger signal gains per change in analyte concentration (593% vs. 136% as shown in Figure 10b). This was consistent with our modeling predictions in Figure 10a and, as expected, the FraCS signals were of greater intensity, which resulted in the ability to easily and reliably differentiate large concentration profiles over multiple orders of magnitude (Figure 10b). This observation was even more apparent when looking at a narrow range of analyte concentrations (1- 10 nM) (Figure 10c). For sensors challenged with solutions of 1 to 10 nM, there was a 335% increase in signal using FraCS versus 52% using the circular electrodes (Figure 10c). As a result, FraCS showed an enhanced ability to discriminate between very small differences in analyte concentrations (even within a single order of magnitude) – a result not possible with the circular electrodes (Figure 10c). As further proof-of-concept validation, we estimated other quantitative metrics of FraCS (i.e. limit of detection) and

experimentally confirmed that an LOD could be improved by using a FraCS-based approach (Figure 10b/c). We tested the density of probe molecules on the surface of FraCS and determined that a 3:1 molar ratio of PNA-SH to MCH-SH was optimal in maximizing the signal output (Figure 11a). Using this surface coverage, we were able to achieve a limit of detection of approximately 100 fM and a dynamic range of 1 pM to 100 nM using complementary DNA oligonucleotides (Figure 11b).

a b c 5 circular Linear circular 2.0 3.0 linear linear 4 Circular

1.5 3 2.0 I (nA) I (nA) I (nA) 1.0 Δ 2 Δ

Current (nA) 1.0 1 0.5

0 0.0 0.0 0 2 4 6 8 10 1 10 100 1 2 4 6 8 10 [DNA] (nM) Concentration (nM) [DNA] (nM)

Figure 10. Nucleic acid quantification using linear apertures. (a) Mathematical modeling of biosensors templated in linear apertures (solid line) versus sensors made with circular templates (dashed line) for the current generated by the sensor as a function of DNA concentration. Quantitative comparisons of sensors with circular apertures (white bars) and linear apertures (black bars) between (b) 1-100 nM target and (c) 1-10 nM target. Data represented are of n=15 different sensors. Columns represent mean and error-bars indicate s.e.m.

a b 25 2000 (-) DNA 20 (+) DNA 1000 15 250 200 10 150

5 Delta I (%) Delta I (nA) 100 0 50 -5 0

1.00 0.75 1nM (+) 1pM (+) PNA-SH Mole Fraction 10nM (+) 10pM (+) 100fM (+) 100nM 100nM(-) (+) 100pM (+)

Figure 11. Sensor surface density and limit of detection (LOD). (a) Comparison of hybridization currents generated from sensors challenged with 100 nM (+) complementary (black bars) and (-) non-complementary DNA targets at varying surface probe densities. (b) LOD determination for sensors challenged with complementary DNA targets between 100 fM -100 nM and 100 nM non- complementary (-) DNA targets. Data represented are of n=20 different sensors. Columns represent mean and error-bars indicate s.e.m.

In addition to the quantification considerations, we set out to validate another key technical advantage of using FraCS for rapid analysis of mRNA. The increase in electrode size in the y-direction allows for very rapid accumulation of the biomarkers of interest on the surface of the detectors (Figure 12). Remarkably, in time-dependent rapid hybridization experiments, FraCS performed even better than anticipated and could detect mRNA transcripts in as little as 15 minutes (Figure 12).

1.5

1.0

Delta I (nA) 0.5

0.0 5 10 15 30 Hyb. Time (min)

Figure 12. Rapid analyte detection using FraCS. Hybridization time course for rapid RNA analysis using LTx assay sensors. Data represented are of n=15 different sensors. Columns represent mean and error-bars indicate s.e.m.

Technical performance of FraCS using donor lung biopsies

Multiplexed chips were created that allowed parallel analysis of the lung assessment markers. Each chip contained up to five sets of FraCS that could be individually functionalized with probes against the marker set of interest. To achieve quantitative analysis of human lung tissue biopsies without an mRNA purification step, we developed a rapid (~5 min) mechanical and chemical lysis method based on non-ionic detergents that were effective in lysing the lung tissue while remaining compatible with FraCS and the electrochemical reporter system. To ensure consistency for signals obtained using unpurified lysate and total RNA from the same donor lung sample, we first demonstrated that the presence of a robust probe self-assembled monolayer (SAM) was sufficient to prevent non-specific adsorption of biomolecules at the electrode surface when compared to a bare gold electrode (signal-to-noise ratio > 2) (Figure 13a).

Next, we compared the FraCS response of both purified and unpurified biopsy samples (Figure 13b). There was a strong, positive correlation of the biomarker responses for both the pure RNA and unpurified lysate (r = 0.96). We then demonstrated that the use of extraneous laboratory equipment was unnecessary for the FraCS based approach (Figure 13c). There was no difference between the non-specific (Drosophila melanogaster) and specific (GAPDH) signals obtained from lung biopsies that were processed using either centrifugation or filtration techniques (p = 0.46) (Figure 13c). We did observe a slight, but significant increase in the non-specific signal of a biopsy that was processed using filtration versus centrifugation (+0.18 nA, p < 0.05).

a b c

10 4 1.5 Centrifugation 8 IL-6 Filtration 3 6 1.0

4 2 ATP11B EGLN1 * 2

Delta I (nA) 0.5 Delta I (nA)

(Normalized (Normalized Current) 1 t IL-10 0 r =0.96

-2 RNA 0 0.0 Bare SAM 1.5 2.0 2.5 3.0 3.5 4.0

Biopsy Lysate (Normalized Current) Dros. GAPDH

Figure 13. Biosensor response in unpurified lysate. (a) Currents generated from the non-specific adsorption of biomolecules at bare- and PNA/MCH modified- (self- assembled monolayer (SAM)) biosensors. (b) Correlation of biomarker signals obtained from purified RNA from a lung biopsy versus unpurified lysate of the same biopsy (r indicates Pearson’s correlation coefficient). (c) GAPDH and Drosophila melanogaster sensors challenged with lung biopsy lysates that were centrifuged (black bars) or syringe filtered (white bars) following tissue lysis. Data represented are of n=15 different sensors. Columns represent mean and error-bars indicate s.e.m. Data were analyzed by two-tailed, student’s t-test. * indicates P < 0.05

An important final aspect of technical validation of the FraCS approach was to investigate signal reproducibility. To study this, we tested two different lung biopsies (one poor- and good-outcome) using multiple sensors on different microchips (Figure 14). There was no difference between the tests, which we confirmed using an ANOVA (Table 5). Our results indicated that there was no statistical difference (p = 0.77 and p = 0.98) in the microchips and the overall variance (% CV) in sensor output ranged from 7% to 2% (Table 5) using the FraCS approach to test a good and poor-outcome donor lung respectively.

a b

5 5

4 4

3 3

2 2 Delta I (nA) Delta I (nA) 1 1

0 0 1 2 3 1 2 3 Microchip Microchip

Figure 14. FraCS signal reproducibility. Box and whisker plots of FraCS IL-6 signals obtained from multiple (n=5) measurements of a (a) good-outcome and (b) poor-outcome donor lung sample on distinct microchips (n=3). Vertical lines show the population means and the whiskers represent the minimum to maximum signals.

Table 5. FraCS signal reproducibility statistics.

Population Statistics ANOVA

2 Donor Lung Meantot (nA) % CV P R

Good-Outcome 1.765 6.883 0.77 0.037 (n=17) Poor-Outcome 1.938 1.936 0.98 0.003 (n=13)

Biological validation of the FraCS approach using donor lung biopsies

Using the FraCS-based approach, we then proceeded to develop probes against the mRNAs of the previously reported donor lung assessment markers (IL-6, IL-10, ATP11B, EGLN1) and control sequences (human GAPDH as a positive control and Drosophila melanogaster GAPDH as a negative control). First, multiple probe sequences for each target mRNA were screened using RNA dot blots to qualitatively select sequences with high affinities (Figure 15). The intensity of each spot was optically evaluated at various purified, total RNA concentrations (300, 600 ng) (Figure 15) and sequences with stronger binding (darker intensity) were selected for PNA synthesis (example for IL-6 and IL-10 shown in Figure 15).

a b

300 ng RNAt 600 ng RNAt 300 ng RNAt 600 ng RNAt

+ control - control + control - control

Figure 15. LTx assay probe validation. RNA dot blots for (a) IL-6 and (b) IL-10 probe sequences detecting target analytes in purified, total RNA (top dots). Complementary (+ control) and non-complementary (- control) sequences are shown in the bottom two dots.

The probes for the LTx-assay were successfully designed (summarized in Table 6) and validated (signal-to-noise ratio > 2.0) using 100 nM (non) complementary synthetic DNA oligonucleotides (Figure 16a-d). To test whether FraCS could accurately measure the levels of the target lung assessment mRNAs in heterogeneous samples, we isolated total RNA from donor lung tissue. For all of the respective LTx-assay probes we observed concentration-dependent signals that were highly reproducible (Figure 17). An analysis of the ratio of two important inflammatory cytokine markers, IL-6 and IL-10, illustrates the excellent precision of the detection strategy, as the ratio of IL-6 to IL-10 remained constant (p = 0.45, One-way ANOVA) at various total RNA concentrations (0.75 – 12.5 ng/µL) (Figure 17c).

Table 6. PNA probe sequences for LTx assay.

Gene Target Species Probe Sequence (N to C)

IL-10 Homo sapiens Cys-O-CGC CGT AGC CTC AGC C

IL-6 Homo sapiens Cys-O-GCC AGT GCC TCT TTG CT

ATP11B Homo sapiens Cys-O-ATC CAC TAC CAG CCC

EGLN1 Homo sapiens Cys-O-TCA ATG TCA GCA AAC

GAPDH Homo sapiens Cys-O-GTT GTC ATA CTT CTC

GAPDH Drosophila melanogaster Cys-O-ACC GAA CTC GTT GTC GTA CC

Legend: ‘Cys’ represents the amino acid cysteine, ‘O’ represents an ethylene glycol linker, and A, T, G, C represent the nucleobases adenine, thymine, guanine, and cytosine respectively.

IL-6 IL-10 a b 2.5 2.5

2.0 2.0

1.5 1.5

1.0 1.0 Delta I (nA) Delta I (nA) 0.5 0.5

0.0 0.0 100nM (-) 100nM (+) 100nM (-) 100nM (+)

ATP11B EGLN1 c d 2.5 2.5

2.0 2.0

1.5 1.5

1.0 1.0 Delta I (nA) Delta I (nA) 0.5 0.5

0.0 0.0 100nM (-) 100nM (+) 100nM (-) 100nM (+)

Figure 16. Validation of LTx probes with DNA targets. Hybridization signals for sensors challenged with 100 nM complementary (+) or non-complementary (-) DNA for (a) IL-6, (b) IL-10, (c) ATP11B, and (d) EGLN1 sensors. Data represented are of n=15 different sensors. Columns represent mean and error-bars indicate s.e.m.

a IL-6 b IL-10 c IL-6 / IL-10 10 10 4

8 8 3 6 6 2 4 4 Delta I (nA) Delta I (nA) 1

2 2 Normalized Ratio

0 0 0 0.75 3 6 12.5 0.75 3 6 12.5 0.75 3 6 12.5

[RNAt] (ng/uL) [RNAt] (ng/uL) [RNAt] (ng/uL)

d GAPDH e ATP11B f ELGN1 10 10 5

8 8 4

6 6 3

4 4 2 Delta I (nA) Delta I (nA) Delta I (nA) 2 2 1

0 0 0 0.75 3 6 12.5 0.75 3 6 12.5 0.75 3 6 12.5

[RNAt] (ng/uL) [RNA ] (ng/uL) [RNA ] (ng/uL) t t

Figure 17. Quantification of RNA markers predictive of lung transplant outcome. Total RNA titration profiles for (a) IL-6, (b) IL-10, (c) IL-6:IL-10, (d) GAPDH, (e) ATP11B, and (f) EGLN1 sensors. Data represented are of n=15 different sensors. Columns represent mean and error-bars indicate s.e.m.

Development and validation of FraCS assay controls

We then validated the suitability of our GAPDH positive control probe. First, we examined the expression levels of GAPDH in various donor biopsies and confirmed that expression of GAPDH was unchanged across several donor lungs (r = 0.06) (Figure 18a). We then determined that the signals obtained at FraCS modified with GAPDH probes were strongly correlated (r = 0.92) with increasing concentrations of total RNA purified from donor lung biopsies (Figure 18b). As such, the GAPDH probe served as a reliable reference control for analyzing donor lung biomarker expression levels.

a b 40 4

30 3 (nA)

20 2

Threshold Ct Threshold 10 1 r =0.06, p=0.8635 Signal GAPDH r =0.92, p=0.0002 0 0 Individual Donor Lungs 0 100 200 300 400 500 [RNA ] (ng/uL) t

Figure 18. GAPDH validation. (a) qPCR-determined GAPDH expression levels (Threshold Ct) obtained from n=11 different donor lungs (Data were analyzed by two- tailed, Pearson correlation). (b) Correlation of GAPDH signals obtained from sensors versus total RNA concentration (Data were analyzed by two-tailed, Pearson correlation) (n=10 different donor lungs).

To obtain proof-of-principle for the use of the FraCS approach to profile assessment markers in lung biopsy samples, we assessed target gene expression levels in tissue from both a mildly injured lung (Figure 19a) and a severely injured lung (Figure 19b). In the case of a severely injured lung (Figure 19b), the LTx-assay accurately displayed increased expression levels of the markers indicative of a poor outcome: ATP11B, IL-6, and IL-6/IL-10.

Good-Outcome Donor Lung Poor-Outcome Donor Lung a b 2.0 2.0

1.5 1.5

1.0 1.0

0.5 0.5 Normalized Current Normalized Current

0.0 0.0

IL-6 IL-6 IL-10 IL-10 EGLN1 ATP11B EGLN1 ATP11B IL-6 / IL-10 IL-6 / IL-10

Figure 19. Donor lung profiling using LTx biosensors. Representative signals obtained from a (a) good-outcome and (b) poor-outcome donor lung. The signals are normalized to GAPDH controls and the non-specific Drosophila melanogaster signal is shown as a dashed line. Data represented are of n=15 different sensors. Columns represent mean and error-bars indicate s.e.m.

2.4 Discussion

In contrast to infectious disease testing, sensors designed to analyze gene expression profiles in human tissues must exhibit an enhanced ability to quantify small differences in target levels. The ability to detect ultra-low levels of target analyte is the goal of infectious disease testing as this allows for early detection and faster administration of medicines. Sensors that are to be used for gene profiling interrogate targets that are expressed on a continuum in human tissues and discerning where on the normal distribution an individual patient falls is crucial. Previous work by our group has been focused on developing sensors for the early detection of disease and, as such, were

optimized to detect low levels of invading organisms (or altered host-state), including bacteria (127, 178, 179), viruses (177), and cancers (172). In order to achieve these superior detection limits, sensors were designed to exhibit maximal signal saturation from a small number of binding target analytes. These sensors, nanostructured microelectrodes (NMEs), exploited small (5 µm in diameter) circular apertures that produced electrodes that protruded into solution and possessed efficient surface roughness to allow for optimal probe display. NMEs were able to achieve detection limits of low attomolar (158), which remains one of the most sensitive tests to date. In their current form, NMEs would not be suitable for use as LTx biosensors. The genes predictive of LTx outcome are predominantly from inflammatory pathways and thus can be expressed at relatively high levels in human cells. Therefore, using NMEs in this application would cause signal saturation for all of the genes studied and it would be impossible to distinguish between good- and poor-outcome lungs. As a result, the goal of this work was to design a new class of sensors that would not saturate at low levels of analyte and would allow for an enhanced ability to discriminate between small differences in gene expression in various donor lungs. To this end, we studied the effects of changing the aperture footprint and resultant metal sensors (Figure 7). By increasing the sensor area, we postulated that it would require more binding events to reach signal saturation and thus the dynamic range of the sensors would be increased. We varied the size and shape of the apertures and electrodes and determined that for circular- and square-based apertures that, although the area of the sensors increased, the morphology was primarily edge growth with recessed interiors (Figure 7). The structure of the apertures was then restricted to produce only edge effects - lines (Figure 7f). The observed edge effects did indeed play a pivotal role in the hybridization signal and for area-matched comparisons it was the perimeter (edge) of the aperture that predicted the signal intensity (Figure 8 and Table 4). In addition to the hybridization signal enhancement, we determined that both total current and current density was higher for the line-based electrodes termed, fractal circuit sensors (FraCS), over our traditional NMEs (Figure 9). This supported the hypothesis that there would be an enhanced ability to quantify targets using FraCS over NMEs. By mathematically and

experimentally testing FraCS versus NMEs there was a notable increase in the dynamic range of the sensors which led to the ability to discern analyte levels within a single order of magnitude – a characteristic not observed with NMEs at high target concentrations (Figure 10). Thus, FraCS were indeed a novel class of biosensors that could be used to quantify expression levels of genomic targets.

The discovery of FraCS for nucleic acid quantification prompted further technical validation studies in order to prove their effectiveness as biosensors. Probe surface density plays an important role in the hybridization efficiency of a probe and target analyte (160, 182). A probe surface that is too dense will limit hybridization efficiency whereas a density that is too sparse will produce very low hybridization signals and require an extended period of time for probe-target binding to occur (182). For the FraCS sensors, we observed that a thiolated probe ratio of approximately 75% was optimal for hybridization efficiency (Figure 11a) and from this we determined that the limit of detection for these sensors was approximately 100 fM (Figure 11b). As expected, the tradeoff between an ability to quantify high expression levels of target (FraCS) and sensitivity (NMEs) was apparent. The observed limit detection for FraCS was significantly higher than that of NMEs (10 aM) (158).

Based on previous diffusion and hybridization studies by our group (173) and others (183), we predicted that biomarker profiles could be collected in less than 30 minutes. In time course experiments, we observed that the FraCS were able to report on the presence of target analytes as quickly as 15 minutes. This is due to the large surface area of functionalized FraCS that provide more binding sites for target mRNA capture and efficient electrocatalysis relative to conventional chip-based sensors, while maintaining a morphology that is optimal for hybridization and electrochemical analysis (181). In addition, the edge effects of electrode growth of linear apertures/FraCS

produced sensors that extend, three-dimensionally, into solution. Taken together, these properties of FraCS would allow for faster reaction kinetics and shorter hybridization times (184).

In addition to quantification considerations, a key requirement of an LTx assay was the ability to identify the target expression levels rapidly and without the use of complex laboratory equipment. In order to facilitate ease-of-use for surgical teams, the LTx assay was required to work with samples in their crudest form. Thus, the goal was to technically validate the LTx assay with crude lysate from lung tissue biopsies with minimal sample processing and handling. Following the development of a combined mechanical and chemical lysis process, we studied the robustness of the self- assembled monolayer (SAM) (Figure 13a). As expected, there was a large signal observed when crude lysates were placed on bare electrodes due to a large number of biomolecules adsorbing to the gold surface. The presence of a robust SAM (PNA and MCH layer) can prevent biofouling of a sensor and we observed complete attenuation of the non-specific adsorption signal (Figure 13a). In addition, we compared the performance of the assay using lysates that were processed via centrifugation or syringe-filtration and found that both techniques produced similar results for non-specific binding (Drosophila melanogaster) and the internal positive control (GAPDH) (Figure 13c). There was a slight, but significant increase in the non-specific signal when using a syringe filter. This could be attributed to the fact that the filtration process could allow more particulate matter to pass through the filter compared to the centrifugation process. Nonetheless, the filtration process represented a means by which the LTx assay could be easily integrated into a simple, automated device that would not require any extraneous laboratory equipment (i.e. centrifuge) or sample preparation. Most importantly, we determined that the gene-specific signals were completely consistent in unpurified, lysed biopsies with that of the corresponding purified RNA (Figure 13b).

Reproducibility is perhaps one of the most important aspects of the technical validation of the LTx assay. By testing the same biopsy on multiple different microchips, we were able to confirm that there were no differences in the signals obtained (Figure 14 and Table 5). Furthermore, for the signals obtained across the different microchips there was 2 - 7% error, which reflects the excellent reproducibility of the FraCS assay. Taken together, we were able to demonstrate, through technical validation of the LTx assay, the defining characteristics (speed, sensitivity, easy-of-use, and reproducibility) of a diagnostic assay amenable to implementation in the transplant setting.

Subsequent to the technical validation of the FraCS-based LTx assay, careful validation of the molecular biology of the test was performed. In 2006 and 2008 a series of published studies demonstrated the association of transplant-specific genetic markers with poor-outcomes (ATP11B, EGLN1, MCPH1, FGFR2) (15) and PGD status (IL-6 and IL-10, IL-8) (86). The results of these works served as the foundation of the LTx assay. We sought to develop sensors that would report on the expression levels of IL-6, IL-10, EGLN1, and ATP11B.

Interleukin-6 (IL-6) is a proinflammatory cytokine that has been shown to play an important role in the differentiation of helper T-cells (TH17) that promotes neutrophil proliferation and migration. In contrast to IL-6, the interleukin-10 (IL-10) cytokine limits and terminates inflammatory responses through the inhibition of T-cells, monocytes, and macrophages via the JAK/stat pathway (Figure 2). These cytokines play an important role in the inflammatory process and are believed to be involved in ischemia- reperfusion injury. Thus, an IL-6:IL-10 ratio reports specifically on an elevated inflammatory status of a donor lung and this metric is highly predictive of outcome. Using qPCR, donor lung tissues with an IL-6 to IL-10 ratio greater than 9.3 were shown

to develop severe PGD and had a 20-fold relative risk of recipient death (86). To date, no other markers have been shown to display such strong predictive ability.

In a follow-up study, ATP11B (ATPase, Class VI, P-type 11B) and EGLN1 (Egl 9 homolog 1) were identified as additional genetic biomarkers that are highly predictive of transplant outcome (15). ATP11B regulates phospholipid-translocating ATPase activity and dysregulation of this protein has been shown to result in altered vesicular transport and play a role in cancer progression (185), yet a specific role in transplant function remains to be elucidated. EGLN1 is an important mediator of oxidoreductase activity in cells and, as such, high expression levels of EGLN1 in a donor lung with PGD III may result in the alveolar damage, lung edema, and severe hypoxia observed in these patients.

The first step towards developing an LTx assay directed at the selected probes was to determine the optimal nucleic acid sequence that displayed superior target binding. By using RNA dot-blot analysis (Figure 15) we were able to observe the sequences with stronger binding affinities to their respective mRNA targets. Probe-target binding efficiency is dictated not only by the specific nucleobases involved in the probe sequences, but also the location (i.e. stem-loop) within the mRNA to which it binds. Having prescreened potential probe sequences to determine the best probes, initial validation using complementary DNA oligonucleotides was performed (Figure 16) prior to testing RNA (Figure 17). As expected, the ratio of IL-6 to IL-10 remained constant at various concentrations of RNA used to challenge the sensors (Figure 17c). The absolute value of hybridization currents for IL-6 and IL-10 was anticipated to increase at higher RNA concentrations; however, it was essential that the measured ratio remain constant over the same concentrations as this was expected to be independent of RNA loading.

Biological validation of the positive loading control, GAPDH, was performed in order to confirm that the behaviour of this standard was appropriate to serve as a normalizing control. GAPDH is an important enzyme involved in the glycolysis pathway and is expressed at high levels in human lung cells (186). Importantly, we confirmed that the GAPDH lung expression levels do not vary greatly from patient to patient (Figure 18a) and that the FraCS-derived GAPDH signals were strongly correlated with the total amount of RNA loaded on a chip (Figure 18b). Taken together, we concluded that by normalizing the LTx gene signals of a given donor lung to the respective GAPDH signal we would be able to (i) correct for any minor size variations in the initial biopsies used and (ii) make direct comparisons when testing various patient samples. Having validated the LTx specific probes and assay controls, it was then possible to complete a gene profile for a good- and poor-outcome donor lung (Figure 19) where we observed a patient-specific pattern for the each of the tested genes. This represents a significant technological advance as, previously, this type of profile would not be available to transplant teams as the standard method of obtaining it, PCR, would not be completed in the allotted time frame of two hours.

2.5 Conclusions

Herein, we were able to develop a novel class of biosensors, FraCS. FraCS demonstrated an enhanced ability to quantify mRNA targets over conventional, NME- based sensors. This novel platform was validated using a set of transplant-relevant reporter probes and successfully showed the key characteristics of an integratable POC device: rapid, sensitive, specific, easy-to-use, and reproducible. Using the FraCS platform, clinical analysis of lung transplant specific biomarkers is now possible, opening the possibility of rapid genomic testing and personalized medicine for the lung.

2.6 Materials and Methods

Microchip fabrication

Microchips were fabricated in-house at the Toronto Nanofabrication Centre (University of Toronto, Toronto, ON) using precoated (5 nm chromium, 50 nm gold, and AZ1600 (positive photoresist)) glass substrates purchased from Telic Company. Standard contact lithography was used to pattern the sensing electrodes and followed by Au and Cr wet etching steps and removal of the positive photoresist etchant mask. SU-8 2002 (negative photoresist) (Microchem Corp.) was then spin-cast (4000 rpm, 40 s) and patterned using contact lithography to create the sensing apertures (circular apertures were 5 µm (protein detection assays) or 25 µm (aperture characterization) in diameter, square apertures were 5 µm x 5 µm, and linear apertures were 5 × 100 µm). Microchips were diced in-house using a standard glasscutter and washed with acetone (Caledon

Labs), isopropyl alcohol (Caledon Labs), and then O2 plasma etched using (Samco RIE System (Samco)).

Biosensor electroplating

Gold electrodes were electrodeposited at room temperature using a Bioanalytical Systems (BASi) epsilon potentiostat with a three-electrode system featuring an Ag/AgCl reference electrode (BASi) and a platinum wire auxiliary electrode. Gold apertures on the glass microchips served as the working electrode and the biosensors were deposited using 50 mM HAuCl4 (Sigma-Aldrich) using D.C. potential amperometry at 0 mV for 30 seconds. For current density calculations, cyclic voltammetry (CV) scans were performed on each electrode from 0 mV to 1500 mV at 100mV/s in 50 mM H2SO4 (Sigma-Aldrich). The area under the reduction peak (400 mV) and the conversion factor (424 µC/cm2) was used to calculate the area of each electrode.

Biosensor functionalization:

PNA probe solution (0.85 µM PNA probe (PNABio), 0.15 µM MCH (Sigma-Aldrich), 10 mM Tris-HCL pH 7.5 (Invitrogen), 1 mM TCEP (Sigma-Aldrich), 100 mM NaCl (Sigma- Aldrich), 1 mM EDTA (Sigma-Aldrich)) was placed on freshly prepared gold electrodes (circular, square, or line apertures; 20 µL probe solution volume) in a humidity chamber and the deposition was allowed to occur overnight at room temperature. Electrodes were thoroughly washed with dH2O then backfilled with 1 mM MCH (Sigma-Aldrich) for at least 2 hours. Electrodes were thoroughly washed before proceeding to hybridization experiments. The PNA probe sequences are summarized in Table 5.

Hybridization

Hybridization reactions were carried out using 20 µL of the various targets (synthetic DNA, purified RNA, and tissue lysate) in 1X SSC (Sigma-Aldrich) for 15 or 30 minutes at 37°C in a humidity chamber. Following hybridization, electrodes were washed thoroughly and prepared for electrochemical measurements. Synthetic DNA oligonucleotides representing the complementary sequence to the respective PNA probes were ordered from Integrated DNA Technologies (IDT).

Electrochemical measurements

7.5 and electrochemical signals were measured in a solution containing 100 µM 3- [Fe(CN)6] (Sigma-Aldrich) and 10 mM Tris-HCl pH 7.5. Differential pulse voltammetry (from 0 mV to -400 mV) was used to measure the current generated from the reduction 3+ of adsorbed [Ru(NH3)6] at the surface of the electrode.

RNA Dot Blot

Total RNA was purified from A549 cells (human non-small lung cell carcinoma; ATCC) using a QIAGEN RNeasy mini kit (QIAGEN) as per manufacture protocol. 300 ng or 600 ng of RNA was then cross-linked to a nitrocellulose membrane (PALL Corp.) in addition to complementary DNA (positive control) and non-complementary DNA (negative control). Synthetic DNA was purchased from Integrated DNA Technologies (IDT). Membranes were then blocked with PBS + 5% BSA and washed with PBS + 0.1% Tween-20. Membranes were incubated with biotinylated probe sequences and developed using enhanced chemiluminescence (ECL) (Thermo Scientific) on a Versadoc System (BioRad).

Lung tissue collection and processing

A biopsy sample (about 2 cm x 1 cm) was obtained from a peripheral part of the donor lung with a mechanical stapler at the end of the cold ischemic period (just prior to the implantation of the lung). The samples were immediately snap-frozen in liquid nitrogen and stored at -80°C until use. All patients provided informed consent and the study was approved by the institutional Human Tissue Committee and Research Ethics Board at the University Health Network, University of Toronto. A portion (~5 mm3) of the frozen tissue sections were ground in liquid nitrogen then placed in lysis buffer (2% Triton-X, 2% NP-40; (Sigma-Aldrich)) with RNase inhibitor (Superase-In (Sigma-Aldrich)) for 5 to

10 minutes. Lysate was subsequently centrifuged at max speed for 3 minutes or syringe filtered and the supernatant was placed on the electrodes (1X SSC (Sigma- Aldrich)).

Mathematical modeling

MATLAB software was used to mathematically model the current generated by the FraCS as a function of DNA concentration (parameters and values summarized below). For a reactant confined to surface of the electrode (ruthenium) the peak current is given by (187):

n2F 2Γ v i = T p 4RT

Where i is the peak current [A], n is the number of electrodes in the redox process, F p is Faradays constant, is the total amount of redox molecule at the electrode surface ΓT [mol], is the scan rate [V s-1], R is the gas constant, and T is temperature [K]. is v ΓT approximated by estimating the total amount of molecules that have reached the surface by diffusion using the equations described by Sheehan et al. (188). The saturation current occurs when surface of the sensor is fully saturated with target molecules. This occurs when a target molecule is bound to each available probe. The maximum number of ruthenium at a sensor is thus given by:

A ΓTmax = ΓPεβ

Where A is the sensor area, ε is the hybridization efficiency, β is the ruthenium hexamine molecules per bound target molecule, and ΓP is the probe density. We assumed that the number of ruthenium molecules bound per target molecule, β, is 20. The parameters used for modeling were: target length (20 bp), spot diameter (15 µm),

line size (10 x 100 µm), probe density (8.7 x 1012 cm-2) (160), hybridization efficiency (30%) (160), Ru/Fe turnover (45X) (170).

Statistical analysis

Statistical calculations and analysis were carried out using Prism 6 (GraphPad) and SPSS (IBM) software. For all statistical calculations, a P-value of less than 0.05 was considered statistically significant.

Chapter 3 Using the FraCS Platform to Predict Lung Transplant Outcomes

Chapter 3 contains materials from the manuscript:

3 Using the FraCS Platform to Predict Lung Transplantation Outcomes

3.1 Abstract

Biomarker profiling is being rapidly incorporated to improve the precision of clinical decision-making in many areas of modern medical practice. This potential improvement has not been transferred to the practice of organ assessment and transplantation due to the fact that previously developed biomarker techniques require an extended period of time to perform, making them useless in the time-sensitive organ assessment process. In this study, we employed our highly quantitative and rapid sensors, fractal circuit sensors (FraCS), to analyze tissue levels of mRNA that correlate with the quality of a donor lung and post-transplant function in humans. We demonstrated that a FraCS- model can be used to predict (AUC = 0.82), with excellent sensitivity (74%) and specificity (91%), the incidence of primary graft dysfunction (PGD) for a donor lung, thus delivering a key predictive value test that could be applied to enhance patient outcomes. This study demonstrates the value of integrating rapid, point-of-care diagnostics into the donor lung assessment process.

3.2 Introduction

Lung transplantation (LTx) is a well-established lifesaving therapeutic option for patients with end-stage lung disease. The procedure has developed rapidly since its inception in the 1980s (189, 190), and clinical outcomes are continually improving (191). Unfortunately, despite best application of current clinical selection and preservation practices, life-threatening primary graft dysfunction (PGD) occurs in up to 30% of all

lung transplant recipients (20, 22, 192-194). PGD accounts for one third of the 30-day mortality following LTx and is associated with an eight times higher mortality rate compared to non-PGD LTx cases (20, 22, 193). PGD is graded from low (0/I) to high (III), with PGD III being associated with significantly poorer patient outcomes.

Because of the fear of PGD, surgeons performing LTx naturally err on the side of caution when selecting suitable donor lungs. This conservative selection practice results in use of only approximately 15% of available donor lungs (13, 195), yet studies show that the utilization rate could be much higher (8, 86). At the same time, a growing number of patients require LTx, and thus there is a high mortality rate (25%) among patients on the lung transplant waitlist (191). Combined, these factors contribute to the growing shortage of donor lungs (196), and the supply-demand gap continues to widen.

The underutilization of this precious resource therefore motivates the development of strategies to improve the classification of donor lungs. Ideally, a classification assay would accurately predict outcomes such as PGD and lung suitability, thereby expanding the usable organ donation pool (7). Currently, lungs donated for transplant are assessed by analysis of lung function, bronchoscopic findings and chest radiographs. These reveal abnormalities such as hypoxia, atelectasis, hemorrhagic contusions, edema, and pulmonary infiltration (19). Unfortunately, these metrics do not accurately, specifically or reliably reflect the injury status of the lung and therefore cannot predict post-transplant lung function or the risk of PGD.

The demands on the rapidity of an improved classification assay are extremely high. For every additional hour over which a lung is stored, graft survival may be further compromised. As a result most surgeons work to stay within a time window for donor

lung assessment of less than 6 hours (197). Ideally a useful classification assay must provide robust and actionable information in much less than this 6-hour timeframe.

With the goal of increasing the availability of precise information relevant to clinical- decision making, an intensive search is underway for molecular biomarkers predictive of LTx outcome. Several studies have identified LTx mRNA biomarkers with expression levels that correlate with post-transplant outcome (15, 86). Using microarrays for initial discovery and subsequent polymerase chain reaction (PCR)-based analysis for confirmation, we have demonstrated that the IL-6/IL-10 mRNA ratio is predictive of lung transplant survival (86). We subsequently demonstrated and validated that a number of additional markers including ATP11B appear to predict the risk and severity of PGD (15).

Having designed a class of nanosensors, fractal circuit sensors (FraCS), that enable quantitative analysis of multiple biomarkers present at relevant concentrations in human lung tissue, we sought to clinically validate a set of outcome-specific predictive genes. Furthermore, with the goal of developing a simple, point-of-care prognostic tool, we investigated the combined effects of several predictive biomarkers in a novel, outcome- driven model of assessment.

The prognostic tool described herein represents proof-of-principle for a personalized medicine approach in the time sensitive transplant surgical environment. The clinical validation of the FraCS systems and the biomarkers used for analysis highlights the value in these novel classes of diagnostic devices. Importantly, this technique has broad implications and could be generalizable to the evaluation of other organs for transplant such as the heart, liver, or kidneys.

3.3 Results

Biological verification of FraCS by quantitative PCR

The LTx-assessment chip enabled by FraCS rapidly assesses predictive mRNA biomarkers in donor lungs from a lung tissue biopsy performed by the surgeons, relying on a simplified sample preparation step to release biomarkers into solution, thus essentially requiring no sample purification (Figure 20).

< 20 min

Lung Tissue mRNA Chip-based detection & analysis

Figure 20. Lung transplant assessment assay. Lung assessment assay workflow. A biopsy is taken from a donated lung, and the cells are homogenized and lysed. mRNA released from the cells is analyzed using a chip based method that delivers a gene expression profile predictive of the outcome of a transplant within 20 minutes of the biopsy.

FraCS made from electrodeposited gold were templated in apertures created in a SU-8 layer patterned over gold contacts on a glass chip. Electrodeposition was performed

using conditions facilitating rapid growth and the generation of spiky, fractal structures extending into solution (Figure 21, SEM images (right panels)). The linear apertures were 5 x 100 microns, a size that was selected to yield sensors with large amounts of surface area (Figure 21). The electrochemical currents collected at these sensors and used for nucleic acid quantitation are generated by a label-free, redox-active, 3+ 3- [Ru(NH3)6] /[Fe(CN)6] reporter system (Figure 5).

20 µm!

3 µm!

Figure 21. Fractal Circuit Sensors (FraCS). A FraCS sensor chip (left) and SEM images of sensors following electrochemical deposition (right). Scale bars are indicated on each SEM image.

We then sought to validate the FraCS system with the current gold standard in for quantitative gene expression profiling – quantitative PCR (qPCR). Twenty-three donor lungs were tested on both the FraCS and qPCR platforms and compared for the expression patterns in PGD 0/I and PGD III+ donor lungs. For each of the genes tested

(IL-6, IL-10, EGLN1, and ATP11B), we observed the same distribution pattern for both FraCS (left panel) and qPCR (right panel) in these donor lungs (Figure 22). To further this observation, we confirmed a significant correlation of the signals from the donor lungs obtained using either FraCS or qPCR for IL-6, IL-10, and ATP11B (Figure 23a-c). EGLN1 displayed a negative correlation (r = -0.20) between the two tests as expected; however, this finding was not significant (Figure 23d). Bland-Altman method analysis of the results obtained using FraCS and qPCR further confirmed the corroboration of both tests as the mean difference between the two tests was zero and there was no signal

a IL-6 b IL-10

1.5 qPCR Relative Expression 0 qPCR Relative Expression 0 1.5 -5 1.0 -5 1.0 -10

0.5 -10 0.5 -15 FraCS PCR FraCS PCR

FraCS Normalized Current Normalized FraCS 0.0 Current Normalized FraCS 0.0 -20

PGD0/I PGDIII+ PGD0/I PGDIII+ PGD0/I PGDIII+ PGD0/I PGDIII+

c EGLN1 d ATP11B

1.0 -1 qPCR Relative Expression 2.0 -1 qPCR Relative Expression

0.8 -2 -2 1.5 0.6 -3 -3 1.0 0.4 -4 -4 0.5 0.2 -5 -5 FraCS PCR FraCS PCR

FraCS Normalized Current Normalized FraCS 0.0 -6 Current Normalized FraCS 0.0 -6

PGD0/I PGDIII+ PGD0/I PGDIII+ PGD0/I PGDIII+ PGD0/I PGDIII+

Figure 22. FraCS vs. Gold Standard, qPCR. Comparison of the FraCS assay response (left y-axis) to qPCR expression levels (right y-axis) of the same biopsy (PGD 0/I (n=11), PGD III+ (n=12)) run on both platforms for (a) IL-6, (b) IL-10, (c) EGLN1, and (d) ATP11B.

bias present in either test (Figure 24 and Table 7) indicating that our FraCS chip-based approach yielded comparable data to qPCR.

IL-6 a b IL-10 5 3.6 r = -0.77, p=0.042 r = -0.87, p=0.012 4 3.4 3 3.2 2

3.0

qPCR Expression qPCR 1 qPCR Expression qPCR

0 2.8 0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 2.0 FraCS Expression FraCS Expression

ATP11B EGLN1 c d 4 5 r = -0.89, p=0.007 r = -0.20, p=0.674 4 3 3 2 2 1

qPCR Expression qPCR Expression qPCR 1

0 0 0 1 2 3 0.0 0.5 1.0 FraCS Expression FraCS Expression Figure 23. qPCR validation. Correlation of signals obtained from qPCR versus FraCS in individual donor lungs (n=7) for (a) IL-6, (b) IL-10, (c) ATP11B, and (d) EGLN1 sensors. Data were analyzed by two-tailed, Pearson correlation.

IL-6 IL-10 ATP11B a b c 2 2 2

1 1 1

0 0 0 Difference Difference Difference -1 -1 -1

-2 -2 -2 0.5 1.0 1.5 2.0 0.5 1.0 1.5 2.0 0.5 1.0 1.5 2.0 2.5 3.0 Average Average Average Figure 24. Bland-Altman analysis of FraCS vs. qPCR. Bland-Altman method comparison (difference versus average) of signals obtained from individual donor lungs (n=7) run on both the FraCS and qPCR platforms for (a) IL-6, (b) IL-10, and (c) ATP11B.

Table 7. Bland-Altman results of FraCS vs. qPCR. Mean Difference Mean 95% Limits of Slope Biomarker Significantly Different Difference Agreement Significant ? From 0 ? IL-6 -0.02 No (p=0.84) -0.58 to 0.53 No (p=0.42) IL-10 -0.21 No (p=0.06) -0.68 to 0.26 No (p=0.52) ATP11B 0.088 No (p=0.61) -0.75 to 0.92 No (p=0.60)

Clinical validation of LTx predictive biomarkers

We examined the expression profiles of IL-6, IL-10, IL-6/IL-10 ratio, EGLN1, and ATP11B in 39 distinct donor lungs that were not previously used in biomarker discovery. The population was separated based on outcome (PGD 0/I and PGD III+) (Figure 25).

We observed a significant difference between the populations for IL-6 (Figure 25a), the IL-6/IL-10 ratio (Figure 25c), and ATP11B (Figure 25d), but not IL-10 (Figure 25b). As expected from the correlational data (Figure 23d), we did not observed a significant difference in the expression levels of EGLN1 when the donor lungs were separated on

IL-6 IL-10 a b 1.5 2.5

2.0 1.0 1.5

1.0 0.5 0.5 Normalized Current Normalized Current

0.0 0.0

PGD0/I PGDIII+ PGD0/I PGDIII+ IL-6 / IL-10 ATP11B c d 3 2.0

1.5 2

1.0

1 0.5 Normalized Current

Normalized Ratio IL-6/IL-10 0 0.0

PGD0/I PGDIII+ PGD0/I PGDIII+

Figure 25. Relative expression of LTx biomarkers. Each circle represents the LTx biomarker signal normalized to GAPDH for an individual donor lung and horizontal lines show the population means of PGD 0/I (n=23) and PGD III+ (n=16) donor lungs for (a) IL-6, (b) IL-10, (c) IL-6/IL-10, and (d) ATP11B. Data were analyzed by a two- tailed Mann-Whitney test. The P-values of each comparison are: IL-6: 0.0007, IL-10: 0.9833, IL-6/IL-10: 0.0027, and ATP11B: 0.0002.

the basis of PGD status (Figure 26). To validate the ability of the LTx biomarkers to predict PGD in transplanted lungs, we performed receiver-operating characteristic (ROC) curve analysis and determined a significant area under the curve (AUC) for IL-6 (0.74) and ATP11B mRNA (0.84), and the IL-6/IL-10 ratio (0.78), but not IL-10 mRNA (0.53) as a predictor alone (Table 8).

EGLN1 1.0

0.8

0.6

0.4

0.2 Normalized Current

0.0 0/I III+ PGD Grade

Figure 26. EGLN1 expression in donor lungs. EGLN1 expression in donor lungs separated based on patient outcome. Each circle represents the LTx EGLN1 signal normalized to GAPDH for an individual donor lung and horizontal lines show the population means of PGD 0/I (n=23), PGD III+ (n=17). Data were analyzed by a two- tailed Mann-Whitney test. The P-values of the comparison is 0.4589.

Table 8. PGD-predictive value of LTx biomarkers.

Biomarker Area under ROC curve p-value

IL-6 0.74 0.0381 IL-10 0.53 0.8177 IL-6/IL-10 0.78 0.0160 ATP11B 0.84 0.0003

Development of the FraCS Prediction Model (FPM)

Knowing that the FraCS sensor chip could provide rapid assessment of PGD biomarkers, we sought to develop a predictive model that incorporated all of the relevant biomarkers. Using the profiles from 32 donor lungs (9 PGD III, 23 PGD 0) in a development group, we performed logistic regression analysis on all marker combinations and determined that a model, referred to as the FraCS Prediction Model (FPM), consisting of the FraCS output for the IL-6/IL-10 ratio, IL-6, and ATP11B was the most predictive with an AUC of 0.96 (p < 0.0001) (Table 9). Using 100 rounds of 9-fold cross-validation with stratification, we estimated the model to perform with an AUC of 0.88 on future samples.

We then validated this model on 20 new, previously untested, samples (10 PGD III), and obtained an AUC of 0.87 (p < 0.0052) (Table 9). The FPM model was refit to all 52 samples to generate a final set of coefficients. This final FPM model had an AUC of 0.88 (p < 0.0001) representing testing on all 52 samples. Cross validation of the final FPM was carried out using 100 rounds of stratified 10-fold cross-validation, which confirmed the significant predictive value of the FPM (AUC: 0.82) (Table 9).

Furthermore, using a cutoff of -0.3575 for the FPM (Figure 27), we observed a diagnostic sensitivity and specificity of predicting PGD of 74% (14/19) and 91% (30/33) respectively (Table 10). Thus, the FPM resulted in positive and negative predictive values (PPV and NPV) of 82% and 86% respectively for the probability of identifying PGD III (poor outcome after transplant) in a donor lung (Table 10).

Table 9. The FraCS Prediction Model (FPM).

Group Area under ROC curve p-value

Development group 0.96 <0.0001 (n = 32, PGD 0/I: n = 23, PGD III: n = 9)

Cross-validation of development group 0.88 <0.05

Validation group 0.87 0.0052 (n = 20, PGD 0/I: n = 10, PGD III; n = 10) All cases 0.88 <0.0001 (n = 52, PGD 0/I: n = 33, PGD III: n = 19)

Cross-validation of all cases 0.82 <0.05

*The FPM was developed by stepwise logistic regression analysis and is expressed by the following equation: log (PGDIII : PGD0/I) = -5.5669 – 0.0484*IL-6 + 2.4370*ATP11B + 2.0469*IL-6/IL-10 e

0 FPM Score FPM -2

-4

PGD 0/I PGD III

Figure 27. FraCS Prediction Model (FPM). FPM values for donor lungs separated based on transplant outcome (presence or absence of PGD). Each circle represents the FPM value for an individual donor lung and horizontal lines show the population means of PGD 0/I (n=33), PGD III (n=19). The dashed line represents a theoretical cutoff value (-0.3575) for the FPM to predict the presence of PGD in a donor lung.

Table 10. Diagnostic characteristics of the FPM.

Sensitivity Specificity PPV NPV

FPM 73.7% (14/19) 90.9% (30/33) 82.4% 85.7% The diagnostic sensitivity, specificity, positive predictive value, and negative predictive value were calculated using -0.3575 as a cutoff in the FPM to distinguish PGD 0/I and III outcomes.

3.4 Discussion

Current decision making for donor lung selection in LTx is based on clinical evaluation of lung function that is largely qualitative (chest radiographs, bronchoscopy, pulmonary secretions, arterial blood gases). Accurate, informative and predictive genomic and proteomic level analysis of donor lungs would prove to be extremely beneficial, yet current techniques prove limiting because of the poor suitability of the required sample- to-answer timeframes in the transplant setting.

The discovery of FraCS and subsequent validation of this approach to meet the demands of a quantitative and timely diagnostic tool for use in donor lung assessment further prompted our investigations into the clinical validation of the platform. With the technical validation of the LTx-specific probes: IL-6, IL-10, ATP11B, and EGLN1, and the ability to obtain timely, gene-related signals in unpurified biopsies, the FraCS platform was compared to the current best-practice in gene analysis. Because PCR is perhaps the most commonly used technique to profile gene expression patterns, it was imperative that our LTx assay performs equally as well as this gold-standard test. In order to validate the LTx assay, we tested the same set of donor lungs on both systems (Figure 22) and found that the distribution of expression patterns were consistent between both platforms. Furthermore, each of our tested genes correlated strongly with

the qPCR results (Figure 23) and when compared using Bland-Altman method analysis, we determined that there was no significant difference between the two tests (Figure 24 and Table 6). In terms of timing the FraCS assay was able to provide a sample-to- answer from a tissue biopsy in less than thirty minutes whereas the qPCR procedure was completed in approximately five hours (combined time of RNA purification (30 minutes), cDNA synthesis (120 minutes), and qPCR (150 minutes)). Together, these data demonstrate that not only does the LTx assay report the same gene-specific results as qPCR, it outperforms the test with respect to easy-of-use and timing.

Having validated the FraCS platform both technically and biologically, we were then able to use it to confirm the biomarker discovery findings first reported several years ago using microarrays and qPCR (15, 86). These studies were the first to identify IL-6, IL-6:IL-10, ATP11B, and EGLN1 as potential predictive genes of PGD and ischemia- reperfusion injury (15, 86). By separating the tested donor lungs based upon the known outcome for each biomarker, we observed a significant upregulation of the genes (IL-6, IL-6/IL-10, and ATP11B) in the PGD III+ donor lungs, with the exception of IL-10 (Figure 25b). This finding was consistent with the previous work that also determined that IL-10 alone was not a predictor of poor outcome; however, as part of the IL-6:IL-10 ratio the metric was significant (86).

The EGLN1 biomarker was not found to be a significant predictor of outcome (Figure 26). This is not due to the validity of the marker itself; rather it is likely an inherent characteristic of the target mRNA. The signals we observed for this marker were low, which may reflect a combination of the size and expression levels of the mRNA. The EGLN1 transcript is quite large (~7100 bp) and is expressed at lower levels in lung tissue compared to the other biomarkers. Thus, we would expected that in the timeframe tested there may not have been a sufficient number of molecules

accumulated on the surface of an electrode to give an accurate electrochemical reading. Future studies directed at improving the signal from this biomarker using the LTx-assay could greatly improve the accuracy and patient outcomes for the lung transplantation process.

Since each marker (IL-6, IL-6/IL-10, and ATP11B) exhibited a statistically significant area under the ROC curve (Table 8), we anticipated that these biomarkers could be used to develop the foundation for a PGD-prediction model. It is likely that each individual biomarker itself would have limited predictive ability, as a donor lung may be consistent with a PGD 0/I outcome for some markers, but not others. By combining several biomarkers, the LTx-assay would then account for the individual response of each biomarker and report a composite profile reflecting the condition of the donor lung overall, as a function of multiple predictive biomarkers. A logistic regression model, the FPM, confirmed previous results that demonstrated that the signal from IL-6, ATP11B, and the ratio of IL-6/IL-10 were highly predictive of PGD. Both cross-validation and an independent validation group of donor lungs confirmed the predictive value of the assay and the FPM (Table 9). Furthermore, this model displayed a high degree of sensitivity and specificity for detecting PGD III in donor lungs (Table 10). Ultimately, this model represents a means by which the raw gene expression output from the LTx-assay biomarkers can be converted into a simple numerical score with a cutoff (-0.3575) that could successfully predict donor lung outcomes (Figure 9) – donor lungs above the cutoff are indicative of PGD III (PPV 82%) and those below the cutoff are likely PGD 0/I donor lungs (NPV 86%) (Table 10). Having a composite numerical score with the FPM will help to facilitate ease-of-use for surgeons by limiting the analysis and interpretation of the data that the transplant teams would have to perform during the time-sensitive transplantation process. The NPV of the FPM (NPV 86%) represents a significant advance over current clinical practice, as all of the lungs used to develop the FPM were assessed as clinically suitable and subsequently transplanted; unfortunately, 19 of 52

(36.5%) patients developed PGD III (NPV 63%). Thus, use of the FPM could provide an improvement of greater than 20% in NPV based on the results of this study. The FPM coefficients and desired cutoff will ultimately change as a larger number of donor lungs will need to be tested in order to improve the accuracy of the FPM; however, with the LTx-assay and FraCS, we have demonstrated the possibility for this type of analysis and profiling to be performed in real time, directly in the operating rooms.

3.5 Conclusions

In conclusion, the FraCS platform was used to biologically validate several important genes predictive of primary graft dysfunction in lung transplantation. We demonstrated that the FraCS approach was superior to the traditional qPCR platform and, in particular, is able to provide a rapid sample-to-answer turnaround that was unattainable with traditional qPCR. Through the rigorous clinical validation of FraCS, we developed a prediction model that accurately predicted the incidence of PGD in a given donor lung with a high degree of sensitivity and specificity. Future studies focused on implementing this profiling approach into transplant procedures and observing patient outcomes will be needed to further validate our results. With these results, we have demonstrated the possibility of a practical, personalized medicine approach in lung transplantation, which will greatly enhance donor lung utilization and patient outcomes.

3.6 Materials and Methods

FraCS preparation

contact lithography was used to pattern the sensing electrodes and followed by Au and Cr wet etching steps and removal of the positive photoresist etchant mask. SU-8 2002 (negative photoresist) (Microchem Corp.) was then spin-cast (4000 rpm, 40 s) and patterned using contact lithography to create the linear apertures for FraCS (5 × 100 µm). Microchips were diced in-house using a standard glasscutter and washed with acetone (Caledon Labs), isopropyl alcohol (Caledon Labs), and then O2 plasma etched using (Samco RIE System (Samco)). FraCS were electrodeposited at room temperature using a Bioanalytical Systems (BASi) epsilon potentiostat with a three- electrode system featuring an Ag/AgCl reference electrode (BASi) and a platinum wire auxiliary electrode. Apertures on the glass microchips served as the working electrode and the biosensors were deposited using 50 mM HAuCl4 (Sigma-Aldrich) using D.C. potential amperometry at 0 mV for 30 seconds.

FraCS functionalization and hybridization

PNA probe solution (0.85 µM PNA probe (PNABio), 0.15 µM MCH (Sigma-Aldrich), 10 mM Tris-HCL pH 7.5 (Invitrogen), 1 mM TCEP (Sigma-Aldrich), 100 mM NaCl (Sigma- Aldrich), 1 mM EDTA (Sigma-Aldrich)) was placed on freshly prepared FraCS (20 µL probe solution volume) in a humidity chamber and the deposition was allowed to occur overnight at room temperature. Electrodes were thoroughly washed with dH2O then backfilled with 1 mM MCH (Sigma-Aldrich) for at least 2 hours. Electrodes were thoroughly washed before proceeding to hybridization experiments. Hybridization reactions were carried out using 20 µL of tissue lysate in 1X SSC (Sigma-Aldrich) for 15 or 30 minutes at 37°C in a humidity chamber. Following hybridization, electrodes were washed thoroughly and prepared for electrochemical measurements.

Electrochemical measurements

All electrochemical measurements were performed on a Bioanalytical Systems (BASi) epsilon potentiostat with a three-electrode system featuring a Ag/AgCl reference electrode (BASi), a platinum wire auxiliary electrode, and the biosensing electrode serving as the working electrode. PNA-modified FraCS were incubated in 100 µM 3+ [Ru(NH3)6] solution (Sigma-Aldrich) for 60 seconds then rinsed in 10 mM Tris-HCl pH 7.5 and electrochemical signals were measured in a solution containing 100 µM 3- [Fe(CN)6] (Sigma-Aldrich) and 10 mM Tris-HCl pH 7.5. Differential pulse voltammetry (from 0 mV to -400 mV) was used to measure the current generated from the reduction 3+ of adsorbed [Ru(NH3)6] at the surface of the electrode.

Quantitative polymerase chain reaction (qPCR)

For RNA analysis, total RNA was purified from lung tissues using a QIAGEN RNeasy mini kit (QIAGEN) according to the manufacturer’s protocol. For synthesis of cDNA (20 µL) from 0.5 µg of total RNA, High Capacity cDNA reverse transcription kit (Applied Biosystems) was used according to the manufacturer’s instructions on an Eppendorf Vapo-protect thermocycler. qPCR reactions were performed on a Applied Biosystems 7500 real-time PCR system (Applied Biosystems) using 2 µL of cDNA, 12.5 µL of SYBR® Green PCR Master Mix (2X) (Applied Biosystems) and 200 nM of gene primers for a total volume of 25 µL. Conditions used for qPCR included an initial step of 15 s at 95°C followed by 40 cycles of 15 s at 95°C and 1 min at 60°C. The gene expression levels of target genes were normalized to the level of GAPDH. Linear regression analysis was used to compare the signals obtained from lung biopsies run on FraCS and qPCR. The equation of the regression line was used to transform qPCR values to FraCS-based signals for subsequent Bland-Altman analysis.

Lung tissue collection and processing

FraCS clinical validation

Signals for lysate samples were GAPDH-normalized and FraCS-current results were plotted using GraphPad Prism 6 and donor lungs were divided into PGD 0/I (n = 23) or PGD III+ (clinically unsuitable for transplant (n = 7) or PGD III after transplant (n = 9)) outcomes. Population means were calculated and a Mann-Whitney U test was used to determine statistical differences in the various groups. For the PGD-prediction model only the PGD 0/I (proven to be non injured by transplant) and PGD III donor lungs (proven to have severe injury by transplant) were used in the calculations (n = 23, 9 respectively). Logistic regression analysis on the currents generated for each individual biomarker was carried out in all combinations of measured markers and the IL-6/IL-10 ratio in order to develop the FraCS Prediction Model. For validation of the FPM, an additional unique set of PGD 0/I (n = 10) and PGD III (n = 10) donor lungs were tested using FraCS and the FPM fit. Cross-validation of the FPM was carried out using 100 rounds of 10-fold (for the complete data set; 9-fold for the development set only) cross

validation with stratification. Diagnostic sensitivity, specificity, NPV, and PPV for the FPM were calculated using the PGD III lungs as true positives and PGD 0/I lungs as true negatives with a cutoff value of -0.3575.

Statistical analysis

Chapter 4 The Development of an Electrochemical Assay for Small Protein Targets using TGFβ1

4 The Development of an Electrochemical Assay for Small Protein Targets Using TGFβ1

4.1 Abstract

Protein biosensors represent an important class of point-of-care diagnostic tools that could be used to positively impact lung transplant outcomes. Significant research has shown that a number of protein targets are predictive of complications associated with the long-term survival of transplant patients. The use of traditional protein detection methods such as an ELISA are not feasible within the transplant timeframe and thus, novel detection assays are required. In this study, we demonstrate the basis for an electrochemical TGFβ1 detection assay. TGFβ1 is an important mediator of lung fibrosis – a key development in chronic lung allograft dysfunction. We demonstrate strategies for the detection of TGFβ1 using a sandwich ELISA–like assay to overcome the small size of the TGFβ1 protein (25-kDa). Furthermore, we show how the use of electrochemistry can overcome the time-intensive washing and subsequent analyte addition steps and can result in the rapid (~60 minute) detection of TGFβ1 in lung transplant perfusate. This work provides the foundation for the development of biosensors the measure small, hard-to-target proteins. The TGFβ1 assay will help to spur the development of other important transplant-relevant protein assays, which will significantly impact the field of transplantation.

4.2 Introduction

Protein biomarkers play an important role in determining the quality of a transplanted organ and the monitoring of allograft function post-transplant. A cytokine profile of a donor lung can not only predict the success of that organ, it can guide surgical teams and clinicians to appropriate therapeutic interventions to improve the injury status of the

lung. Several cytokines have been identified in the transplant process as predictive of patient outcomes, IL-8 (198), or targets for organ treatment and repair, IL-10 (89). Other important biomarkers, including transforming growth factor beta 1 (TGFβ1), have been shown to affect transplant outcomes, but lack further characterization as potential biomarkers due to an absence of practical protein detection assays.

TGFβ1 is a 25-kDa protein produced by most cells in the human body and plays an important role in cellular growth and development (199). As a natural growth factor, TGFβ1 promotes the division and proliferation of cells within the body. Although this functionality of TGFβ1 is required in normal development, a disease state can arise as a consequence of TGFβ1 signaling. Idiopathic pulmonary fibrosis (IPF) and restrictive allograft syndrome (RAS) are two severe complications of lung function with a similar disease manifestation (200). Aberrant proliferation of human lung fibroblast cells can lead to the obliteration of lung airways and fibrosis of the alveoli (32). The diagnosis of RAS occurs in approximately 25-35% of all chronic lung allograft rejection patients and imparts a very poor prognostic outlook for a patient (approximately 12 months) (32). Several treatment options for RAS have been proposed, including the use of pirfinidone as it influences the production of TGFβ1 (201, 202). Thus, by monitoring the levels of TGFβ1 in a donor lung, surgical teams will be better equipped to: (i) identify early onset fibrosis and (ii) monitor the effectiveness of therapeutic interventions, such as pirfinidone treatment, which can be directed towards controlling the fibrotic process.

TGFβ1 is produced in an inactive form and requires several processing steps prior to biological activation. TGFβ1 is a member of the large latent TGFβ complex (LLC) alongside TGFβ propeptide and latent TGFβ binding proteins (203). The propeptide form of TGFβ is termed the latency-associated protein (LAP) and the non-covalent association of LAP with TGFβ1 prevents signaling from TGFβ1. The activation of

TGFβ1 requires the dissociation of LAP from TGFβ1 and, as a result, levels of dissociated LAP are an indirect measure of the active from of TGFβ1 (204). Thus, diagnostic assays directed at monitoring TGFβ1 levels can measure activated TGFβ1 itself or use LAP as a surrogate for TGFβ1 (205). Importantly, for assays that are primarily based on relative size of reporters, LAP is a better target as it has a mass of approximately 50-kDa.

The discovery of novel protein biomarkers will lead to significant improvements in the transplantation process, but only if there exists technology capable of reporting the clinical levels of these biomarkers in a non-disruptive and rapid manner. To this end, the current platforms used in protein biomarker discovery (ELISA, flow cytometry, SPR), are not easily integrated into the transplant setting. These traditional techniques require bulky, labor-intensive equipment, sample-processing/purification steps, and often produce results outside of the allowable transplant timeframe (> 2 hours). As such, integration of protein-based biomarker monitoring in transplantation remains limited.

Our lab has developed a novel protein detection platform that has been used to rapidly identify clinically relevant levels of the ovarian cancer biomarker (CA-125) (174). This assay exploits the steric blocking effects on the redox reporter ferrocyanide due to the presence of the 125-kDa CA-125 protein on the surface of gold electrodes (174). The large size of CA-125 makes it an attractive target for this type of blocking assay; however, the ability to detect smaller proteins remains to be elucidated. Herein, we describe the development of platform for the electrochemical detection of small proteins using a ferrocyanide blocking. We demonstrate that this assay is capable of measuring LAP concentrations in transplant relevant media.

4.3 Results

TGFβ1 assay technical validation

An assay designed to detect small (< 50-kDa) proteins was proposed that was modeled after a sandwich ELISA (Figure 28). The presence of the antibody/antigen complex (“Positive Complex”) was sufficient to sterically hinder the access of ferrocyanide molecules to the surface of an electrode resulting in the attenuation of the oxidation current (dashed DPV trace, Figure 29b). Proof-of-concept testing was completed that demonstrated that the normalized current from the capture surface alone, absence of target protein (“-LAP”), and absence of secondary antibody (“Negative Complex”) were insufficient to block the ferrocyanide oxidation reaction whereas the positive complex produced a decrease in the total amount of current of about 75 nA (Figure 30).

LAP LAP

LAP 3+/4+ [Fe(CN)6] Electrochemical Scanning

Capture Negative Positive Surface Complex Complex Figure 28. Protein detection using ferricyanide/ferrocyanide blocking. Gold sensors are functionalized with a capture antibody (red, “Capture Surface”) then challenged with the protein of interest, LAP (blue), and a secondary antibody (green) to that protein. The presence of the protein alone is insufficient to sterically block the surface (“Negative Complex”) whereas the presence of the protein and secondary antibody (“Positive Complex”) is sufficiently large. Electrochemical measurements are made to measure protein binding at the electrode surface using a ferricyanide/ferrocyanide reporter system.

a b 0.0 0.0

LAP

LAP -0.5 -0.5

Negative Complex

Current (µA) Current (µA) Positive Complex Neg. Complex -1.0 -1.0 Pos. Complex 50 150 250 350 50 150 250 350 Potential (mV) Potential (mV)

Figure 29. Electrochemical scanning using the ferricyanide/ferrocyanide blocking assay. Representative electrochemical scans for the oxidation of ferrocyanide at the electrode surface for a capture antibody alone (dashed line) and in the presence of target analyte (solid line) for a (a) negative complex and (b) positive complex.

LAP

LAP LAP

Capture Positive Positive Negative Complex Surface Complex Complex

100 I (nA)

Δ 50

- LAP

Cap. Surface Neg. Complex Pos. Complex

Figure 30. Electrochemical LAP detection. Currents obtained for the oxidation of ferrocyanide at the electrode surface for the capture antibody alone (Cap. Surface), absence of LAP (-LAP), presence of LAP only (Neg. Complex), and presence of LAP and secondary antibody (Pos. Complex). Data represented are of n=20 different sensors. Columns represent mean and error-bars indicate s.e.m.

We then compared the electrochemical LAP detection method against a traditional HRP-ELISA (Figure 31). For the same concentrations of target protein, LAP, the electrochemical assay produced signals that displayed greater linearity/dynamic range (10 pg/mL to 100 pg/mL) and better sensitivity (< 10 pg/mL) (Figure 31a) when compared to the ELISA. As expected, the ELISA demonstrated the characteristic signal saturation above 1 ng/mL (Figure 31b). Taken together, these results confirmed that the biological reaction conditions (Ab/Ag binding) were conserved between the electrochemical and ELISA-based detection methods, yet the electrochemical tests proved superior in diagnostic characteristics.

a b 5 2 2 800 r =0.98 r =0.99 4 900 3 450nm 1000 2 OD OD Current (nA) 1100 1

1200 0 0.001 0.010 0.100 1 10 100 1000 0.001 0.010 0.100 1 10 100 1000

[LAP] (ng/mL) [LAP] (ng/mL)

Figure 31. LAP limit of detection (LOD) and comparison to ELISA. (a) Currents obtained for the oxidation of ferrocyanide at the electrode surface for various concentrations of LAP. Each point represents n=20 different sensors and error-bars indicate s.e.m. (b) OD 450 nm measurements for the same concentrations of LAP using an ELISA-based technique. r2 represents the goodness of fit for each graph using non-linear regression. The dashed line represents the signal obtained from a negative control sample.

We then sought to perform several assay steps concurrently in an interrogation solution (“InterrSol”). InterrSol contained the LAP secondary antibody, ferrocyanide reporter, and LAP protein (Figure 32a). By allowing the components of the InterrSol solution to hybridize with the electrodes for 60 minutes prior to electrochemical scanning, we observed the characteristic current attenuation of approximately 200 nA only in the presence of target protein (Figure 32b).

a b

Fe2+/3+

LAP t = 60 min. 2+/3+ 1.4 LAP LAP Fe

2+/3+ 2+/3+ Fe InterrSol Fe LAP 1.2 Fe2+/3+ LAP

LAP 1.0 2+/3+ 2+/3+ Fe Fe Current (µA)

0.8 Cap.Surface

Cap. Surface InterrSol - LAP InterrSol + LAP

Figure 32. Timing improvements of LAP detection using the ferricyanide/ferrocyanide-blocking assay. (a) Representation of the improved LAP assay timing by co-hybridizing LAP, secondary antibody, and the ferricyanide/ferrocyanide in the same solution (“InterrSol”) with the capture surface. (b) Electrochemical measurements made using the InterrSol approach for the capture surface, absence of LAP, and presence of 100 ng/mL LAP. Data represented are from n=20 different sensors and error-bars indicate s.e.m. The dashed line indicates the limit of detection of the LAP blocking assay.

100

Detection of TGFβ1 in transplant-relevant media

With the proof-of-concept testing and technical validation completed for LAP protein sensors, we then sought to validate these platforms in biologically relevant media, STEEN Solution. For the LAP assay, when the hybridization media was changed from PBS to STEEN, the integrity of the assay was upheld as we observed a characteristic decrease in the presence of the blocking antibody/antigen complex (“+LAP”) compared to the capture surface alone and absence of protein (“-LAP”) (Figure 33). The current attenuation (~20 nA) of the positive complex in STEEN solution (Figure 33) was slightly lower than that observed using PBS (~100 - 200 nA (Figures 30 and 32b)) indicating a loss of sensitivity in the more complex media.

60 Current (nA)

- LAP + LAP

Cap. Surface

Figure 33. Validation of LAP detection in lung perfusate (STEEN) solution. Electrochemical measurements made using the LAP blocking assay approach for the capture surface, absence of LAP, and presence of 100 ng/mL LAP in lung perfusate solution (STEEN). Data represented are from n=20 different sensors and error-bars indicate s.e.m.

101

4.4 Discussion

Whereas nucleic acid tests provide valuable insight into the genome of an individual donor lung, protein-based assays provide another key resource to the understanding of the molecular mechanisms behind current and future organ function. Cytokine assays add an additional source of biological samples compared to the tissue biopsy-based nucleic acid tests – lung perfusate. During the organ procurement process, donor fluids are flushed from the lung replaced with an acellular lung-specific media, STEEN Solution™. Precise monitoring of soluble cytokines in lung perfusate can be used to predict acute and chronic lung allograft dysfunction (62). Using a similar NME-derived electrochemical platform, we sought to develop a novel set of assays that would complement our FraCS-based nucleic acid test. Nucleic acid sensors are built upon the principles of negative charge accumulation on an electrode surface due to the presence of the negatively charged phosphate backbone. Because the amino acid components of proteins can vary dramatically and their respective charges are pH dependent, it is difficult to employ the same ruthenium-ferricyanide assay as used in nucleic acid detection. Previous research by our group has demonstrated that protein-based detection is possible using the NME platform by measuring the presence of bound proteins using the redox potential of iron (ferricyanide and ferrocyanide) (174). In this study, a ferrocyanide molecule was sufficiently small to rapidly diffuse to the surface of an electrode that had been functionalized with a capture antibody; however, the presence of a large antigen, CA-125, was sufficiently large to impede the diffusion of ferrocyanide and as a result diminish the respective oxidation of the redox reporter (174). The target of interest for the proof-of-concept testing used in this study was transforming growth factor beta (TGFβ) as it plays an important role in restrictive allograft syndrome (RAS) (200). It is believed the TGFβ plays an important role in the differentiation of stem cells and as such, maybe a key contributor to the advanced fibrotic process indicative of RAS (200). The activation of TGFβ is a multistep process and this makes active TGFβ difficult to measure in solution; however, the monitoring of

102

an integral component of the TGFβ-complex, the latency associated peptide (LAP), provides a good surrogate for activated TGFβ (204). LAP is a 51-kDa protein and this size of protein is insufficient to attenuate the ferrocyanide signal in a blocking-based electrochemical assay. To circumvent this, we created a sandwich ELISA-like complex on the surface of an electrode consisting of a capture antibody, LAP, and secondary antibody (Figure 28). Using this configuration we observed sufficient blocking only in the presence of LAP and secondary antibody (Figure 30). The gold standard for protein quantification is an ELISA. In a head-to-head comparison test, we found that our electrochemical LAP detection assay provided a better limit of detection and dynamic range over a conventional ELISA run under the same conditions (Figure 31). The inclusion of the prolonged enzymatic HRP-amplification step within the ELISA may contribute to the observed signal saturation beyond 1 ng/mL. In contrast, the electrochemical detection assay does not contain an extended signal amplification period and thus the assay is able to distinguish much higher concentrations of LAP compared to the ELISA. In order to become a feasible diagnostic tool for use within the transplant setting, the electrochemical assay must provide clinical answers in a fraction of the time of an ELISA. To overcome this challenge, we exploited the fact that the electrochemical assay takes place on the surface of the electrodes and is independent of the molecules in solution. That is, any unbound secondary antibodies remain in solution (if no wash step is present) and do not sterically hinder the diffusion of the ferrocyanide molecules to the electrode surface. Thus, we determined that we could eliminate several wash steps in addition to sequential protein, secondary antibody, electrochemical reporter hybridization steps by combining them all into a single interrogation solution (InterrSol) (Figure 32). Thus, the LAP assay time was greatly shortened (comparable to only a single-step ELISA) and only required approximately sixty minutes to perform versus the multi-hour time requirement of an ELISA.

103

Having met the technical requirements of a specific and rapid LAP detection assay, we sought to biologically validate the assay in transplant relevant media (STEEN Solution). The assay performed well when the sampling matrix was increased from PBS to the more complex STEEN (Figure 33) demonstrating the robustness of these sensors. This was anticipated as the in the nucleic acid LTx assay, the development of a robust SAM can provide excellent specificity in complex matrices and the use of STEEN solution is far less complex when compared to tissue lysate. As expected, there was a slight loss in signal/sensitivity for the assay run using the more complex STEEN solution. Previous work by our group has demonstrated that the degree of nanostructuring of NMEs can be tailored to specific sensitivity ranges (158) and, as such, future work using the TGFβ1 assay will explore improving the sensitivity. Taken together, we were able to demonstrate that this approach, using InterrSol, provided a rapid and specific assay for TGFβ1 detection in transplant-relevant media.

4.5 Conclusions

We have developed a template for the detection of small proteins with nanostructured microelectrodes using a ferricyanide/ferrocyanide-blocking assay in a rapid, sandwich ELISA-like assay. As proof-of-concept, we developed this assay using the transplant- relevant reporter TGFβ1 (LAP) and demonstrated the utility of this test in lung-specific media, STEEN solution. In the future, studies that investigate and determine prognostic levels of TGFβ1 (LAP) in lung perfusate solution will be required to validate this approach. As such, large, clinical studies that involves test a large number of donor lungs will be conducted. Enhancements to the transplant procedure, made possible by rapid protein-based diagnostics such as the work herein, will greatly improve the confidence of transplant teams when assessing donor lungs for transplant and, consequently, patient outcomes.

104

4.6 Materials and Methods

Microchip fabrication

Microchips were fabricated in-house at the Toronto Nanofabrication Centre (University of Toronto, Toronto, ON) using precoated (5 nm chromium, 50 nm gold, and AZ1600 (positive photoresist)) glass substrates purchased from Telic Company. Standard contact lithography was used to pattern the sensing electrodes and followed by Au and Cr wet etching steps and removal of the positive photoresist etchant mask. SU-8 2002 (negative photoresist) (Microchem Corp.) was then spin-cast (4000 rpm, 40 s) and patterned using contact lithography to create the 5 µm circular sensing apertures. Microchips were diced in-house using a standard glasscutter and washed with acetone

(Caledon Labs), isopropyl alcohol (Caledon Labs), and then O2 plasma etched using (Samco RIE System (Samco)).

Biosensor electroplating

105

Biosensor functionalization

10 mM cysteamine (Sigma-Aldrich) was placed on freshly prepared gold electrodes (circular apertures; 20 µL probe solution volume) in a humidity chamber and the deposition was allowed to occur overnight at room temperature. Electrodes were thoroughly washed with dH2O then incubated with 2.5% v/v glutaraldehyde (Sigma- Aldrich) for 60 minutes. Sensors were washed thoroughly and then incubated with LAP capture monoclonal antibody (MAB2461; R&D Systems) at 5 µg/mL in PBS (Invitrogen) for 60 minutes. Electrodes were thoroughly washed before proceeding to hybridization experiments.

Hybridization

Hybridization reactions were carried out using 20 µL of various concentrations of recombinant LAP (246-LP; R&D Systems) and 2 µg/mL of LAP secondary monoclonal antibody (BAM2462; R&D Systems) in PBS (Invitrogen) or STEEN Solution™ (XVIVO Perfusion) for 60 minutes at room temperature. Following hybridization, electrodes were washed thoroughly and prepared for electrochemical measurements.

Electrochemical measurements

All electrochemical measurements were performed on a Bioanalytical Systems (BASi) epsilon potentiostat with a three-electrode system featuring a Ag/AgCl reference electrode (BASi), a platinum wire auxiliary electrode, and the biosensing electrode 3- serving as the working electrode. Electrodes were incubated in 2.5 mM [Fe(CN)6] and 4- 2.5 mM [Fe(CN)6] (Sigma-Aldrich) for 30 seconds. Differential pulse voltammetry (from 0 mV to 400 mV) was used to measure the current generated from the oxidation of ferrocyanide at the surface of the electrode. For “Interrsol” experiments, recombinant

106

LAP, LAP secondary antibody, and ferricyanide/ferrocyanide were all present in the reaction solution for 60 minutes prior to DPV scanning.

Enzyme-linked immunosorbent assay (ELISA)

High-binding, 96-well Costar® plates (Corning Life Sciences) were coated with 5 µg/mL of LAP capture antibody in PBS, overnight, at 4°C. Plates were thoroughly washed then blocked with a solution of 1% (w/v) of BSA (Sigma-Aldrich) in PBS for 60 minutes at room temperature and shaken at 500 rpm. Plates were washed thoroughly and incubated various concentrations of recombinant LAP followed by 2 µg/mL of LAP secondary antibody in PBS + 0.1% BSA for 60 minutes at room temperature and 500 rpm. Plates were subsequently washed and incubated with streptavidin-HRP (Cell Signaling) for 30 minutes at room temperature and 500 rpm. Following washing, 3,3’5,5’-tetramethybenzidine (TMB) (Cell Signaling) was added to each well for 5-15 minutes and protected from light. To stop the reaction, an equal volume of 1.0 N H2SO4 was added to each well and the absorbance at 450 nm was read using a (Spectramax M2 (Molecular Devices)).

Statistical analysis

107

Chapter 5 Detection of Small Peptides – An Electrochemical Assay for Endothelin-1 Analysis

Chapter 5 contains materials from the unpublished manuscript:

Sage, A. T., Bai, X., Liu, M., Keshavjee, S., and Kelley, S.O. Using the inherent chemistry of the endothelin-1 peptide to develop a rapid assay for lung transplantation. Manuscript submitted for publication. (2015).

108

5 Detection of Small Peptides – an Electrochemical Assay for Endothelin-1 Analysis

5.1 Abstract

The detection of endothelin-1 (ET-1) in lung transplant samples would provide surgeons a valuable diagnostic tool for the prediction of lung transplant outcome. Levels of ET-1 have been shown to be increased donor lung bronchoalveolar lavage and lung perfusate samples and are indicative of a poor prognosis. Measuring ET-1 prior to transplantation could allow transplant teams to make more informed decisions as to the clinical outcomes of a lung; however, current diagnostic methods for measuring ET-1 are impractical for use in the transplant setting. Herein, we describe a novel, electrochemical detection platform for the detection of ET-1. We utilize the inherent biochemistry of the ET-1 peptide to develop a competitive ELISA-like assay that is capable of reporting ET-1 levels with a high degree of accuracy. Furthermore, we demonstrate that this approach is suitable using lung perfusate media. Thus, we have opened the possibility of rapid, ET-1 testing that can be easily integrated into the transplant setting. This assay will provide transplant teams with a greater depth of knowledge of the status of a donor lung and have a significant impact on patient outcomes.

5.2 Introduction

Lung transplantation (LTx) is a life-saving procedure for patients suffering from end- stage lung disease. At present, donor lungs are assessed for transplant suitability based on several physiological parameters including donor/organ medical history and pulmonary compliance measures. These physiological metrics do not reliably predict recipient outcomes post-transplant. Thus, the inclusion of lung-specific biomarker tests,

109

prior to transplantation, that could accurately predict LTx outcomes would be of great benefit to patients and transplant teams.

Ex-vivo lung perfusion (EVLP) is a novel technique that has been developed to improve the LTx procedure by affording more time for transplant teams to assess and treat a donor lung under normothermic conditions (7). In addition, EVLP can provide a means by which donor lungs can be treated therapeutically without the detrimental effects of the host immune system (7). As such, EVLP can allow for the discovery, validation, and monitoring of predictive biomolecules in EVLP perfusate. In studies using EVLP, circulating levels of the endothelin-1 (ET-1) peptide have been shown to be predictive of donor lung function (206).

Endothelin-1 (ET-1) is an important chemokine expressed in lung tissue and plays a key role in vasoconstriction (207) and fibroblast proliferation (208). Although the effects of ET-1 are widespread throughout the body, the production of ET-1 occurs primarily in lung tissue (209). ET-1 is released by lung endothelial cells and counteracts the vasodilative effects of nitric oxide (NO) and prostacyclin. Through the preferential binding of the ET-1 receptor in the underlying smooth muscle cells of the vasculature (210), ET-1 signaling activates phospholipase C, which promotes an increase in intracellular calcium and leads to the vasoconstrictive effects of ET-1 (211). Studies have shown that the ET-1 peptide plays a pivotal role in cardiovascular disease and, in particular, hypertension (212). Recently ET-1 has been implicated in pulmonary arterial hypertension (PAH), which is one of the leading causes of end-stage lung disease leading to transplantation (2, 3). The balance of vasoactive mediators is disrupted in PAH, which leads to chronic pulmonary vasoconstriction and subsequent hypertension (213-215). Failing response to therapeutic intervention, patients with PAH become candidates for transplantation, which will help to improve respiratory capacity and the quality of life for the patient.

110

The effects of ET-1 signaling are not limited to end stage lung disease. Post-transplant, ET-1 pathway activation can become a significant risk factor for both acute and chronic lung injury. Primary graft dysfunction (PGD) is a severe form of acute rejection and can occur in approximately 30% of LTx cases. Recent work has demonstrated strong correlation between ET-1 levels and the development of PGD through the disruption of the alveolar-capillary barrier (206). In addition to vasoconstrictive properties, ET-1 is a profibrotic mediator through the activation of PKC, Akt, and the MAP kinases (211). Fibroblast proliferation is a key contributor to the narrowing of the bronchioles and represents a major characteristic of chronic lung allograft dysfunction (CLAD) (200). Bronchiolitis obliterans syndrome (BOS) is the major form of CLAD and represents the principal cause of late graft loss (216). A diagnosis of BOS occurs in over 30% of all transplant patients after three years post-transplant and has a five-year mortality rate of approximately 60-70% (216). Recent studies have shown that mesenchymal stem cells (MSC) play an important role in the progression of fibrosis associated with BOS and that their proliferation, differentiation, and migration is governed by ET-1 signaling (217). Serum concentrations of ET-1 have been shown to be predictive of BOS (37) and, as such, ET-1 is a powerful biomarker that can be also used to predict long-term survival in transplant patients.

Unlike other lung protein biomarkers, ET-1 is a small 21 amino acid peptide. It displays a high degree of homology to the other members of the ET peptide family, ET-2 and ET- 3 (218) and contains an N-terminal cysteine residue. ET-1 is first synthesized as a 203 amino acid precursor, preproendothelin, then cleaved into a smaller 38 amino acid peptide, big-ET-1 (219). Finally, active ET-1 is produced through the activity of the endothelin-converting enzyme (ECE). Synthesis of ET-1 can be stimulated by a variety of sources including mechanical forces, hypoxia, and other cytokines (220), which may contribute to the increased ET-1 levels observed in lung transplantation.

111

Given the unique role that ET-1 plays in both lung function and transplant survival rates, it is a valuable target for clinical diagnostics. Current detection methods for ET-1 are based primarily on the enzyme-linked immunosorbent assay (ELISA) protocol (37, 206, 217). The typical workflow of an ET-1 ELISA can take upwards of 4 hours and requires significant user input. Yet, to be clinically relevant within the decision-making processes of LTx, these assays must provide sample-to-answer times of less than two hours. As such, integration of ET-1 testing into the transplant setting remains elusive. Therefore, this work sets out to develop a novel ET-1 biosensor based on electrochemical detection methods that display sensitive, automatable, and rapid readout properties. By monitoring ET-1 levels during the transplant process, a new level of long-term survival prediction can be assessed in addition to preemptive targeted therapeutic strategies that, together, will improve the quality of life for the transplant patient.

5.3 Results

ET-1 assay technical validation

By exploiting the amino acid sequence of the endothelin-1 peptide, we developed an electrochemical assay to monitor the presence of endogenous ET-1 using an approach similar to that of a competitive ELISA (Figure 34). In brief, a ferrocyanide molecule is sufficiently small to rapidly diffuse to the surface of an modified electrode; however, the presence of a large, blocking protein (ex. ET-1 antibody) can sufficiently impede the diffusion of ferrocyanide to the electrode surface (Figure 34a). As a result, the respective oxidation current of the redox reporter is diminished as shown by signal attenuation during a differential pulse voltammetry (DPV) electrochemical scan.

We technically validated the scheme by comparing the DPV scans of a sensor with no blocking, complete blocking, and partial blocking due to ET-1 peptide in solution (Figure 35a). We observed a characteristic attenuation of current in the case of complete blocking (Figure 35a, dashed line) compared to no blocking (Figure 35a solid line). As

112

expected, the case where the presence of ET-1 peptide in solution could compete for ET-1 antibodies, we observed only partial blocking (Figure 35a, dotted line). To confirm that we could quantify the results in Figure 35a, we then calculated the relative % blocking of an electrode and found approximately 85% blocking with no ET-1 peptide in solution compared to 48% blocking with ET-1 in solution (Figure 35b). A titration profile for the signal attenuation as a function of ET-1 antibody concentration was conducted and showed a good dynamic range between 0 and 10 ng/mL of antibody (Figure 36). Above 10 ng/mL, there was saturation of these sensors (1 ng/mL synthetic ET-1 peptide on electrode surface). In subsequent proof-of-concept work, we employed sensors that had a higher surface density of synthetic ET-1 peptide (up to 10 µg/mL) to limit the saturation of the sensors thus improving the dynamic range. A competitive ELISA was performed (Figure 37) to validate this detection platform. We observed an expected decrease in OD450nm with a corresponding increase in endogenous ET-1 concentration (Figure 37). In order to demonstrate the utility of this platform, we varied the size of each sensor from small to large (30 to 120 seconds electrodeposition) (Figure 38). As expected, the degree to which a sensor was blocked (% available surface) was a product of the electrode size. The largest sensors (120 seconds) exhibited limited antibody concentration dependence whereas the small sensors (30 seconds) reached saturation for low levels of antibody (1000 pg/mL) and the medium sized sensors (60 seconds) were in between (Figure 38).

113

Figure 34. ET-1 detection using ferricyanide/ferrocyanide-blocking assay. (a) Gold sensors (yellow) are functionalized using the N-terminal cysteine residue of the ET-1 peptide (blue) under reducing conditions. With no ET-1 peptide present in solution (a), ET-1 antibodies (pink) sterically hinder the electrode surface and prevent the oxidation of ferrocyanide (red). (b) The presence of endogenous ET-1 in solution (blue) binds and sequesters a population of ET-1 antibodies that would otherwise bind ET-1 on the electrode surface, thus reducing the steric hindrance of ferrocyanide electron transfer at the electrode surface.

114

a b 150 0

100 -200 + αET-1

+ αET-1 & ET-1 Delta I (%) 50 -400

Current (nA) - αET-1

-600 0 0 100 200 300 400 - + +αET-1 Potential (mV) - - +ET-1

Figure 35. Validation of ET-1 detection scheme. (a) Differential pulse voltammetry scans of currents recorded for various degrees of sensor blocking arising from: no antibody (solid line), ET-1 antibodies (dashed line), and both ET-1 peptide and antibody (dotted line) in solution. (b) Representative quantifications of DPV currents with and without antibody or ET-1 peptide present in solution. Each point represents n=20 different sensors and error-bars indicate s.e.m.

100 Available Surface (%) 120 0 0.001 0.01 0.1 1 10 [αET-1] (µg/mL)

Figure 36. ET-1 biosensor characteristics. Currents obtained (reported as % available surface) for the oxidation of ferrocyanide at the electrode surface for various concentrations of ET-1 antibody bound to surface-adhered ET-1 peptide (1 ng/mL). Each point represents n=20 different sensors and error-bars indicate s.e.m.

115

0.7 r2 = 0.75, P<0.05

0.6

450nm 0.5 OD OD 0.4

0.3 0.0 0.5 1.0 [ET-1] /µg/mL [ET-1] (µg/mL)

Figure 37. Comparison of ET-1 assay to ELISA. OD 450 nm measurements for various concentrations of endogenous ET-1 using an ELISA-based technique. r2 represents the goodness of fit using non-linear regression. The concentration of ET-1 antibody was 1 µg/mL.

0 30s 60s 120s 50

100 Available Surface (%) 150 0 1 10 100 1000 10000 [αET-1] (pg/mL)

Figure 38. Effect of sensor size on ET-1 LOD. Currents obtained (reported as % available surface) for the oxidation of ferrocyanide at the electrode surface for various concentrations of ET-1 antibody bound to surface-adhered ET-1 peptide (1 ng/mL) for sensors that were electrodeposited for 30 (triangles), 60 (inverted triangles), or 120 (circles) seconds.

116

Biological validation of the ET-1 assay using STEEN solution:

We first performed a preliminary trial using PBS where a standard titration of ET-1 antibody was run alongside two unknown concentrations of endogenous ET-1 peptide (Figure 39a). Using the same approach, we repeated the trial using STEEN solution in lieu of PBS (Figure 39b). Linear regression was used to derive an equation that represented the antibody concentration as a function of surface blocking using a standard dilution of ET-1 antibodies (equation shown in upper right-hand corner of each panel). We then calculated a theoretical antibody concentration from the observed surface blocking effects of the unknown samples (x1 and x2) (Table 11). For the PBS trial, we observed 72% and 56% surface blocking for x1 and x2 compared to 52% and 31% in STEEN solution (Table 11). By comparing the calculated antibody concentration to the actual, known loaded antibody concentration we were able to estimate the endogenous ET-1 concentration in the spiked samples. The calculated

ET-1 concentrations were 528 and 194 ng/mL and 481 and 241 ng/mL for x1 and x2 in PBS and STEEN solution respectively (Summarized in Table 11). We observed an average of 14% error in the PBS assay and 3.7% in the spiked STEEN assay for the difference between the experimentally derived ET-1 concentrations versus the actual ET-1 concentrations (500 and 250 ng/mL) in solution.

117

a b 00 0 y y= =- -84.6 48.0 x x+ + 95.6 94.7 y = - 84.6 x + 95.6

5050 50 Available Surface (%) Available Surface (%) Available Surface (%) x x 100100 x11 x2 2 100 x1 x2 0.00.0 0.50.5 1.01.0 0.0 0.5 1.0

[ α[αET-1]ET-1] (/µg/mLg/mL) [αET-1] (µg/mL)

Figure 39. Endogenous ET-1 detection. Currents obtained (reported as % available surface) for the oxidation of ferrocyanide at the electrode surface for the detection of endogenous ET-1 in (a) PBS and (b) STEEN solution. The equation for the standard curve (open circles, solid line) is shown in the upper right quadrant of each graph. The dashed lines represent the observed % available surface for two unknown ET-1 peptide concentrations, x1 and x2, extrapolated to theoretical anti-ET-1 concentrations.

Table 11. Estimated endogenous ET-1 concentrations. [αET-1] [ET-1] [ET-1] Unknown % Available Buffer Calculated Calculated Actual Sample Surface (µg/mL) (ng/mL)* (ng/mL)

x1 72.0 0.472 528 500 PBS

x2 56.0 0.806 194 250

x1 51.7 0.519 481 500 STEEN

x2 31.4 0.759 241 250

*Calculated [ET-1] is obtained from the following equation: [ET-1]calc. = [αET-1] added – [αET-1] calculated where [αET-1] added is 1 µg/mL and [αET-1] calculated is obtained by solving the equation of the line in Fig. 39.

118

5.4 Discussion

Endothelin-1 is a small peptide that acts as a potent vasoconstrictor (212). Circulating levels of ET-1 have been shown to be a strong predictor of acute and chronic lung injury (37, 206, 217). In addition to our LAP detection assay, we sought to include the ET-1 peptide in our LTx diagnostic toolkit. ET-1 is unique in that we were able to exploit the biochemical properties of this peptide to develop an electrochemical assay based on the competitive-ELISA method. As a twenty-one amino acid peptide with an N-terminal cysteine residue, ET-1 bears remarkable similarities (size and anchoring thiol) to the PNA probes used in the nucleic acid assay (Table 6). Thus, we hypothesized that we would be able to functionalize a gold electrode with synthetic ET-1 self-assembled monolayer in a very similar manner to PNA. The presence of the ET-1 peptide in solution would compete with the surface-bound ET-1 for the binding of ET-1 antibodies (Figure 34b). That is, a portion of the antibody would be bound to solution ET-1 (endogenous levels) and the remaining portion would bind to the synthetic, surface bound ET-1. Thus, we expected to observe changes in ferrocyanide blocking signals due to differential ET-1 antibody levels bound to the electrode surface in the presence of endogenous ET-1.

We technically validated the ET-1 detection scheme (Figure 35) and characterized the sensor by titrating the amount of ET-1 antibody detected on the surface of an electrode (Figure 36). In addition, we were able to cross-validate the technique of ET-1 detection using a competitive ELISA (Figure 37). Notably, we performed proof-of-concept testing of ET-1 detection in PBS (Figure 39a). By concurrently running a standard curve of ET- 1 antibody concentrations, we were able to extrapolate theoretical ET-1 levels in, unknown, spiked samples with a high degree of accuracy (Figure 39 and Table 11). As such, we were able to demonstrate the ability to quickly (~75 minute) detect the ET-1 peptide. Moving forward, a greater emphasis will be placed on improving the detection limits and speed of this assay by investigating microfluidic techniques that concentrate ET-1 in the sample.

119

Having met the technical requirements of a specific and rapid peptide detection assay, we sought to biologically validate the ET-1 assay in lung perfusate media. EVLP perfusate solution, STEEN solution, is an acellular matrix that represents a simplified medium for rapid and sensitive biological analysis. During EVLP, diagnostic biomarkers such ET-1 are present and can accumulate in STEEN solution (198, 206). Thus, we were able to successfully extrapolate ET-1 peptide levels in EVLP samples with a high degree of accuracy. The integrity of the assay was upheld as the complexity of the sampling matrix was increased from PBS to STEEN (Figures 39b and Table 11). This was anticipated as both the LAP and nucleic acid LTx assays, we observed that the development of a robust SAM could provide excellent specificity. Importantly, the ability to fine-tune each sensor to a particular sensitivity range was demonstrated by changing the size of each sensor (Figure 38); however, further study is needed to determine the specific sensing levels that would be required in the clinical setting.

5.5 Conclusions

In conclusion, we have developed a novel sensing platform capable of detecting very short peptide sequences using a competitive electrochemical assay. This indirect approach serves as a strong foundation for determining endogenous ET-1 concentrations in lung perfusate and will be of great benefit to transplant teams for the prediction of long-term patient outcomes. Future work will explore efforts that further improve the speed and subsequent timing of the ET-1 assay in order to facilitate its clinical implementation. In addition, studies that are focused on determining and quantifying an absolute cutoff for the ET-1 levels associated with lung and patient outcomes will be investigated. By monitoring ET-1 levels during the transplant process, a new level of biomarker-based patient survival prediction is now possible and this

120

information can be used to guide transplant teams towards targeted therapeutic strategies that, together, will improve the quality of life for the transplant patient.

5.6 Materials and Methods

Microchip fabrication

Microchips were fabricated in-house at the Toronto Nanofabrication Centre (University of Toronto, Toronto, ON) using precoated (5 nm chromium, 50 nm gold, and AZ1600 (positive photoresist)) glass substrates purchased from Telic Company. Standard contact lithography was used to pattern the sensing electrodes and followed by Au and Cr wet etching steps and removal of the positive photoresist etchant mask. SU-8 2002 (negative photoresist) (Microchem Corp.) was then spin-cast (4000 rpm, 40 s) and patterned using contact lithography to create the 5 µm circular sensing apertures. Microchips were diced in-house using a standard glasscutter and washed with acetone

(Caledon Labs), isopropyl alcohol (Caledon Labs), and then O2 plasma etched using (Samco RIE System (Samco)).

Biosensor electroplating

121

Biosensor functionalization

5 µg/mL of synthetic ET-1 peptide (Sigma-Aldrich) was placed on freshly prepared gold electrodes (circular apertures; 20 µL probe solution volume) under reducing conditions in a humidity chamber overnight at room temperature. Electrodes were thoroughly washed with dH2O then backfilled with 1 mM MCH (Sigma-Aldrich) for at least 2 hours. Electrodes were thoroughly washed before proceeding to hybridization experiments.

Hybridization

Hybridization reactions were carried out using 20 µL of 1 µg/mL ET-1 polyclonal antibody (ab48251; Abcam) in PBS (Invitrogen) or STEEN Solution™ (XVIVO Perfusion) for 60 minutes at room temperature. Following hybridization, electrodes were washed thoroughly and prepared for electrochemical measurements.

Electrochemical measurements

Enzyme-linked immunosorbent assay (ELISA)

High-binding, 96-well Costar® plates (Corning Life Sciences) were coated with 10 µg/mL of synthetic ET-1 peptide in PBS, overnight, at 4°C. Plates were thoroughly

122

washed then blocked with a solution of 1% (w/v) of BSA (Sigma-Aldrich) in PBS for 60 minutes at room temperature and shaken at 500 rpm. For the ET-1 assay, a preincubation of various concentrations of synthetic ET-1 peptide and 1 µg/mL of ET-1 antibody was carried out in a separate reaction tube for 45 minutes prior to being added to the ELISA plate for 45 minutes at room temperature and 500 rpm. Plates were subsequently washed and incubated with streptavidin-HRP (Cell Signaling) for 30 minutes at room temperature and 500 rpm. Following washing, 3,3’5,5’- tetramethybenzidine (TMB) (Cell Signaling) was added to each well for 5-15 minutes and protected from light. To stop the reaction, an equal volume of 1.0 N H2SO4 was added to each well and the absorbance at 450 nm was read using a (Spectramax M2 (Molecular Devices)).

ET-1 concentration determination

ET-1 concentrations were calculated by extrapolating x-values (anti-ET-1 antibody concentrations) from experimentally derived y-values (% available surface) based on the equation of the line from a standard curve of anti-ET-1 concentration dilutions. The calculated anti-ET-1 concentration was then subtracted from the actual, added, anti-ET- 1 concentration to derive the endogenous ET-1 peptide concentration bound to anti-ET- 1 antibodies in solution.

Statistical analysis

123

Chapter 6 Conclusions and Future Outlook

124

6 Conclusions and Future Outlook

6.1 Ph.D. Research Summary and Perspective

The sensors described in this work represent a significant advance towards the use of rapid diagnostics in lung transplantation. Over the next decade, the market for personalized medicine in transplantation is poised for significant growth (221). Importantly, panel-based testing of interleukins in a complex scoring system is predicted to become a major driver in the immunology/transplant diagnostic field with short time to realization (221). These facts, taken together, support the view that works such as this on real-time point-of-care lung assessment offer the potential to improve clinical practice and patient outcomes in significant ways.

In recent years, point-of-care diagnostics have been focused on developing devices for the detection of infectious disease; however, there is a growing body of work that demonstrates a significant role of these technologies in many other fields including transplantation. The procedures surrounding lung transplantation have evolved greatly over the past few decades and are continually improving. The introduction of point-of- care diagnostics in this field could greatly change our current clinical approach. The use of the transplant assays developed in this thesis could potentially aid surgeons in the decision-making process when selecting donor lungs and represents a major advance towards personalized transplant medicine for the lung. The information provided by these tests could be acted upon to treat the donor lung in a directed manner to repair a specific diagnosed injury and studies have shown that various treatment regimes can be effective in improving donor lung function (222). Furthermore, the LTx-assays herein could be used to guide therapy and monitor response to treatment in donor lungs that may have been previously discarded - thereby expanding the number of available organs. Current research establishes that various biomarkers can be used to select suitable donor lungs that would otherwise be rejected

125

using current clinical practice standards (223). It is estimated that over 40% of the discarded lungs may in fact be suitable for clinical transplantation (8), thus the rapid sample-to-answer turnaround provided by the these assays could have broad implications for the organ selection process. With this technology, organs can now be diagnosed, treated, repaired, and confirmed to be improved in a transplant relevant timeline – a feat unachievable with current technologies.

Moving forward, large, multi-center, prospective trials will be needed to further validate our results. These studies will focus on implementing this profiling approach into transplant procedures and observing patient outcomes. Additionally, standardization of the sample collection process (including biopsy size, location, etc.) will help to improve the precision of the nucleic acid assay. Further studies are required for the proteomic assays to carefully identify the specific cutoffs that discriminate donor lung outcomes. In time, the breadth of the panel of LTx-assay biomarkers can be expanded to include the future discovery and validation of additional genomic, proteomic, and metabolomic biomarkers. Highly multiplexed, multi-analyte tests will have tremendous predictive values and electrochemical biosensing approaches have been shown to be amenable to large-scale multiplexing (approximately one hundred analytes simultaneously) (179). Using this approach, the prognostic value of the LTx-assays will increase in accuracy resulting in extremely high negative and positive predictive test values.

Beyond lung transplantation, rapid diagnostics will play an important role in additional solid organ transplants. Currently, kidney transplantation is one of the most common forms of solid organ transplantation with over one thousand procedures performed in Canada in 2012 (Summarized in Table 12, from (2)). Similarly, there have been around 500 liver and 150 heart transplants performed in that time (Summarized in Table 12, from (2)). Common amongst all forms of organ transplantation are challenges

126

surrounding organ utilization. Unlike lungs, the utilization rates for these other procedures are higher (Summarized in Table 12, from (3)). Thus, there are potentially thousands of organs and subsequent patients that would benefit from enhanced testing each year. Further investigation into the various predictive biomarkers for other solid organ procedures will provide a starting point for the adaptation of the LTx assays described here. The translation of the LTx assays to other organ transplantation fields is afforded the advantage of the potential overlap between prognostic biomarkers and very similar input sources (i.e. tissue biopsies), which should allow for timely product development.

Although the assays in this work were developed for use within the transplantation setting, the biomarkers and pathways these tests monitor have implications that extend beyond organ assessment. Cytokines and inflammatory biomarkers play a pivotal role in many diseases including diabetes, rheumatoid arthritis, and cardiovascular disease. As such, the tests described in this work could be modified and adapted for novel applications. Similar to the transplantation requirements, the monitoring of cytokines for other diseases would require sensors that are highly quantitative and easily integratable within the health care setting. Future studies that optimize the FraCS assay for use with novel samples (ex. whole blood, cerebral spinal fluid, or urine) would be required to transform these tests for other applications.

Table 12. Solid organ transplantation in Canada and the United States.

Solid Organ Number of Transplants 2012 Utilization Rate Total Organs (approx.) Transplanted Canada/USA (%) Canada/USA

Lung 200/2,000 20 1,000/10,000 (11,000)

Kidney 1,250/17,000 75 1675/22,675 (24,350)

Heart 150/2,500 33 450/7,500 (8,000)

Liver 500/6,000 75 675/8,000 (8,675)

127

Additionally, the protein-based assays developed in this work could be applied to the field of stem cell therapy. Primary stem cell cultures are an emerging field of medical treatment; however, translation into current practice remains limited due challenges in the growth and production of these cells. Cultured primary cells tend to be heterogeneous and dynamically complex systems that, without proper control, can develop into off-target cell populations. The accumulation of signaling proteins, including TGFβ1, produced by off-target cell populations can inhibit the yield of desired cell populations. Strategies that minimize the impact of emerging mature cell populations can enable the expansion of hematopoietic progenitors at higher cell seeding densities and longer culture times. Direct analysis of TGFβ1 and other soluble factors important to stem cell production could play a key role in commercial stem cell production. The electrochemical assays described in this thesis would represent a significant advance over current protein detection technologies, as they are fully automatable and integratable into a stem cell bioreactor system. As such, adaptation of the protein assays herein may have important implications in stem cell therapies

6.2 Concluding Remarks

As the field of nanotechnology continues to develop at a tremendous pace, it is apparent that molecular diagnostics will see considerable change over the coming decades. For existing tests, we should see a shift away from the concept of centralized laboratories in favour of rapid, in-house (office) tests. This trend will provide patients with faster access to information and better treatments as physicians will be able to know ‘instantly’ which treatment to provide and thus avoid prescribing generalized medications/therapies as a stopgap until the results of the current tests arrive. In addition, the advent of novel technologies will open up the possibility of testing for previously untestable conditions, diseases, and elusive biomarkers. Since 1983, there has been significant growth in the field of lung transplantation; however, the integration of rapid molecular diagnostic tools has been slow to adapt. By demonstrating the power

128

of predictive nucleic acid and proteomic tests in this work, it is our hope that this will be groundswell needed to prompt the development of rapid transplant diagnostics. By identifying short and long-term donor lung outcomes, pre-implantation, this work demonstrates a powerful new approach to organ screening, testing, and treatment for the future.

129

Chapter 7 References

130

7 References

7.1 List of Contributed Works

A. T. Sage, X. Bai, M. Liu, S. Keshavjee, S. O. Kelley, Using the inherent chemistry of the endothelin-1 peptide to develop a rapid assay for lung transplantation. Manuscript submitted for publication. (2015).

A. T. Sage, J. D. Besant, L. Mahmoudian, M. Poudineh, X. Bai, R. Zamel, M. Hsin, E. H. Sargent, M. Cypel, M. Liu, S. Keshavjee, S. O. Kelley, Fractal circuit sensors enable rapid quantification of biomarkers for donor lung assessment for transplantation. Science Advances, in press. (2015).

Y. G. Zhou, Y. Wan, A. T. Sage, M. Poudineh, S. O. Kelley, Effect of microelectrode structure on electrocatalysis at nucleic acid-modified sensors. Langmuir 30, 14322- 14328 (2014).

A. T. Sage, J. D. Besant, B. Lam, E. H. Sargent, S. O. Kelley, Ultrasensitive electrochemical biomolecular detection using nanostructured microelectrodes. Acc Chem Res 47, 2417-2425 (2014).

M. Poudineh, R. M. Mohamadi, A. Sage, L. Mahmoudian, E. H. Sargent, S. O. Kelley, Three-dimensional, sharp-tipped electrodes concentrate applied fields to enable direct electrical release of intact biomarkers from cells. Lab Chip 14, 1785-1790 (2014).

B. Lam, J. Das, R. D. Holmes, L. Live, A. Sage, E. H. Sargent, S. O. Kelley, Solution- based circuits enable rapid and multiplexed pathogen detection. Nat Commun 4, 2001 (2013).

B. Lam, R. D. Holmes, J. Das, M. Poudineh, A. Sage, E. H. Sargent, S. O. Kelley, Optimized templates for bottom-up growth of high-performance integrated biomolecular detectors. Lab Chip 13, 2569-2575 (2013).

131

7.2 References

1. J. R. Benfield, J. C. Wain, The history of lung transplantation. Chest Surg Clin N Am 10, 189-199, xi (2000).

2. CIHI, "Canadian Organ Replacement Register Annual Report: Treatment of End- Stage Organ Failure in Canada, 2003 to 2012," (CIHI, Ottawa, ON, 2014).

3. OPTN/SRTR, "OPTN/SRTR 2012 Annual Data Report," (Department of Health and Human Services, Health Resources and Services Administration, Rockville, MD, 2014).

4. CDC, "National Center for Health Statistics. Deaths: Final Data for 2009," National Vital Statistics Report No. 3 (2012).

5. CDC, "Current Cigarette Smoking Among Adults - United States. 2005-2012," Morbidity and Mortality Weekly Report No. 02 (2014).

6. M. Cypel, M. Sato, E. Yildirim, W. Karolak, F. Chen, J. Yeung, C. Boasquevisque, V. Leist, L. G. Singer, K. Yasufuku, M. Deperrot, T. K. Waddell, S. Keshavjee, A. Pierre, Initial experience with lung donation after cardiocirculatory death in Canada. J Heart Lung Transplant 28, 753-758 (2009).

7. M. Cypel, J. C. Yeung, M. Liu, M. Anraku, F. Chen, W. Karolak, M. Sato, J. Laratta, S. Azad, M. Madonik, C. W. Chow, C. Chaparro, M. Hutcheon, L. G. Singer, A. S. Slutsky, K. Yasufuku, M. de Perrot, A. F. Pierre, T. K. Waddell, S. Keshavjee, Normothermic ex vivo lung perfusion in clinical lung transplantation. N Engl J Med 364, 1431-1440 (2011).

8. L. B. Ware, Y. Wang, X. Fang, M. Warnock, T. Sakuma, T. S. Hall, M. Matthay, Assessment of lungs rejected for transplantation and implications for donor selection. Lancet 360, 619-620 (2002).

9. C. H. Kang, M. Anraku, M. Cypel, M. Sato, J. Yeung, S. A. Gharib, A. F. Pierre, M. de Perrot, T. K. Waddell, M. Liu, S. Keshavjee, Transcriptional signatures in donor lungs from donation after cardiac death vs after brain death: a functional pathway analysis. J Heart Lung Transplant 30, 289-298 (2011).

132

10. D. P. Mason, S. C. Murthy, G. V. Gonzalez-Stawinski, M. M. Budev, A. C. Mehta, A. M. McNeill, G. B. Pettersson, Early experience with lung transplantation using donors after cardiac death. J Heart Lung Transplant 27, 561-563 (2008).

11. G. I. Snell, B. J. Levvey, T. Oto, R. McEgan, D. Pilcher, A. Davies, S. Marasco, F. Rosenfeldt, Early lung transplantation success utilizing controlled donation after cardiac death donors. Am J Transplant 8, 1282-1289 (2008).

12. D. P. Mason, L. Thuita, J. M. Alster, S. C. Murthy, M. M. Budev, A. C. Mehta, G. B. Pettersson, E. H. Blackstone, Should lung transplantation be performed using donation after cardiac death? The United States experience. J Thorac Cardiovasc Surg 136, 1061-1066 (2008).

13. J. E. Tuttle-Newhall, S. M. Krishnan, M. F. Levy, V. McBride, J. P. Orlowski, R. S. Sung, Organ donation and utilization in the United States: 1998-2007. Am J Transplant 9, 879-893 (2009).

14. M. Cypel, M. Liu, M. Rubacha, J. C. Yeung, S. Hirayama, M. Anraku, M. Sato, J. Medin, B. L. Davidson, M. de Perrot, T. K. Waddell, A. S. Slutsky, S. Keshavjee, Functional repair of human donor lungs by IL-10 gene therapy. Sci Transl Med 1, 4ra9 (2009).

15. M. Anraku, M. J. Cameron, T. K. Waddell, M. Liu, T. Arenovich, M. Sato, M. Cypel, A. F. Pierre, M. de Perrot, D. J. Kelvin, S. Keshavjee, Impact of human donor lung gene expression profiles on survival after lung transplantation: a case-control study. Am J Transplant 8, 2140-2148 (2008).

16. Y. Suzuki, E. Cantu, J. D. Christie, Primary graft dysfunction. Semin Respir Crit Care Med 34, 305-319 (2013).

17. M. de Perrot, M. Liu, T. K. Waddell, S. Keshavjee, Ischemia-reperfusion-induced lung injury. Am J Respir Crit Care Med 167, 490-511 (2003).

18. J. C. Lee, J. D. Christie, S. Keshavjee, Primary graft dysfunction: definition, risk factors, short- and long-term outcomes. Semin Respir Crit Care Med 31, 161-171 (2010).

19. J. D. Christie, M. Carby, R. Bag, P. Corris, M. Hertz, D. Weill, I. W. G. o. P. L. G. Dysfunction, Report of the ISHLT Working Group on Primary Lung Graft Dysfunction part II: definition. A consensus statement of the International Society

133

for Heart and Lung Transplantation. J Heart Lung Transplant 24, 1454-1459 (2005).

20. E. P. Trulock, J. D. Christie, L. B. Edwards, M. M. Boucek, P. Aurora, D. O. Taylor, F. Dobbels, A. O. Rahmel, B. M. Keck, M. I. Hertz, Registry of the International Society for Heart and Lung Transplantation: twenty-fourth official adult lung and heart-lung transplantation report-2007. J Heart Lung Transplant 26, 782-795 (2007).

21. J. D. Christie, D. Van Raemdonck, M. de Perrot, M. Barr, S. Keshavjee, S. Arcasoy, J. Orens, I. W. G. o. P. L. G. Dysfunction, Report of the ISHLT Working Group on Primary Lung Graft Dysfunction part I: introduction and methods. J Heart Lung Transplant 24, 1451-1453 (2005).

22. J. C. Lee, J. D. Christie, Primary graft dysfunction. Proc Am Thorac Soc 6, 39-46 (2009).

23. S. M. Arcasoy, A. Fisher, R. R. Hachem, M. Scavuzzo, L. B. Ware, I. W. G. o. P. L. G. Dysfunction, Report of the ISHLT Working Group on Primary Lung Graft Dysfunction part V: predictors and outcomes. J Heart Lung Transplant 24, 1483- 1488 (2005).

24. C. S. Herrington, M. E. Prekker, A. K. Arrington, D. Susanto, J. W. Baltzell, L. L. Studenski, D. M. Radosevich, R. F. Kelly, S. J. Shumway, M. I. Hertz, H. B. Bittner, P. S. Dahlberg, A randomized, placebo-controlled trial of aprotinin to reduce primary graft dysfunction following lung transplantation. Clin Transplant 25, 90-96 (2011).

25. S. Keshavjee, R. D. Davis, M. R. Zamora, M. de Perrot, G. A. Patterson, A randomized, placebo-controlled trial of complement inhibition in ischemia- reperfusion injury after lung transplantation in human beings. J Thorac Cardiovasc Surg 129, 423-428 (2005).

26. M. O. Meade, J. T. Granton, A. Matte-Martyn, K. McRae, B. Weaver, P. Cripps, S. H. Keshavjee, P. Toronto Lung Transplant, A randomized trial of inhaled nitric oxide to prevent ischemia-reperfusion injury after lung transplantation. Am J Respir Crit Care Med 167, 1483-1489 (2003).

27. M. L. Barr, S. M. Kawut, T. P. Whelan, R. Girgis, H. Bottcher, J. Sonett, W. Vigneswaran, D. M. Follette, P. A. Corris, I. W. G. o. P. L. G. Dysfunction, Report of the ISHLT Working Group on Primary Lung Graft Dysfunction part IV:

134

recipient-related risk factors and markers. J Heart Lung Transplant 24, 1468- 1482 (2005).

28. J. M. Diamond, J. C. Lee, S. M. Kawut, R. J. Shah, A. R. Localio, S. L. Bellamy, D. J. Lederer, E. Cantu, B. A. Kohl, V. N. Lama, S. M. Bhorade, M. Crespo, E. Demissie, J. Sonett, K. Wille, J. Orens, A. S. Shah, A. Weinacker, S. Arcasoy, P. D. Shah, D. S. Wilkes, L. B. Ware, S. M. Palmer, J. D. Christie, G. Lung Transplant Outcomes, Clinical risk factors for primary graft dysfunction after lung transplantation. Am J Respir Crit Care Med 187, 527-534 (2013).

29. J. D. Christie, L. B. Edwards, A. Y. Kucheryavaya, C. Benden, F. Dobbels, R. Kirk, A. O. Rahmel, J. Stehlik, M. I. Hertz, The Registry of the International Society for Heart and Lung Transplantation: Twenty-eighth Adult Lung and Heart- Lung Transplant Report--2011. J Heart Lung Transplant 30, 1104-1122 (2011).

30. M. Sato, Chronic lung allograft dysfunction after lung transplantation: the moving target. Gen Thorac Cardiovasc Surg 61, 67-78 (2013).

31. K. C. Meyer, G. Raghu, G. M. Verleden, P. A. Corris, P. Aurora, K. C. Wilson, J. Brozek, A. R. Glanville, I. A. E. B. T. F. Committee, I. A. E. B. T. F. Committee, An international ISHLT/ATS/ERS clinical practice guideline: diagnosis and management of bronchiolitis obliterans syndrome. Eur Respir J 44, 1479-1503 (2014).

32. M. Sato, T. K. Waddell, U. Wagnetz, H. C. Roberts, D. M. Hwang, A. Haroon, D. Wagnetz, C. Chaparro, L. G. Singer, M. A. Hutcheon, S. Keshavjee, Restrictive allograft syndrome (RAS): a novel form of chronic lung allograft dysfunction. J Heart Lung Transplant 30, 735-742 (2011).

33. G. M. Verleden, R. Vos, S. E. Verleden, W. De Wever, S. I. De Vleeschauwer, A. Willems-Widyastuti, H. Scheers, L. J. Dupont, D. E. Van Raemdonck, B. M. Vanaudenaerde, Survival determinants in lung transplant patients with chronic allograft dysfunction. Transplantation 92, 703-708 (2011).

34. G. Devouassoux, C. Drouet, I. Pin, C. Brambilla, E. Brambilla, P. E. Colle, C. Pison, G. Grenoble Lung Transplant, Alveolar neutrophilia is a predictor for the bronchiolitis obliterans syndrome, and increases with degree of severity. Transpl Immunol 10, 303-310 (2002).

35. A. Elssner, F. Jaumann, S. Dobmann, J. Behr, M. Schwaiblmair, H. Reichenspurner, H. Furst, J. Briegel, C. Vogelmeier, Elevated levels of interleukin-8 and transforming growth factor-beta in bronchoalveolar lavage fluid

135

from patients with bronchiolitis obliterans syndrome: proinflammatory role of bronchial epithelial cells. Munich Lung Transplant Group. Transplantation 70, 362-367 (2000).

36. A. Magnan, J. L. Mege, J. C. Escallier, J. Brisse, C. Capo, M. Reynaud, P. Thomas, B. Meric, L. Garbe, M. Badier, L. Viard, P. Bongrand, R. Giudicelli, D. Metras, P. Fuentes, D. Vervloet, M. Noirclerc, Balance between alveolar macrophage IL-6 and TGF-beta in lung-transplant recipients. Marseille and Montreal Lung Transplantation Group. Am J Respir Crit Care Med 153, 1431- 1436 (1996).

37. M. Salama, P. Jaksch, O. Andrukhova, S. Taghavi, W. Klepetko, S. Aharinejad, Endothelin-1 is a useful biomarker for early detection of bronchiolitis obliterans in lung transplant recipients. J Thorac Cardiovasc Surg 140, 1422-1427 (2010).

38. M. Fattal-German, I. Frachon, J. Cerrina, F. L. Ladurie, F. Lecerf, P. Dartevelle, S. Berrih-Aknin, Particular phenotypic profile of blood lymphocytes during obliterative bronchiolitis syndrome following lung transplantation. Transpl Immunol 2, 243-251 (1994).

39. J. E. Fildes, N. Yonan, C. T. Leonard, Natural killer cells and lung transplantation, roles in rejection, infection, and tolerance. Transpl Immunol 19, 1-11 (2008).

40. J. E. Fildes, N. Yonan, K. Tunstall, A. H. Walker, L. Griffiths-Davies, P. Bishop, C. T. Leonard, Natural killer cells in peripheral blood and lung tissue are associated with chronic rejection after lung transplantation. J Heart Lung Transplant 27, 203-207 (2008).

41. G. Devouassoux, C. Pison, C. Drouet, I. Pin, C. Brambilla, E. Brambilla, Early lung leukocyte infiltration, HLA and adhesion molecule expression predict chronic rejection. Transpl Immunol 8, 229-236 (2001).

42. M. Fasano, S. Yousem, J. Jagirdar, MIB-1 as a predictor of response in lung allografts with moderate acute cellular rejection. Am J Surg Pathol 22, 749-754 (1998).

43. S. M. Bhorade, H. Chen, L. Molinero, C. Liao, E. R. Garrity, W. T. Vigneswaran, R. Shilling, A. Sperling, A. Chong, M. L. Alegre, Decreased percentage of CD4+FoxP3+ cells in bronchoalveolar lavage from lung transplant recipients correlates with development of bronchiolitis obliterans syndrome. Transplantation 90, 540-546 (2010).

136

44. L. Badri, S. Murray, L. X. Liu, N. M. Walker, A. Flint, A. Wadhwa, K. M. Chan, G. B. Toews, D. J. Pinsky, F. J. Martinez, V. N. Lama, Mesenchymal stromal cells in bronchoalveolar lavage as predictors of bronchiolitis obliterans syndrome. Am J Respir Crit Care Med 183, 1062-1070 (2011).

45. A. L. Gregson, A. Hoji, R. Saggar, D. J. Ross, B. M. Kubak, B. D. Jamieson, S. S. Weigt, J. P. Lynch, 3rd, A. Ardehali, J. A. Belperio, O. O. Yang, Bronchoalveolar immunologic profile of acute human lung transplant allograft rejection. Transplantation 85, 1056-1059 (2008).

46. A. Van Muylem, C. Knoop, M. Estenne, Early detection of chronic pulmonary allograft dysfunction by exhaled biomarkers. Am J Respir Crit Care Med 175, 731-736 (2007).

47. M. Iversen, C. M. Burton, S. Vand, L. Skovfoged, J. Carlsen, N. Milman, C. B. Andersen, M. Rasmussen, M. Tvede, Aspergillus infection in lung transplant patients: incidence and prognosis. Eur J Clin Microbiol Infect Dis 26, 879-886 (2007).

48. A. C. Pasqualotto, M. O. Xavier, L. B. Sanchez, C. D. de Oliveira Costa, S. M. Schio, S. M. Camargo, J. J. Camargo, T. C. Sukiennik, L. C. Severo, Diagnosis of invasive aspergillosis in lung transplant recipients by detection of galactomannan in the bronchoalveolar lavage fluid. Transplantation 90, 306-311 (2010).

49. P. Tabarsi, A. Soraghi, M. Marjani, P. Zandian, P. Baghaei, K. Najafizadeh, A. Droudinia, S. A. Sarrafzadeh, P. Javanmard, D. Mansouri, Comparison of serum and bronchoalveolar lavage galactomannan in diagnosing invasive aspergillosis in solid-organ transplant recipients. Exp Clin Transplant 10, 278-281 (2012).

50. M. D. Stone, S. B. Harvey, G. L. Nelsestuen, C. Reilly, M. I. Hertz, C. H. Wendt, Elevated peptides in lung lavage fluid associated with bronchiolitis obliterans syndrome. PLoS One 9, e84471 (2014).

51. C. Sammons, C. T. Doligalski, Utility of procalcitonin as a biomarker for rejection and differentiation of infectious complications in lung transplant recipients. Ann Pharmacother 48, 116-122 (2014).

52. X. Y. Yu, Y. Wang, H. Zhong, Q. L. Dou, Y. L. Song, H. Wen, Diagnostic value of serum procalcitonin in solid organ transplant recipients: a systematic review and meta-analysis. Transplant Proc 46, 26-32 (2014).

137

53. C. Reilly, T. Cervenka, M. I. Hertz, T. Becker, C. H. Wendt, Human neutrophil peptide in lung chronic allograft dysfunction. Biomarkers 16, 663-669 (2011).

54. D. S. Wilkes, K. M. Heidler, M. Niemeier, G. R. Schwenk, P. N. Mathur, W. M. Breite, O. W. Cummings, J. C. Weissler, Increased bronchoalveolar IgG2/IgG1 ratio is a marker for human lung allograft rejection. J Investig Med 42, 652-659 (1994).

55. C. Ward, E. H. Walters, L. Zheng, H. Whitford, T. J. Williams, G. I. Snell, Increased soluble CD14 in bronchoalveolar lavage fluid of stable lung transplant recipients. Eur Respir J 19, 472-478 (2002).

56. M. Reynaud-Gaubert, V. Marin, X. Thirion, C. Farnarier, P. Thomas, M. Badier, P. Bongrand, R. Giudicelli, P. Fuentes, Upregulation of chemokines in bronchoalveolar lavage fluid as a predictive marker of post-transplant airway obliteration. J Heart Lung Transplant 21, 721-730 (2002).

57. F. Meloni, N. Solari, S. Miserere, M. Morosini, A. Cascina, C. Klersy, E. Arbustini, C. Pellegrini, M. Vigano, A. M. Fietta, Chemokine redundancy in BOS pathogenesis. A possible role also for the CC chemokines: MIP3-beta, MIP3- alpha, MDC and their specific receptors. Transpl Immunol 18, 275-280 (2008).

58. P. Jaksch, S. Taghavi, W. Klepetko, M. Salama, Pretransplant serum human chitinase-like glycoprotein YKL-40 concentrations independently predict bronchiolitis obliterans development in lung transplant recipients. J Thorac Cardiovasc Surg 148, 273-281 (2014).

59. F. Meloni, R. Salvini, A. M. Bardoni, I. Passadore, N. Solari, P. Vitulo, T. Oggionni, M. Vigano, E. Pozzi, A. M. Fietta, Bronchoalveolar lavage fluid proteome in bronchiolitis obliterans syndrome: possible role for surfactant protein A in disease onset. J Heart Lung Transplant 26, 1135-1143 (2007).

60. F. D'Ovidio, H. Kaneda, C. Chaparro, M. Mura, D. Lederer, S. Di Angelo, H. Takahashi, C. Gutierrez, M. Hutcheon, L. G. Singer, T. K. Waddell, J. Floros, M. Liu, S. Keshavjee, Pilot study exploring lung allograft surfactant protein A (SP-A) expression in association with lung transplant outcome. Am J Transplant 13, 2722-2729 (2013).

61. P. M. Fisichella, C. S. Davis, E. Lowery, L. Ramirez, R. L. Gamelli, E. J. Kovacs, Aspiration, localized pulmonary inflammation, and predictors of early-onset bronchiolitis obliterans syndrome after lung transplantation. J Am Coll Surg 217, 90-100; discussion 100-101 (2013).

138

62. T. Saito, H. Takahashi, H. Kaneda, M. Binnie, S. Azad, M. Sato, T. K. Waddell, M. Cypel, M. Liu, S. Keshavjee, Impact of cytokine expression in the pre- implanted donor lung on the development of chronic lung allograft dysfunction subtypes. Am J Transplant 13, 3192-3201 (2013).

63. S. S. Weigt, A. Derhovanessian, E. Liao, S. Hu, A. L. Gregson, B. M. Kubak, R. Saggar, R. Saggar, V. Plachevskiy, M. C. Fishbein, J. P. Lynch, 3rd, A. Ardehali, D. J. Ross, H. J. Wang, R. M. Elashoff, J. A. Belperio, CXCR3 chemokine ligands during respiratory viral infections predict lung allograft dysfunction. Am J Transplant 12, 477-484 (2012).

64. S. Cantisan, R. Lara, M. Montejo, J. Redel, A. Rodriguez-Benot, J. Gutierrez- Aroca, M. Gonzalez-Padilla, L. Bueno, A. Rivero, R. Solana, J. Torre-Cisneros, Pretransplant interferon-gamma secretion by CMV-specific CD8+ T cells informs the risk of CMV replication after transplantation. Am J Transplant 13, 738-745 (2013).

65. F. R. Rega, B. M. Vanaudenaerde, W. A. Wuyts, N. C. Jannis, G. M. Verleden, T. E. Lerut, D. E. Van Raemdonck, IL-1beta in bronchial lavage fluid is a non- invasive marker that predicts the viability of the pulmonary graft from the non- heart-beating donor. J Heart Lung Transplant 24, 20-28 (2005).

66. A. S. Shah, M. S. Leffell, D. Lucas, A. A. Zachary, Elevated pretransplantation soluble CD30 is associated with decreased early allograft function after human lung transplantation. Hum Immunol 70, 101-103 (2009).

67. J. D. Christie, S. Bellamy, L. B. Ware, D. Lederer, D. Hadjiliadis, J. Lee, N. Robinson, A. R. Localio, K. Wille, V. Lama, S. Palmer, J. Orens, A. Weinacker, M. Crespo, E. Demissie, S. E. Kimmel, S. M. Kawut, Construct validity of the definition of primary graft dysfunction after lung transplantation. J Heart Lung Transplant 29, 1231-1239 (2010).

68. R. J. Shah, S. L. Bellamy, A. R. Localio, N. Wickersham, J. M. Diamond, A. Weinacker, V. N. Lama, S. Bhorade, J. A. Belperio, M. Crespo, E. Demissie, S. M. Kawut, K. M. Wille, D. J. Lederer, J. C. Lee, S. M. Palmer, J. Orens, J. Reynolds, A. Shah, D. S. Wilkes, L. B. Ware, J. D. Christie, A panel of lung injury biomarkers enhances the definition of primary graft dysfunction (PGD) after lung transplantation. J Heart Lung Transplant 31, 942-949 (2012).

69. M. De Perrot, Y. Sekine, S. Fischer, T. K. Waddell, K. McRae, M. Liu, D. A. Wigle, S. Keshavjee, Interleukin-8 release during early reperfusion predicts graft

139

function in human lung transplantation. Am J Respir Crit Care Med 165, 211-215 (2002).

70. N. Satoda, T. Shoji, Y. Wu, T. Fujinaga, F. Chen, A. Aoyama, J. T. Zhang, A. Takahashi, T. Okamoto, I. Matsumoto, H. Sakai, Y. Li, X. Zhao, T. Manabe, E. Kobayashi, S. Sakaguchi, H. Wada, H. Ohe, S. Uemoto, J. Tottori, T. Bando, H. Date, T. Koshiba, Value of FOXP3 expression in peripheral blood as rejection marker after miniature swine lung transplantation. J Heart Lung Transplant 27, 1293-1301 (2008).

71. X. Qiao, K. Jiang, J. Nie, K. Fan, Z. Zheng, J. Wang, J. Li, Increased expression of Tim-3 and its ligand Galectin-9 in rat allografts during acute rejection episodes. Biochem Biophys Res Commun 445, 542-548 (2014).

72. N. A. Francalancia, S. C. Wang, N. L. Thai, R. Aeba, R. L. Simmons, S. A. Yousem, R. L. Hardesty, B. P. Griffith, Graft cytokine mRNA activity in rat single lung transplants by reverse transcription-polymerase chain reaction: effect of cyclosporine. J Heart Lung Transplant 11, 1041-1045 (1992).

73. D. Krustrup, C. B. Madsen, M. Iversen, L. Engelholm, L. P. Ryder, C. B. Andersen, The number of regulatory T cells in transbronchial lung allograft biopsies is related to FoxP3 mRNA levels in bronchoalveolar lavage fluid and to the degree of acute cellular rejection. Transpl Immunol 29, 71-75 (2013).

74. C. B. Madsen, A. Norgaard, M. Iversen, L. P. Ryder, Elevated mRNA levels of CTLA-4, FoxP3, and granzyme B in BAL, but not in blood, during acute rejection of lung allografts. Transpl Immunol 24, 26-32 (2010).

75. V. J. Gimino, J. D. Lande, T. R. Berryman, R. A. King, M. I. Hertz, Gene expression profiling of bronchoalveolar lavage cells in acute lung rejection. Am J Respir Crit Care Med 168, 1237-1242 (2003).

76. D. J. Ross, A. Moudgil, A. Bagga, M. Toyoda, A. M. Marchevsky, R. M. Kass, S. C. Jordan, Lung allograft dysfunction correlates with gamma-interferon gene expression in bronchoalveolar lavage. J Heart Lung Transplant 18, 627-636 (1999).

77. A. Moudgil, A. Bagga, M. Toyoda, E. Nicolaidou, S. C. Jordan, D. Ross, Expression of gamma-IFN mRNA in bronchoalveolar lavage fluid correlates with early acute allograft rejection in lung transplant recipients. Clin Transplant 13, 201-207 (1999).

140

78. R. Shi, J. Yang, A. Jaramillo, N. S. Steward, A. Aloush, E. P. Trulock, G. Alexander Patterson, M. Suthanthiran, T. Mohanakumar, Correlation between interleukin-15 and granzyme B expression and acute lung allograft rejection. Transpl Immunol 12, 103-108 (2004).

79. B. M. Vanaudenaerde, L. J. Dupont, W. A. Wuyts, E. K. Verbeken, I. Meyts, D. M. Bullens, E. Dilissen, L. Luyts, D. E. Van Raemdonck, G. M. Verleden, The role of interleukin-17 during acute rejection after lung transplantation. Eur Respir J 27, 779-787 (2006).

80. J. M. Charpin, M. Stern, G. Lebrun, P. Aubin, D. Grenet, D. Israel-Biet, Increased endothelin-1 associated with bacterial infection in lung transplant recipients. Transplantation 71, 1840-1847 (2001).

81. B. M. Vanaudenaerde, S. I. De Vleeschauwer, R. Vos, I. Meyts, D. M. Bullens, V. Reynders, W. A. Wuyts, D. E. Van Raemdonck, L. J. Dupont, G. M. Verleden, The role of the IL23/IL17 axis in bronchiolitis obliterans syndrome after lung transplantation. Am J Transplant 8, 1911-1920 (2008).

82. L. M. Eerola, H. S. Alho, P. K. Maasilta, K. A. Inkinen, A. L. Harjula, S. H. Litmanen, U. S. Salminen, Matrix metalloproteinase induction in post-transplant obliterative bronchiolitis. J Heart Lung Transplant 24, 426-432 (2005).

83. J. M. Charpin, J. Valcke, L. Kettaneh, B. Epardeau, M. Stern, D. Israel-Biet, Peaks of transforming growth factor-beta mRNA in alveolar cells of lung transplant recipients as an early marker of chronic rejection. Transplantation 65, 752-755 (1998).

84. D. Jonigk, M. Merk, K. Hussein, L. Maegel, K. Theophile, M. Muth, U. Lehmann, C. L. Bockmeyer, M. Mengel, J. Gottlieb, T. Welte, A. Haverich, H. Golpon, H. Kreipe, F. Laenger, Obliterative airway remodeling: molecular evidence for shared pathways in transplanted and native lungs. Am J Pathol 178, 599-608 (2011).

85. A. Elssner, F. Jaumann, W. P. Wolf, M. Schwaiblmair, J. Behr, H. Furst, H. Reichenspurner, J. Briegel, J. Niedermeyer, C. Vogelmeier, Bronchial epithelial cell B7-1 and B7-2 mRNA expression after lung transplantation: a role in allograft rejection? Eur Respir J 20, 165-169 (2002).

86. H. Kaneda, T. K. Waddell, M. de Perrot, X. H. Bai, C. Gutierrez, T. Arenovich, C. Chaparro, M. Liu, S. Keshavjee, Pre-implantation multiple cytokine mRNA

141

expression analysis of donor lung grafts predicts survival after lung transplantation in humans. Am J Transplant 6, 544-551 (2006).

87. K. Jiang, L. Cheng, J. Wang, J. Li, J. Nie, Heme oxygenase-1 expression in rats with acute lung rejection and implication. J Huazhong Univ Sci Technolog Med Sci 29, 84-87 (2009).

88. B. V. Erne, W. Jungraithmayr, J. Buschmann, S. Arni, W. Weder, I. Inci, Effect of N-acetylcysteine on acute allograft rejection after rat lung transplantation. Ann Thorac Surg 95, 1021-1027 (2013).

89. S. Hirayama, M. Sato, S. Loisel-Meyer, Y. Matsuda, H. Oishi, Z. Guan, T. Saito, J. Yeung, M. Cypel, D. M. Hwang, J. A. Medin, M. Liu, S. Keshavjee, Lentivirus IL-10 gene therapy down-regulates IL-17 and attenuates mouse orthotopic lung allograft rejection. Am J Transplant 13, 1586-1593 (2013).

90. J. Xu, E. Torres, A. L. Mora, H. Shim, A. Ramirez, D. Neujahr, K. L. Brigham, M. Rojas, Attenuation of obliterative bronchiolitis by a CXCR4 antagonist in the murine heterotopic tracheal transplant model. J Heart Lung Transplant 27, 1302- 1310 (2008).

91. A. M. Caliendo, D. N. Gilbert, C. C. Ginocchio, K. E. Hanson, L. May, T. C. Quinn, F. C. Tenover, D. Alland, A. J. Blaschke, R. A. Bonomo, K. C. Carroll, M. J. Ferraro, L. R. Hirschhorn, W. P. Joseph, T. Karchmer, A. T. MacIntyre, L. B. Reller, A. F. Jackson, A. Infectious Diseases Society of, Better tests, better care: improved diagnostics for infectious diseases. Clin Infect Dis 57 Suppl 3, S139- 170 (2013).

92. J. D. Lande, J. Patil, N. Li, T. R. Berryman, R. A. King, M. I. Hertz, Novel insights into lung transplant rejection by microarray analysis. Proc Am Thorac Soc 4, 44- 51 (2007).

93. L. A. Weintraub, M. M. Sarwal, Microarrays: a monitoring tool for transplant patients? Transpl Int 19, 775-788 (2006).

94. Y. Yuan, J. Zhang, H. Zhang, X. Yang, Silver nanoparticle based label-free colorimetric immunosensor for rapid detection of neurogenin 1. Analyst 137, 496- 501 (2012).

142

95. M. Askari, J. P. Alarie, M. Moreno-Bondi, T. Vo-Dinh, Application of an antibody biochip for p53 detection and cancer diagnosis. Biotechnol Prog 17, 543-552 (2001).

96. C. F. Fronczek, D. J. You, J. Y. Yoon, Single-pipetting microfluidic assay device for rapid detection of Salmonella from poultry package. Biosens Bioelectron 40, 342-349 (2013).

97. R. Fan, O. Vermesh, A. Srivastava, B. K. Yen, L. Qin, H. Ahmad, G. A. Kwong, C. C. Liu, J. Gould, L. Hood, J. R. Heath, Integrated barcode chips for rapid, multiplexed analysis of proteins in microliter quantities of blood. Nat Biotechnol 26, 1373-1378 (2008).

98. H. Xu, J. Chen, J. Birrenkott, J. X. Zhao, S. Takalkar, K. Baryeh, G. Liu, Gold- nanoparticle-decorated silica nanorods for sensitive visual detection of proteins. Anal Chem 86, 7351-7359 (2014).

99. Y. K. Oh, H. A. Joung, H. S. Han, H. J. Suk, M. G. Kim, A three-line lateral flow assay strip for the measurement of C-reactive protein covering a broad physiological concentration range in human sera. Biosens Bioelectron 61, 285- 289 (2014).

100. C. D. Chin, Y. K. Cheung, T. Laksanasopin, M. M. Modena, S. Y. Chin, A. A. Sridhara, D. Steinmiller, V. Linder, J. Mushingantahe, G. Umviligihozo, E. Karita, L. Mwambarangwe, S. L. Braunstein, J. van de Wijgert, R. Sahabo, J. E. Justman, W. El-Sadr, S. K. Sia, Mobile device for disease diagnosis and data tracking in resource-limited settings. Clin Chem 59, 629-640 (2013).

101. J. M. Nam, C. S. Thaxton, C. A. Mirkin, Nanoparticle-based bio-bar codes for the ultrasensitive detection of proteins. Science 301, 1884-1886 (2003).

102. S. Martic, M. K. Rains, H. B. Kraatz, Probing copper/tau protein interactions electrochemically. Anal Biochem 442, 130-137 (2013).

103. E. Stern, J. F. Klemic, D. A. Routenberg, P. N. Wyrembak, D. B. Turner-Evans, A. D. Hamilton, D. A. LaVan, T. M. Fahmy, M. A. Reed, Label-free immunodetection with CMOS-compatible semiconducting nanowires. Nature 445, 519-522 (2007).

104. E. Y. Kim, J. Stanton, B. T. Korber, K. Krebs, D. Bogdan, K. Kunstman, S. Wu, J. P. Phair, C. A. Mirkin, S. M. Wolinsky, Detection of HIV-1 p24 Gag in plasma by

143

a nanoparticle-based bio-barcode-amplification method. Nanomedicine (Lond) 3, 293-303 (2008).

105. C. S. Thaxton, R. Elghanian, A. D. Thomas, S. I. Stoeva, J. S. Lee, N. D. Smith, A. J. Schaeffer, H. Klocker, W. Horninger, G. Bartsch, C. A. Mirkin, Nanoparticle- based bio-barcode assay redefines "undetectable" PSA and biochemical recurrence after radical prostatectomy. Proc Natl Acad Sci U S A 106, 18437- 18442 (2009).

106. L. M. Lai, I. Y. Goon, K. Chuah, M. Lim, F. Braet, R. Amal, J. J. Gooding, The biochemiresistor: an ultrasensitive biosensor for small organic molecules. Angew Chem Int Ed Engl 51, 6456-6459 (2012).

107. A. K. Shah, M. M. Hill, M. J. Shiddiky, M. Trau, Electrochemical detection of glycan and protein epitopes of glycoproteins in serum. Analyst 139, 5970-5976 (2014).

108. S. A. Hong, J. Kwon, D. Kim, S. Yang, A rapid, sensitive and selective electrochemical biosensor with concanavalin A for the preemptive detection of norovirus. Biosens Bioelectron 64, 338-344 (2015).

109. D. Liu, P. Luo, W. Sun, L. Zhang, Z. Wang, Detection of beta-glucans using an amperometric biosensor based on high-affinity interaction between Dectin-1 and beta-glucans. Anal Biochem 404, 14-20 (2010).

110. T. Kong, R. Su, B. Zhang, Q. Zhang, G. Cheng, CMOS-compatible, label-free silicon-nanowire biosensors to detect cardiac troponin I for acute myocardial infarction diagnosis. Biosens Bioelectron 34, 267-272 (2012).

111. F. Y. Yeh, T. Y. Liu, I. H. Tseng, C. W. Yang, L. C. Lu, C. S. Lin, Gold nanoparticles conjugates-amplified aptamer immunosensing screen-printed carbon electrode strips for thrombin detection. Biosens Bioelectron 61, 336-343 (2014).

112. J. A. Ho, H. C. Chang, N. Y. Shih, L. C. Wu, Y. F. Chang, C. C. Chen, C. Chou, Diagnostic detection of human lung cancer-associated antigen using a gold nanoparticle-based electrochemical immunosensor. Anal Chem 82, 5944-5950 (2010).

144

113. S. O. Kelley, C. A. Mirkin, D. R. Walt, R. F. Ismagilov, M. Toner, E. H. Sargent, Advancing the speed, sensitivity and accuracy of biomolecular detection using multi-length-scale engineering. Nat Nanotechnol 9, 969-980 (2014).

114. D. M. Rissin, C. W. Kan, T. G. Campbell, S. C. Howes, D. R. Fournier, L. Song, T. Piech, P. P. Patel, L. Chang, A. J. Rivnak, E. P. Ferrell, J. D. Randall, G. K. Provuncher, D. R. Walt, D. C. Duffy, Single-molecule enzyme-linked immunosorbent assay detects serum proteins at subfemtomolar concentrations. Nat Biotechnol 28, 595-599 (2010).

115. R. M. Mohamadi, J. D. Besant, A. Mepham, B. Green, L. Mahmoudian, T. Gibbs, I. Ivanov, A. Malvea, J. Stojcic, A. L. Allan, L. E. Lowes, E. H. Sargent, R. K. Nam, S. O. Kelley, Nanoparticle-mediated binning and profiling of heterogeneous circulating tumor cell subpopulations. Angew Chem Int Ed Engl 54, 139-143 (2015).

116. E. Ozkumur, A. M. Shah, J. C. Ciciliano, B. L. Emmink, D. T. Miyamoto, E. Brachtel, M. Yu, P. I. Chen, B. Morgan, J. Trautwein, A. Kimura, S. Sengupta, S. L. Stott, N. M. Karabacak, T. A. Barber, J. R. Walsh, K. Smith, P. S. Spuhler, J. P. Sullivan, R. J. Lee, D. T. Ting, X. Luo, A. T. Shaw, A. Bardia, L. V. Sequist, D. N. Louis, S. Maheswaran, R. Kapur, D. A. Haber, M. Toner, Inertial focusing for tumor antigen-dependent and -independent sorting of rare circulating tumor cells. Sci Transl Med 5, 179ra147 (2013).

117. L. M. Zanoli, R. D'Agata, G. Spoto, Functionalized gold nanoparticles for ultrasensitive DNA detection. Anal Bioanal Chem 402, 1759-1771 (2012).

118. Z. H. Mo, X. L. Wei, Toward hybridization assays without PCR using universal nanoamplicons. Anal Bioanal Chem 386, 2219-2223 (2006).

119. H. Yang, A. Hui, G. Pampalakis, L. Soleymani, F. F. Liu, E. H. Sargent, S. O. Kelley, Direct, electronic microRNA detection for the rapid determination of differential expression profiles. Angew Chem Int Ed Engl 48, 8461-8464 (2009).

120. J. Zhao, S. Tang, J. Storhoff, S. Marla, Y. P. Bao, X. Wang, E. Y. Wong, V. Ragupathy, Z. Ye, I. K. Hewlett, Multiplexed, rapid detection of H5N1 using a PCR-free nanoparticle-based genomic microarray assay. BMC Biotechnol 10, 74 (2010).

121. P. Valentini, R. Fiammengo, S. Sabella, M. Gariboldi, G. Maiorano, R. Cingolani, P. P. Pompa, Gold-nanoparticle-based colorimetric discrimination of cancer-

145

related point mutations with picomolar sensitivity. ACS Nano 7, 5530-5538 (2013).

122. K. J. Liu, T. H. Wang, PCR-free, microfluidic single molecule analysis of circulating nucleic acids in lung cancer patient serum. Conf Proc IEEE Eng Med Biol Soc 2011, 8392-8395 (2011).

123. S. Dubus, J. F. Gravel, B. Le Drogoff, P. Nobert, T. Veres, D. Boudreau, PCR- free DNA detection using a magnetic bead-supported polymeric transducer and microelectromagnetic traps. Anal Chem 78, 4457-4464 (2006).

124. Y. V. Gerasimova, D. M. Kolpashchikov, Folding of 16S rRNA in a signal- producing structure for the detection of bacteria. Angew Chem Int Ed Engl 52, 10586-10588 (2013).

125. J. Y. Choi, Y. T. Kim, T. S. Seo, Polymerase chain reaction-free variable-number tandem repeat typing using gold nanoparticle-DNA monoconjugates. ACS Nano 7, 2627-2633 (2013).

126. L. Sun, J. Irudayaraj, PCR-free quantification of multiple splice variants in a cancer gene by surface-enhanced Raman spectroscopy. J Phys Chem B 113, 14021-14025 (2009).

127. B. Lam, Z. Fang, E. H. Sargent, S. O. Kelley, Polymerase chain reaction-free, sample-to-answer bacterial detection in 30 minutes with integrated cell lysis. Anal Chem 84, 21-25 (2012).

128. L. Esfandiari, M. Lorenzini, G. Kocharyan, H. G. Monbouquette, J. J. Schmidt, Sequence-specific DNA detection at 10 fM by electromechanical signal transduction. Anal Chem 86, 9638-9643 (2014).

129. R. Tavallaie, S. R. De Almeida, J. J. Gooding, Toward biosensors for the detection of circulating microRNA as a cancer biomarker: an overview of the challenges and successes. Wiley Interdiscip Rev Nanomed Nanobiotechnol, (2014).

130. W. Zhu, X. Su, X. Gao, Z. Dai, X. Zou, A label-free and PCR-free electrochemical assay for multiplexed microRNA profiles by ligase chain reaction coupling with quantum dots barcodes. Biosens Bioelectron 53, 414-419 (2014).

146

131. Y. Wen, H. Pei, Y. Shen, J. Xi, M. Lin, N. Lu, X. Shen, J. Li, C. Fan, DNA Nanostructure-based Interfacial engineering for PCR-free ultrasensitive electrochemical analysis of microRNA. Sci Rep 2, 867 (2012).

132. Z. Gao, A highly sensitive electrochemical assay for microRNA expression profiling. Analyst 137, 1674-1679 (2012).

133. G. Zheng, L. Qin, C. A. Mirkin, Spectroscopically enhancing electrical nanotraps. Angew Chem Int Ed Engl 47, 1938-1941 (2008).

134. X. Chen, S. Roy, Y. Peng, Z. Gao, Electrical sensor array for polymerase chain reaction-free messenger RNA expression profiling. Anal Chem 82, 5958-5964 (2010).

135. A. C. Lee, Z. Dai, B. Chen, H. Wu, J. Wang, A. Zhang, L. Zhang, T. M. Lim, Y. Lin, Electrochemical branched-DNA assay for polymerase chain reaction-free detection and quantification of oncogenes in messenger RNA. Anal Chem 80, 9402-9410 (2008).

136. A. P. Turner, J. C. Pickup, Diabetes mellitus: biosensors for research and management. Biosensors 1, 85-115 (1985).

137. Y. Xiang, Y. Lu, Using personal glucose meters and functional DNA sensors to quantify a variety of analytical targets. Nat Chem 3, 697-703 (2011).

138. C. Ding, Q. Zhang, J. M. Lin, S. S. Zhang, Electrochemical detection of DNA hybridization based on bio-bar code method. Biosens Bioelectron 24, 3140-3143 (2009).

139. R. Gasparac, B. J. Taft, M. A. Lapierre-Devlin, A. D. Lazareck, J. M. Xu, S. O. Kelley, Ultrasensitive electrocatalytic DNA detection at two- and three- dimensional nanoelectrodes. J Am Chem Soc 126, 12270-12271 (2004).

140. C. Fan, K. W. Plaxco, A. J. Heeger, Electrochemical interrogation of conformational changes as a reagentless method for the sequence-specific detection of DNA. Proc Natl Acad Sci U S A 100, 9134-9137 (2003).

141. M. Lin, Y. Wen, L. Li, H. Pei, G. Liu, H. Song, X. Zuo, C. Fan, Q. Huang, Target- responsive, DNA nanostructure-based E-DNA sensor for microRNA analysis. Anal Chem 86, 2285-2288 (2014).

147

142. H. Z. Yu, Y. Li, L. M. Ou, Reading disc-based bioassays with standard computer drives. Acc Chem Res 46, 258-268 (2013).

143. J. Zhang, S. Song, L. Zhang, L. Wang, H. Wu, D. Pan, C. Fan, Sequence- specific detection of femtomolar DNA via a chronocoulometric DNA sensor (CDS): effects of nanoparticle-mediated amplification and nanoscale control of DNA assembly at electrodes. J Am Chem Soc 128, 8575-8580 (2006).

144. J. Wang, G. Liu, A. Merkoci, Electrochemical coding technology for simultaneous detection of multiple DNA targets. J Am Chem Soc 125, 3214-3215 (2003).

145. V. M. Pierce, R. L. Hodinka, Comparison of the GenMark Diagnostics eSensor respiratory viral panel to real-time PCR for detection of respiratory viruses in children. J Clin Microbiol 50, 3458-3465 (2012).

146. S. B. Griesemer, R. Holmberg, C. G. Cooney, N. Thakore, A. Gindlesperger, C. Knickerbocker, D. P. Chandler, K. St George, Automated, simple, and efficient influenza RNA extraction from clinical respiratory swabs using TruTip and epMotion. J Clin Virol 58, 138-143 (2013).

147. D. Bechlian, A. Honstettre, M. Terrier, C. Brest, C. Malenfant, M. J. Mozziconacci, C. Chabannon, RNA extracted from blood samples with a rapid automated procedure is fit for molecular diagnosis or minimal residual disease monitoring in patients with a variety of malignant blood disorders. Biopreserv Biobank 7, 123-128 (2009).

148. J. Loeffler, P. Swatoch, D. Akhawi-Araghi, H. Hebart, H. Einsele, Automated RNA extraction by MagNA Pure followed by rapid quantification of cytokine and chemokine gene expression with use of fluorescence resonance energy transfer. Clin Chem 49, 955-958 (2003).

149. L. Van Heirstraeten, P. Spang, C. Schwind, K. S. Drese, M. Ritzi-Lehnert, B. Nieto, M. Camps, B. Landgraf, F. Guasch, A. H. Corbera, J. Samitier, H. Goossens, S. Malhotra-Kumar, T. Roeser, Integrated DNA and RNA extraction and purification on an automated microfluidic cassette from bacterial and viral pathogens causing community-acquired lower respiratory tract infections. Lab Chip 14, 1519-1526 (2014).

150. G. Xu, T. M. Hsieh, D. Y. Lee, E. M. Ali, H. Xie, X. L. Looi, E. S. Koay, M. H. Li, J. Y. Ying, A self-contained all-in-one cartridge for sample preparation and real-time PCR in rapid influenza diagnosis. Lab Chip 10, 3103-3111 (2010).

148

151. M. Gueudin, J. C. Plantier, V. Lemee, M. P. Schmitt, L. Chartier, T. Bourlet, A. Ruffault, F. Damond, M. Vray, F. Simon, Evaluation of the Roche Cobas TaqMan and Abbott RealTime extraction-quantification systems for HIV-1 subtypes. J Acquir Immune Defic Syndr 44, 500-505 (2007).

152. M. Madi, A. Hamilton, D. Squirrell, V. Mioulet, P. Evans, M. Lee, D. P. King, Rapid detection of foot-and-mouth disease virus using a field-portable nucleic acid extraction and real-time PCR amplification platform. Vet J 193, 67-72 (2012).

153. C. Dye, Global epidemiology of tuberculosis. Lancet 367, 938-940 (2006).

154. D. Hillemann, S. Rusch-Gerdes, C. Boehme, E. Richter, Rapid molecular detection of extrapulmonary tuberculosis by the automated GeneXpert MTB/RIF system. J Clin Microbiol 49, 1202-1205 (2011).

155. J. Du, Z. Huang, Q. Luo, G. Xiong, X. Xu, W. Li, X. Liu, J. Li, Rapid diagnosis of pleural tuberculosis by Xpert MTB/RIF assay using pleural biopsy and pleural fluid specimens. J Res Med Sci 20, 26-31 (2015).

156. B. V. A. Devogelaere, K.; De Baere, I.; Holemans, P.; Ivens, T.; Claes, B.; Rondelez, E.; Vandercruyssen, G.; Van der Auwera, M.; Kockx, M.; Vanden Bempt, I.; Vandenbroucke, I.; Maertens, G.; Sablon, E., in 104th Annual Meeting of the American Association for Cancer Research. (Cancer Research, Washington, DC. Philadelphia (PA), 2013), vol. 73(8 Suppl), pp. 4213.

157. Z. Fang, L. Soleymani, G. Pampalakis, M. Yoshimoto, J. A. Squire, E. H. Sargent, S. O. Kelley, Direct profiling of cancer biomarkers in tumor tissue using a multiplexed nanostructured microelectrode integrated circuit. ACS Nano 3, 3207-3213 (2009).

158. L. Soleymani, Z. Fang, E. H. Sargent, S. O. Kelley, Programming the detection limits of biosensors through controlled nanostructuring. Nat Nanotechnol 4, 844- 848 (2009).

159. D. A. LaVan, P. M. George, R. Langer, Simple, three-dimensional microfabrication of electrodeposited structures. Angew Chem Int Ed Engl 42, 1262-1265 (2003).

160. X. Bin, E. H. Sargent, S. O. Kelley, Nanostructuring of sensors determines the efficiency of biomolecular capture. Anal Chem 82, 5928-5931 (2010).

149

161. E. Vasilyeva, B. Lam, Z. Fang, M. D. Minden, E. H. Sargent, S. O. Kelley, Direct genetic analysis of ten cancer cells: tuning sensor structure and molecular probe design for efficient mRNA capture. Angew Chem Int Ed Engl 50, 4137-4141 (2011).

162. B. Lam, R. D. Holmes, J. Das, M. Poudineh, A. Sage, E. H. Sargent, S. O. Kelley, Optimized templates for bottom-up growth of high-performance integrated biomolecular detectors. Lab Chip 13, 2569-2575 (2013).

163. J. C. Love, D. B. Wolfe, R. Haasch, M. L. Chabinyc, K. E. Paul, G. M. Whitesides, R. G. Nuzzo, Formation and structure of self-assembled monolayers of alkanethiolates on palladium. J Am Chem Soc 125, 2597-2609 (2003).

164. P. E. Laibinis, J. J. Hickman, M. S. Wrighton, G. M. Whitesides, Orthogonal self- assembled monolayers: alkanethiols on gold and alkane carboxylic acids on alumina. Science 245, 845-847 (1989).

165. A. B. Steel, T. M. Herne, M. J. Tarlov, Electrochemical quantitation of DNA immobilized on gold. Anal Chem 70, 4670-4677 (1998).

166. A. B. Steel, T. M. Herne, M. J. Tarlov, Electrostatic interactions of redox cations with surface-immobilized and solution DNA. Bioconjug Chem 10, 419-423 (1999).

167. A. B. Steel, R. L. Levicky, T. M. Herne, M. J. Tarlov, Immobilization of nucleic acids at solid surfaces: effect of oligonucleotide length on layer assembly. Biophys J 79, 975-981 (2000).

168. M. Steichen, Y. Decrem, E. Godfroid, C. Buess-Herman, Electrochemical DNA hybridization detection using peptide nucleic acids and [Ru(NH3)6]3+ on gold electrodes. Biosens Bioelectron 22, 2237-2243 (2007).

169. M. A. Lapierre, M. O'Keefe, B. J. Taft, S. O. Kelley, Electrocatalytic detection of pathogenic DNA sequences and antibiotic resistance markers. Anal Chem 75, 6327-6333 (2003).

170. M. A. Lapierre-Devlin, C. L. Asher, B. J. Taft, R. Gasparac, M. A. Roberts, S. O. Kelley, Amplified electrocatalysis at DNA-modified nanowires. Nano Lett 5, 1051- 1055 (2005).

171. C. C. Pritchard, H. H. Cheng, M. Tewari, MicroRNA profiling: approaches and considerations. Nat Rev Genet 13, 358-369 (2012).

150

172. I. Ivanov, J. Stojcic, A. Stanimirovic, E. Sargent, R. K. Nam, S. O. Kelley, Chip- based nanostructured sensors enable accurate identification and classification of circulating tumor cells in prostate cancer patient blood samples. Anal Chem 85, 398-403 (2013).

173. L. Soleymani, Z. Fang, B. Lam, X. Bin, E. Vasilyeva, A. J. Ross, E. H. Sargent, S. O. Kelley, Hierarchical nanotextured microelectrodes overcome the molecular transport barrier to achieve rapid, direct bacterial detection. ACS Nano 5, 3360- 3366 (2011).

174. J. Das, S. O. Kelley, Protein detection using arrayed microsensor chips: tuning sensor footprint to achieve ultrasensitive readout of CA-125 in serum and whole blood. Anal Chem 83, 1167-1172 (2011).

175. M. Moscovici, A. Bhimji, S. O. Kelley, Rapid and specific electrochemical detection of prostate cancer cells using an aperture sensor array. Lab Chip 13, 940-946 (2013).

176. J. Das, K. B. Cederquist, A. A. Zaragoza, P. E. Lee, E. H. Sargent, S. O. Kelley, An ultrasensitive universal detector based on neutralizer displacement. Nat Chem 4, 642-648 (2012).

177. A. Bhimji, A. A. Zaragoza, L. S. Live, S. O. Kelley, Electrochemical enzyme- linked immunosorbent assay featuring proximal reagent generation: detection of human immunodeficiency virus antibodies in clinical samples. Anal Chem 85, 6813-6819 (2013).

178. J. D. Besant, J. Das, E. H. Sargent, S. O. Kelley, Proximal bacterial lysis and detection in nanoliter wells using electrochemistry. ACS Nano 7, 8183-8189 (2013).

179. B. Lam, J. Das, R. D. Holmes, L. Live, A. Sage, E. H. Sargent, S. O. Kelley, Solution-based circuits enable rapid and multiplexed pathogen detection. Nat Commun 4, 2001 (2013).

180. D. R. Walt, Chemistry. Miniature analytical methods for medical diagnostics. Science 308, 217-219 (2005).

181. Y. G. Zhou, Y. Wan, A. T. Sage, M. Poudineh, S. O. Kelley, Effect of microelectrode structure on electrocatalysis at nucleic acid-modified sensors. Langmuir 30, 14322-14328 (2014).

151

182. H. D. Hill, J. E. Millstone, M. J. Banholzer, C. A. Mirkin, The role radius of curvature plays in thiolated oligonucleotide loading on gold nanoparticles. ACS Nano 3, 418-424 (2009).

183. T. M. Squires, R. J. Messinger, S. R. Manalis, Making it stick: convection, reaction and diffusion in surface-based biosensors. Nat Biotechnol 26, 417-426 (2008).

184. L. Soleymani, Z. Fang, X. Sun, H. Yang, B. J. Taft, E. H. Sargent, S. O. Kelley, Nanostructuring of patterned microelectrodes to enhance the sensitivity of electrochemical nucleic acids detection. Angew Chem Int Ed Engl 48, 8457-8460 (2009).

185. M. Moreno-Smith, J. B. Halder, P. S. Meltzer, T. A. Gonda, L. S. Mangala, R. Rupaimoole, C. Lu, A. S. Nagaraja, K. M. Gharpure, Y. Kang, C. Rodriguez- Aguayo, P. E. Vivas-Mejia, B. Zand, R. Schmandt, H. Wang, R. R. Langley, N. B. Jennings, C. Ivan, J. E. Coffin, G. N. Armaiz, J. Bottsford-Miller, S. B. Kim, M. S. Halleck, M. J. Hendrix, W. Bornman, M. Bar-Eli, J. S. Lee, Z. H. Siddik, G. Lopez- Berestein, A. K. Sood, ATP11B mediates platinum resistance in ovarian cancer. J Clin Invest 123, 2119-2130 (2013).

186. R. D. Barber, D. W. Harmer, R. A. Coleman, B. J. Clark, GAPDH as a housekeeping gene: analysis of GAPDH mRNA expression in a panel of 72 human tissues. Physiol Genomics 21, 389-395 (2005).

187. A. P. A. Brown, F.C., Cyclic and differential pulse voltammetric behavior of reactants confined to the electrode surface. Anal Chem 49, 1589-1595 (1977).

188. P. E. Sheehan, L. J. Whitman, Detection limits for nanoscale biosensors. Nano Lett 5, 803-807 (2005).

189. Toronto Lung Transplant Group. Unilateral lung transplantation for pulmonary fibrosis. N Engl J Med 314, 1140-1145 (1986).

190. S. M. Studer, R. D. Levy, K. McNeil, J. B. Orens, Lung transplant outcomes: a review of survival, graft function, physiology, health-related quality of life and cost-effectiveness. Eur Respir J 24, 674-685 (2004).

191. J. De Meester, J. M. Smits, G. G. Persijn, A. Haverich, Listing for lung transplantation: life expectancy and transplant effect, stratified by type of end-

152

stage lung disease, the Eurotransplant experience. J Heart Lung Transplant 20, 518-524 (2001).

192. J. D. Christie, L. B. Edwards, P. Aurora, F. Dobbels, R. Kirk, A. O. Rahmel, D. O. Taylor, A. Y. Kucheryavaya, M. I. Hertz, Registry of the International Society for Heart and Lung Transplantation: twenty-fifth official adult lung and heart/lung transplantation report--2008. J Heart Lung Transplant 27, 957-969 (2008).

193. J. D. Christie, J. S. Sager, S. E. Kimmel, V. N. Ahya, C. Gaughan, N. P. Blumenthal, R. M. Kotloff, Impact of primary graft failure on outcomes following lung transplantation. Chest 127, 161-165 (2005).

194. B. F. Meyers, M. de la Morena, S. C. Sweet, E. P. Trulock, T. J. Guthrie, E. N. Mendeloff, C. Huddleston, J. D. Cooper, G. A. Patterson, Primary graft dysfunction and other selected complications of lung transplantation: A single- center experience of 983 patients. J Thorac Cardiovasc Surg 129, 1421-1429 (2005).

195. J. D. Punch, D. H. Hayes, F. B. LaPorte, V. McBride, M. S. Seely, Organ donation and utilization in the United States, 1996-2005. Am J Transplant 7, 1327-1338 (2007).

196. S. C. Charman, L. D. Sharples, K. D. McNeil, J. Wallwork, Assessment of survival benefit after lung transplantation by patient diagnosis. J Heart Lung Transplant 21, 226-232 (2002).

197. G. Thabut, H. Mal, J. Cerrina, P. Dartevelle, C. Dromer, J. F. Velly, M. Stern, P. Loirat, G. Leseche, M. Bertocchi, J. F. Mornex, A. Haloun, P. Despins, C. Pison, D. Blin, M. Reynaud-Gaubert, Graft ischemic time and outcome of lung transplantation: a multicenter analysis. Am J Respir Crit Care Med 171, 786-791 (2005).

198. T. N. Machuca, M. Cypel, J. C. Yeung, R. Bonato, R. Zamel, M. Chen, S. Azad, M. K. Hsin, T. Saito, Z. Guan, T. K. Waddell, M. Liu, S. Keshavjee, Protein expression profiling predicts graft performance in clinical ex vivo lung perfusion. Ann Surg 261, 591-597 (2015).

199. D. A. Lawrence, Transforming growth factor-beta: a general review. Eur Cytokine Netw 7, 363-374 (1996).

153

200. S. E. Verleden, E. Vandermeulen, D. Ruttens, R. Vos, A. Vaneylen, L. J. Dupont, D. E. Van Raemdonck, B. M. Vanaudenaerde, G. M. Verleden, Neutrophilic reversible allograft dysfunction (NRAD) and restrictive allograft syndrome (RAS). Semin Respir Crit Care Med 34, 352-360 (2013).

201. H. Nakazato, H. Oku, S. Yamane, Y. Tsuruta, R. Suzuki, A novel anti-fibrotic agent pirfenidone suppresses tumor necrosis factor-alpha at the translational level. Eur J Pharmacol 446, 177-185 (2002).

202. P. W. Noble, C. Albera, W. Z. Bradford, U. Costabel, M. K. Glassberg, D. Kardatzke, T. E. King, Jr., L. Lancaster, S. A. Sahn, J. Szwarcberg, D. Valeyre, R. M. du Bois, C. S. Group, Pirfenidone in patients with idiopathic pulmonary fibrosis (CAPACITY): two randomised trials. Lancet 377, 1760-1769 (2011).

203. K. Miyazono, U. Hellman, C. Wernstedt, C. H. Heldin, Latent high molecular weight complex of transforming growth factor beta 1. Purification from human platelets and structural characterization. J Biol Chem 263, 6407-6415 (1988).

204. E. Csaszar, K. Chen, J. Caldwell, W. Chan, P. W. Zandstra, Real-time monitoring and control of soluble signaling factors enables enhanced progenitor cell outputs from human cord blood stem cell cultures. Biotechnol Bioeng 111, 1258-1264 (2014).

205. R. M. Lyons, J. Keski-Oja, H. L. Moses, Proteolytic activation of latent transforming growth factor-beta from fibroblast-conditioned medium. J Cell Biol 106, 1659-1665 (1988).

206. T. N. Machuca, M. Cypel, Y. Zhao, H. Grasemann, F. Tavasoli, J. C. Yeung, R. Bonato, M. Chen, R. Zamel, Y. M. Chun, Z. Guan, M. de Perrot, T. K. Waddell, M. Liu, S. Keshavjee, The role of the endothelin-1 pathway as a biomarker for donor lung assessment in clinical ex vivo lung perfusion. J Heart Lung Transplant, (2015).

207. A. H. Chester, M. H. Yacoub, The role of endothelin-1 in pulmonary arterial hypertension. Glob Cardiol Sci Pract 2014, 62-78 (2014).

208. M. Uguccioni, L. Pulsatelli, B. Grigolo, A. Facchini, L. Fasano, C. Cinti, M. Fabbri, G. Gasbarrini, R. Meliconi, Endothelin-1 in idiopathic pulmonary fibrosis. J Clin Pathol 48, 330-334 (1995).

154

209. H. Matsumoto, N. Suzuki, H. Onda, M. Fujino, Abundance of endothelin-3 in rat intestine, pituitary gland and brain. Biochem Biophys Res Commun 164, 74-80 (1989).

210. C. Frelin, D. Guedin, Why are circulating concentrations of endothelin-1 so low? Cardiovasc Res 28, 1613-1622 (1994).

211. M. E. Ivey, N. Osman, P. J. Little, Endothelin-1 signalling in vascular smooth muscle: pathways controlling cellular functions associated with atherosclerosis. Atherosclerosis 199, 237-247 (2008).

212. H. Martynowicz, A. Janus, D. Nowacki, G. Mazur, The role of chemokines in hypertension. Adv Clin Exp Med 23, 319-325 (2014).

213. B. W. Christman, C. D. McPherson, J. H. Newman, G. A. King, G. R. Bernard, B. M. Groves, J. E. Loyd, An imbalance between the excretion of thromboxane and prostacyclin metabolites in pulmonary hypertension. N Engl J Med 327, 70-75 (1992).

214. V. V. McLaughlin, M. D. McGoon, Pulmonary arterial hypertension. Circulation 114, 1417-1431 (2006).

215. R. M. Tuder, C. D. Cool, M. W. Geraci, J. Wang, S. H. Abman, L. Wright, D. Badesch, N. F. Voelkel, Prostacyclin synthase expression is decreased in lungs from patients with severe pulmonary hypertension. Am J Respir Crit Care Med 159, 1925-1932 (1999).

216. D. O. Taylor, L. B. Edwards, M. M. Boucek, E. P. Trulock, P. Aurora, J. Christie, F. Dobbels, A. O. Rahmel, B. M. Keck, M. I. Hertz, Registry of the International Society for Heart and Lung Transplantation: twenty-fourth official adult heart transplant report--2007. J Heart Lung Transplant 26, 769-781 (2007).

217. M. Salama, O. Andrukhova, P. Jaksch, S. Taghavi, W. Kelpetko, G. Dekan, S. Aharinejad, Endothelin-1 governs proliferation and migration of bronchoalveolar lavage-derived lung mesenchymal stem cells in bronchiolitis obliterans syndrome. Transplantation 92, 155-162 (2011).

218. A. Inoue, M. Yanagisawa, S. Kimura, Y. Kasuya, T. Miyauchi, K. Goto, T. Masaki, The human endothelin family: three structurally and pharmacologically distinct isopeptides predicted by three separate genes. Proc Natl Acad Sci U S A 86, 2863-2867 (1989).

155

219. J. B. Denault, A. Claing, P. D'Orleans-Juste, T. Sawamura, T. Kido, T. Masaki, R. Leduc, Processing of proendothelin-1 by human furin convertase. FEBS Lett 362, 276-280 (1995).

220. N. Galie, A. Manes, A. Branzi, The endothelin system in pulmonary arterial hypertension. Cardiovasc Res 61, 227-237 (2004).

221. S. M. Kulkarni, Philip; Furstenthal, Laura; Evers, Matthias., "Personalized Medicine-the path forward," Pharmaceuitcal and Medical Products Practice (2013).

222. M. Liu, A. S. Slutsky, Anti-inflammatory therapies: application of molecular biology techniques in intensive care medicine. Intensive Care Med 23, 718-731 (1997).

223. M. Cypel, H. Kaneda, J. C. Yeung, M. Anraku, K. Yasufuku, M. de Perrot, A. Pierre, T. K. Waddell, M. Liu, S. Keshavjee, Increased levels of interleukin-1beta and tumor necrosis factor-alpha in donor lungs rejected for transplantation. J Heart Lung Transplant 30, 452-459 (2011).