UC Riverside UC Riverside Electronic Theses and Dissertations

Title Quantitative Proteomic Analysis of ATP-binding Proteins

Permalink https://escholarship.org/uc/item/7kz1z0vz

Author Miao, Weili

Publication Date 2019

Peer reviewed|Thesis/dissertation

eScholarship.org Powered by the California Digital Library University of California

UNIVERSITY OF CALIFORNIA RIVERSIDE

Quantitative Proteomic Analysis of ATP-binding Proteins

A Dissertation submitted in partial satisfaction of the requirements for the degree of

Doctor of Philosophy

Chemistry

Weili Miao

June 2019

Dissertation Committee:

Dr. Yinsheng Wang, Chairperson Dr. Ryan R. Julian Dr. Jingsong Zhang

The Dissertation of Weili Miao is approved:

Committee Chairperson

University of California, Riverside ACKNOWLEDGMENTS

The past five years are one of the most critical periods of my life. It witnessed the growth of my knowledge, broaden my experience and deepen my understanding in philosophy. I could never finish my Ph.D. smoothly without the guidance, support, help and efforts from a lot of people.

First of all, my deepest gratitude and respect is to my graduate advisor, Prof. Yinsheng

Wang, who provided the opportunity, guidance and financial support in all my projects for my PhD study. During these five years, he gave me many constructive ideas both for my research and future career plan. His endless patience, generosity and the enthusiasm towards work helped me overcome many difficulties during my Ph.D. Without his help and support, it would not be possible for me to complete my doctoral study successfully.

I would like to thank all the members of my dissertation committee, Prof. Ryan R. Julian and Prof. Jingsong Zhang. Their review and evaluation of this work as well as their valuable suggestions and generous help during my PhD study are deeply appreciated. I would also like to thank Prof. Wenwan Zhong, Prof. Huiwang Ai, Prof Quan Cheng and

Prof. Richard Hooley for all their patient help and instructions of the curricula in graduate school. I would like to thank Dr. Songqin Pan in the Proteomics Core for his help and advises in mass spectrometers. I would like to thank the staff members in Mass

Spectrometry Facility, Dr. Jie Zhou and Mr. Ron New for their help with Thermo TSQ vantage use. I would like to thank Prof. Xuemei Chen, Prof. Joseph Genereux, Prof. Min

Xue and Prof. Pingyun Feng for the help with other facility use.

I also appreciate all the help and friendship from current and former members of the

Wang group. Especially, I would like to thank Dr. Lei Guo for her help and guidance with proteomic study when I first started my research here. I would like to thank Dr. Xiaogang

Jiang and Dr Yongsheng Xiao for their help on the instruments set up. I would like to express my sincere thanks to Dr. Lin Li and Dr. Xiaoxia Dai for their help as well as constructive advices for my molecular biology studies, Tianyu Qi for her help in experiments. I would like to thank Dr. Shuo Liu for her help in presentation, Dr. Nathan E.

Price for his help in writing, Shuli Zhai for her lab management and all other former and current group members: Dr. Tao Bing, Dr. Lijuan Fu, Dr. Qian Cai, Dr. Pengcheng Wang,

Dr. Preston Williams, Eric Stephens, Zi Wang, Dr. Nicole Williams, Dr. Ji Jiang, Dr. Yang

Yu, Ming Huang, Dr. Jun Wu, Yuxiang Cui, Jiabin Wu, Lok Ming Tam, Gwendolyn

Gonzalez, David Bade, Xuejiao Dong, Dr. Jiapeng Leng. Dr. Tianlu Wang, Ying Tan, Su

Guo, Dr. Hua Du, Ross Furash, Jiekai Yin, Jun Yuan, Zi Gao, Dr. Xiaochuan Liu, Dr. Feng

Tan, Dr. Xiaomei He, Yinan Wang, Yenyu Yang, Feng Tang, Dr. Yuxiang Sun, Dr. Kailin

Yu and Dr. Quanqing Zhang.

I am also appreciative of fellow graduate students or postdocs in the Department of

Chemistry and Environmental Toxicology and Graduate Program, especially from the Xue,

Chen and Julian group. I’d like to thank Dr. Yonghui Zhao, Dr. Zhonghan Li, Dylan Riggs,

Tyler Lambeth for the helpful discussion, technical support and friendship.

I would also like to thank staff from the Department of Chemistry for providing prompt assistance of administrative matters. They are Ms. Christina Youhas, Dr. Kevin Simpson,

Ms. Barbara Outzen, Ms. Tina Enriquez, Ms. Natasha Gonzales, Mr. Jaime Matute and Mr.

Prisciliano Saavedra. I would also thank the financial support for my PhD study by the

Department of Chemistry, UC Riverside.

I would like to thank my husband, Dr. Chengyu Mao. Without his company, support, love, encouragement and guidance, I would not be able to complete my PhD study.

Finally, I would like to express my deepest gratitude to my parents Mingjun Miao and

Jinlian Ni, my sister Ni Miao for their dedication, love and continuous support., and all of other family members for everything they have done for me.

Best wishes to them all!

COPYRIGHT ACKNOWLEDGEMENTS The text and figures in Chapter 2, in part or full, are a reprint of the material as it appears in Analytical Chemistry 2016, 88, 9773-9779. The coauthor (Dr. Yinsheng Wang) listed in that publication directed and supervised the research that forms the basis of this chapter. The text and figures in Chapter 3, in part or full, are a reprint of the material as it appears in Analytical Chemistry 2019, 91, 3209-3214 and Journal of Proteome Research, April 2019, in press. The coauthor (Dr. Yinsheng Wang) listed in that publication directed and supervised the research that forms the basis of this chapter. The text and figures in Chapter 4, in part or full, are a reprint of the material as it appears in Journal of Proteome Research, 2019, 18, 2279-2286. The coauthor (Dr. Yinsheng Wang) listed in that publication directed and supervised the research that forms the basis of this chapter. The text and figures in Chapter 5, in part or full, are a reprint of the material as it appears in Analytical Chemistry 2018, 90, 6835-6842. The coauthor (Dr. Yinsheng Wang) listed in that publication directed and supervised the research that forms the basis of this chapter. The text and figures in Chapter 7, in part or full, are a reprint of the material as it appears in Analytical Chemistry 2018, 90, 11751-11755. The coauthor (Dr. Yinsheng Wang) listed in that publication directed and supervised the research that forms the basis of this chapter.

vii

DEDICATION

To my family!

viii

ABSTRACT OF THE DISSERTATION

Quantitative Proteomic Analysis of ATP-binding Proteins

Weili Miao

Doctor of Philosophy, Graduate Program in Chemistry University of California, Riverside, June 2019 Dr. Yinsheng Wang, Chairperson

Targeted proteomics techniques, which rely on multiple-reaction monitoring (MRM) or parallel-reaction monitoring (PRM), have become extensively employed in quantitative proteomics studies. In this dissertation, we developed and employed targeted proteomics methods to the analyze comprehensively ATP-binding proteins, including kinases, heat shock proteins and helicases.

In Chapter 2, we expanded the kinome MRM library to include ~80% of the human kinome and employed this library, together with an ATP-affinity probe, for profiling comprehensively alterations of the kinases upon treatment with methylglyoxal. Our results led to the quantification of 328 unique kinases and the novel discovery of kinases involved in diabetes signaling pathways.

In Chapter 3, we developed a PRM-based targeted proteomic method to monitor the protein expression of ~ 80% of the human kinome. By employing this method, together with the method in Chapter 2, we assessed the alterations in protein expression and ATP

ix binding affinities of over 300 kinases in cultured human cells elicited by three FDA- approved small-molecule kinase inhibitors (dabrafenib, vemurafenib and imatinib). We identified CHK1 and MAP2K5 as novel target kinases for imatinib and vemurafenib, respectively. We also employed the PRM-based targeted proteomic method to examine the reprogramming of the human kinome during colorectal cancer (CRC) metastasis in Chapter

4 and discovered phosphoribosyl pyrophosphate synthetase 2 (PRPS2) as a promoter for

CRC metastasis.

In Chapter 5, we developed a PRM-based targeted proteomic method to quantitatively assess 70% of the human heat shock proteome and applied this method to analyze differential expression of heat shock proteins in three matched primary/metastatic pairs of melanoma cell lines. We were able to discover DNAJB4 as a suppressor for melanoma metastasis. We also found the expression of DNAJB4 was stimulated by three HSP90 inhibitors treatment in Chapter 6. We then demonstrated that the elevated expression of

DNAJB4 occurs, in part, through an epitranscriptomic mechanism.

In Chapter 7, we developed a PRM-based targeted proteomic method that allows for the quantification of >80% of the human helicase proteome. By employing this method, together with a CRISPR-Cas9 genome editing method, we discovered that helicases may constitute an important group of client proteins for HSP90.

TABLE OF CONTENTS

ACKNOWLEDGMENTS ...... iv

DEDICATION ...... viii

ABSTRACT OF THE DISSERTATION ...... ix

TABLE OF CONTENTS ...... xi

LIST OF FIGURES ...... xvi

LIST OF TABLES ...... xxix

Chapter 1. General Overview ...... 1

1.1 Introduction ...... 1

1.2 MS-based quantitative proteomics ...... 3

1.2.1 Label-free quantification ...... 3

1.2.2 SILAC ...... 3

1.2.3 Isotope-coded affinity tag (ICAT) ...... 5

1.2.4 Isobaric tag for relative and absolute quantitation (iTRAQ) ...... 6

1.3 Protein quantification in discovery and targeted MS mode ...... 7

1.3.1 Discovery proteomics ...... 7

1.3.2 Targeted proteomics ...... 9

1.3.3 Scheduled targeted proteomics ...... 10

1.4 ATP-binding proteins ...... 11

1.4.1 Human kinome analysis ...... 12

1.4.2 Other ATP-binding proteins analysis ...... 13

1.5 Scope of the dissertation ...... 14

References ...... 18

Chapter 2. A High-throughput Targeted Proteomic Approach for Comprehensive Profiling of Methylglyoxal-induced Perturbations of the Human Kinome ...... 30

Introduction ...... 30

Materials and Methods ...... 32

Results ...... 39

1. Development of an MRM Assay for Human Kinome Profiling ...... 40

2. Methylglyoxal-induced alterations of human kinome in HEK293T cells ...... 43

3. MG led to the down-regulation of RTKs...... 45

Discussion ...... 47

References ...... 49

Chapter 3. Comprehensive Analysis of the Human Kinome Perturbed by Clinical Kinase

Inhibitors ...... 64

Introduction ...... 64

Materials and Methods ...... 66

Results ...... 75

1. Targeted Proteomic Methods for Assessing the Proteome-Wide Alterations in Protein

Expression and ATP-binding Affinity of Kinases in Response to Inhibitor

Treatment ...... 75

2. Profound Alterations in Protein Expression and ATP Binding Affinities of Kinases

Elicited by Kinase Inhibitors ...... 76

xii

3. CHK1 is a Novel Target Kinase of Imatinib ...... 79

4. Vemurafenib Suppresses the ATP-binding Affinity of MAP2K5 ...... 79

Conclusions ...... 81

References ...... 84

Chapter 4. Targeted Quantitative Kinome Analysis Identifies PRPS2 as a Promoter for

Colorectal Cancer Metastasis ...... 106

Introduction ...... 106

Materials and methods ...... 107

Results ...... 113

1. Differential Expression of Kinase Proteins in Primary and Metastatic Human CRC

Cells ...... 113

2. PRPS2 Drives CRC Metastasis ...... 114

3. PRPS2 Modulates the Expression of MMP-9 and E-cadherin ...... 115

4. PRPS2 and Myc...... 116

Conclusions ...... 117

References ...... 118

Chapter 5. A Targeted Proteomic Approach for Heat Shock Proteins Reveals DNAJB4 as a Suppressor for Melanoma Metastasis ...... 134

Introduction ...... 134

Materials and Methods ...... 135

Results ...... 141

xiii

1. Development of a High-throughput PRM Method for the Quantitative Analysis of

Heat Shock Proteome ...... 141

2. Scheduled LC-PRM Analysis Revealed Differential Expression of Heat Shock

Proteins during Melanoma Metastasis...... 143

3. DNAJB4 is Commonly Down-regulated in Metastatic Melanoma Cells and It

Modulates the Invasive Capabilities of Cultured Melanoma Cells ...... 145

4. DNAJB4 Suppresses Melanoma Cell Invasion by Regulating Matrix

Metalloproteinases (MMPs)...... 146

Discussion ...... 148

References ...... 151

Chapter 6. HSP90 Inhibitors Stimulate DNAJB4 Protein Expression through an

Epitranscriptomic Mechanism ...... 167

Introduction ...... 167

Materials and methods ...... 168

Results ...... 177

1. Treatment with HSP90 inhibitors led to elevated expression of a number of heat shock

proteins through a post-transcriptional mechanism ...... 177

2. HSP90 inhibitor-stimulated increase in DNAJB4 protein expression involves an

m6A-based epitranscriptomic mechanism...... 180

3. m6A modification at site 114 on 5′-UTR promotes the translation of DNAJB4 ..... 182

4. Heat shock stress stimulated DNAJB4 protein expression through an m6A-based

epitranscriptomic mechanism...... 183

xiv

Discussion ...... 184

References ...... 185

Chapter 7. Identification of Helicase Proteins as Clients for HSP90 ...... 216

Introduction ...... 216

Materials and Methods ...... 217

Results ...... 222

References ...... 228

LIST OF FIGURES

Figure 1.1. General procedures of label-free (a) and SILAC (b) experiment for quantitative proteomics ...... 23

Figure 1.2. ICAT quantitative proteomics. (a) The structure of ICAT reagent (adapted from Nat Protoc. 2007, 1, 2650). (b) General procedures of ICAT experiment for quantitative proteomics ...... 24

Figure 1.3. iTRAQ quantitative proteomics. (a) The structure of iTRAQ reagent. (b) General procedures of iTRAQ experiment for quantitative proteomics ...... 25

Figure 1.4. Typical MS/MS scan events in mass spectrometer for discovery proteomics in DDA (adopted from Mol Cell Proteomics. 2009, 8, 2759.) (a) and DIA (adopted from Mol Cell Proteomics. 2012, 11, O111.016717.) (b) mode...... 26

Figure 1.5. Targeted proteomics analysis. (A) MRM analysis on a triple quadrupole (QQQ) mass spectrometer and (b) PRM analysis on a Q Exactive mass spectrometer ...... 27

Figure 1.6. Scheduled MRM or PRM analysis. Schematic diagrams illustrate the empirical determination of iRT scale and conversion of retention times for targeted peptides into iRTs (a) and calculation of the predicted retention time from reference peptides (b) ...... 28

Figure 1.7. The dendrogram of the human kinome (adopted from Science. 2002, 298, 1912.) ...... 29

Figure 2.1. A targeted proteomic approach for interrogating the human kinome. (a) The working principle of ICAP probe. (b) The chemical structure of the ICAP probe. (c) Linear fit between iRT values of desthiobiotin- and desthiobiotin-C3-labeled peptides. (d) Venn diagram showing the kinome coverage obtained from DDA and MRM analysis ...... 52

Figure 2.2. Experimental strategy for human kinome analysis with the use of ICAP probe and LC-MRM. Shown is the forward labeling experiment, where the light and heavy desthiobiotin-C3-ATP probe were treated with lysate of MG-treated cells and control cells, respectively ...... 53

Figure 2.3. MG-induced alterations of human kinome in HEK293T cells ...... 54

Figure 2.4. Representative MRM results for EGFR. (a) Correlation between iRT in library and measured RT on a TSQ Vantage mass spectrometer. (b) MRM traces for peptide ITDFGLAK#LLGAEEK from EGFR derived from forward and reverse labeling experiments. (c) MS/MS of a representative peptide ITDFGLAK#LLGAEEK from EGFR (the y5, y7 and y8 ions labeled in red represent the 3 transitions used for MRM analysis).

xvi

(d) Relative abundances of three fragment ions monitored in MRM analysis compared to those in MS/MS acquired from shotgun proteomic analysis. ‘dotp’ represents dot product among relative abundances of 3 fragment ions observed in MRM run and DDA run. ‘F’ and ‘R’ refer to data obtained from forward and reverse labeling experiments. In forward labeling, the light and heavy desthiobiotin-C3-ATP probes were treated with the lysates of MG-treated and control HEK293T cells, respectively. The opposite was conducted for reverse-labeling experiments ...... 55

Figure 2.5. The quantification results for kinases involved in the MAPK pathway (a) and receptor tyrosine kinases (b) that were altered upon MG treatment. The data represent the mean and standard deviation of results obtained from five labeling experiments, which included three forward and two reverse labeling experiments. In forward labeling, the light and heavy desthiobiotin-C3-ATP probes were treated with the lysates of MG-treated and control HEK293T cells, respectively. The opposite was conducted for reverse labeling experiments ...... 57

Figure 2.6. MG treatment led to diminished level of EGFR protein. (a) Changes in mRNA levels of EGFR, IGF2R and FGFR1 genes after treatment with different doses of MG (n = 3). (b) Quantification results showing the dose-dependent changes in expression levels of EGFR protein after MG treatment (n = 3) and representative Western blot result. (c) Quantification results showing the time-dependent changes in expression levels of EGFR protein after MG treatment (n = 3) and representative Western blot result. (d) Quantification results showing that the decrease in expression level of EGFR induced by MG treatment could be rescued by pre-treatment with NAC, but not AG or MET. Shown also are the Western blot result. All p values were calculated by using unpaired, two-tailed t-test, and the p-values in (d) referred to the comparisons with the results obtained from cells treated with MG alone. The concentration of AG, NAC, and MET used to pretreat cells were all 1 mM ...... 59

Figure 2.7. MG treatment led to diminished level of IGF2R protein. (a) Quantification results showing the dose-dependent alterations in expression levels of IGF2R protein after MG treatment (n = 3) and representative Western blot result. (b) The decrease in expression level of IGF2R protein induced by MG treatment could be rescued by pre-incubation of cells with NAC, but not AG or MET (n = 3). Shown also are the representative Western blot result. All p values were calculated by using unpaired, two-tailed t-test, and the p- values in (b) referred to the comparisons with the results obtained from cells treated with MG alone. The concentration of AG, NAC, and MET used to pretreat cells were all 1 mM ...... 61

Figure 3.1. PRM- and MRM-based targeted proteomic approaches for interrogating the human kinome. (a) The chemical structures of the isotope-coded ATP affinity probes. (b) Experimental strategy for PRM- and MRM-based targeted proteomic approaches. (c) Venn diagrams displaying the numbers of kinases included in the PRM kinome library and those that could be quantified by the PRM method (left), and the overlaps of kinases (middle)

xvii and protein kinases (right) between the PRM- and MRM-based kinome libraries ...... 87

Figure 3.2. PRM- and MRM-based targeted proteomic approaches for interrogating the perturbations in expression and ATP binding affinity of kinases induced by kinase inhibitor treatment. (a) A kinome dendrogram depicting the kinases that are included in our PRM and MRM kinome spectral libraries. The kinome dendrogram was adapted with permission from Cell Signaling Technology (http://www.cellsignal.com). (b) Venn diagrams depicting the overlaps of the number of kinases that were quantified by the PRM- and MRM-based kinome profiling methods for cellular samples obtained from treatments with the three kinase inhibitors. (c) Scatter plots showing the lack of correlation between the ratios of kinases in inhibitor-treated over control DMSO-treated cells obtained from the PRM and MRM methods, or the absence of apparent correlation of alterations in kinase protein expression triggered by dabrafenib and vemurafenib treatments ...... 89

Figure 3.3. Differential protein expression (a) and ATP-binding affinities (b) of kinase proteins in K-562 cells induced by imatinib treatment. The kinase protein expression data represent the mean of results obtained from three forward and two reverse SILAC labeling experiments, and the ATP-binding affinity results reflect five forward and six reverse ATP probe labeling experiments. Blue, red, and grey bars represent those kinases with ratios (in imatinib-treated/control cells) that are < 0.67, > 1.5, and between 0.67 and 1.5, respectively ...... 91

Figure 3.4. Western blot analyses for validating the protein expression levels of kinases in K-562 cells with or without imatinib treatment. (a) Images from Western blot analyses of the expression levels of select kinases in K-562 cells with or without imatinib treatment. (b) Representative PRM traces for monitoring the expression levels of the same kinases as shown in (a). (c) The quantification results for the ratios of kinase proteins in imatinib- treated cells over untreated cells. Error bars represent standard deviations ...... 93

Figure 3.5. The alterations in expression levels of kinase proteins in M14 cells after treatment with two small-molecule BRAF inhibitors, dabrafenib (a) and vemurafenib (b). The cells were treated with 100 nM inhibitor for 24 h. Displayed are the ratios of expression of kinase proteins in BRAF inhibitor-treated over mock-treated M14 cells, where the X- axis was plotted in log10 scale. The data represent the average ratios obtained from two forward and two reverse SILAC labeling experiments for dabrafenib treatment and two forward and one reverse SILAC labeling experiments for vemurafenib treatment. The red and blue bars designate those kinases that were up- and down-regulated, respectively, by at least 1.5-fold upon the inhibitor treatment ...... 94

Figure 3.6. Western blot analyses for validating the protein expression levels of kinases in M14 cells with or without dabrafenib/vemurafenib treatment. (a) Images from Western blot analyses of the expression levels of representative kinases in M14 cells with or without

xviii dabrafenib or vemurafenib treatment. The quantification results for the ratios of kinase proteins in dabrafenib- and vemurafenib-treated cells over control untreated cells are shown in (b) and (c), respectively. The data represent the mean ± S. D. of the quantification results (n = 3) ...... 96

Figure 3.7. Heat map for time-dependent changes in kinase protein expression in M14 cells following treatment with 100 nM dabrafenib. The data represent the means of the results obtained from at least one forward and one reverse SILAC labeling results ...... 98

Figure 3.8. Differential ATP binding affinity of kinase proteins in M14 cells upon a 24-hr treatment with 100 nM dabrafenib (a) and vemurafenib (b). The data represent the means of the results obtained from two forward and two reverse ATP affinity probe labeling results for dabrafenib treatment, and two forward and one reverse ATP affinity probe labeling results for vemurafenib treatment ...... 99

Figure 3.9. Imatinib inhibits the activity of CHK1. (a) Representative MRM and PRM traces for the quantifications of CHK1. (b) Western blot for the validation of the protein expression and activity of CHK1 in K-562 cells with and without imatinib treatment. (c) Clonogenic survival assay results showing the effect of imatinib on sensitizing MDA-MB- 231 cells toward NCS. Error bars represent standard deviation. *, 0.01 < p < 0.05; **, 0.001 < p < 0.01; #, no significant difference. The p-values were calculated against the control using two-tailed, unpaired Student’s t-test ...... 101

Figure 3.10. Vemurafenib binds MAP2K5. (a) MRM traces for tryptic peptides of MAP2K5 in M14 cells with or without vemurafenib treatment. (b) Western blot for monitoring the expression levels of MAP2K5 and ERK5, and the phosphorylation level of ERK5. (c) Quantitative comparison of ratios of protein expression and activity of MAP2K5 obtained from Western blot analyses, and ATP-binding affinity obtained from MRM analyses. The data represent the mean ± S. D. of the quantification results (n = 3). (d) The Venn diagram showing the overlap between the cell-based and in vitro kinome profiling by vemurafenib treatment. (e) A scatter plot displaying the correlation between the ratios (vemurafenib treat/control) obtained from in cellulo (in M14 cells) and in vitro (M14 cell lysate) experiments, respectively. (f) MRM traces of MAP2K5 and SRC obtained from in vitro kinome profiling assay ...... 102

Figure 3.11. Relative growth of M14 cells upon a 24-hr treatment with the indicated concentrations of BIX-02188, an inhibitor for MAP2K5 ...... 104

Figure 3.12. The experimental strategy of using MRM-based targeted proteomic approach for probing the in vitro (in the whole cell lysate of M14 cells) ATP-binding affinity of kinases upon treatment with a kinase inhibitor, vemurafenib ...... 105

Figure 3.13. Differential ATP binding affinities of kinase proteins in lysates of M14 cells with and without a 2-hr pre-treatment with 100 nM vemurafenib. The data represent the

xix means of the results obtained from two forward and one reverse ATP probe labeling results ...... 106

Figure 4.1. A PRM-based targeted proteomic approach for interrogating the perturbations in protein expression levels of kinases during CRC metastasis. (a) Experimental strategy for PRM-based targeted proteomic approach. (b) A Venn diagram displaying the overlap between quantified kinases from the forward and reverse SILAC labelings of the SW480/SW620 pair of CRC cells. (c) Correlation between the ratios of kinase protein expression in SW480/SW620 cells obtained from forward and reverse SILAC labeling experiments ...... 121

Figure 4.2. Differential expression of kinase proteins in paired SW480/SW620 CRC cells. Blue, red, and grey bars represent those kinases with ratios (SW480/SW620) that are < 0.67, > 1.5, and between 0.67 and 1.5, respectively ...... 123

Figure 4.3. AK2, PRPS2 IGF2R, EGFR and CHEK1 are regulated in metastatic CRC cells. (a) PRM traces for the quantifications of AK2 and PRPS2 proteins in SW480/SW620 cells. (a) Western blot for the validation of the expression levels of AK2 and PRPS proteins in the paired CRC cells. (c) Quantitative comparison of the ratios of AK2 obtained from PRM and Western blot analysis. (d) Quantitative comparison of the ratios of PRPS proteins obtained from PRM (for PRPS1 and PRPS2) and Western blot analysis (for PRPS1/2/3). (e) Western blot for the validation of the expression levels of IGF2R, EGFR and CHEK1 in the paired CRC cells. (f) Quantitative comparison of ratios of IGF2R, EGFR and CHEK1 obtained from PRM and Western blot analysis. The data represent the mean ± S. D. of the quantification results (n=3) ...... 124

Figure 4.4. Analysis of PRPS2 gene. (a) KEGG pathway analysis of up-regulated kinases during CRC metastasis. Shown are the top ten up-regulated pathways. (b) Box-and-whisker plot showing upregulated PRPS2 mRNA expressions in CRC cell lines in the CCLE database (from 56 CRC cell lines and 1019 Pan-Cancer cell lines). The p values were calculated based on unpaired, two-tailed Student’s t-test: ***, p < 0.001. Shown by the whiskers extending outside of the box are the maximum and minimum z-scores of PRPS2 expression in CRC cells. The displayed boxes contain the interquartile z-scores of PRPS2 expression obtained from CRC cells ...... 126

Figure 4.5. PRPS2 modulates the migratory and invasive capacities of CRC cells. Western blot results showing the ectopic overexpression of PRPS2 in in SW480 cells (a) and shRNA-mediated knock-down of PRPS2 in SW620 cells (b). (c) The migratory and invasive abilities of SW480 primary colorectal cancer cells were increased upon ectopic overexpression of PRPS2. Shown in (d) and (e) are the quantification results of migratory and invasive abilities of SW480 primary colorectal cancer cells upon ectopic overexpression of PRPS2 gene, and those of SW620 metastatic colorectal cancer cells upon shRNA-mediated stable knock-down of PRPS2 gene, respectively. The data represent the mean ± S. D. of the quantification results (n=3). The p values were calculated using

xx unpaired, two-tailed Student’s t-test: #, p ≥ 0.05; *, 0.01 ≤ p < 0.05; **, 0.001 ≤ p < 0.01; ***, p < 0.001 ...... 128

Figure 4.6. PRPS2 regulates the expression and enzymatic activity of MMP-9. (a) Gelatin zymography assay showing the changes in activities of secreted MMP-2 and MMP-9 upon ectopic overexpression of PRPS2 in SW480 cells. (b) Modulation of activities of secreted MMP-2 and MMP-9 by PRPS2 in SW480 cells. (c) RT-qPCR results showing the changes in mRNA levels of CDH2 (encoding N-cadherin), CDH1 (encoding E-cadherin), MMP2 and MMP9 genes in SW480 cells upon ectopic overexpression of PRPS2. (d) Gelatin zymography assay showing the changes in activities of secreted MMP-2 and MMP-9 after shRNA-mediated stable knock-down of PRPS2 gene in SW620 cells. (e) Modulation of activities of secreted MMP-2 and MMP-9 by PRPS2 in SW620 cells. (f) RT-qPCR results showing the modulation in mRNA levels of CDH2, CDH1, MMP2 and MMP9 genes in SW620 cells upon siRNA-mediated knockdown of PRPS2. The data represent the mean ± S. D. of the quantification results (n=3). The p values were calculated based on unpaired, two-tailed Student’s t-test: #, p ≥ 0.05; *, 0.01 ≤ p < 0.05; **, 0.001 ≤ p < 0.01; ***, p < 0.001 ...... 130

Figure 4.7. Myc drives the expression of PRPS2 in CRC cells and patients. (a) Quantitative comparison of mRNA expression between Myc and PRPS2 genes obtained from 56 CRC cell lines in the CCLE database. (b) Quantitative comparison of mRNA expression between Myc and PRPS2 genes obtained from 435 CRC patients in TCGA database ...... 132

Figure 5.1. PRM-based targeted proteomic approach for interrogating the human heat shock proteome. (a) A Venn diagram displaying the numbers of heat shock proteins included in the PRM library and those that could be quantified by the PRM method. (b) A pie chart depicting the protein coverage of different groups of heat shock proteins. (c) Experimental strategy for PRM-based targeted proteomic approach ...... 154

Figure 5.2. Performances of PRM-based targeted proteomic approach for interrogating the perturbations in expression of heat shock proteins during melanoma metastasis. (a) Differential expression of heat shock proteins in WM-115/WM-266-4 paired melanoma cells. (b) A Venn diagram displaying the overlap between quantified heat shock proteins from the forward and reverse SILAC labelings of WM-115/WM-266-4 paired melanoma cells. (c) Correlation between the ratios obtained from forward and reverse SILAC labeling experiments. (d) A heatmap showing the differences in expression of heat shock proteins in 3 pairs of primary/metastatic melanoma cell lines. Genes were clustered according to Euclidean distance. The data in (a) and (d) represent the mean of the results obtained from one forward and one reverse SILAC labeling results ...... 156

Figure 5.3. Differential expression of heat shock proteins in paired IGR-39/IGR-37 (a) and WM793/1205Lu (b) primary/metastatic melanoma cells. The data represent the mean of results obtained from one forward and one reverse SILAC labeling experiments ...... 158

xxi

Figure 5.4. DNAJB4 is down-regulated in metastatic melanoma cells. (a) Western blot for the validation of the expression levels of DNAJB4 in the three pairs of melanoma cells. 115, 266-4, 39, 37, 793 and 1205 denote WM-115, WM-266-4, IGR-39, IGR-37, WM-793 and 1205Lu cells. (b) PRM traces for the quantification of DNAJB4 protein in 3 pairs of melanoma cells from forward and reverse SILAC labeling experiments. (c) Quantitative comparison of ratios of DNAJB4 obtained from PRM and Western blot analysis (n=3). (d) Kaplan-Meier survival analysis showing that higher levels of expression of DNAJB4 gene confers better prognosis of melanoma patients ...... 159

Figure 5.5. DNAJB4 modulates the invasive capacities of melanoma cells. (a) Western blot results showing the siRNA-mediated knock-down of DNAJB4 in three lines of primary melanoma cells. (b-c) Quantification data showing the effects of expression levels of DNAJB4 on the migratory (b) and invasive (c) abilities of melanoma cells. The data represent the mean and standard deviation of results obtained from three parallel experiments. O.E. represents ectopic overexpression ...... 161

Figure 5.6. DNAJB4 modulates the enzymatic activities of MMP-2 and MMP-9. (a) Representative gelatin zymography assay result showing the changes in activities of secreted MMP-2 and MMP-9 in WM-793 cells upon treatment with control, non-targeting siRNA or siRNA targeting DNAJB4. (b) Quantification results showing the modulation of activities of secreted MMP-2 and MMP-9 in primary melanoma cells upon siRNA-induced knockdown of DNAJB4 (left) or in metastatic melanoma cells upon ectopic overexpression of DNAJB4 (right). The data represent the mean and standard deviation of results obtained from three parallel experiments, and were normalized to the results obtained for primary cells treated with control non-targeting siRNA (left) or metastatic melanoma cells treated with control empty vector (right) ...... 163

Figure 5.7. qRT-PCR showing the modulation in mRNA levels of MMP2 and MMP9 genes in primary melanoma cells upon siRNA-mediated knockdown of DNAJB4 or metastatic melanoma cells upon ectopic expression of DNAJB4 ...... 165

Figure 6.1. A schematic diagram showing the forward SILAC-based proteomic workflow for examining the alterations in expression levels of heat shock proteins after HSP90 inhibitor treatment ...... 187

Figure 6.2. HSP90 inhibitors induced substantial reprogramming of the heat shock proteome. (a) A heat map showing the alterations in expression levels of heat shock proteins in M14 cells upon a 24-hr treatment with 100 nM of ganetespib, AT13387, or 17- DMAG. The data represent the mean of results obtained from two forward and two reverse SILAC labeling experiments. (b) Representative PRM traces for monitoring the relative expression levels of several heat shock proteins with or without HSP90 inhibitor treatment. (c) Western blot for validating the expression levels of select heat shock proteins after

xxii treatment with the three HSP90 inhibitors. (d) Quantification data for the differences in expression levels of heat shock proteins in M14 cells with or without HSP90 inhibitor treatment, as obtained from Western blot (n = 3) and LC-PRM analysis (n = 4, two forward and two reverse SILAC labelings) ...... 188

Figure 6.3. Expression of heat shock proteins in HeLa and HEK293T cells after a 24-hr treatment with ganetespib. Representative Western blot images and quantification results for the expression of HSP70 and DNAJB4 proteins in HeLa (a) and HEK293T (b) cells. The quantification data in (a) and (b) represent the mean ± S. D. of results from three independent experiments ...... 190

Figure 6.4. RT-qPCR confirms the transcriptional and post-transcriptional mechanisms of increased expression of DNAJB4. (a) Real-time quantitative PCR for monitoring the mRNA expression levels of several genes encoding heat shock proteins in M14 cells at different time points following exposure to 100 nM ganetespib. GAPDH and HPRT1 genes were employed as the controls. (b-c) RT-qPCR results show that the mRNA level of DNAJB4 in M14 cells exhibited a progressive increase in the polysome fraction following ganetespib treatment. Data were normalized to the mRNA level of GAPDH (b) or HPRT1 (c) gene. The quantification data represent the mean S. D. of results from three independent experiments. The p values were calculated based on unpaired, two-tailed Student’s t-test: #, p > 0.05; *, 0.01 ≤ p < 0.05; **, 0.001 ≤ p < 0.01; ***, p < 0.001 ...... 191

Figure 6.5. Ganetespib treatment increases the translation efficiency and protein synthesis of DNAJB4. (a) RT-qPCR results show that the mRNA level of DNAJB4 exhibited a progressive increase in the polysome fraction following ganetespib treatment in M14 cells. Data were normalized to the mRNA level of HPRT1 gene. (b) Traces for fractionation of polysome isolated from untreated M14 cells or M14 cells at 6 or 24 hr following treatment with 100 nM ganetespib. (c) A scatter plot shows the correlation between the expression change and protein synthesis (fold turnover) of heat shock proteins in M14 cells after ganetespib treatment. (d) m6A RIP-qPCR result shows the increased m6A level in DNAJB4 mRNA in M14 cells after a 6-hr treatment with ganetespib. The data were normalized to the mRNA level of HPRT1 gene. The quantification data represent the mean ± S. D. of results from three independent experiments. The p values referred to comparison between 0 h and the indicated time points following HSP90 inhibitor treatment and were calculated based on unpaired, two-tailed Student’s t-test: #, p > 0.05; *, 0.01 ≤ p < 0.05; **, 0.001 ≤ p < 0.01; ***, p < 0.001 ...... 192

Figure 6.6. Pulse-chase SILAC coupled with LC-PRM analysis for the quantification of newly synthesized heat shock proteins in M14 cells after treatment with ganetespib. A schematic diagram showing the workflow for the four pulse chase experiments during 0-6 h following ganetespib treatment ...... 194

xxiii

Figure 6.7. Pulse-chase SILAC labeling, together with LC-PRM analysis, reveals the changes in newly synthesized heat shock proteins in M14 cells during the course of 0-6 hr following treatment with 100 nM ganetespib. A bar graph shows the fold turnover of heat shock proteins in inhibitor-treated over mock-treated (with DMSO) cells during the course of 0-6 hr following treatment with 100 nM ganetespib. The data represent the mean of results obtained from one forward and one reverse SILAC labeling experiments. Red and blue bars represent those heat shock proteins that display at least a 1.5-fold increase and decrease, respectively, in protein synthesis in M14 cells upon ganetespib treatment relative to control (with DMSO treatment). The Y-axis was plotted in log10 scale ...... 195

Figure 6.8. Time-dependent changes in expression levels of HSP70, DNAJB4, METTL3, FTO, ALKBH5, and YTHDF3 in M14 cells following treatment with 100 nM of the indicated HSP90 inhibitors. Western blot images and the quantification data showing the alterations in expression levels of the indicated proteins at different time points following treatment with ganetespib (a), AT13387 (b), or 17-DMAG (c). Shown are the ratios of expression of the indicated proteins over β-actin, and further normalized to the ratios obtained for the control cells without HSP90 inhibitor treatment. The data represent the mean ± S. D. of results from three independent experiments. The p values referred to comparison between 0 h and 24 h treatment and were calculated using unpaired, two-tailed Student’s t-test: #, p > 0.05; *, 0.01 ≤ p < 0.05; **, 0.001 ≤ p < 0.01; ***, p < 0.001 .... 194 Figure 6.9. The ganetespib-stimulated expression of DNAJB4 protein involves writer and eraser proteins of m6A. (a) Western blot for monitoring the expression levels of DNAJB4, Flag-tagged FTO and ALKBH5, and endogenous FTO in M14 cells transfected with control plasmid or plasmids for the ectopic expression of Flag-FTO or ALKBH5 at different time intervals following treatment with 100 nM ganetespib. (b) Quantification data based on Western blot analysis in (a). (c) Western blot for monitoring the expression levels of DNAJB4 in M14 cells treated with control non-targeting siRNA (siCtrl), siMETTL3-1 or siMETTL3-2 at different time points following treatment with 100 nM ganetespib. (d) Quantification data based on Western blot analysis in (c). β-actin was employed as the loading control in (a) and (c). Shown in (b) and (d) are the ratios of expression of DNAJB4 protein over β-actin, and further normalized to the ratios obtained for the control cells without ganetespib treatment. The quantification data in (b) and (d) represent the mean ± S. D. of results from three independent experiments. The p values referred to comparisons between control cells and cells with ectopic overexpression of the indicated genes (b), or between controls and siRNA-mediated knock-down of the indicated genes (d). The p values were calculated using unpaired, two-tailed Student’s t-test: #, p > 0.05; *, 0.01 ≤ p < 0.05; **, 0.001 ≤ p < 0.01; ***, p < 0.001 ...... 198

Figure 6.10. The ganetespib-stimulated expression of DNAJB4 protein involves eraser protein of m6A, ALKBH5. (a) Western blot for monitoring the expression levels of DNAJB4 protein in M14 cells treated with control non-targeting siRNA (siCtrl), siALKBH5-1 or siALKBH5-2 at different time points following treatment with 100 nM ganetespib. (b) Quantification data based on Western blot analysis in (a). β-actin was employed as the loading control in (a). Shown in (b) is the ratios of expression of DNAJB4

xxiv protein over actin, and further normalized to the ratios obtained for the control cells without ganetespib treatment. The quantification data in (b) represent the mean ± S. D. of results from three independent experiments. The p values were calculated based on unpaired, two- tailed Student’s t-test: #, p > 0.05; *, 0.01 ≤ p < 0.05; **, 0.001 ≤ p < 0.01; ***, p < 0.001 ...... 200

Figure 6.11. The ganetespib-stimulated expression of DNAJB4 protein involves reader proteins of m6A. (a) Western blot for monitoring the expression levels of DNAJB4 in HEK293T cells or the isogenic cells with the YTHDF1, YTHDF2, and YTHDF3 genes being ablated by the CRISPR-Cas9 genome editing method at different time intervals following exposure with 100 nM ganetespib. (b) Quantification data based on Western blot analysis in (a). (c) Western blot for monitoring the expression levels of DNAJB4 in YTHDF3-deficient HEK293T cells complemented with an empty pRK7 plasmid (control) or pRK7-YTHDF3 at different time points following exposure to 100 nM ganetespib. (d) Quantification data based on Western blot analysis in (c). β-actin was employed as the loading control in (a) and (c). Shown in (b) and (d) are the ratios of expression of DNAJB4 protein over β-actin, and further normalized to the ratios obtained for the control cells without ganetespib treatment. The quantification data in (b) and (d) represent the mean ± S. D. of results from three independent experiments. The p values referred to comparisons between HEK293T cells and the isogenic cells with YTHDF1/2/3 genes being individually ablated (b), or between complementation with control and YTHDF3 plasmid (d). The p values were calculated using unpaired, two-tailed Student’s t-test: #, p > 0.05; *, 0.01 ≤ p < 0.05; **, 0.001 ≤ p < 0.01; ***, p < 0.001 ...... 202

Figure 6.12. CRISPR-Cas9-mediated targeting of YTHDF1, YTHDF2, and YTHDF3 genes in HEK293T cells. Displayed are the Sanger sequencing data for confirming the out-of- frame deletions found in the three genes. The successful depletion of YTHDF1 and YTHDF3 were also confirmed by Western blot analysis (See Figure 5a). We were not able to validate the knockout of YTHDF2 gene by Western blot owing to the lack of highly specific antibody ...... 204

Figure 6.13. ALKBH5 modulates the expression levels of some heat shock proteins during ganetespib treatment. (a) A bar graph shows the changes in expression of heat shock proteins with or without overexpression of ALKBH5 in M14 cells after a 24-hr treatment with 100 nM ganetespib. The data represent the mean of results obtained from one forward and one reverse SILAC labeling experiments. Blue bars indicate those heat shock proteins that display at least a 1.5-fold decrease in protein expression in M14 cells upon ectopic overexpression of ALKBH5 (with 2 µg of ALKBH5 plasmid) relative to control. (b) Western blot for monitoring the expression levels of HSPB1, Flag-tagged FTO and ALKBH5, and endogenous FTO in M14 cells transfected with control plasmid or plasmids for the ectopic expression of Flag-FTO or ALKBH5 at different time intervals following exposure with 100 nM ganetespid. (c) Quantification data of HSPB1 based on Western blot

xxv analysis in (b). The quantification data in (c) represent the mean ± S. D. of results from three independent experiments ...... 205

Figure 6.14. Ganetespib treatment resulted in increased m6A levels in the 5 -UTR of DNAJB4 mRNA, and methylation at one of these sites led to elevated translation efficiency of DNAJB4. (a) Displayed is a bar graph showing that the activity of m6A demethylase is decreased in M14 cells after a 6-hr treatment with 100 nM ganetespib (n = 6). (b) A diagram showing the 5 -UTR of the human DNAJB4 mRNA and the adenosine sites monitored by the SELECT assay. The adenosine sites marked in red and black are the m6A motif sites and the negative control sites, respectively. (c) Relative template abundances after elongation and ligation measured by SELECT (n = 3). A total of seven sites were chosen, four from m6A motif site (GAC) and three from UAA or CAA sites (as negative control). ‘Neg’ represents negative control. The p-values were calculated versus the mean value of the three negative controls. (d) Relative luciferase activities for wild-type (WT) 5′-UTR of DNAJB4 gene and the corresponding A→C mutants (n=3). Firefly luciferase activity was normalized to that of Renilla luciferase and further normalized to the wild-type construct. The data represent the mean ± S. D. of results obtained from three separate experiments. The p values referred to comparison with the wild-type plasmid, and were calculated using unpaired, two-tailed Student’s t-test: #, p > 0.05; *, 0.01 ≤ p < 0.05; **, 0.001 ≤ p < 0.01; ***, p < 0.001 ...... 207

Figure 6.15. Heat shock stress stimulated elevated expression of DNAJB4, which involves writer, eraser and reader proteins of m6A. Shown in (a), (c) and (e) are Western blot images displaying the expression levels of DNAJB4 in M14 cells following heat shock treatment (HS, at 42.0 C for 60 min), or the same cells after siRNA-mediated knockdown of METTL3 (siMETTL3-1) (a), ectopic overexpression of ALKBH5 (c), or the isogenic HEK293T cells with YTHDF1, YTHDF2, or YTHDF3 genes being knocked out by the CRSIPR-Cas9 genomic editing method (e). The relevant quantification data are presented in (b), (d) and (f). -actin was employed as the loading control in (a), (c) and (e). Shown in (b) and (f) are the ratios of expression of DNAJB4 protein over β-actin, and in (d) these ratios were further normalized to the ratios obtained for the cells at t=0 post heat shock. The quantification data in (b), (d) and (f) represent the mean ± S. D. of results from three independent experiments. The p values referred to comparison between control cells and siRNA-mediated knock-down (b) or ectopic expression (d) of the indicated genes, or HEK293T cells and the isogenic cells with YTHDF1/2/3 genes being individually ablated (f). The p values were calculated based on unpaired, two-tailed Student’s t-test: *, 0.01 ≤ p < 0.05; **, 0.001 ≤ p < 0.01; ***, p < 0.001 ...... 209

Figure 6.16. Table of content figure ...... 211

Figure 7.1. A PRM-based targeted proteomic approach for interrogating the human helicase proteome. (a) A Venn diagram depicting the numbers of helicases included in the PRM library and those that could be quantified by the PRM method. (b) A bar graph showing the coverages of different groups of helicases in the PRM library. (c) The

xxvi experimental strategy, involving the use of forward SILAC labeling together with the PRM-based targeted proteomic analysis, for assessing the alterations in expression of helicase proteins in human cells upon treatment with the two HSP90 inhibitors ...... 230

Figure 7.2. The alterations in expression levels of helicase proteins in M14 cells after treatment with two small-molecule HSP90 inhibitors, onalespib (a) and alvespimycin (b). The cells were treated with 100 nM inhibitor for 24 h. Displayed are the ratios of expression of helicase proteins in HSP90 inhibitor-treated over mock-treated M14 cells, where the Y- axis was plotted in log10 scale. The data represent the average ratios obtained from two biological replicates (i.e. one forward and one reverse SILAC labeling experiments). The red and blue bars designate those helicases that were up- and down-regulated, respectively, by at least 1.5-fold upon the inhibitor treatment ...... 232

Figure 7.3. Extracted-ion chromatograms for monitoring representative helicases (MCM4, MCM7, RUVBL1 and EIF4A1) in M14 cells with or without a 24-h treatment with 100 nM onalespib. The peptide sequences and the transitions employed for plotting the ion chromatograms are listed in the figure ...... 234

Figure 7.4. Analytical performance of the PRM method. (a) A scatter plot displaying the correlation between the ratios obtained from forward and reverse SILAC labeling experiments (left), and a Venn diagram showing the overlap between quantified helicases from the forward and reverse SILAC labeling experiments in M14 cells with or without alvespimycin treatment (right). (b) A scatter plot showing the correlation between the alterations in expression levels of helicase proteins in M14 cells treated with onalespib or alvespimycin. (c) Western blot for the validation of the expression levels of representative helicases in M14 cells with vs. without HSP90 inhibitor treatment, where ARAF was used as a positive control. (d) PRM traces for the relative quantifications of representative helicases. (e-f) Quantitative comparisons of the ratios obtained from PRM (n = 2, one forward and one reverse SILAC labelings) and Western blot analyses (n = 3) for representative helicases in M14 cells with vs. without a 24-h treatment with 100 nM of onalespib (e) or alvespimycin (f) ...... 235

Figure 7.5. Targeted integration of a tandem affinity tag to the C-terminus of HSP90 and affinity pull-down in conjunction with LC-PRM analysis for assessing the interaction between HSP90β and helicase proteins. (a) Western blot confirmed the targeted integration of the tandem affinity tag to endogenous HSP90β protein with the CRISPR-Cas9 method. (b) Experimental strategy for combining forward SILAC labeling with the LC-PRM-based targeted proteomic approach for the identification of cellular proteins that can interact with HSP90β. (c) Representative PRM traces showing the relative quantification results of MCM4 and MCM7 from the anti-Flag pull-down mixture in HEK293T cells with or without the integration of tandem affinity tag to the C-terminus of HSP90 protein from both forward and reverse SILAC labeling experiments. (d) A Venn diagram depicting the number of helicases that could bind with HSP90β, i.e., those helicases that could be

xxvii enriched from affinity pull-down from Flag-HSP90 cells, and that could be down-regulated upon HSP90 inhibitor treatment ...... 237

Figure 7.6. Validation for the incorporation of Flag-tag in endogenous HSP90β protein. (a) Immunoprecipitation followed by Western blot analysis for validating the interactions between HSP90β and ARAF, and the lack of interaction between HSP90β and EGFR. (b) Immunoprecipitation followed by LC-MS/MS analysis led to the identification of HSP90β with a 65% sequence coverage. Amino acid sequences highlighted in red were identified from LC-MS/MS analysis ...... 239

Figure 7.7. Bar graph depicting the ratios of helicase proteins obtained from affinity down from lysates expressing Flag-tagged HSP90β vs. the corresponding pull-down from lysate of parental HEK293T cells. Red bars designate those helicases with at least a 1.5-fold enrichment from lysates of HEK293T cells with a tandem affinity tag being incorporated to the C-terminus of endogenous HSP90β protein over the parental HEK293T cells ..... 241

Figure 7.8. Helicases as putative client proteins of HSP90. The inner and middle layers represent those helicases that were both enriched from the lysates of HSP90β-tagged cells and down-regulated in M14 cells upon treatment with both and one of the two HSP90 inhibitors, respectively. The outer layer designates those helicases that were enriched from the lysate of the HSP90β-tagged cells but whose expression levels in M14 cells were not modulated by either of the two HSP90 inhibitors ...... 242

Figure 7.9. Interactions between HSP90 and helicases. (a) Correlation between the ratios obtained from affinity pull-down and onalespib treatment. (b) Immunoprecipitation followed by Western blot analysis for validating the interactions between HSP90β and MCM4/MCM7 ...... 243

xxviii

LIST OF TABLES

Table 2.1. Sequence for qRT-PCR primers ...... 62

Table 2.2. iRT comparison between desthiobiotin-labeled and desthiobiotin-C3-labeled peptides ...... 63

Table 4.1. Sequences for RT-qPCR primers ...... 133

Table 5.1. Sequences for RT-qPCR primers ...... 166

Table 6.1. LC-MS and MS/MS in the data-dependent acquisition (DDA) mode for confirming the equi-mass mixing of light- and heavy-labeled lysates in SILAC experiments ...... 212

Table 6.2. Sequences for RT-qPCR primers ...... 213

Table 6.3. Primer sequences for SELECT method ...... 214

Table 6.4. Primer sequences for mutagenesis amplification from pGL3-DNAJB4-5′UTR. The C and G in bold indicate the mutation sites, where A is mutated to C ...... 215

xxix

Chapter 1

General Overview

1.1 Introduction

The central dogma of molecular biology involves DNA replication, DNA transcription and RNA translation, and this flow of genetic information produces protein as the final product (1). Proteins participate in and regulate a variety of cellular processes (2).

Therefore, information on differential protein expression and post-translational modifications can provide deeper understanding of exogenous stimuli or diseases (3).

Consequently, a brand-new discipline referred to as “Proteomics” was born. So far, several different aspects of proteomics have been well-established, including protein identification, protein quantification, post-translational modifications, protein-protein interactions, and protein-small molecule interactions (4).

An important aspect of studying proteins is protein quantification, and to this end the

Western blot approach was developed for quantification of a single protein (5). However, due to low throughput of the method, different global approaches have been developed to perform protein analysis on a proteomic scale (6). Global analysis of protein expression was initially achieved by high-resolution two-dimensional (2D) gel electrophoresis (7).

After SDS electrophoresis (8) followed by silver or Coomassie blue staining (9), expression of proteins were analyzed by software (10). However, the accuracy of quantification could

1 be compromised by co-localization of multiple proteins at the same spots due to sample complexity. More methods were developed for large-scale protein quantification.

With the advances in mass spectrometry instrumentation, quantitative proteomics was applied for large-scale protein quantification. Due to the sequence-based peptide identification and quantification of peptides, a mass spectrometry (MS)-based approach provides higher throughput and increased reliability for protein identification and quantification, compared to the conventional proteomics methods (11).

So far, MS-based proteomics has been focused on three aspects of protein identification and quantification. First, quantitative comparison of protein expression has been extensively applied to biomarker discovery (12). Second, the high specificity and accuracy of MS-based proteomics make it a powerful approach for studying post-translational modifications (PTM) (13, 14). Third, MS-based proteomics approaches have been widely used to support protein interaction studies and biological network mapping (15, 16).

Network mapping is not limited to protein-protein interactions but also involves the interactions of proteins with small molecules (17), RNA or DNA (18).

In this chapter, I will first introduce the commonly used MS-based strategies to study quantitative proteomics, including label-free, metabolic labeling and chemical labeling quantification. Then, I will discuss the discovery and targeted proteomics, as well as their applications in ATP-binding protein analysis, which is the main focus of my Ph.D. work.

1.2 MS-based quantitative proteomics

Several different methods are utilized to achieve MS-based quantification of proteins

(19). One approach is called “label-free quantification”, where spectra of the same peptide from two separate mass spectrometric analyses are compared (20). A more advanced approach uses stable isotope-labeling by amino acids in cell culture (SILAC) and quantify the relevant proteins, this method eliminates systematic variation from sample processing and MS measurement (21). Thus, relative abundance of protein expression from different samples can be determined.

1.2.1 Label-free quantification

Label-free quantification serves as a highly economical and rapid proteomic analysis.

Samples are separately digested and analyzed by MS during the label-free quantification procedure. Peptide quantification is achieved by comparing the signal intensity in MS or spectral counts of MS/MS between different samples (Figure 1.1a) (22, 23). The overall accuracy of this approach is reasonable; however, the quality of the result is compromised by cell culture conditions, differences in sample preparation, digestion efficiency or mass spectrometric measurements. In addition, the dynamic range is relatively small due to less precision of smaller peptide ratios compared to larger ratios (24).

1.2.2 SILAC

To overcome the disadvantages brought by label-free quantification, stable isotope labeling by amino acids in cell culture (SILAC) was introduced as a metabolic labeling method for protein quantification (21). Cells are cultured in media with the addition of

3 stable-isotope labeled amino acids, typically lysine and arginine (25). Although arginine is non-essential, it is known to be acquired from diet and has been shown to be essential for the growth of some cell lines (26). Cells will take up these amino acids from the cell culture media for protein synthesis, resulting in stable isotope labeling of the proteome (27).

Because trypsin cleaves primarily at the carboxyl side of lysine and arginine, all the tryptic peptides, except some C-terminal peptides, should carry at least one isotope-labeled amino acid. After 6-8 cell doublings, more than 99% of endogenous proteins will carry isotope- labeled arginine or lysine. Subsequently, a pair of differentially labeled cells from different physiological states or treatments will be harvested and lysed. After mixing equal amounts of the resulting cell lysates they are reduced, alkylated, and enzymatically digested prior to

MS analysis. The resulting peptide mixture is subjected to LC-MS/MS analysis and quantification is based on the relative MS signal intensity of a chosen peptide pair, which, due to isotope labeling, has a clearly defined m/z shift (Figure 1.1b).

The advantage of the SILAC labeling method over previous label-free quantification is that cell lysates from different stages of cells can be combined immediately after cell lysis, which eliminates the variability in sample preparation and mass spectrometry detection.

Therefore, SILAC labeling is particularly suitable for assessing subtle changes in protein expression of cells under internal or external stimulus (28). However, the number of available labels in the SILAC method is limited. So far, a maximum of three independent labeling experiments can be simultaneously conducted in a single SILAC experiment (29).

Furthermore, the SILAC method is limited to cell culture conditions and is very difficult to adapt for the quantification of proteins in tissue or organism samples.

1.2.3 Isotope-coded affinity tag (ICAT)

In addition to metabolic labeling – such as SILAC – isotope-coded affinity tag (ICAT) is one of the most commonly used chemical isotope labeling methods for protein quantification in a cells, tissue or organisms (30). To achieve ICAT labeling, the reagent contains three elements: an affinity tag, a light or heavy isotope-labeled linker, and a thiol- reactive group which will specifically react with protein residues (Figure 1.2a) (31). With this design, cysteines primarily react with the thiol reactive group and are labeled (32).

After lysis of different stages of cells, the two different cell lysates are labeled with either light or heavy-isotope forms of the ICAT probe and after the covalent attachment to the cysteine residues, they are mixed in equal amount. The resulting peptides are isolated by affinity enrichment with avidin and analyzed by LC-MS/MS after tryptic digestion.

Quantification of a chosen peptide is achieved via comparison of the MS signal intensities from the peptide pairs (Figure 1.2b).

Compared to the metabolic labeling method, the ICAT-labeling strategy is more practical for complex clinical samples analysis, including the analysis of proteins from tissues and organisms (33). However, ICAT labeling could lead to incomplete proteome coverage, caused by the selective reaction to the cysteine residue (34). Thus, an alternative method called dimethyl-labeling was developed, which introduces stable isotopic tags by methylation of all primary amino groups on lysine residues or peptide N-terminals (35). and this strategy was widely adopted for global proteome quantification (36). Nevertheless, the current ICAT- or dimethyl-based labeling methods can reach up to only 3-plex (37). A

5 quantification method with higher through-put is required for the analysis of large-scale samples.

1.2.4 Isobaric tag for relative and absolute quantitation (iTRAQ)

Chemical labeling, such as isobaric tag for relative and absolute quantification (iTRAQ), can be implemented for high-throughput quantitative proteomics (38). An isobaric tag encompasses an amine-reactive NHS groups, a mass normalizer (balance group) and a mass tag (reporter group) (Figure 1.3a). Labeling by iTRAQ is performed at the peptide level and separately labeled peptide mixtures are then combined and analyzed by MS

(Figure 1.3b). Due to identical nominal mass and chemical properties of the isobaric tag labeling, each group of labeled peptides co-elutes during the chromatographic separation.

MS analysis utilizing CID or HCD fragmentation breaks the covalent bond between the balance group and reporter group, and the charge is retained on the fragments from the reporter group. Thus, the resulting reporter ions include at least four isotope-coded variants with different masses, allowing for the quantification of relative protein abundance from multiple experimental conditions.

Similar to iTRAQ, Thermo Fishier Scientific has released a product of tandem mass tags

(TMT) isobaric mass tagging kit. This TMT-labeling kit can provide simultaneous quantification of 10 samples and PTM quantification can be easily achieved with TMT labeling as well (39). However, due to the isolation interference problem induced by the co-isolation and co-fragmentation of near isobaric ions together with the target ions (40,

41), both iTRAQ and TMT suffer from ratio underestimation at the MS2 level (42). To

6 resolve the problem, the Mann group has recently developed an easily abstractable sulfoxide-based isobaric-tag (EASI-tag), which dissociates at low collision energy and generates peptide-coupled, interference-free reporter ions with high yield (43). Despite efforts to improve the quantification accuracy of isobaric tags, combination of samples late in the workflow is prone to additional experimental errors and renders the precision of the quantification low when comparing to SILAC labeling methods.

1.3 Protein quantification in discovery and targeted MS modes

There are two fundamental approaches for MS-based proteomic acquisition, namely, discovery and targeted proteomics.

1.3.1 Discovery proteomics

Discovery proteomics, also known as shotgun proteomics, is the most widely used approach in quantitative proteomics. Shotgun proteomic studies, include data-dependent and data-independent acquisition methods, offering the opportunity for systems-wide analysis.(44) In the data-dependent acquisition (DDA) mode, the mass spectrometer continuously repeats a data acquisition cycle consisting of a full-scan mass spectrum, and the nth (typically 10-20th) most abundant ions in a mass spectrum are selected for fragmentation automatically by the instrument data system in each acquisition cycle.(45,

46) Due to the screening of the top n ions, DDA may miss important targets with low abundance (Figure 1.4a).

To resolve the problem with low abundant targets, a concept of data-independent acquisition (DIA) was raised in 2004 by Yates lab (47) and began to be widely employed

7 as sequential window acquisition of all theoretical mass spectra (SWATH-MS) for protein expression and PTM quantification from 2012 (48, 49). Instead of scanning for the top n events, the DIA method is based on sequential isolation and fragmentation of relatively large precursor ion windows (10-m/z) until a desired mass range has been covered (Figure

1.4b). Therefore, tandem mass spectra would cover the information for all the peptides in the desired mass range. Fragment-ion intensities are extracted in tandem mass spectra for peptide quantification, which provides benefits over quantification directly from MS scans, including increased signal-to-noise ratio, sensitivity, selectivity and dynamic range (47).

However, due to the large precursor window, the detection accuracy could be compromised.

Discovery proteomics does not require any prior knowledge of the proteome sample’s composition, and thus provides the potential to uncover novel PTM sites and proteins. With discovery proteomics, researchers have achieved identification of more than 10,000 protein groups and over 30,000 phosphorylation sites from human or mouse origin (50).

Nevertheless, some important proteins such as kinases and membrane proteins, which have relatively low abundance compared to the entire proteome, may have lower chance for detection. In addition, the space for the MS to analyze is quite huge due to the complexity of the entire proteome. Therefore, the MS may lead to the detection of a slightly different proteins and compromise the reproducibility of detection between runs (51). Thus, the inadequate sensitivity and reproducibility of discovery proteomics severely hamper their utility in biomarker discovery and clinical studies.

1.3.2 Targeted proteomics

Alternatively, targeted proteomics, which relies on multiple-reaction monitoring (MRM) or parallel-reaction monitoring (PRM) has gained increasing popularity in recent years.(52,

53) MRM analysis is often performed on a triple quadrupole (QQQ) MS instrument. The first and third quadrupoles function as mass filters; the second one serves as the collision cell (Figure 1.5a). The pair of product ions filtered in the third quadrupole and its corresponding precursor ion selected in the first quadrupole is considered a transition pair.

Each transition pair is predetermined (54) and can be defined either from synthetic peptides

(55), or from high-quality MS/MS spectra generated by large-scale discovery proteomics analysis (56). Generally, at least three transition pairs are required to accurately quantify a single peptide.

The MRM-based targeted MS analysis allows rapid and continuous monitoring of target ions, leading to high-throughput detection of peptides. Additionally, the two ion filtering steps enhance the sensitivity and specificity for peptide detection by about 100-fold compared with MS analysis in a DDA-based discovery mode (57). Furthermore, the specific transitions filtered for each peptide provide much higher reproducibility than shotgun proteomics. However, as the filter window is normally set as 0.7 Da, MRM is more suitable for analyzing the samples after clean-up. Complex samples like whole cell lysate digests would introduce false positives from the background signal of similar molecular weight (58).

PRM has also been applied for peptide identification and quantification; PRM allows parallel detection of all transitions in a single analysis, consequently providing higher throughput (53). Additionally, this method can utilize the Orbitrap detector in Q Exactive or Orbitrap Fusion mass spectrometer, allowing high resolution, high accuracy for mass detection, and minimizing the background signal (59). Thus, the specificity for peptide identification and accuracy for quantification is much better compared to MRM-based MS analysis. Furthermore, PRM allows detection of a wider dynamic range than MRM in the presence of a yeast background matrix (60). Considering the high throughput, specificity and accuracy of PRM-based peptide identification and quantification, PRM-based targeted proteomics has become a widely used technique for protein quantification (12) and PTM detection (61).

1.3.3 Scheduled targeted proteomics

Due to instrument limitations, the events that can be monitored per cycle time are limited.

Therefore, for targeted proteomics, the total number of transitions (MRM) and precursor ions (PRM) monitored in each run is restricted. To achieve high-throughput quantification, we adopted either scheduled MRM or PRM. To this end, we utilized iRT, which is an empirically determined retention time scale for peptides, to determine the retention time of targeted peptides (62).

The iRT scale is measured and normalized to the assigned iRT scores from reference peptides, which should have relative high abundance that can be easily detected by mass spectrometers. To prevent interference of peptide quantification, the reference peptides

10 should be different from natural protein sequences. Therefore, we employed ten abundant peptides from tryptic digested bovine serum albumin (BSA), which are well distributed across the chromatographic elution time. The iRT of the earliest and latest reference peptides are assigned as 0 and 100, respectively. The iRT calculation of all other reference and targeted peptides is determined by the retention time of each peptide from the full-MS scan and the linear regression obtained from the measured retention time versus iRT of the

0 and 100 reference peptides (Figure 1.6a). In order to acquire the slope and intercept of the linear regression on the current instrument for the scheduled targeted proteomics analysis, reference peptides must be analyzed in advance. Predicted retention times of all targeted peptides can be transformed from their iRT scores using the linear regression obtained from the reference peptides. (Figure 1.6b).

1.4 ATP-binding proteins

Adenosine triphosphate (ATP) is an abundant nucleoside triphosphate that often functions as a form of intracellular energy transfer (63). ATP binds to numerous proteins called “ATP-binding proteins” that play pivotal roles in many cellular processes like metabolism, synthesis, active transport and cell signaling. There are almost 1500 proteins assigned as ATP-binding proteins in the UniProt database; however, the whole picture of the ATP-binding proteome remains unclear. Biotin-conjugated acyl-ATP probes have been developed and widely employed for large-scale identification of ATP-binding proteins (64,

65). In mammalian cells, 349 ATP-binding proteins were identified in HL-60 cells using a desthiobiotin-conjugated ATP-affinity probe (66).

1.4.1 Human kinome analysis

Kinases are one of the most important enzyme families in eukaryotic cells, and they mediate cellular protein and lipid phosphorylation to regulate downstream signaling cascades (Figure 1.7) (67). Aberrant expression and/or activation of kinases are closely associated with disease development and resistance toward cancer therapy (68). Therefore, kinases have become one of the most intensively pursued enzyme super families as drug targets for cancer chemotherapy and more than 30 small-molecule kinase inhibitors have been approved by the Food and Drug Administration (FDA) for cancer chemotherapy (69).

Furthermore, kinase inhibitors have been extensively developed for more than 400 kinases, covering the majority of the human kinome (70-72). Thus, comprehensive analysis of kinase protein expression and ATP-binding affinity would be important for disease treatment and drug discovery.

Traditional methods for measuring kinase expression rely primarily on antibody-based immunoassays, which have high specificity and sensitivity (73). However, this method requires high quality antibodies and can only detect the expression of one kinase at a time.

Therefore, we developed a PRM-based proteomics method for targeted analysis of expression of the human kinome at the entire proteome scale (74). This method has been widely applied for studying kinome reprogramming during cancer metastasis and development of therapeutic resistance.

Considering the extremely low levels of expression of a lot of kinases, enrichment methods for kinases have been extensively pursued in the last decade. As mentioned above,

12 kinase inhibitors have been developed for more than 400 kinases and those inhibitors can bind to one specific or a family of kinases through covalent or non-covalent interactions.

To utilize these binding interactions, researchers immobilized multi-kinase inhibitors onto a solid phase and applied multi-inhibitors beads (kinobeads) to enrich kinases before further analysis (75, 76). Kinobeads have been widely adopted for the enrichment of the human kinome and have uncovered putative targets for diseases (77, 78). However, currently inhibitors target only 400 kinases compared to the total, more than 600 kinases in human kinome; hence, more than 30% of kinases cannot be enriched, which may lead to loss of some important information. Therefore, an ATP-affinity probe was employed for kinase enrichment (79, 80). Almost all of the protein kinases employ ATP as the phosphate donor to phosphorylate their substrate, and thus, exhibit strong binding toward ATP (81).

Another advantage of this approach is that LC-MS/MS analysis of the resulting biotin- labeled peptides determines not only the identities of the labeled proteins but also the labeling sites, or the ATP-binding sites of enriched kinases. Both kinobeads and ATP- affinity probes are heavily employed for kinase enrichment followed by LC-MS/MS analysis.

1.4.2 Other ATP-binding proteins analysis

In addition to kinases, other ATP-binding proteins, including ATP-binding cassette transporters (ABC transporters), chaperones (mainly heat shock proteins), and helicases are involved in a variety of pivotal cellular processes as well. For example, numerous ABC transporters catalyze the hydrolysis of ATP to provide energy required for translocation of

13 various substrates across cell membranes (82), and helicases are ATPases which are capable of unwinding both duplex and more complex structures of nucleic acids (83).

The ATP-affinity probe is able to pull down those ATP-binding proteins for LC-MS/MS analysis and has been applied to discover the ATP-binding sites in known ATP-binding proteins (84, 85). In addition, the ATP-affinity probe is also employed for discovering unknown ATP-binding proteins and target proteins of kinase inhibitors (86). In another similar technique, ATP was immobilized onto solid beads to enrich and discover the soluble ATP-binding proteome in plant mitochondria (87).

Similar to our analysis of kinases, we developed PRM-based proteomics methods for targeted analysis of the expression of human heat shock proteins and helicases at the entire proteome scale (12, 16). Those targeted analysis methods have been applied for cancer metastasis, therapeutic resistance and protein-protein interaction studies (12, 16, 88).

1.5 Scope of the dissertation

ATP-binding proteins encompass kinases, ATP-binding cassette transporters (ABC transporters), chaperones (mainly heat shock proteins), and helicases; these proteins play pivotal roles in many cellular processes such as metabolism, synthesis, active transport and cell signaling. In this dissertation, we applied targeted proteomics, together with an ATP- affinity probe pull-down method to quantify the expression and ATP-binding affinities of

ATP-binding proteins.

In Chapter 2, we expanded the kinome MRM library to cover 474 proteins of the human kinome (80%) and employed this library, together with an ATP-affinity probe, for

14 comprehensively profiling the alterations of these kinases upon treatment with methylglyoxal in HEK293T human embryonic kidney cells. Methylglyoxal is a glycolysis byproduct that is present at elevated levels in blood and tissues of diabetic patients and is thought to contribute to diabetic complications. Our results led to the quantification of 328 unique kinases. In particular, we found that methylglyoxal treatment gave rise to altered expression of a number of kinases in the MAPK pathway and diminished expression of several receptor tyrosine kinases, including epidermal growth factor receptor (EGFR), insulin growth factor 2 receptor (IGF2R), fibroblast growth factor receptor 1 (FGFR1), etc.

Furthermore, we demonstrated that the diminished expression of EGFR occurred through a mechanism that is distinct from the reduced expression of IGF2R and FGFR1.

In Chapter 3, we developed a PRM-based targeted proteomic method to monitor the protein expression of ~ 80% of the human kinome. By employing this method, together with the proteome-wide interrogation of the ATP binding affinities of kinases, we assessed the alterations in protein expression and ATP binding activities of over 300 kinases in cultured human cells, which were elicited by three FDA-approved small-molecule kinase inhibitors (dabrafenib, vemurafenib and imatinib). The results revealed profound changes in protein expression and ATP-binding affinities of many kinases in cells upon treatment with these inhibitors. Moreover, we identified a number of kinases whose activities could be suppressed or activated by these inhibitors, and we identified CHK1 and MAP2K5 as novel target kinases for imatinib and vemurafenib, respectively.

In Chapter 4, we utilized the PRM-based targeted proteomic method developed in

Chapter 3 to examine the reprogramming of the human kinome during colorectal cancer

(CRC) metastasis. We were able to quantify the relative expression of 299 kinase proteins in a pair of matched primary/metastatic CRC cell lines. Among the differentially expressed kinases, we observed phosphoribosyl pyrophosphate synthetase 2 (PRPS2) promotes the migration and invasion of cultured CRC cells by regulating the activity of matrix metalloproteinase 9 (MMP-9) and the expression of E-cadherin. Moreover, we found that up-regulation of PRPS2 in metastatic CRC cells could be induced by the MYC proto- oncogene.

In Chapter 5, we developed a targeted proteomic method, relying on LC-MS/MS in the

PRM mode, for assessing quantitatively the human heat shock proteome. The method facilitated the coverage of approximately 70% of the human heat shock proteome, and displayed much better throughput and sensitivity than the shotgun proteomic approach. We also applied the PRM method for assessing the differential expression of heat shock proteins in three matched primary/metastatic pairs of melanoma cell lines. We were able to quantify ~45 heat shock proteins in each pair of cell lines, and the quantification results revealed that DNAJB4 is down-regulated in the three lines of metastatic melanoma cells relative to the corresponding primary melanoma cells. Interrogation of The Cancer

Genome Atlas data showed that lower levels of DNAJB4 expression conferred poorer prognosis in melanoma patients. Moreover, we found that DNAJB4 suppresses the invasion of cultured melanoma cells through diminished expression and activities of matrix metalloproteinases 2 and 9 (MMP-2 and MMP-9).

In Chapter 6, we employed the unbiased quantitative proteomic method mentioned in

Chapter 5 and uncovered that treatment with three HSP90 inhibitors could result in elevated

16 expression of a large number of heat shock proteins, including HSPA1 and DNAJB4. We also demonstrated that HSP90 inhibitor-mediated increase in expression of DNAJB4 occurs, in part, through an epitranscriptomic mechanism and is substantially modulated by the writer (METTL3), eraser (ALKBH5 and, to a lesser degree, FTO), and reader

(YTHDF1-3) proteins of N6-methyladenosine (m6A). Furthermore, exposure to ganetespib led to elevated m6A levels at all four m6A motif sites in the 5'-UTR of DNAJB4 mRNA, and a luciferase reporter assay showed that methylation at one of these sites (adenosine

114) in the 5'-UTR promotes translation of the reporter gene mRNA. These results indicate an m6A regulatory mechanism is engaged upon heat shock treatment.

In Chapter 7, we developed a PRM-based targeted proteomic method that allows for the quantification of >80% of the human helicase proteome. By employing this method, we demonstrated that a large number of helicase proteins exhibited diminished expression in cultured human cells upon treatment with two small-molecule inhibitors of HSP90. We further introduced a tandem affinity tag to the C-terminus of endogenous HSP90 protein by using the CRISPR-Cas9 genome editing method. Affinity purification followed by LC-

PRM analysis revealed enrichment of 40 out of the 66 quantified helicases from the lysate of cells expressing tagged HSP90, indicating helicases may constitute an important group of client proteins for HSP90.

References

1. Crick F. Central Dogma of Molecular Biology. Nature. 1970;227:561. 2. Lopez-Mejia IC, Fajas L. Cell cycle regulation of mitochondrial function. Curr Opin Cell Biol. 2015;33:19-25. 3. Masuda M, Yamada T. Signaling pathway profiling by reverse-phase protein array for personalized cancer medicine. Biochim Biophys Acta. 2015;1854:651-7. 4. Kenyon GL, DeMarini DM, Fuchs E, Galas DJ, Kirsch JF, Leyh TS, et al. Defining the Mandate of Proteomics in the Post-Genomics Era: Workshop Report. Mol Cell Proteomics. 2002;1:763. 5. Towbin H, Staehelin T, Gordon J. Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc Natl Acad Sci USA. 1979;76:4350-4. 6. Patterson SD, Aebersold RH. Proteomics: the first decade and beyond. Nat Genet. 2003;33:311. 7. Anderson NL, Hofmann JP, Gemmell A, Taylor J. Global approaches to quantitative analysis of gene-expression patterns observed by use of two-dimensional gel electrophoresis. Clin Chem. 1984;30:2031. 8. Rabilloud T, Chevallet M, Luche S, Lelong C. Two-dimensional gel electrophoresis in proteomics: Past, present and future. J Proteomics. 2010;73:2064-77. 9. Switzer RC, Merril CR, Shifrin S. A highly sensitive silver stain for detecting proteins and peptides in polyacrylamide gels. Anal Biochem. 1979;98:231-7. 10. Berth M, Moser FM, Kolbe M, Bernhardt J. The state of the art in the analysis of two- dimensional gel electrophoresis images. Appl Microbiol Biotechnol. 2007;76:1223-43. 11. Aebersold R, Mann M. Mass spectrometry-based proteomics. Nature. 2003;422:198. 12. Miao W, Li L, Wang Y. A targeted proteomic approach for heat shock proteins reveals DNAJB4 as a suppressor for melanoma metastasis. Anal Chem. 2018;90:6835-42. 13. Miao W, Zhang C, Cai Y, Zhang Y, Lu H. Fast solid-phase extraction of N-linked glycopeptides by amine-functionalized mesoporous silica nanoparticles. Analyst. 2016;141:2435- 40. 14. Villén J, Gygi SP. The SCX/IMAC enrichment approach for global phosphorylation analysis by mass spectrometry. Nat Protoc. 2008;3:1630. 15. Williams P, Li L, Dong X, Wang Y. Identification of SLIRP as a G Quadruplex-Binding Protein. J Am Chem Soc. 2017;139:12426-9. 16. Miao W, Li L, Wang Y. Identification of helicase proteins as clients for HSP90. Anal Chem. 2018;90:11751-5. 17. Ong S-E, Schenone M, Margolin AA, Li X, Do K, Doud MK, et al. Identifying the proteins to which small-molecule probes and drugs bind in cells. Proc Natl Acad Sci USA. 2009;106:4617. 18. Mittler G, Butter F, Mann M. A SILAC-based DNA protein interaction screen that identifies candidate binding proteins to functional DNA elements. Genome Res. 2009;19:284-93. 19. Kaake RM, Wang X, Huang L. Profiling of Protein Interaction Networks of Protein Complexes Using Affinity Purification and Quantitative Mass Spectrometry. Mol Cell Proteomics. 2010;9:1650. 20. Zhu W, Smith JW, Huang CM. Mass spectrometry-based label-free quantitative proteomics. J Biomed Biotechnol. 2010;2010:840518.

21. Ong S-E, Blagoev B, Kratchmarova I, Kristensen DB, Steen H, Pandey A, et al. Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics. Mol Cell Proteomics. 2002;1:376-86. 22. Liu H, Sadygov RG, Yates JR. A Model for Random Sampling and Estimation of Relative Protein Abundance in Shotgun Proteomics. Anal Chem. 2004;76:4193-201. 23. Florens L, Carozza MJ, Swanson SK, Fournier M, Coleman MK, Workman JL, et al. Analyzing chromatin remodeling complexes using shotgun proteomics and normalized spectral abundance factors. Methods. 2006;40:303-11. 24. Oda Y, Huang K, Cross FR, Cowburn D, Chait BT. Accurate quantitation of protein expression and site-specific phosphorylation. Proc Natl Acad Sci USA. 1999;96:6591. 25. Ong S-E, Kratchmarova I, Mann M. Properties of 13C-Substituted Arginine in Stable Isotope Labeling by Amino Acids in Cell Culture (SILAC). J Proteome Res. 2003;2:173-81. 26. Scott L, Lamb J, Smith S, Wheatley DN. Single amino acid (arginine) deprivation: rapid and selective death of cultured transformed and malignant cells. Br J Cancer. 2000;83:800. 27. Ong S-E, Mann M. A practical recipe for stable isotope labeling by amino acids in cell culture (SILAC). Nat Protoc. 2007;1:2650. 28. Mann M. Functional and quantitative proteomics using SILAC. Nat Rev Mol Cell Biol. 2006;7:952. 29. Bantscheff M, Schirle M, Sweetman G, Rick J, Kuster B. Quantitative mass spectrometry in proteomics: a critical review. Anal Bioanal Chem. 2007;389:1017-31. 30. Shiio Y, Aebersold R. Quantitative proteome analysis using isotope-coded affinity tags and mass spectrometry. Nat Protoc. 2006;1:139. 31. Gygi SP, Rist B, Gerber SA, Turecek F, Gelb MH, Aebersold R. Quantitative analysis of complex protein mixtures using isotope-coded affinity tags. Nat Biotechnol. 1999;17:994. 32. Leitner A, Lindner W. Current chemical tagging strategies for proteome analysis by mass spectrometry. J Chromatogr B. 2004;813:1-26. 33. Li C, Hong Y, Tan YX, Zhou H, Ai JH, Li SJ, et al. Accurate qualitative and quantitative proteomic analysis of clinical hepatocellular carcinoma using laser capture microdissection coupled with isotope-coded affinity tag and two-dimensional liquid chromatography mass spectrometry. Mol Cell Proteomics. 2004;3:399-409. 34. Zhang H, Yan W, Aebersold R. Chemical probes and tandem mass spectrometry: a strategy for the quantitative analysis of proteomes and subproteomes. Curr Opin Chem Biol. 2004;8:66-75. 35. Kovanich D, Cappadona S, Raijmakers R, Mohammed S, Scholten A, Heck AJR. Applications of stable isotope dimethyl labeling in quantitative proteomics. Anal Bioanal Chem. 2012;404:991-1009. 36. Pan Y, Cheng K, Mao J, Liu F, Liu J, Ye M, et al. Quantitative proteomics reveals the kinetics of trypsin-catalyzed protein digestion. Anal Bioanal Chem. 2014;406:6247-56. 37. Zhang Y, Xie X, Zhao X, Tian F, Lv J, Ying W, et al. Systems analysis of singly and multiply O-glycosylated peptides in the human serum glycoproteome via EThcD and HCD mass spectrometry. J Proteomics. 2018;170:14-27. 38. Wiese S, Reidegeld KA, Meyer HE, Warscheid B. Protein labeling by iTRAQ: A new tool for quantitative mass spectrometry in proteome research. Proteomics. 2007;7:340-50. 39. Bai B, Tan H, Pagala VR, High AA, Ichhaporia VP, Hendershot L, et al. Deep Profiling of Proteome and Phosphoproteome by Isobaric Labeling, Extensive Liquid Chromatography, and Mass Spectrometry. Methods Enzymol. 2017;585:377-95. 40. Ting L, Rad R, Gygi SP, Haas W. MS3 eliminates ratio distortion in isobaric multiplexed quantitative proteomics. Nat methods. 2011;8:937-40.

41. Bantscheff M, Boesche M, Eberhard D, Matthieson T, Sweetman G, Kuster B. Robust and sensitive iTRAQ quantification on an LTQ Orbitrap mass spectrometer. Mol Cell Proteomics. 2008;7:1702-13. 42. Roumeliotis TI, Weisser H, Choudhary JS. Evaluation of a Dual Isolation Width Acquisition Method for Isobaric Labeling Ratio Decompression. J Proteome Res. 2018. 43. Virreira Winter S, Meier F, Wichmann C, Cox J, Mann M, Meissner F. EASI-tag enables accurate multiplexed and interference-free MS2-based proteome quantification. Nat Methods. 2018;15:527-30. 44. Meissner F, Mann M. Quantitative shotgun proteomics: considerations for a high-quality workflow in immunology. Nat Immunol. 2014;15:112. 45. Canterbury JD, Merrihew GE, MacCoss MJ, Goodlett DR, Shaffer SA. Comparison of data acquisition strategies on quadrupole ion trap instrumentation for shotgun proteomics. J Am Soc Mass Spectrom. 2014;25:2048-59. 46. Olsen JV, Schwartz JC, Griep-Raming J, Nielsen ML, Damoc E, Denisov E, et al. A dual pressure linear ion trap orbitrap instrument with very high sequencing speed. Mol Cell Proteomics. 2009;8:2759. 47. Venable JD, Dong M-Q, Wohlschlegel J, Dillin A, Yates Iii JR. Automated approach for quantitative analysis of complex peptide mixtures from tandem mass spectra. Nat Methods. 2004;1:39. 48. Gillet LC, Navarro P, Tate S, Röst H, Selevsek N, Reiter L, et al. Targeted Data Extraction of the MS/MS Spectra Generated by Data-independent Acquisition: A New Concept for Consistent and Accurate Proteome Analysis. Mol Cell Proteomics. 2012;11:O111.016717. 49. Liu Y, Hüttenhain R, Surinova S, Gillet LCJ, Mouritsen J, Brunner R, et al. Quantitative measurements of N-linked glycoproteins in human plasma by SWATH-MS. Proteomics. 2013;13:1247-56. 50. Huttlin EL, Jedrychowski MP, Elias JE, Goswami T, Rad R, Beausoleil SA, et al. A Tissue- Specific Atlas of Mouse Protein Phosphorylation and Expression. Cell. 2010;143:1174-89. 51. Tabb DL, Vega-Montoto L, Rudnick PA, Variyath AM, Ham A-JL, Bunk DM, et al. Repeatability and reproducibility in proteomic identifications by liquid chromatography-tandem mass spectrometry. J Proteome Res. 2010;9:761-76. 52. Marx V. Targeted proteomics. Nat Methods. 2012;10:19. 53. Doerr A. Targeting with PRM. Nat Methods. 2012;9:950. 54. Doerr A. Mass spectrometry–based targeted proteomics. Nat Methods. 2012;10:23. 55. Hüttenhain R, Surinova S, Ossola R, Sun Z, Campbell D, Cerciello F, et al. N-Glycoprotein SRMAtlas. Mol Cell Proteomics. 2013;12(4):1005. 56. Prakash A, Tomazela DM, Frewen B, MacLean B, Merrihew G, Peterman S, et al. Expediting the Development of Targeted SRM Assays: Using Data from Shotgun Proteomics to Automate Method Development. J Proteome Res. 2009;8:2733-9. 57. Lange V, Picotti P, Domon B, Aebersold R. Selected reaction monitoring for quantitative proteomics: a tutorial. Mol Syst Biol. 2008;4:222-. 58. Reiter L, Rinner O, Picotti P, Hüttenhain R, Beck M, Brusniak M-Y, et al. mProphet: automated data processing and statistical validation for large-scale SRM experiments. Nat Methods. 2011;8:430. 59. Rauniyar N. Parallel Reaction Monitoring: A Targeted Experiment Performed Using High Resolution and High Mass Accuracy Mass Spectrometry. Int J Mol Sci. 2015;16:28566-81. 60. Peterson AC, Russell JD, Bailey DJ, Westphall MS, Coon JJ. Parallel Reaction Monitoring for High Resolution and High Mass Accuracy Quantitative, Targeted Proteomics. Mol Cell Proteomics. 2012;11:1475.

61. Dekker LJM, Zeneyedpour L, Snoeijers S, Joore J, Leenstra S, Luider TM. Determination of Site-Specific Phosphorylation Ratios in Proteins with Targeted Mass Spectrometry. J Proteome Res. 2018;17:1654-63. 62. Escher C, Reiter L, MacLean B, Ossola R, Herzog F, Chilton J, et al. Using iRT, a normalized retention time for more targeted measurement of peptides. Proteomics. 2012;12:1111- 21. 63. Knowles JR. Enzyme-Catalyzed Phosphoryl Transfer Reactions. Annu Rev Biochem. 1980;49:877-919. 64. Qiu H, Wang Y. Probing Adenosine Nucleotide-Binding Proteins with an Affinity-Labeled Nucleotide Probe and Mass Spectrometry. Anal Chem. 2007;79:5547-56. 65. Patricelli MP, Szardenings AK, Liyanage M, Nomanbhoy TK, Wu M, Weissig H, et al. Functional interrogation of the kinome using nucleotide acyl phosphates. Biochem. 2007;46:350- 8. 66. Xiao Y, Guo L, Jiang X, Wang Y. Proteome-wide discovery and characterizations of nucleotide-binding proteins with affinity-labeled chemical probes. Anal Chem. 2013;85:3198-206. 67. Manning G, Whyte DB, Martinez R, Hunter T, Sudarsanam S. The Protein Kinase Complement of the Human Genome. Science. 2002;298:1912. 68. Blume-Jensen P, Hunter T. Oncogenic kinase signalling. Nature. 2001;411:355-65. 69. Wu P, Nielsen TE, Clausen MH. FDA-approved small-molecule kinase inhibitors. Trends Pharmacol Sci. 2015;36:422-39. 70. Davis MI, Hunt JP, Herrgard S, Ciceri P, Wodicka LM, Pallares G, et al. Comprehensive analysis of kinase inhibitor selectivity. Nat Biotechnol 2011;29:1046-51. 71. Fabian MA, Biggs WH, 3rd, Treiber DK, Atteridge CE, Azimioara MD, Benedetti MG, et al. A small molecule-kinase interaction map for clinical kinase inhibitors. Nat Biotechnol 2005;23:329-36. 72. Karaman MW, Herrgard S, Treiber DK, Gallant P, Atteridge CE, Campbell BT, et al. A quantitative analysis of kinase inhibitor selectivity. Nat Biotechnol. 2008;26:127. 73. Pulford K, Lamant L, Morris SW, Butler LH, Wood KM, Stroud D, et al. Detection of Anaplastic Lymphoma Kinase (ALK) and Nucleolar Protein Nucleophosmin (NPM)-ALK Proteins in Normal and Neoplastic Cells With the Monoclonal Antibody ALK1. Blood. 1997;89:1394. 74. Miao W, Guo L, Wang Y. Imatinib-Induced Changes in Protein Expression and ATP- Binding Affinities of Kinases in Chronic Myelocytic Leukemia Cells. Anal Chem. 2019;91:3209- 14. 75. Daub H, Olsen JV, Bairlein M, Gnad F, Oppermann FS, Körner R, et al. Kinase-Selective Enrichment Enables Quantitative Phosphoproteomics of the Kinome across the Cell Cycle. Mol Cell. 2008;31:438-48. 76. Bantscheff M, Eberhard D, Abraham Y, Bastuck S, Boesche M, Hobson S, et al. Quantitative chemical proteomics reveals mechanisms of action of clinical ABL kinase inhibitors. Nat Biotechnol. 2007;25:1035. 77. Stuhlmiller TJ, Miller SM, Zawistowski JS, Nakamura K, Beltran AS, Duncan JS, et al. Inhibition of lapatinib-induced kinome reprogramming in ERBB2-positive breast cancer by targeting BET family bromodomains. Cell Rep. 2015;11:390-404. 78. Duncan JS, Whittle MC, Nakamura K, Abell AN, Midland AA, Zawistowski JS, et al. Dynamic reprogramming of the kinome in response to targeted MEK inhibition in triple-negative breast cancer. Cell. 2012;149:307-21. 79. Miao W, Xiao Y, Guo L, Jiang X, Huang M, Wang Y. A high-throughput targeted proteomic approach for comprehensive profiling of methylglyoxal-induced perturbations of the human kinome. Anal Chem. 2016;88:9773-9.

80. Xiao Y, Guo L, Wang Y. A targeted quantitative proteomics strategy for global kinome profiling of cancer cells and tissues. Mol Cell Proteomics. 2014;13:1065-75. 81. D S. The NTP phosphate donor in kinase reactions: is ATP a monopolist? Acta Biochim Pol. 1996;43:9-23. 82. Chène P. ATPases as drug targets: learning from their structure. Nat Rev, Drug Discov. 2002;1:665. 83. Singleton MR, Dillingham MS, Wigley DB. Structure and mechanism of helicases and nucleic acid translocases. Annu Rev Biochem. 2007;76:23-50. 84. Xiao Y, Guo L, Wang Y. Isotope-coded ATP probe for quantitative affinity profiling of ATP-binding proteins. Anal Chem. 2013;85:7478-86. 85. Xiao Y, Guo L, Jiang X, Wang Y. Proteome-Wide Discovery and Characterizations of Nucleotide-Binding Proteins with Affinity-Labeled Chemical Probes. Anal Chem. 2013;85:3198- 206. 86. Adachi J, Kishida M, Watanabe S, Hashimoto Y, Fukamizu K, Tomonaga T. Proteome- wide discovery of unknown ATP-binding proteins and kinase inhibitor target proteins using an ATP probe. J Proteome Res. 2014;13:5461-70. 87. Ito J, Heazlewood JL, Millar AH. Analysis of the Soluble ATP-Binding Proteome of Plant Mitochondria Identifies New Proteins and Nucleotide Triphosphate Interactions within the Matrix. J Proteome Res. 2006;5:3459-69. 88. Miao W, Fan M, Huang M, Li JJ, Wang Y. Targeted profiling of heat shock proteome in radioresistant breast cancer cells. Chem Res Toxicol. 2019;32:326-32.

Figure 1.1. General procedures of label-free (a) and SILAC (b) experiment for quantitative proteomics.

a b State A State B State A State B

Light Amino Acids Heavy Amino Acids

Lysis Lysis Combine, mix at 1:1 ratio

Tryptic digestion Tryptic digestion Tryptic digestion

LC-MS/MS LC-MS/MS LC-MS/MS

Data analysis Data analysis

Figure 1.2. ICAT quantitative proteomics. (a) The structure of ICAT reagent (adapted from

Nat Protoc. 2007, 1, 2650). (b) General procedures of ICAT experiment for quantitative proteomics.

b State A State B

Lysis Lysis

React with light React with heavy ICAT ICAT

Combine, mix at 1:1 ratio

Tryptic digestion

LC-MS/MS

Data analysis

Figure 1.3. iTRAQ quantitative proteomics. (a) The structure of iTRAQ reagent. (b)

General procedures of iTRAQ experiment for quantitative proteomics.

a Peptide Reactive Group Reporter Mass 114-117

Balance Mass 31-28 Isobaric Tag Mass 145 b State 1 State 2 State 3 State 4

Lysis, Reduction, Alkylation & Enzymatic Digestion

React with React with React with React with ITRAQ 114 ITRAQ 115 ITRAQ 116 ITRAQ 117

Combine, mix at 1:1 ratio

LC-MS/MS

Data analysis

Figure 1.4. Typical MS/MS scan events in mass spectrometer for discovery proteomics in DDA (adopted from Mol Cell Proteomics. 2009, 8, 2759.) (a) and DIA (adopted from

Mol Cell Proteomics. 2012, 11, O111.016717.) (b) mode.

Figure 1.5. Targeted proteomics analysis. (A) MRM analysis on a triple quadrupole (QQQ) mass spectrometer and (b) PRM analysis on a Q Exactive mass spectrometer.

a Q1 Q2 Q3 100 100 80 80

60 60

40 40 Relative

20 20 Abundance

Precursor Fragmentation Fragment 0 0 Selection Selection Retention Time

b Orbitrap Analyzer Q1 HCD

Precursor Selection Fragmentation

Known RT

Intensity MeasuredRT

Retention Time iRT

RT RT1 RTX RT2 iRTX = [(RTX - RT1) / (RT2 - RT1)] * 100 iRT 0 x 100

b Reference peptides

Known iRT MeasuredRT

iRT RT = a (iRT) + b

Figure 1.7. The dendrogram of the human kinome (adopted from Science. 2002, 298,

1912.).

Chapter 2

A High-throughput Targeted Proteomic Approach for

Comprehensive Profiling of Methylglyoxal-induced

Perturbations of the Human Kinome

Introduction

Kinases are responsible for catalyzing the phosphorylation of numerous biological molecules (1), and they play crucial roles in cell signaling and regulating cell proliferation and metabolism (2). Aberrant expression and/or activation of kinases have been found in cancer and other human diseases (3). However, the extremely low levels of expression of many kinases hamper global kinome studies. Recently published results showed that ATP- affinity probes could be employed for the covalent labeling, enrichment, and subsequent mass spectrometric identification and quantification of kinases with excellent specificity and sensitivity (Figure 2.1a) (4, 5).

Targeted proteomics technique, which relies on multiple-reaction monitoring (MRM) on a triple-quadrupole mass spectrometer, has become extensively employed in quantitative proteomics studies (6). Compared to data-dependent analysis (DDA)-based discovery mode, the MRM-based targeted proteomic method exhibits a sensitivity that is two orders of magnitude higher towards peptide detection (7). We recently developed an

MRM-based method for high-throughput assessment of kinase expression and activation

(8). The previous version of the kinome library encompassed 320 unique peptides from

242 kinases, among which approximately 200 were protein kinases. Importantly, this library contains information about the integrated retention times (iRT)(9) and unique MRM transitions for kinase peptides. Along with the use of desthiobiotin-containing isotope- coded ATP-affinity probe (ICAP) or stable isotope labeling by amino acids in cell culture

(SILAC), the method enables quantitative profiling of the global kinome (10, 11). The previous version of the library, however, only encompasses approximately 40% of the human kinome, which leaves a large portion of kinases inaccessible (12). In addition, a similar number of kinases were identified with the use of kinome beads for kinase enrichment (13-16). Thus, monitoring the majority of the human kinome necessitates the expansion of the MRM kinome library.

Methylglyoxal (MG) is formed from non-enzymatic and enzymatic degradation of triose phosphates produced in the glycolysis pathway (i.e. dihydroxyacetone phosphate and glyceraldehyde-3-phosphate), and from the oxidation of hydroxyacetone and aminoacetone (17). As a result, hyperglycemia leads to elevated formation of MG, where the serum concentrations of MG increase by 5-6- and 2-3-fold in patients with type-1 and type-2 diabetes mellitus, respectively.(18) As a highly reactive -ketoaldehyde, MG reacts with arginine, lysine and cysteine residues in proteins to yield advanced glycation end- products (AGEs), and AGE formation is considered a potential link between hyperglycemia and chronic diabetic complications (17). MG can also induces the formation of N2(1-carboxyethyl)-2’-deoxyguanosine (N2-CEdG) as the major stable DNA adduct and cause DNA damage (19, 20). However, no study has been conducted about how MG

31 perturbs cell signaling by altering the expression or activation of kinases at the global kinome level.

In this study, we expanded substantially the MRM-based kinome library by interrogating the global kinomes of 9 additional human cell lines and by using synthetic desthiobiotinylated kinase peptides. We then employed an MRM-based targeted proteomic method with the expanded kinome library to profile the perturbations of kinome of

HEK293T cells upon treatment with methylglyoxal. More than 300 unique kinases were quantified; a number of kinases in the MAPK pathway and several receptor tyrosine kinases were found to be substantially altered in response to methylglyoxal treatment.

Together, we expanded successfully our MRM kinome library to achieve an unprecedented coverage of the human kinome and our study also provided important insights into the roles of kinase signaling in the development of diabetic complications.

Materials and Methods

Cell culture

MCF-7 and DU-145 cells (ATCC) were cultured in Eagle's Minimum Essential

Medium. GM-00637, GM-04429 (obtained from Prof. Gerd P. Pfeifer), GM15876A

(provided by Prof. Karlene Cimprich) (21), HCT-116, U2OS, and HEK293T (ATCC) cells were cultured in Dulbecco's Modified Eagle Medium. CEM cells (ATCC) were cultured in RPMI 1640 Medium. All cell culture media were supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA) and penicillin (100 IU/mL). The cells were maintained

7 at 37°C in a humidified atmosphere containing 5% CO2. Approximately 2×10 cells were

32 harvested, washed with cold PBS for three times, and lysed in a 1-mL lysis buffer, which contained 0.7% CHAPS, 50 mM HEPES (pH 7.4), 0.5 mM EDTA, 100 mM NaCl, and 10

µL protease inhibitor cocktail on ice for 30 min. The cell lysates were centrifuged at

16,000g at 4°C for 30 min and the resulting supernatants collected.

Labeling of ATP-binding proteins, tryptic digestion, and affinity purification of desthiobiotin-C3-labeled peptides

Approximately 2×107 cells were harvested, washed with cold PBS for three times, and lysed in a 1 mL lysis buffer, which contained 0.7% CHAPS, 50 mM HEPES (pH 7.4), 0.5 mM EDTA, 100 mM NaCl, and 10 µL (1:100) protease inhibitor cocktail on ice for 30 min.

The cell lysates were centrifuged at 16,000g at 4°C for 30 min and the resulting supernatants were collected. Endogenous nucleotides in the resulting protein extract was removed by gel filtration using a NAP-25 column (Amersham Biosciences). Cell lysates were subsequently eluted into a buffer containing 50 mM HEPES (pH 7.4), 75 mM NaCl, and 5% glycerol. The amounts of proteins in the lysates were quantified using Quick Start

Bradford Protein Assay (Bio-Rad). Prior to the labeling reaction, MgCl2, MnCl2, and

CaCl2 were added to the concentrated cell lysate until their final concentrations reached 50,

5 and 5 mM, respectively.

Approximately 1 mg cell lysate in a 1-mL solution was treated with the light desthiobiotin-C3-ATP affinity probe (Figure 1a) at a final concentration of 100 µM.

Labeling reactions were conducted at room temperature with gentle shaking for 2.5 h. After the reaction, the remaining probes in the cell lysates were removed by buffer exchange

33 with 50 mM NH4HCO3 (pH 8.5) using Amicon Ultra-4 filters (10,000 NMWL, Millipore).

After addition of 8 M urea for protein denaturation, dithiothreitol and iodoacetamide for cysteine reduction and alkylation, the labeled proteins were digested with modified sequencing-grade trypsin (Roche Applied Science) at an enzyme/substrate ratio of 1:100 in 50 mM NH4HCO3 (pH 8.5) at 37°C overnight. The peptide mixture was subsequently dried in a Speed-vac and redissolved in 1 mL PBS buffer (100 mM potassium phosphate and 0.15 M NaCl, pH 7.5), to which solution was subsequently added 300 µL avidin- agarose resin (Sigma-Aldrich). The mixture was then incubated at room temperature for 1 h with gentle shaking. The agarose resin was washed sequentially with 3 mL PBS buffer and 3 mL H2O to remove unbound peptides, and the desthiobiotin-conjugated peptides were subsequently eluted with 1% TFA in CH3CN/H2O (7:3, v/v) at 75°C. The resulting enriched peptide samples were desalted by employing OMIX C18 pipet tips (Agilent

Technologies) and analyzed on an LTQ-Orbitrap Velos mass spectrometer (Thermo Fisher

Scientiﬁc) for discovery-mode analysis.

LC-MS/MS analysis on an LTQ-Orbitrap Velos mass spectrometer

Probe pull down samples were analyzed by on-line 2D-LC-MS/MS and synthetic peptides were analyzed by 1D-LC-MS/MS on an LTQ-Orbitrap Velos mass spectrometer equipped with a nanoelectrospray ionization source. The fully automated 8-cycle on-line two-dimensional LC-MS/MS was set up as described(22) with minor modifications.

Briefly, samples were automatically loaded from a 48-well microplate autosampler using an EASY-nLC II system (Thermo Fisher Scientific) at 3 µL/min onto a biphasic precolumn

(150 µm i.d.) comprised of a 3.5-cm column packed with 5 µm C18 120 Å reversed-phase

34 material (ReproSil-Pur 120 C18-AQ, Dr. Maisch) for 1D-LC-MS/MS, and packed for another 3.5-cm with Luna 5 µm SCX 100 Å strong cation-exchange resin (Phenomenex,

Torrance, CA) for on-line 2D-LC-MS/MS. The biphasic trapping column was connected to a 20-cm fused-silica analytical column (PicoTip Emitter, New Objective, 75 µm i.d.) with 3 µm C18 beads (ReproSil-Pur 120 C18-AQ, Dr. Maisch). For on-line 2D separation, ammonium acetate at concentrations of 0, 5, 10, 20, 50, 100, 200, 500 and 1000 mM were then sequentially injected using a 48-well autosampler from the sample vial to elute bound peptides from the precolumn to the analytical column with reversed-phase separation. The peptides were then separated with a 180-min linear gradient of 2-35% acetonitrile in 0.1% formic acid and at a flow rate of 250 nL/min. The LTQ-Orbitrap Velos was operated in a data-dependent scan mode. Full-scan mass spectra were acquired using the Orbitrap analyzer with a resolution of 60,000 with lock mass option enabled for the ion of m/z

445.120025(23). Up to 20 most abundant ions found in MS with charge state of 2 and above were sequentially isolated and sequenced in the linear ion-trap with a normalized collision energy of 35, an activation Q value of 0.25, and an activation time of 10 ms.

Retention time (RT) extraction and iRT calculation for kinase peptides

To calculate the iRT score (9) for each kinase peptide in the MRM kinome library, 10 peptides derived from the tryptic digestion mixture of bovine serum albumin (BSA) were selected to constitute the reference peptides for a new iRT scale. iRT values of these 10

BSA standard peptides were calculated using their empirically measured RT from shotgun

LC-MS/MS analysis by setting iRT scores of peptides AEFVEVTK and

DAFLGSFLYEYSR as 0 and 100, respectively. The BSA peptide mixture was then added

35 to desthiobiotin-C3-labeled kinase peptide mixture from the lysates of GM04429,

GM00637, HEK293T, GM15876A, DU-145, HCT116, MCF7, U2OS and CEM cells and measured by 1D-LC-MS/MS on the LTQ Orbitrap Velos with a 130-min linear gradient.

RTs were extracted for all BSA standard peptides as well as kinase peptides using the

Skyline MS1 filtering workflow.(24) The transformed iRT values for all the newly identified kinase peptides were calculated based on linear regression of iRT and experimentally measured RT of peptides with previously determined iRT score. With the accumulation of iRT scores for targeted kinase peptides, it is not necessary to include the

BSA standard mixture for RT extraction and iRT calculation. Any common kinase peptides that yield a regression with a R2 value of 0.99 or higher were used as standard peptides to calculate the new iRT score for newly identified kinase peptides by Skyline.

LC-MRM Analysis

All LC-MRM experiments were carried out on a TSQ Vantage triple-quadrupole mass spectrometer equipped with a nanoelectrospray ionization source coupled to an

EASY-nLC II system (Thermo Scientific). Samples were automatically loaded onto a 4- cm trapping column (150 µm i.d.) packed with 5 µm 120 Å reversed phase C18 material

(ReproSil-Pur 120 C18-AQ, Dr. Maisch) at 3 µL/min. The trapping column was coupled to a 20-cm fused silica analytical column (PicoTip Emitter, New Objective, 75 µm i.d.) with 3 µm C18 beads (ReproSil-Pur 120 C18-AQ, Dr. Maisch). The peptides were then separated with a 130-min linear gradient of 2-35% acetonitrile in 0.1% formic acid and at a flow rate of 230 nL/min. The spray voltage was 1.9 kV. Q1 and Q3 resolutions were 0.7

Da and the cycle time was 5 s.

To enable automated, multiplexed MRM analyses, we generated an interactive

Skyline spectral library file containing tandem mass spectra of all unique desthiobiotin-

C3 conjugated kinase peptides from 15 cell line samples and synthetic peptides, along with their iRT values (9), which were acquired from discovery-mode analysis on an LTQ

Orbitrap Velos mass spectrometer, using Skyline (version 3.5) (25). Collision energies were calculated using a linear equation that was specific to the TSQ Vantage instrument according to the Skyline default setting. BSA standard mixtures were analyzed in unscheduled MRM-mode prior to the analysis of the enriched desthiobiotin-C3 conjugated peptides. The linear predictor of empirical RT from iRT for targeted kinase peptides was then determined by the linear regression of RTs of BSA standard peptides obtained for the current chromatography setup. This linear predictor was rechecked between every five MRM sample analyses by injecting another BSA standard mixture.

Three or four pairs of transitions were monitored for each light/heavy desthiobiotin-C3- labeled peptide in quantification experiments. These targeted transitions were monitored in five separate injections for each sample in scheduled MRM mode with a retention time window of 8 min.

All raw files were processed using Skyline (version 3.5) for the generation of extracted-ion chromatograms and peak integration. The targeted peptides were first manually checked to ensure the overlaid chromatographic profiles of multiple fragment ions derived from light and heavy forms of the same peptide. The data were then processed to ensure that the distribution of the relative intensities of multiple transitions associated with the same precursor ion correlates with the theoretical distribution from

37 kinome MS/MS spectral library entry. The sum of peak area from all transitions of light- or heavy-labeled peptides was used for quantification.

Western blot

HEK293T cells were cultured in 75 cm2 cell culture flasks until 40-50% confluency and treated with methylglyoxal. Cells were lysed following the aforementioned procedures.

The concentrations of the resulting protein lysates were determined by Bradford Assay

(Bio-Rad). The whole cell lysate for each sample (10 μg) was denatured by boiling in

Laemmli loading buffer and then resolved by using SDS-PAGE. Subsequently, the proteins were transferred onto the nitrocellulose membrane at 4 °C overnight. The resulting membrane was blocked with PBS-T (PBS with 0.1% Tween 20) containing 5% milk (Bio-Rad) at 4°C for 6 h. Next, the membrane was incubated with primary antibody at 4°C overnight and then secondary antibody at room temperature for 1 h. After thorough washing with PBS-T buffer, the HRP signals were detected with Pierce ECL Western

Blotting Substrate (Thermo).

Phospho-FRS2- (Tyr196) antibody (Cell Signaling # 3864, with a 1:1000 dilution),

Phospho-FGFR (Tyr653/654) antibody (Cell Signaling #3471, with a 1:1000 dilution),

EGFR 1005 antibody (Santa Cruz Biotechnology, sc-03, with a 1:10000 dilution), IGF2R

H-20 antibody (Santa Cruz Biotechnology, sc-14408, with a dilution ratio of 1:1000), and

FRS2 H-91 antibody (Santa Cruz Biotechnology, sc-14408, with a dilution ratio of 1:1000) were employed as primary antibodies. Horseradish peroxidase-conjugated anti-rabbit IgG and Alexa Fluor® 647 donkey anti-goat IgG (H+L) were used as secondary antibodies.

Membranes were also probed with anti-actin antibody (Cell Signaling #4967, at a dilution ratio of 1:10000) to confirm equal protein loading. Cells were treated with 100 ng/mL

FGF21 (Sigma-Aldrich) for 20 min before harvesting for Western blot analysis of phospho-

FRS2- and phospho-FGFR.

Real-time PCR

HEK293T cells were seeded in 12-well plates at 50% conﬂuence level. Total RNA was extracted from cells using TRI Reagent (Sigma). Approximately 1 μg RNA was reverse transcribed by employing M-MLV reverse transcriptase (Promega) and an oligo(dT)16 primer. After incubating at 42°C for 60 min, the reverse transcriptase was deactivated by heating at 85°C for 5 min. Quantitative real-time PCR was performed using iQ SYBR

Green Supermix kit (Bio-Rad) on a Bio-Rad iCycler system (Bio-Rad), and the running conditions were at 95°C for 3 min and 45 cycles at 95°C for 15 sec, 55°C for 30 sec, and

72°C for 45 sec. The comparative cycle threshold (Ct) method (ΔΔCt) was used for the relative quantification of gene the qRT-PCR expression. Primers are listed in Table 2.1.

The mRNA level of each gene was normalized to that of the internal control (GAPDH).

Results

The major objectives of the present study are to expand the MRM kinome library for encompassing the majority of the human kinome and to employ the MRM-based targeted proteomic approach to explore the kinase-mediated signaling events that are perturbed by

MG.

1. Development of an MRM Assay for Human Kinome Profiling

Expansion of MRM Kinome Library – We first purified, by using avidin agarose beads, the desthiobiotin-C3-labeled kinase peptides from the tryptic digestion mixtures of lysates of 9 human cell lines (i.e. GM-00637, GM-04429, GM15876A, HCT-116, U2OS, CEM,

DU-145, HEK293T and MCF-7) that were labeled with the light desthiobiotin-C3-ATP probe (Figure 2.1b). The resulting desthiobiotin-C3-labeled peptides were analyzed on an

LTQ Orbitrap Velos mass spectrometer in the DDA mode, which included a total of more than 200 LC-MS/MS runs. The tandem mass spectra and the retention time information of all desthiobiotin-C3-labeled kinase peptides were imported into Skyline to expand the initial kinome library that was constructed from LC-MS/MS data acquired from 6 human cell lines (i.e. K562, IMR-90, HeLa-S3, Jurkat-T, WM-115, and WM-266-4).(8) A maximum of five peptides were selected for each kinase, and the three or four most abundant y-ions observed in MS/MS acquired in the DDA mode were employed for MRM monitoring of each peptide.

Peptide selection - Owing to its relatively high reactivity, the desthiobiotin-C3-ATP probe may react with, apart from the lysine residue(s) located at the ATP-binding site, lysine residues in other proteins through electrostatic interactions (4, 5). Thus, we classified the kinase peptides from the 15 cell lines in our library into two groups on the basis of local amino acid sequences surrounding the probe-labeled lysine, as described previously (8).

The first group of peptides harbor at least one of the three ATP-binding motifs

(HRDXKXXN, VAXK or GXXXXGK, with ‘X’ being any amino acid) (4, 5), whereas the second group includes those peptides that do not contain any of the three conserved

40 motifs but were identified at least five times in our shotgun proteomics experiments.

Although the latter group of peptides do not reside in any of the known conserved ATP- binding motifs and their involvement in ATP binding remains to be determined, their high frequencies of identification in shotgun proteomic studies render these peptides suitable candidates for quantifying the corresponding kinases.

With the use of the ICAP reagents (Figure 2.1b), a large portion of protein kinases were labeled on the lysine residues located in the ATP-binding motifs that are highly conserved.

As a result, some of the identified desthiobiotin-C3-labeled peptides may be assigned to multiple protein kinases. To avoid ambiguity in kinase quantification with the MRM method, we inspected each identified desthiobiotin-C3-labeled peptide in the MRM kinome library and deleted those that could be attributed to multiple kinases.

Synthetic peptides – After combining all unique kinase peptides from the 15 cell lines, the MRM library contains 640 distinct peptides derived from 370 kinases (by using DAVID

Bioinformatics Resources) (26, 27), which cover about 60% of the human kinome. To further expand the library, we checked the sequences of the protein kinases (12) that were not identified from the aforementioned shotgun proteomic analyses. We then designed 192 unique desthiobiotin-labeled peptides that encompass at least one ATP-binding motif of

163 protein kinases, and these peptides were custom-synthesized. The synthetic peptides were then analyzed by LC-MS and MS/MS in the DDA mode, and the acquired data were processed by MaxQuant and incorporated into the Skyline library, as described above. By doing so, we substantially expanded our MRM library to contain 818 unique peptides from

474 distinct human kinases. These include 409 protein kinases, 14 lipid kinases, and 51

41 other kinases (e.g. nucleoside kinases and carbohydrate kinases). Thus, our expanded

MRM kinome library has a coverage of approximately 80% of the human kinome which contains a total of 518 protein kinases (12).

Retention time calibration – Considering that at least three transitions are needed for each of light- and heavy-probe-labeled peptide, quantitative measurements of 818 kinase peptides by MRM necessitates more than 5000 MRM transitions. To achieve this level of multiplexed detection, it is essential to employ scheduled MRM analysis where the mass spectrometer is programmed to detect only a limited number of peptides in each pre- defined retention time window. Therefore, accurate retention time (RT) prediction for the kinase peptides is essential for our multiplexed MRM-based kinome assay. To this end, we calculated the iRT value for each peptide on our target list following a previously published method.(9) Based on retention time information of targeted peptides analyzed on an LTQ

Orbitrap Velos coupled with EASY-nLC II system, we used 10 BSA peptides as standards and successfully converted empirically determined retention times of 98% (803 out of 818) kinase peptides into normalized iRT scores, which reflect their conserved elution order.

Since the iRT value represents an inherent attribute of the hydrophobicity of a peptide, it can also serve as another parameter to validate the quantification results from MRM assay, where any significant deviation from the linear plot of iRT versus the measured RT could be attributed to false-positive detection.

It is worth noting that the custom-synthesized peptides carry a desthiobiotin-labeled lysine, whereas the kinase peptides arising from ATP affinity probe labeling contain a desthiobiotin-C3-labeled lysine. Peptides harboring these two types of modified lysine may

42 display different retention times in LC; thus, it is important to establish the relationship between iRT values for peptides carrying desthiobiotin and desthiobiotin-C3 modifications.

To this end, we conducted another set of pull-down experiments with the use of desthiobiotin-ATP probe, which led to the identification of approximately 1000 desthiobiotin-labeled peptides. With the iRT values of the desthiobiotin-labeled kinase peptides and the existing iRT values of the corresponding desthiobiotin-C3-labeled kinase peptides in the library, a linear relationship in iRT values between the desthiobiotin-labeled peptides and the desthiobiotin-C3-labeled counterparts could be established (Figure 2.1c,

Table 2.2). By employing this linear equation and the iRT values for all the desthiobiotin- labeled synthetic peptides, the iRT values for the corresponding desthiobiotin-C3-labeled peptides could be determined and integrated into the MRM library.

Library test - With the above MRM-based assay design, we set out to assess the kinome coverage of the method using the lysate of GM00637 cells. It turned out that the MRM- based kinome analysis led to the detection of 470 unique peptides from 313 distinct human kinases in 5 LC-MS/MS runs. On the other hand, the shotgun proteomic approach using online 2D-LC-MS/MS with a 9-step gradient during SCX fractionation only gave rise to the identification of 174 kinases. This result demonstrates that the MRM-based kinome assay provides superior coverage of the human kinome over the DDA method (Figure 2.1d).

2. Methylglyoxal-induced alterations of human kinome in HEK293T cells

We applied the MRM-based kinome library, along with our ICAP probes, to assess the perturbations of the kinome of HEK293T cells upon a 24-hr treatment with 200 M

43 methylglyoxal (Figure 2.2). In this vein, it is of note that the estimated rate of MG formation in tissues of healthy human subjects was 125 M per day (28). In addition, previous studies showed that only about 2-3% of MG in the culture medium is incorporated into mammalian cells (29, 30). Thus, the concentration of the MG used in the present study is physiologically relevant.

The results from the MRM-based targeted proteomic method led to the quantification of a total of 328 unique kinases, including 288 protein kinases, 11 lipid kinases and 29 other kinases, which cover approximately 55% of the human kinome (Figure 2.3). In the viewpoint that approximately 400 protein kinases are expressed in a single cell (14), more than 70% of the expressed protein kinases could be quantified with the MRM-based kinome assay. All the quantified kinase peptides exhibit an excellent linear fit between the observed retention time and the iRT in the library (Figure 2.4a). Similar relative abundances of 3 transitions obtained from MRM and shotgun proteomic analyses illustrate the robust quantification of the kinase peptides by MRM analysis (Figure 2.4c, d). More than 90% of the quantified kinase peptides displayed a consistent trend in forward and reverse labeling experiments in MRM-based targeted analysis (representative results for

EGFR are shown in Figure 2.4b), demonstrating the excellent performance of this method.

Among the 328 quantified kinases, 40 and 9 were down- and up-regulated (with at least

1.5-fold change), respectively, in HEK293T cells following a 24-hr treatment with 200 µM

MG. In addition, 11 out of 49 perturbed kinases are involved in MAPK pathway, rendering it the major pathway perturbed by MG treatment (Figure 2.5a). In this vein, the MG- induced alteration of MAPK signaling pathway was reported previously (31, 32).

Furthermore, 8 receptor tyrosine kinases (RTKs), including EPHA2, EPHA4, EGFR,

IGF2R, FGFR1, FGFR2, FGFR3 and NTRK2, were found to be altered upon MG treatment and all of them except FGFR2 were down-regulated (Figure 2.5b).

3. MG led to the down-regulation of RTKs

EGFR, IGF2R, and FGFR1 were found to be related to diabetes (33-35) and were all revealed by our global kinome profiling data to be down-regulated upon MG treatment

(Figure 2.5b). Thus, we focused on these kinases to explore the mechanisms through which their alterations are induced by MG treatment. To this end, we first examined whether the diminished levels of EGFR, IGF2R and FGFR1 proteins are due to reduced mRNA expression of the corresponding genes. Our real-time PCR results showed that only the mRNA level of EGFR gene was decreased with the dose of MG, whereas no apparent change in mRNA level was observed for IGF2R or FGFR gene (Figure 2.6a).

We next validated the MRM findings by using Western blot analysis and assessed whether the MG-induced decrease of the three receptor tyrosine kinases could be rescued by pre-incubation of cells with scavengers of -ketoaldehydes and an antioxidant. Indeed, our Western blot results revealed that EGFR protein displayed a time- and dose-dependent decrease in expression level in HEK293T cells upon MG treatment (Figure 2.6b&c).

Aminoguanidine (AG) and metformin (MET) are scavengers of -ketoaldehydes, and thus they inhibit the formation of MG-induced advanced glycation end-products (36, 37). N- acetyl-L-cysteine (NAC), as an antioxidant, inhibits ROS-dependent apoptosis (38). To exploit further the mechanisms underlying the MG-induced decline in EGFR protein level,

45 we pretreated HEK293T cells with 1 mM AG, NAC or MET for 7 hr prior to a 24-hr treatment with 500 µM MG. Our results showed that pretreatment with NAC, but not AG or MET, could significantly restore the expression level of EGFR (Figure 2.6d).

In line with the MRM data, Western blot result also revealed a dose-dependent decrease in IGF2R protein level in HEK293T cells upon MG treatment (Figure 2.7a). In addition, the MG-induced decrease in IGF2R expression could be rescued by pretreatment of

HEK293T cells with NAC, AG or MET (Figure 2.7b). FGFR activation with the use of recombinant fibroblast growth factor 21 has been proposed for the treatment of diabetes

(39), and our MRM kinome assay revealed the diminished expression/activation of FGFR1 upon MG treatment. Because of the relatively low level of expression of FGFR1 in

HEK293T cells, we were not able to detect FGFR1 protein in these cells using Western blot (data not shown). Instead, we employed Western blot to monitor the effects of MG treatment on FGF21-mediated FGFR1 activation by assessing its autophosphorylation on

Tyr653/654(40) and the phosphorylation of its substrate protein, FRS2 (41). Indeed we found that MG treatment resulted in diminished phosphorylation of FGFR1 and FRS2.

Moreover, the reduced phosphorylation of these two proteins could again be rescued by preincubation of HEK293T cells with NAC, AG or MET (Figure S5). Together, the above results suggest that MG elicited the decreased levels of FGFR1 and IGF2R through a common mechanism, which differs from that underlying the MG-induced down-regulation of EGFR.

Discussion

Reversible phosphorylation of proteins and other biomolecules by kinases and phosphatases constitute one of the most important and best studied pathways in cell signaling. Thus, high-throughput methods for global profiling of the human kinome will enable systematic interrogation of kinase-mediated molecular events conferred by intracellular signaling and extracellular cues. However, existing methods for profiling the global human kinome are hampered by the limited kinome coverage. In this present study, we made significant advances toward this challenge by establishing an MRM-based targeted proteomic approach with an unprecedented coverage of the human kinome. In particular, our expanded kinome library covers approximately 80% of human kinome and contains 474 human kinases (488 unique IPIs), among which 409, 14 and 51 were protein kinases, lipid kinases, and other kinases, respectively.

This method was successfully employed to assess the alterations of the kinome in

HEK293T cells after MG treatment, which allowed for the quantification of 328 kinases.

Among these quantified kinases, 49 were perturbed, 8 of which were RTKs. Further experiments on selected receptor tyrosine kinases showed that MG led to a dose-dependent decrease in the level of EGFR protein via a mechanism that differs from the diminished expression of IGF2R and FGFR1 proteins. While MG treatment led to a dose-dependent decline in mRNA level of EGFR gene, the mRNA expression of IGF2R or FGFR1 gene was not perturbed by MG exposure. In addition, while the MG-triggered decrease in all three receptor tyrosine kinases could be restored by preincubation of cells with an antioxidant (i.e. NAC), scavengers of -ketoaldehydes (i.e. AG and MET) could only

47 rescue the MG-induced reduction of IGF2R and FGFR1 proteins, but not that of EGFR.

The lack of effect of AG or MET on restoring the MG-elicited decrease in EGFR protein expression suggests that MG may bind directly to some cell-surface proteins, which may stimulate the production of ROS and ultimately lead to transcriptional down-regulation of

EGFR gene. The detailed molecular events underlying these processes, however, await further investigation. In addition, the capability of NAC in restoring the MG-induced decreases in all three receptor tyrosine kinases is in keeping with the previous proposal that hyperglycemia-induced ROS activates many pathways of tissue damage in diabetic patients (42, 43). Our results suggest that the inhibition of RTKs may contribute to the development of diabetic complications, and restoration of the functions of RTKs may provide new venues for the therapeutic interventions of diabetic complications.

References

1. Cheng HC, Qi RZ, Paudel H, Zhu HJ. Regulation and function of protein kinases and phosphatases. Enzyme Res. 2011;2011:794089. 2. Lemmon MA, Schlessinger J. Cell signaling by receptor tyrosine kinases. Cell. 2010;141(7):1117-34. 3. Blume-Jensen P, Hunter T. Oncogenic kinase signalling. Nature. 2001;411(6835):355-65. 4. Xiao Y, Guo L, Jiang X, Wang Y. Proteome-wide discovery and characterizations of nucleotide-binding proteins with affinity-labeled chemical probes. Anal Chem. 2013;85(6):3198- 206. 5. Xiao Y, Guo L, Wang Y. Isotope-coded ATP probe for quantitative affinity profiling of ATP-binding proteins. Anal Chem. 2013;85(15):7478-86. 6. Picotti P, Aebersold R. Selected reaction monitoring-based proteomics: workflows, potential, pitfalls and future directions. Nat Meth. 2012;9(6):555-66. 7. Lange V, Picotti P, Domon B, Aebersold R. Selected reaction monitoring for quantitative proteomics: a tutorial. Mol Syst Biol. 2008;4:222. 8. Xiao Y, Guo L, Wang Y. A Targeted Quantitative Proteomics Strategy for Global Kinome Profiling of Cancer Cells and Tissues. Mol Cell Proteomics. 2014;13(4):1065-75. 9. Escher C, Reiter L, MacLean B, Ossola R, Herzog F, Chilton J, et al. Using iRT, a normalized retention time for more targeted measurement of peptides. Proteomics. 2012;12(8):1111-21. 10. Guo L, Xiao Y, Fan M, Li JJ, Wang Y. Profiling global kinome signatures of the radioresistant MCF-7/C6 breast cancer cells using MRM-based targeted proteomics. J Proteome Res. 2015;14(1):193-201. 11. Guo L, Xiao Y, Wang Y. Application of adenosine triphosphate affinity probe and scheduled multiple-reaction monitoring analysis for profiling global kinome in human cells in response to arsenite treatment. Anal Chem. 2014;86(21):10700-7. 12. Manning G, Whyte DB, Martinez R, Hunter T, Sudarsanam S. The Protein Kinase Complement of the Human Genome. Science. 2002;298(5600):1912-34. 13. Stuhlmiller TJ, Miller SM, Zawistowski JS, Nakamura K, Beltran AS, Duncan JS, et al. Inhibition of Lapatinib-Induced Kinome Reprogramming in ERBB2-Positive Breast Cancer by Targeting BET Family Bromodomains. Cell Rep. 2015;11(3):390-404. 14. Duncan JS, Whittle MC, Nakamura K, Abell AN, Midland AA, Zawistowski JS, et al. Dynamic reprogramming of the kinome in response to targeted MEK inhibition in triple-negative breast cancer. Cell. 2012;149(2):307-21. 15. Daub H, Olsen JV, Bairlein M, Gnad F, Oppermann FS, Korner R, et al. Kinase-selective enrichment enables quantitative phosphoproteomics of the kinome across the cell cycle. Mol Cell. 2008;31(3):438-48. 16. Oppermann FS, Gnad F, Olsen JV, Hornberger R, Greff Z, Kéri G, et al. Large-scale Proteomics Analysis of the Human Kinome. Mol Cell Proteomics. 2009;8(7):1751-64. 17. Thornalley PJ. Pharmacology of methylglyoxal: formation, modification of proteins and nucleic acids, and enzymatic detoxification--a role in pathogenesis and antiproliferative chemotherapy. Gen Pharmacol. 1996;27(4):565-73. 18. McLellan AC, Thornalley PJ, Benn J, Sonksen PH. Glyoxalase system in clinical diabetes mellitus and correlation with diabetic complications. Clin Sci. 1994;87(1):21-9.

19. Yuan B, Cao H, Jiang Y, Hong H, Wang Y. Efficient and accurate bypass of N2-(1- carboxyethyl)-2'-deoxyguanosine by DinB DNA polymerase in vitro and in vivo. Proc Natl Acad Sci U S A. 2008;105(25):8679-84. 20. Murata-Kamiya N, Kamiya H. Methylglyoxal, an endogenous aldehyde, crosslinks DNA polymerase and the substrate DNA. Nucl Acids Res. 2001;29(16):3433-8. 21. Bomgarden RD, Lupardus PJ, Soni DV, Yee MC, Ford JM, Cimprich KA. Opposing effects of the UV lesion repair protein XPA and UV bypass polymerase η on ATR checkpoint signaling. EMBO J. 2006;25(11):2605-14. 22. Taylor P, Nielsen PA, Trelle MB, Hørning OB, Andersen MB, Vorm O, et al. Automated 2D Peptide Separation on a 1D Nano-LC-MS System. J Proteome Res. 2009;8(3):1610-6. 23. Olsen JV, de Godoy LMF, Li G, Macek B, Mortensen P, Pesch R, et al. Parts per Million Mass Accuracy on an Orbitrap Mass Spectrometer via Lock Mass Injection into a C-trap. Molecular & Cellular Proteomics. 2005;4(12):2010-21. 24. Marx V. Targeted proteomics. Nat Meth. 2013;10(1):19-22. 25. MacLean B, Tomazela DM, Shulman N, Chambers M, Finney GL, Frewen B, et al. Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics. 2010;26(7):966-8. 26. Huang DW, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protocols. 2008;4(1):44-57. 27. Huang DW, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 2009;37(1):1-13. 28. Phillips SA, Thornalley PJ. The formation of methylglyoxal from triose phosphates. Eur J Biochem. 1993;212(1):101-5. 29. Che W, Asahi M, Takahashi M, Kaneto H, Okado A, Higashiyama S, et al. Selective induction of heparin-binding epidermal growth factor-like growth factor by methylglyoxal and 3- deoxyglucosone in rat aortic smooth muscle cells. The involvement of reactive oxygen species formation and a possible implication for atherogenesis in diabetes. J Biol Chem. 1997;272(29):18453-9. 30. Riboulet-Chavey A, Pierron A, Durand I, Murdaca J, Giudicelli J, Van Obberghen E. Methylglyoxal Impairs the Insulin Signaling Pathways Independently of the Formation of Intracellular Reactive Oxygen Species. Diabetes. 2006;55(5):1289-99. 31. Miyata S, Miyazaki H, Liu B-F, Fukunaga M, Hamada Y, Ueyama S, et al. Activation of MAP kinase superfamily signaling pathways by methylglyoxal. Int Congress Series. 2002;1245:87- 9. 32. Liu B-F, Miyata S, Hirota Y, Higo S, Miyazaki H, Fukunaga M, et al. Methylglyoxal induces apoptosis through activation of p38 mitogen-activated protein kinase in rat mesangial cells. Kidney Int. 2003;63(3):947-57. 33. McCann JA, Xu YQ, Frechette R, Guazzarotti L, Polychronakos C. The Insulin-Like Growth Factor-II Receptor Gene Is Associated with Type 1 Diabetes: Evidence of a Maternal Effect. J Clin Endocrinol Metab. 2004;89(11):5700-6. 34. Buteau J, Foisy S, Joly E, Prentki M. Glucagon-Like Peptide 1 Induces Pancreatic β-Cell Proliferation Via Transactivation of the Epidermal Growth Factor Receptor. Diabetes. 2003;52(1):124-32. 35. Wu A-L, Kolumam G, Stawicki S, Chen Y, Li J, Zavala-Solorio J, et al. Amelioration of Type 2 Diabetes by Antibody-Mediated Activation of Fibroblast Growth Factor Receptor 1. Sci Transl Med. 2011;3(113):113ra26-ra26.

36. Thornalley PJ, Yurek-George A, Argirov OK. Kinetics and mechanism of the reaction of aminoguanidine with the alpha-oxoaldehydes glyoxal, methylglyoxal, and 3-deoxyglucosone under physiological conditions. Biochem Pharmacol. 2000;60(1):55-65. 37. Ruggiero-Lopez D, Lecomte M, Moinet G, Patereau G, Lagarde M, Wiernsperger N. Reaction of metformin with dicarbonyl compounds. Possible implication in the inhibition of advanced glycation end product formation. Biochem Pharmacol. 1999;58(11):1765-73. 38. Curtin JF, Donovan M, Cotter TG. Regulation and measurement of oxidative stress in apoptosis. J Immunol Meth. 2002;265(1–2):49-72. 39. Kharitonenkov A, Shiyanova TL, Koester A, Ford AM, Micanovic R, Galbreath EJ, et al. FGF-21 as a novel metabolic regulator. J Clin Invest. 2005;115(6):1627-35. 40. Furdui CM, Lew ED, Schlessinger J, Anderson KS. Autophosphorylation of FGFR1 Kinase Is Mediated by a Sequential and Precisely Ordered Reaction. Mol Cell. 2006;21(5):711-7. 41. Kouhara H, Hadari YR, Spivak-Kroizman T, Schilling J, Bar-Sagi D, Lax I, et al. A Lipid- Anchored Grb2-Binding Protein That Links FGF-Receptor Activation to the Ras/MAPK Signaling Pathway. Cell. 1997;89(5):693-702. 42. Brownlee M. Biochemistry and molecular cell biology of diabetic complications. Nature. 2001;414(6865):813-20. 43. Brownlee M. The pathobiology of diabetic complications: a unifying mechanism. Diabetes. 2005;54(6):1615-25.

Figure 1 Figure 2.1. A targeted proteomic approach for interrogating the human kinome. (a) The a working principle of ICAP probe. (b) TheH chemical2N structure of the ICAP probe. (c) Linear O N N HN NH O X X O O O X N fit between iRT valuesO P ofO PdesthiobiotinO P O O -N and desthiobiotin-C3-labeled peptides. (d) Venn N H X X OH OH OH X O diagram showing the kinome coverageHO obtainedOH from DDA and MRM analysis. Desthiobiotin-C3-ATP Light: X = H Heavy: X = D a c b O O Desthiobiotin linker C O P O ADP OH

H2N-Lys ATP-binding protein

H2N-Lys O O ATP-binding Desthiobiotin linker C O P O ADP protein OH

Amide Bond Formation

O Desthiobiotin linker C NH-Lys O ATP-binding Figure 1 HO P O ADP protein OH b d a H2N O N N HN NH O X X O O O X N O P O P O P O O N N H X X OH OH OH X O HO OH Desthiobiotin-C3-ATP Light: X = H Heavy: X = D

b O O Desthiobiotin linker C O P O ADP OH

H2N-Lys ATP-binding protein

H2N-Lys O O ATP-binding Desthiobiotin linker C O P O ADP protein OH

Amide Bond Formation

O Desthiobiotin linker C NH-Lys O ATP-binding HO P O ADP protein OH

Figure 2.3. MG-induced alterations of human kinome in HEK293T cells.

AKT1 CHAK2 RON NEK5 CDK13 PAK1 AKT2 EEF2K ROR1 NEK7 CDK14 PAK2 AKT3 MTOR ROS1 NEK8 CDK17 SLK BCL1 RIOK1 SSK1 CDK19 SPAK NEK9 BTF2 SMG1 STK10/LOK SYK NME3 Atypical CDK2 CCNB1 AMPK1 TK1 CDK20 STK11 TK NME7 BRSK2 CCND1 CDK7 STK23 TYK2 NME8 CAMK1 DMPK CDK9 STK24 YES1 NPR1 CAMK1D STE DMPK2 CDKL1 STK3 ZAP70 NTKL EXOSC10 CAMK2A ALK4 CDKL3 STK38 PAN3 GRK2 CAMK2D ALK5 CLK2 STK4 PFKL CAMK2G GRK3 CLK3 TAOK1 ALK7 CAMK4 PFKP GRK5 CLK4 TAOK2 ARAF CAMKK1 PGK1 GRK6 CRKL TNIK BRAF CAMKK2 PGK2 LATS1 DCK YSK1 CARK CASK PI4K2A LATS2 DCLK3 ABL1 ILK CHEK1 PI4KB MAST3 ERCC2 BMX IRAK1 CHEK2 PIK3C2A MASTL ERK5 CAD IRAK4 DAPK1 TKL PIK3C3 MSK1 GSK3A CKB LIMK2 MSK2 MAPKAPK1A CMGC GSK3B CSK LRRK1 PIK3CA PKN1 MAPKAPK1B MAPK1 EGFR LRRK2 PIK3CD PKN2 MAPKAPK1C MAPK14 EPHA1 MLK4 PIP4K2A PRKAA1 MAPKAPK2 MAPK15 EPHA2 RIPK1 PIP4K2B PRKAA2 MAPKAPK3 MAPK3 EPHA4 ZAK PIP4K2C PRKACA MAPKAPK5 MAPK4 EPHA5 AAK1 PIP5K3 PRKACG MARK4 MAPK6 EPHA7 ADK PKMYT1 PRKAG1 MELK MAPK7 EPHB2 AGK CAMK PLK1 MYLK AURKA PRKAG2 MAPK8 EPHB3 PLK2 PRKCA NIM1K MNAT1 EPHB4 AURKB AGC PLK4 PRKCB NUAK1 ERBB2 AURKC PRPF4B PNKP PRKCD NUAK2 PRPK ERBB4 CHUK POLR2A PRKCG OBSCN PRPS1 FER/TYK3 CMPK1 Other POLR2B PRKCH PASK PRPS2 FGFR1 CPNE3 POLR2E PRKCI PHKG2 PRPSAP2 FGFR2 EIF2AK1 PRKD2 PKDCC SRPK1 FGFR3 EIF2AK2 PP4C

PRKD3 SIK1 CIT TK FGFR4 EIF2AK3 PXK PRKDC SIK3 MAP2K1 FGR EIF2AK4 RBKS PRKX SNRK MAP2K3 FLT1 ENG RO60 ROCK1 STK33 MAP2K4 FLT3 ERN1 SBK1 ROCK2 TSSK2 MAP2K5 HCK FAM20C SRC RPS6KC1 TSSK4 MAP2K6 IGF2R FASTKD2 STK35 RSK4 TSSK6 MAP2K7 INSRR Other GAK TBK1 SGK1 TTN MAP3K1 JAK1 GALK1 TBRG4 SGK2 CSNK1A1 MAP3K11 JAK2 GOGA5 TJP2 CSNK1A1L SGK3 MAP3K15 LCK HK1 TLK1 CSNK1G1 IKBKAP STK14A MAP3K2 LYK5 TLK2 CSNK2A1 STE MAP3K3 LYN IKBKB STK14B TRIB1 STK38L CSNK2A2 MAP3K4 MATK IP6K1 TRIM28

YANK3 CK1 CSNK2B MAP3K5 MET IPMK TWF1 AK1 VRK1 MAP3K6 MUSK IPPK TYMK AK2 VRK2 MAP3K7 NTRK2 ITGAE UCK2 ATM ACF1 MAP4K1 NTRK3 LGMN ATR CABLES1 MAP4K2 PDGFRA NAGK UHMK1 BCKDK CDC42BPA MAP4K3 PTK2 NEK1 ULK1 BRD2 CDK10 MASK PTK7 NEK2 ULK2 Atypical ULK3 CHAK1 CMGC CDK12 OXSR1 RET NEK4 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 RMG/Control RMG/Control RMG/Control RMG/Control RMG/Control RMG/Control

Figure 2.4. Representative MRM results for EGFR. (a) Correlation between iRT in library and measured RT on a TSQ Vantage mass spectrometer. (b) MRM traces for peptide

ITDFGLAK#LLGAEEK from EGFR derived from forward and reverse labeling experiments. (c) MS/MS of a representative peptide ITDFGLAK#LLGAEEK from EGFR

(the y5, y7 and y8 ions labeled in red represent the 3 transitions used for MRM analysis).

MG-treated and control HEK293T cells, respectively. The opposite was conducted for reverse-labeling experiments.

Figure 3

a bc Light probe labeling Heavy probe labeling 160 Forward Reverse 140 y = 0.7072x + 43.446 1 1 R² = 0.9985 120 0.8 0.8 100 0.6 0.6 80 0.4 0.4

60 Measured RT Measured

40 0.2 0.2 Relative Abundance Relative 20 0 0 -20 0 20 40 60 80 100 120 140 160 110.5 112.5 110 112 iRT Retention Time (min)

bc d y5

y7 =0.89

y8 =0.86 dotp 1.01 dotp y5 0.8

b10

y8 0.6 y7 y6 b Intensity 9 2+ 0.4 y12 b8 b11 y4 b13 y10 y3

y2 0.2 Normalized Peak Area PeakNormalized 0 Library R1 R2 m/z Library F R

Figure 4

a 2.5

1.5

Treated/Control 1

- MG 0.5

2.5

1.5

Treated/Control -

0.5 MG

Figure 2.6. MG treatment led to diminished level of EGFR protein. (a) Changes in mRNA levels of EGFR, IGF2R and FGFR1 genes after treatment with different doses of MG (n =

3). (b) Quantification results showing the dose-dependent changes in expression levels of

EGFR protein after MG treatment (n = 3) and representative Western blot result. (c)

Quantification results showing the time-dependent changes in expression levels of EGFR protein after MG treatment (n = 3) and representative Western blot result. (d)

Quantification results showing that the decrease in expression level of EGFR induced by

MG treatment could be rescued by pre-treatment with NAC, but not AG or MET. Shown also are the Western blot result. All p values were calculated by using unpaired, two-tailed t-test, and the p-values in (d) referred to the comparisons with the results obtained from cells treated with MG alone. The concentration of AG, NAC, and MET used to pretreat cells were all 1 mM.

Figure 5 a b MG (M) 0 50 200 5001000

EGFR Control 200 M MG 500 M MG -Actin 1.2

1.2 1.01 p=0.399 p=0.011 p = = p0.002 1.01 0.8

p=0.001 p = = p0.00003 0.8 p=0.000003 0.6 0.6 0.4

0.4 Relative mRNA Level mRNA Relative 0.2 EGFR Level Relative 0.2

0 0 EGFR IGF2R FGFR1 01 502 2003 5004 10005 MG (M)

c d

NAC MET Pre-treatment − − AG Time (hr) 0 4 10 24 MG (500 M) − + + + +

EGFR EGFR

-Actin -Actin

1.2 1.2 p=0.013 1.01 1.01 p=0.0011 p=0.16 0.8 p=0.0004 0.8 p=0.46 0.6 0.6 0.4 0.4 0.2

Relative EGFR Level Relative 0.2 Relative EGFR Level Relative 0 0 0 4 10 24 Pre-treatment −1 −2 AG3 NAC4 MET5 1Treatment2 Time3 (hr)4 MG (500 M) − + + + +

Figure 2.7. MG treatment led to diminished level of IGF2R protein. (a) Quantification results showing the dose-dependent alterations in expression levels of IGF2R protein after

MG treatment (n = 3) and representative Western blot result. (b) The decrease in expression level of IGF2R protein induced by MG treatment could be rescued by pre-incubation of Figure 6 cells with NAC, but not AG or MET (n = 3). Shown also are the representative Western blot result. All p values were calculated by using unpaired, two-tailed t-test, and the p- values in (b) referred to the comparisons with the results obtained from cells treated with

MG alone. The concentration of AG, NAC, and MET used to pretreat cells were all 1 mM.

a b

Pre-treatment − − AG NAC MET MG (M) 0 50 200 500 1000 MG (500 M) − + + + + IGF2R IGF2R

- Actin -Actin

p=0.0007 1.2 1.2 p=0.169 p=0.008 p=0.004 1.01 1.01 p=0.001 0.8 0.8 p=0.002 0.6 0.6 p=0.00007 0.4 0.4

Relative IGF2RLevel Relative 0.2 0.2 Relative IGF2RLevel Relative

0 0 0 1 502 2003 4500 51000 Pre-treatment − − AG NAC MET MG ( M) 1 2 3 4 5  MG (500 M) − + + + +

Table 2.1. Sequence for qRT-PCR primers.

Gene Forward Primer Reverse Primer Name EGFR 5’-AAAGTTTGCCAAGGCACGAGT-3’ 5’- AGGACATAACCAGCCACCTCCT-3’ IGF2R 5’-CCATGCAGAACCAGAGCAGAAT-3’ 5’- TCCATTTTTATCCACTGCACACACT-3’ FGFR1 5’-AAGTCGGACGCAACAGAGAAAGAC-3’ 5’- CTTGGAGGCATACTCCACGA-3’ GAPDH 5’- GTGGAGTCCACTGGCGTCTTC-3’ 5’-CTGATGATCTTGAGGCTGTTGTCA-3’

Table 2.2. iRT comparison between desthiobiotin-labeled and desthiobiotin-C3-labeled peptides.

Peptide sequence iRT of desthiobiotin-labeled iRT of desthiobiotin-C3- peptides labeled peptides AVIMGAPGSGK#GTVSSR 30.27 25.66 EAFSLFDK#DGDGTITTK 79.45 75.09 GGPNIITLADIVK#DPVSR 110.31 106.18 GTFGK#VILVK 63.41 59.80 GTFGK#VILVR 66.77 63.14 NIIHGSDSVK#SAEK 14.27 7.92 TFIAIK#PDGVQR 57.95 52.31 VMLGETNPADSK#PGTIR 39.61 37.72 DIK#GANILR 50.62 46.37 AVDDGVNTFK#VLTR 70.70 65.79

Chapter 3

Comprehensive Analysis of the Human Kinome Perturbed by

Clinical Kinase Inhibitors

Introduction Kinases play crucial roles in cell signaling (1), and small-molecule kinase inhibitors have been widely employed as valuable tools for investigating kinase-mediated cell signaling for decades (2). In addition, aberrant activation of kinase signaling is widely associated with, and sometimes leads to the development of leukemia and many types of solid tumors (2-4). As a result, kinase inhibitors have recently become a very important class of drugs for anti-cancer therapy, where more than 30 small-molecule kinase inhibitors have been approved by the Food and Drug Administration (FDA) for treating different types of malignancies (2-4).

Appropriate use of small-molecule kinase inhibitors in cell signaling research and in cancer chemotherapy requires the knowledge about whether other kinases are also targeted by these inhibitors. In this respect, most kinase inhibitors are designed to disrupt, directly or indirectly, the ATP-binding capabilities of the target kinases, where the ATP-binding domains are highly conserved among kinases (2, 4). Hence, despite with substantial efforts in optimizing the structures of kinase inhibitors for selective binding toward the intended kinase, the inhibitors may also perturb the ATP binding affinities and suppress the activities of other kinases (2, 4, 5). In addition, cancer cells may respond to inhibitor treatment by

64 reprograming their kinomes through altering the protein expression levels of kinases (6).

The knowledge about kinome reprogramming elicited by a kinase inhibitor and about its off-targets is important for understanding more completely the mechanisms through which these inhibitors confer therapeutic efficacy, side effects and resistance. Such knowledge is also crucial for the accurate interpretation of data when these inhibitors are used in cell signaling research. Furthermore, on the grounds that the drug safety and pharmacological properties of these FDA-approved kinase inhibitors are well-established (7), revealing previously unrecognized kinase targets for these inhibitors may allow for repurposing of these inhibitors for treating other human diseases.

The human kinome is encoded with 518 genes (8), and many kinases are expressed at very low levels. Thus, the investigation about kinome reprogramming in response to kinase inhibitor treatment and the detailed characterizations of the target kinases of an inhibitor in live cells entail high-throughput and highly sensitive analytical methods for assessing independently the protein expression and ATP-binding affinity of protein kinases.

In the present study, we address the above critical needs by developing a novel targeted quantitative proteomic approach for monitoring the changes in protein expression, which, together with the previously reported targeted quantitative proteomic method for gauging the alterations in ATP-binding affinities (9-12), of ~300 non-redundant kinases in cells upon treatment with three FDA-approved kinase inhibitors. Our results confirmed previously reported kinase targets for these inhibitors, uncovered many novel putative targets, and revealed profound changes in expression levels of kinase proteins in cancer cells upon treatment with these kinase inhibitors.

Materials and Methods

Cell culture

GM-00637 (obtained from Prof. Gerd P. Pfeifer), GM15876A (provided by Prof.

Karlene Cimprich) (13), HCT-116, and HEK293T (ATCC), MCF-7, MDA-MB-231

(obtained from Prof. Jian-Jian Li), and M14 (National Cancer Institute) cells were cultured in Dulbecco's modified eagle medium (DMEM). DU-145, WM-115 and WM-266-4 cells

(ATCC) were cultured in Eagle's minimum essential medium (EMEM). HL-60, Jurkat-T and CEM cells (ATCC) were cultured in RPMI 1640 medium. K-562 cells (ATCC) were cultured in Iscove's modified Dulbecco's medium (IMDM). All culture media were supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA) and penicillin (100

IU/mL). The cells were maintained at 37°C in a humidified atmosphere containing 5% CO2.

Approximately 2×107 cells were harvested, washed with ice-cold PBS for three times, and lysed by incubating on ice for 30 min with CelLytic M (Sigma) cell lysis reagent containing

1% protease inhibitor cocktail. The cell lysates were centrifuged at 9,000g at 4°C for 30 min, and the resulting supernatants collected. For SILAC experiments (14), M14 and K-

13 15 562 cells were cultured in medium containing unlabeled lysine and arginine, or [ C6, N2]-

13 lysine and [ C6]-arginine for at least 2 weeks to promote complete incorporation of the stable isotope-labeled amino acids.

Preparation of the desthiobiotinylated nucleotide affinity probe, desthiobiotin labeling and affinity purification of ATP-binding proteins

The desthiobiotinylated nucleotide affinity probes were prepared according to previously published procedures(15, 16). Approximately 2×107 cells were harvested, washed with ice-cold PBS for three times, and lysed in a 1-mL lysis buffer, which contained 0.7% CHAPS, 50 mM HEPES (pH 7.4), 0.5 mM EDTA, 100 mM NaCl, and 10

µL (1:100) protease inhibitor cocktail on ice for 30 min. The cell lysates were centrifuged at 16,000g at 4°C for 30 min and the supernatants collected. Endogenous nucleotides in the resultant protein extract were removed by gel filtration using a NAP-25 column

(Amersham Biosciences). Cell lysates were subsequently eluted into a buffer containing

50 mM HEPES (pH 7.4), 75 mM NaCl, and 5% glycerol. The amounts of proteins in the lysates were quantified using Quick Start Bradford Protein Assay (Bio-Rad). Prior to the labeling reaction, MgCl2, MnCl2, and CaCl2 were added to the concentrated cell lysate until their final concentrations reached 50, 5 and 5 mM, respectively.

For pulling down kinases and other ATP-binding proteins, approximately 1 mg cell lysate in a 1-mL solution was incubated, at room temperature with gentle shaking for 2.5 h, with the light desthiobiotin-C3-ATP affinity probe at a final concentration of 100 µM.

To the resulting mixture was subsequently added 300 µL avidin-agarose resin (Sigma-

Aldrich), and the mixture was then incubated, with gentle shaking, at room temperature for

1 h. The agarose resin was washed sequentially with 3 mL PBS buffer and 3 mL H2O to remove unbound proteins, and the desthiobiotin-conjugated proteins were subsequently eluted with 1% TFA in CH3CN/H2O (7:3, v/v) at 75°C.

Tryptic digestion of enriched ATP-binding proteins and whole-cell protein lysates, and LC-MS/MS analyses in the data-dependent acquisition (DDA) mode

The above enriched ATP-binding proteins from the lysates of eight human cell lines

(i.e. CEM, Du-145, GM00637, GM15876A, HCT-116, HEK293T, HL-60, and Jurkat-T) were washed with 8 M urea for protein denaturation, and dithiothreitol and iodoacetamide for cysteine reduction and alkylation, respectively. The proteins were subsequently digested with modified MS-grade trypsin (Pierce) at an enzyme/substrate ratio of 1:100 in

50 mM NH4HCO3 (pH 8.5) at 37°C overnight. The peptide mixture was subsequently dried in a Speed-vac, desalted with OMIX C18 pipette tips (Agilent Technologies), and analyzed by LC-MS and MS/MS on a Q Exactive Plus quadruple-Orbitrap mass spectrometer

(Thermo Fisher Scientiﬁc) in the DDA mode.

The mass spectrometer was coupled with an EASY-nLC 1200 system, and the samples were automatically loaded onto a 4-cm trapping column (150 µm i.d.) packed with

ReproSil-Pur 120 C18-AQ resin (5 µm in particle size and 120 Å in pore size, Dr. Maisch

GmbH HPLC) at 3 µL/min. The trapping column was coupled to a 20-cm fused silica analytical column (PicoTip Emitter, New Objective, 75 µm i.d.) packed with ReproSil-Pur

120 C18-AQ resin (3 µm in particle size and 120 Å in pore size, Dr. Maisch GmbH HPLC).

The peptides were then separated using a 140-min linear gradient of 9-38% acetonitrile in

0.1% formic acid and at a flow rate of 300 nL/min. The spray voltage was 1.8 kV. Full- scan mass spectra were acquired in the range of m/z 350-1500 using the Orbitrap analyzer at a resolution of 70,000. Up to 25 most abundant ions found in MS with a charge state of

2 or higher were sequentially isolated and collisionally activated in the HCD cell at a collision energy of 27 to yield MS/MS.

The whole cell lysates prepared from three human cell lines (i.e. WM-115, WM-266-4 and HEK293T) were again denatured by urea, followed by cysteine reduction/alkylation, and tryptic digestion, as described above. The resulting peptide mixtures were subsequently dried in a Speed-vac, desalted using OMIX C18 pipette tips (Agilent Technologies), and subjected to LC-MS and MS/MS analyses on an LTQ Orbitrap Velos mass spectrometer

(Thermo Fisher Scientiﬁc) in the DDA mode. Up to 20 most abundant ions found in MS were sequentially isolated and collisionally activated in the linear ion trap at a normalized collision energy of 35 to yield MS/MS.

Database search

Maxquant, Version 1.5.2.8, was used to analyze the LC-MS and MS/MS data for protein identification (17). The maximum number of miss-cleavages for trypsin was two per peptide. Cysteine carbamidomethylation was set as a ﬁxed modification. Methionine oxidation and serine, threonine and tyrosine phosphorylation were set as variable modiﬁcations. The tolerances in mass accuracy was 20 ppm for MS, and 20 ppm and 0.5

Da for MS/MS acquired on the Q Exactive Plus files and LTQ Orbitrap Velos, respectively.

The maximum false discovery rates (FDRs) were set to 0.01 at both peptide and protein levels, and the minimum required peptide length was six amino acids. The total identified proteins were then filtered by the DAVID (version 6.7) bioinformatic tool with the Gene

Ontology (GO) term of kinase (18).

Development of a PRM assay for human kinome profiling

Maxquant search of the LC-MS/MS data obtained from the aforementioned shotgun proteomic experiments, which included more than 200 LC-MS/MS runs, led to the identification of 11,879 protein groups. The tandem mass spectra and retention time information for all kinase peptides except those labeled with desthiobiotin-C3 were imported into Skyline for constructing the kinome PRM library. Because some kinases share highly similar sequences, we inspected manually all kinase peptides and incorporated only those peptides that can be uniquely assigned to specific kinases into the library, with a maximum of four unique peptides being included for any given kinase. In doing so, the current PRM kinome library contained 1050 unique peptides representing 478 non- redundant human kinases, which encompassed 395 protein kinases, 21 lipid kinases, and

62 other kinases. Thus, our PRM kinome library contains approximately 80% of the human kinome that has a total of 518 protein kinases (19).

To achieve high-throughput detection of kinase peptides, we adopted scheduled PRM analysis, where the mass spectrometer was programmed to acquire the MS/MS for the precursor ions of a limited number of peptides in each pre-defined retention time window.

Therefore, accurate retention time (RT) information for each kinase peptide was required for our PRM-based kinome assay. To this end, we calculated the normalized RT (iRT) value for each peptide on our target list following a previously published method with the use of tryptic digestion mixture of bovine serum albumin (BSA) as standards (20). The linear RT vs. iRT relationship was redefined between every 5-7 LC-MRM and LC-PRM runs by reinjecting the tryptic digestion mixture of BSA. Since the iRT value represents an

70 intrinsic property (i.e. hydrophobicity) of a peptide, it was also employed as a criterion to validate the results obtained from the PRM assay, where a marked deviation of the observed RT from the predicted RT is considered a false-positive detection.

Sample preparation, MRM and PRM analyses

For assessing the alterations in ATP binding affinities of kinases in cultured cells induced by kinase inhibitors, we treated K-562 cells with 1 μM imatinib, and M14 cells with 100 nM dabrafenib or vemurafenib for 24 h. The cells were subsequently harvested by centrifugation and lysed, following the aforementioned procedures. The removal of endogenous nucleotides from protein lysate, labeling with desthiobiotin-conjugated ATP- affinity probe, tryptic digestion, and affinity purification of desthiobiotin-labeled peptides were performed following previously published procedures (20). The resultant peptide mixture was subjected to LC-MS/MS analysis on a TSQ Vantage triple-quadrupole mass spectrometer, where the instrument was operated in the scheduled MRM mode and coupled with an Easy-nLC II system. The peptides were separated with a 130-min linear gradient of 2-35% acetonitrile in 0.1% formic acid and at a flow rate of 230 nL/min. The spray voltage was 1.9 kV. Q1 and Q3 resolutions were 0.7 Da and the cycle time was 5 s.

For assessing how vemurafenib modulates the ATP binding affinities of kinases in cell lysate, we removed endogenous nucleotides from the whole cell protein lysate of M14 cells, and treated the resultant lysate (1 mg/mL) with 100 nM vemurafenib (or DMSO control) at room temperature for 2 h prior to labeling with desthiobiotin-conjugated ATP-affinity

71 probe. The samples were then processed and subjected to LC-MRM analysis in the same way as described above.

To assess the differential expression of protein kinases in inhibitor-treated cells, both forward and reverse SILAC labeling experiments were conducted. In this context, the lysates of light-labeled, inhibitor-treated cells and heavy-labeled, mock-treated cells (with

DMSO) were combined at 1:1 ratio (by mass) in the forward SILAC experiments, whereas the reverse SILAC experiments were performed in the opposite way. The whole cell lysates were subsequently digested following the above-described procedures and analyzed on a

Q Exactive Plus quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientiﬁc) in the scheduled PRM-mode.

All raw data acquired from LC-MRM and LC-PRM analyses were processed using

Skyline (version 3.5(21)) for the generation of extracted-ion chromatograms and peak integration. The targeted peptides were first manually checked to ensure the overlaid chromatographic profiles of multiple fragment ions derived from light and heavy forms of the same peptide. The data were then processed to ensure that the distribution of the relative intensities of multiple transitions associated with the same precursor ion correlates with the theoretical distribution in the kinome MS/MS spectral library. The sum of peak area from all transitions of light or heavy forms of peptides was used for quantification. In the samples prepared from the lysates of dabrafenib-treated M14 cells, we observed, from LC-MRM analysis, several kinase peptides displaying large light/heavy ratios (>1.5) in both forward and reverse ATP affinity probe labeling experiments. These were likely due to

72 interferences from other unmodified peptides, and the results from these kinase peptides were excluded from further analysis.

Western blot

K-562, MDA-MB-231, and M14 cells were cultured in a 6-well plate and the cells were lysed at 40-50% confluency following the above-described procedures. For Western blot analysis of p-ERK5, the cells were treated with 100 ng/mL FGF21 (Sigma-Aldrich) for 20 min and then harvested for cell lysis. The concentrations of proteins in the resulting lysates were determined by using Bradford Assay (Bio-Rad), and 10 μg protein lysate was denatured by boiling in Laemmli loading buffer and resolved by SDS-PAGE. The proteins were subsequently transferred onto a nitrocellulose membrane at 4°C overnight. The resulting membrane was blocked with PBS-T (PBS with 0.1% Tween 20) containing 5% milk (Bio-Rad) at 4°C for 6 h. The membrane was then incubated with primary antibody at 4°C overnight and subsequently with secondary antibody at room temperature for 1 h.

After thorough washing with PBS-T, the HRP signal was detected with Pierce ECL

Western Blotting Substrate (Thermo).

Antibodies recognizing human AK1 (Santa Cruz Biotechnology, sc-165981, 1:1000 dilution), ARAF (Santa Cruz Biotechnology, sc-166771, 1:4000 dilution), BRAF (Santa

Cruz Biotechnology, sc-5284, 1:4000 dilution), CCND3 (Santa Cruz Biotechnology, sc-

453, 1:2000 dilution), CHK1 (Cell Signaling Technology, 2360S, 1:2000 dilution), p-

CHK1 S296 (Abcam, ab79758, 1:1000 dilution), EGFR (Santa Cruz Biotechnology, sc-03, with a 1:10000 dilution), ERK5 (Santa Cruz Biotechnology, sc-393405, with a 1:1000

73 dilution), p-ERK5 (T218/Y220, Santa Cruz Biotechnology, sc-135761, with a 1:1000 dilution), IGF2R (Santa Cruz Biotechnology, sc-14408, 1:1000 dilution), JAK3 (Thermo

Fishier Scientific, AHO1572, 1:1000 dilution), MAP3K5 (a.k.a. ASK1, Santa Cruz

Biotechnology, sc-5294, 1:2000 dilution), MAP2K5 (a.k.a. MEK5, Santa Cruz

Biotechnology, sc-365198, 1:2000 dilution), MAP3K3 (a.k.a. MEKK3, Santa Cruz

Biotechnology, sc-136260, 1:1000 dilution), MEK1 (Santa Cruz Biotechnology, sc-6250,

1:2000 dilution), p-MEK1 (Santa Cruz Biotechnology, sc-136542, 1:1000 dilution), mTOR (Cell Signaling Technology, 2972S, 1:1000 dilution), N-cadherin (Thermo Fishier

Scientific, 33-3900, 1:1000 dilution), SCYL3 (Santa Cruz Biotechnology, sc-398328,

1:2000 dilution), Slug (Thermo Fishier Scientific, PA5-11922, 1:1000 dilution), and

STK26 (Abcam, ab52491, 1:20000 dilution) were employed as primary antibodies.

Horseradish peroxidase-conjugated anti-rabbit IgG, IRDye® 680LT Goat anti-Mouse IgG and donkey anti-goat IgG-HRP (Santa Cruz Biotechnology, sc-2020, 1:10000 dilution) were used as secondary antibodies. Membranes were also probed with -actin antibody

(Cell Signaling #4967, 1:10000 dilution) to confirm equal protein loading.

Clonogenic survival assay

MDA-MB-231 cells were seeded in six-well plates at densities of 100-300 cells per well. The cells were exposed to various doses of neocarzinostatin (NCS, 0-100 ng/mL) and a fixed dose of imatinib (1 μM) in DMEM medium and cultured for 10 days. The colonies were then ﬁxed with 6% glutaraldehyde and stained with 0.5% crystal violet. Colonies with more than 50 cells were counted under a microscope (22).

Results

1. Targeted Proteomic Methods for Assessing the Proteome-Wide Alterations in

Protein Expression and ATP-binding Affinity of Kinases in Response to Inhibitor

Treatment

We embarked on the development of targeted quantitative proteomic methods to examine, at the entire proteome scale, the changes in protein expression levels and activities of kinases in cultured human cells following treatment with kinase inhibitors. We showed recently that labeling with isotope-coded ATP affinity probes, in conjunction with liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis in the multiple- reaction monitoring (MRM) mode, afforded a high-throughput and highly sensitive assessment about the ATP binding affinities of over 300 unique kinases in cultured human cells (Figure 3.1a) (10).

To examine the alterations in kinase protein expression elicited by small-molecule kinase inhibitors, we developed, for the first time, a parallel-reaction monitoring (PRM) method (23) for the proteome-wide interrogation of the protein expression levels of kinases

(Figure 3.1b). To this end, we first established a Skyline (24) PRM spectral library based on LC-MS/MS data acquired from shotgun proteomic analyses of whole-cell lysates or enriched ATP-binding proteins from the lysates of a number of cell lines (see Materials and Methods for details). The PRM library encompasses a total of 1050 unique peptides derived from 478 non-redundant human kinases. By employing this PRM method, in combination with stable isotope labeling by amino acids in cell culture (SILAC), we were

75 able to obtain quantitative information about the expression levels of approximately 300 unique kinases in cultured human cells. A recent RNA-Seq study showed that about 400 kinases are expressed in a single cell line (6); thus, our method allows for the quantitative assessment of approximately 75% of the expressed kinome.

2. Profound Alterations in Protein Expression and ATP Binding Affinities of Kinases

Elicited by Kinase Inhibitors

We next combined the PRM- and MRM-based methods to assess the perturbations in protein expression levels and activities of kinases upon treatment with three FDA-approved small-molecule kinase inhibitors. In this respect, dabrafenib and vemurafenib were approved for treating melanoma in patients carrying a mutant BRAF (V600E or V600K)

(2), and we used M14 melanoma cells, which contain the V600E mutant form of BRAF, for kinome profiling experiments. Imatinib, which inhibits the activity of c-ABL kinase

(25), is used for treating chronic myelogenous leukemia (CML) in patients with the BCR-

ABL fusion gene (25). We employed the K-562 human CML cells with the BCR-ABL oncogene (26) for the experiment.

Our PRM- and MRM-based kinome spectral libraries commonly contain 350 non- redundant kinases (Figure 3.1c and Figure 3.2a). A combination of the PRM and MRM data, therefore, permits independent assessments about how the expression levels and ATP- binding activities of kinases are modulated by each kinase inhibitor. Along this line, the two methods individually facilitated the quantification of approximately 300 unique kinases upon each inhibitor treatment, and 150 could be quantified by both methods (Figure

3.2b). A comparison of kinase ratios (with/without kinase inhibitor treatment) obtained from PRM and MRM analyses revealed the lack of apparent correlation between the inhibitor-induced alterations in kinase protein expression and the corresponding changes in ATP binding affinities (Figure 3.2c). This finding underscores the different extents of

ATP binding affinities, and hence the varying degrees of activation, of kinases upon inhibitor treatment.

Our quantification results showed that exposure to each inhibitor led to pronounced reprogramming of the kinome in cells by inhibiting the ATP-binding affinities of many kinases and concomitantly stimulating the expression of many kinases (Figures 3.3a). For instance, substantially more kinases display diminished ATP binding affinities than those with heightened ATP binding affinities (65 vs. 8), whereas a much larger number of kinases exhibited elevated protein expression than those with reduced expression upon imatinib treatment (48 vs. 9, Figure 3.3b). The augmented expressions of AK1, CCND3 and SCYL3 in K-562 cells upon imatinib treatment were validated by Western blot analysis (Figure

3.4).

While both dabrafenib and vemurafenib were intended to target the V600E and V600K mutants of BRAF, treatment with these two inhibitors led to markedly different reprograming of the human kinome. We found that a 24-hr treatment with 100 nM vemurafenib led to the up- and down-regulations of 58 and 5 kinases, respectively (Figure

3.5a), whereas the corresponding treatment with dabrafenib resulted in the up- and down- regulations of 27 and 42 kinases, respectively (Figure 3.5b). We confirmed, by using

Western blot analyses, the expression levels of 11 quantified kinases (mTOR, IGF2R,

JAK3, CHK1, AK1, STK26, ARAF, BRAF, MAP2K1, MAP3K3 and MAP3K5) in M14 cells with or without dabrafenib or vemurafenib treatment. The results demonstrated that the PRM method afforded robust quantifications of the protein expression levels of these kinases (Figure 3.6).

The PRM-based quantification data showed that dabrafenib distinguishes from the other three kinase inhibitors in that its administration led to the decreased expression of a large number of kinases (Figure 3.5b). We next performed a time-dependent experiment, where we treated M14 cells with 100 nM dabrafenib for shorter durations (i.e. 4 and 12 h) and examined the protein expression levels of kinases at these two time points. Our result showed that many kinases exhibited augmented protein expression at 4 h after treatment and some kinases started to display decreased expression after 12 h (Figure 3.7), suggesting that the decreased expression of most kinases occurred after 12 h of dabrafenib exposure.

By using the efficiency of desthiobiotin-ATP probe labeling as a proxy for ATP- binding affinity. We found that, our results confirmed ABL and BRAF as target kinases for imatinib and dabrafenib, respectively. In addition, GAK and ARAF, which were previously identified as direct targets for imatinib and debrafenib, respectively (27, 28), were quantified with decreased ATP-binding affinity. In keeping with previous findings, we observed that vemurafenib exposure resulted in decreased ATP binding affinities of

ARAF, BRAF and ZAK (Figure 3.8) (29, 30).

Aside from discovering those kinases exhibiting reduced ATP-binding affinities upon inhibitor treatment, we were also able to detect kinases displaying augmented binding

78 affinities toward ATP upon treatments with kinase inhibitors. Among the kinases with increased ATP-binding affinities, CRAF is known to be activated via a paradoxical pathway through the dabrafenib-mediated inhibition of one protomer of BRAF in the

BRAF-CRAF heterodimer (30).

3. CHK1 is a Novel Target Kinase of Imatinib

Among the putative kinase targets of imatinib, CHK1 displays decreased ATP-binding affinity, which we validated by Western blot analysis (Figure 3.9). In this vein, our results showed that imatinib treatment led to a marked attenuation in the kinase activity of CHK1

(as manifested by reduced auto-phosphorylation at Ser296), albeit with no appreciable change in its protein expression level (Figure 3.9). In response to DNA double strand break

(DSB) formation in cells, CHK1 coordinates DNA damage response signaling and cell cycle checkpoint control (31). Hence, we next asked whether the cellular sensitivity toward neocarzinostatin (NCS), a radiomimetic agent that induces DNA DSBs, could be modulated by co-treatment with imatinib. Our result from clonogenic survival assay indeed showed that imatinib rendered MDA-MB-231 cells more sensitive toward NCS (Figure

3.9c).

4. Vemurafenib Suppresses the ATP-binding Affinity of MAP2K5

In addition to ARAF, BRAF and ZAK, our results showed that vemurafenib treatment led to the diminished ATP binding affinities of several other kinases, including

MAP2K5 (Figures 3.10a-c). In this vein, the inhibition of MAP2K5 by vemurafenib is even more pronounced than that of BRAF (Figure 3.10a), and similar inhibition of MAP2K5

79 was previously observed for PLX-4720, another BRAF inhibitor and a structural analog of vemurafenib.(32) In agreement with the proteomic data, results from Western blot analysis showed that treatment of M14 cells with vemurafenib led to a marked diminution in the kinase activity of MAP2K5 (as reflected by reduced p-ERK5), albeit with no appreciable change in protein expression level of MAP2K5 (Figure 3.10). However, the proliferation of the M14 cells was not inhibited upon treatment with up to 1 µM BIX-02188, a small- molecule inhibitor for MAP2K5 (Figure 3.11). Thus, inhibition of MAP2K5 does not constitute a major mechanism for the cytotoxic effects of vemurafenib in BRAF-V600E mutant melanoma cells.

For comparison, we also examined the changes in ATP binding affinities of kinases in the whole-cell lysate of M14 cells upon treatment with vemurafenib (Figure 3.12). We were able to quantify 268 kinases, among which 7 exhibited reduced ATP binding affinities after 100 nM vemurafenib treatment (Figure 3.13). Among these kinases, ARAF, ZAK and

MAP2K5 exhibited markedly lower ATP binding affinities in the lysate treated with vemurafenib (Figures 3.10d-e & 3.13). In addition, the in vitro assay with the use of cell lysate provides information about other kinases exhibiting lower ATP binding affinities upon treatment with 100 nM vemurafenib, including SRC (Figure 3.10f), IP6K1, FER, and

PRPS1, whereas these four kinases did not display reduced ATP binding affinities in the above cell-based experiments. On the other hand, AK6, ABR, BRAF, CHUK, FGFR3, and PEAK1 displayed reduced ATP binding affinities in cellular experiments; however, all of them except CHUK1 were not detected in the in-vitro assay. The failure to detect these five kinases could be attributed to the lack of stability of these kinases, where the in-vitro

80 experiment involved a 2-h incubation of the lysate with vemurafenib prior to ATP affinity probe labeling. The differences in vemurafenib-elicited variations in ATP binding affinities, as revealed from the in cellulo and in vitro assays, could arise from the presence of transient and unstable protein-protein interactions that may modulate the kinase’s capability in binding to ATP.

Discussion

In this study, we made significant advances in quantitative analysis of the human kinome by developing a novel PRM-based method for quantifying the protein expression of kinases and by combining this method with our previously reported MRM-based targeted proteomic approaches for assessing independently the protein expression and ATP-binding affinity of kinases. In particular, our PRM and MRM kinome libraries each encompasses approximately 80% of the human kinome, with 350 kinases being commonly included in the two libraries (Figure 2a). By employing the two targeted proteomic approaches, we explored the alterations in protein expression and activities of kinases in cultured human cells upon treatment with three FDA-approved small-molecule kinase inhibitors.

Previously published methods for profiling the kinase inhibitor-induced alterations of the human kinome in cells rely on the use of affinity resin immobilized with multiple kinase inhibitors, which allowed for the interrogation of approximately 200 kinases (6). The binding of kinases to inhibitor-immobilized beads, similar as the ATP affinity probe method employed here (9, 10, 12), is influenced by both the protein expression level and the activation of kinases (6). To our knowledge, our PRM-based targeted proteomic approach facilitated, for the first time, the assessment of the alterations in protein

81 expression levels of kinases upon kinase inhibitor treatment. Our results revealed pronounced changes in expression levels of a number of kinases upon treatment with each of the three kinase inhibitors.

We were able to quantify the perturbations in expression levels and activities of approximately 300 kinases in cells upon exposure to each kinase inhibitor by employing

ATP-affinity pull-down coupled with LC-MRM analysis and LC-PRM analysis of the tryptic digestion mixture of whole cell lysate. Analysis of kinase protein expression and

ATP-binding affinity data together led to the discovery of CHK1 as a novel target for imatinib, and we found that imatinib could sensitize cancer cells toward NCS, a radiomimetic agent capable of inducing DNA double strand breaks. In addition, the ATP binding affinity of MAP2K5 is inhibited by vemurafenib in both in cellulo and in vitro experiments, suggesting MAP2K5 is a novel binding target for vemurafenib. Consistent with the results obtained from quantitative proteomic analyses, Western blot analysis revealed that exposure of M14 cells to vemurafenib led to diminished phosphorylation of

ERK5.

Aside from uncovering kinases that are inhibited by small molecules, our kinome profiling method also led to the revelation of those kinases with heightened ATP-binding affinities upon treatment with these small molecules. While so far all small molecule-based, kinase-targeting therapeutic approaches exploit the inhibition of kinase activities, some human diseases arise from diminished kinase activities, e.g. cancer and diabetes (33-35).

Thus, the proteome-wide characterizations of alterations in ATP binding affinities of kinases upon treatment with these kinase-targeting small molecules may also lead to the

82 identification of those kinases whose activities could be stimulated by these molecules, thereby broadening the therapeutic applications of these small-molecule drugs.

References

1. Lemmon MA, Schlessinger J. Cell signaling by receptor tyrosine kinases. Cell. 2010;141(7):1117-34. 2. Wu P, Nielsen TE, Clausen MH. FDA-approved small-molecule kinase inhibitors. Trends Pharmacol Sci. 2015;36(7):422-39. 3. Druker BJ. Perspectives on the development of a molecularly targeted agent. Cancer Cell. 2002;1(1):31-6. 4. Dancey J, Sausville EA. Issues and progress with protein kinase inhibitors for cancer treatment. Nat Rev Drug Discov. 2003;2(4):296-313. 5. Davies SP, Reddy H, Caivano M, Cohen P. Specificity and mechanism of action of some commonly used protein kinase inhibitors. Biochem J. 2000;351(1):95. 6. Duncan JS, Whittle MC, Nakamura K, Abell AN, Midland AA, Zawistowski JS, et al. Dynamic reprogramming of the kinome in response to targeted MEK inhibition in triple-negative breast cancer. Cell. 2012;149(2):307-21. 7. Chong CR, Sullivan DJ, Jr. New uses for old drugs. Nature. 2007;448(7154):645-6. 8. Manning G, Whyte DB, Martinez R, Hunter T, Sudarsanam S. The Protein Kinase Complement of the Human Genome. Science. 2002;298(5600):1912-34. 9. Patricelli MP, Nomanbhoy TK, Wu J, Brown H, Zhou D, Zhang J, et al. In situ kinase profiling reveals functionally relevant properties of native kinases. Chem Biol. 2011;18(6):699- 710. 10. Miao W, Xiao Y, Guo L, Jiang X, Huang M, Wang Y. A high-throughput targeted proteomic approach for comprehensive profiling of methylglyoxal-induced perturbations of the human kinome. Anal Chem. 2016;88:9773-9. 11. Patricelli MP, Szardenings AK, Liyanage M, Nomanbhoy TK, Wu M, Weissig H, et al. Functional interrogation of the kinome using nucleotide acyl phosphates. Biochemistry. 2007;46(2):350-8. 12. Xiao Y, Guo L, Wang Y. A targeted quantitative proteomics strategy for global kinome profiling of cancer cells and tissues. Mol Cell Proteomics. 2014;13(4):1065-75. 13. Bomgarden RD, Lupardus PJ, Soni DV, Yee MC, Ford JM, Cimprich KA. Opposing effects of the UV lesion repair protein XPA and UV bypass polymerase η on ATR checkpoint signaling. EMBO J. 2006;25:2605-14. 14. Ong S-E, Blagoev B, Kratchmarova I, Kristensen DB, Steen H, Pandey A, et al. Stable Isotope Labeling by Amino Acids in Cell Culture, SILAC, as a Simple and Accurate Approach to Expression Proteomics. Mol Cell Proteomics. 2002;1(5):376-86. 15. Xiao Y, Guo L, Jiang X, Wang Y. Proteome-wide discovery and characterizations of nucleotide-binding proteins with affinity-labeled chemical probes. Anal Chem. 2013;85(6):3198- 206. 16. Xiao Y, Guo L, Wang Y. Isotope-coded ATP probe for quantitative affinity profiling of ATP-binding proteins. Anal Chem. 2013;85(15):7478-86. 17. Cox J, Mann M. MaxQuant enables high peptide identification rates, individualized p.p.b.- range mass accuracies and proteome-wide protein quantification. Nat Biotechnol. 2008;26:1367- 72. 18. Huang DW, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protocols. 2008;4(1):44-57. 19. Manning G, Whyte DB, Martinez R, Hunter T, Sudarsanam S. The Protein Kinase Complement of the Human Genome. Science. 2002;298:1912.

20. Escher C, Reiter L, MacLean B, Ossola R, Herzog F, Chilton J, et al. Using iRT, a normalized retention time for more targeted measurement of peptides. Proteomics. 2012;12:1111- 21. 21. MacLean B, Tomazela DM, Shulman N, Chambers M, Finney GL, Frewen B, et al. Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics. 2010;26:966-8. 22. Franken NA, Rodermond HM, Stap J, Haveman J, van Bree C. Clonogenic assay of cells in vitro. Nat Protoc. 2006;1(5):2315-9. 23. Peterson AC, Russell JD, Bailey DJ, Westphall MS, Coon JJ. Parallel reaction monitoring for high resolution and high mass accuracy quantitative, targeted proteomics. Mol Cell Proteomics. 2012;11(11):1475-88. 24. MacLean B, Tomazela DM, Shulman N, Chambers M, Finney GL, Frewen B, et al. Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics. 2010;26(7):966-8. 25. Dar AC, Lopez MS, Shokat KM. Small Molecule Recognition of c-Src via the Imatinib- Binding Conformation. Chemistry & biology. 2008;15(10):1015-22. 26. Milani G, Lana T, Bresolin S, Aveic S, Pasto A, Frasson C, et al. Expression Profiling of Circulating Microvesicles Reveals Intercellular Transmission of Oncogenic Pathways. Mol Cancer Res. 2017;15(6):683-95. 27. Fabian MA, Biggs WH, 3rd, Treiber DK, Atteridge CE, Azimioara MD, Benedetti MG, et al. A small molecule-kinase interaction map for clinical kinase inhibitors. Nat Biotechnol. 2005;23(3):329-36. 28. Peng SB, Henry JR, Kaufman MD, Lu WP, Smith BD, Vogeti S, et al. Inhibition of RAF isoforms and active dimers by LY3009120 leads to anti-tumor activities in RAS or BRAF mutant cancers. Cancer Cell. 2015;28(3):384-98. 29. Vin H, Ojeda SS, Ching G, Leung ML, Chitsazzadeh V, Dwyer DW, et al. BRAF inhibitors suppress apoptosis through off-target inhibition of JNK signaling. Elife. 2013;2:e00969. 30. Hatzivassiliou G, Song K, Yen I, Brandhuber BJ, Anderson DJ, Alvarado R, et al. RAF inhibitors prime wild-type RAF to activate the MAPK pathway and enhance growth. Nature. 2010;464(7287):431-5. 31. Liu Q, Guntuku S, Cui XS, Matsuoka S, Cortez D, Tamai K, et al. Chk1 is an essential kinase that is regulated by Atr and required for the G(2)/M DNA damage checkpoint. Genes Dev. 2000;14(12):1448-59. 32. Davis MI, Hunt JP, Herrgard S, Ciceri P, Wodicka LM, Pallares G, et al. Comprehensive analysis of kinase inhibitor selectivity. Nat Biotechnol. 2011;29:1046. 33. Kissil JL, Feinstein E, Cohen O, Jones PA, Tsai YC, Knowles MA, et al. DAP-kinase loss of expression in various carcinoma and B-cell lymphoma cell lines: possible implications for role as tumor suppressor gene. Oncogene. 1997;15(4):403-7. 34. Kulkarni RN, Bruning JC, Winnay JN, Postic C, Magnuson MA, Kahn CR. Tissue-specific knockout of the insulin receptor in pancreatic beta cells creates an insulin secretory defect similar to that in type 2 diabetes. Cell. 1999;96(3):329-39. 35. Gao Y, Ge G, Ji H. LKB1 in lung cancerigenesis: a serine/threonine kinase as tumor suppressor. Protein Cell. 2011;2(2):99-107.

Figure 3.1. PRM- and MRM-based targeted proteomic approaches for interrogating the human kinome. (a) The chemical structures of the isotope-coded ATP affinity probes. (b)

Experimental strategy for PRM- and MRM-based targeted proteomic approaches. (c) Venn diagrams displaying the numbers of kinases included in the PRM kinome library and those that could be quantified by the PRM method (left), and the overlaps of kinases (middle) and protein kinases (right) between the PRM- and MRM-based kinome libraries.

86 a O NH2 N N HN NH O X X O O O X O P O P O P O O N N N H X X OH OH OH X O HO OH b X = H, Light; X=D, Heavy

Kinase Inhibitor DMSO Kinase Inhibitor DMSO

Light Amino Acids Heavy Amino Acids Label with Label with Light ATP Probe Heavy ATP Probe Combine, Mix at 1:1 Combine, Mix at 1:1 ratio ratio

Tryptic Digestion, Tryptic Digestion Avidin Enrichment

LC-MS/MS (PRM) LC-MS/MS (MRM)

Data Analysis Data Analysis c Human Kinome (518 protein kinases, ~600 total kinases) Kinases Protein kinases

Quantified kinases (~ 124 350 128 111 304 93 300 total kinases)

Library Kinases (395 MRM library PRM library MRM library PRM library protein kinases, 478 total kinases)

MRM methods, or the absence of apparent correlation of alterations in kinase protein expression triggered by dabrafenib and vemurafenib treatments.

a ADRBK1 BRSK2 DYRK1A ABL1 CLP1 AAK1 ADRBK2 CAMK2G DYRK1B ABL2 CMPK1 PFKM AKT1 CAMKK2 GSK3A BTK CPNE3 PFKP CAD PI4K2B AKT2 CHEK1 GSK3B DLG1 CSK PI4KB CCND3 CHEK2 MAPK1 DTYMK EXOSC10 DDR1 PIK3C2A MAPKAPK2 MAPK12 EPHB4 EIF2AK2 PIK3C2B GRK6 MAPKAPK3 MAPK14 EPHB6 ERCC2 PIK3CB LATS1 MAPKAPK5 MAPK15 FGFR4 ERCC3 PIK3CD MASTL MARK2 IGF2R FASTKD1 PIK3R2 PKN1 MAPK3 MELK TK LYN FASTKD2 PIK3R3 PKN2 MAPK8 MKNK1 PDGFRB FASTKD3 PIK3R4

PKN3 MAPK9 CMGC CAMK MNAT1 PTK2 FASTKD5 PIKFYVE PRKAA1 NLK OBSCN PTK2B FKBP1A PIP4K2A PRKACA PHKA1 PRPF4B PTK7 FPGT PIP4K2C PRKACG PHKA2 PRPK SYK GAK PKMYT1 PRKAG1 PRPS1 TK1 PLK1 PRKAG2 PHKB GALK1 PRPS2 TK2 PNKP PRKCD PHKG2 GALK2 AGC TNK1 POLR2A SIK2 PRPSAP1 GNE PRKCE TNK2 POLR2B PRKCI STK11 PRPSAP2 TYK2 GTF2H1 POLR2C PRKD2 STK33 SRPK1 YES1 GTF2H3 POLR2D PRKD3 TRIO SRPK2 ACVR1 GTF2H4 POLR2E PRKDC CSNK1A1 CIT ARAF GUK1 POLR2G ROCK1 CSNK1D MAP2K1 BRAF HK1 POLR2H ROCK2 CSNK1G2 MAP2K2 ILK HK2 POLR2L RPS6KA1 CSNK1G3 IRAK1 HKDC1 POMK MAP2K4 RPS6KA3 CSNK2A1 IRAK4 HUS1 PPIP5K1 MAP2K7 RPS6KB1 CSNK2A2 KSR1 IKBKAP Other PPIP5K2

CK1 MAP3K1

RPS6KB2 CSNK2B LIMK2 IKBKB PPP4C Other TKL LMTK2 PXK RPS6KC1 TTBK2 MAP3K11 IPPK RAF1 RP2 STK38L VRK1 MAP3K4 IRPK1 RIPK2 SCYL1 ADCK4 VRK2 MAP3K7 RIPK4 ITPK1 SCYL2 AK1 MAP4K1 CDC42BPA TGFBR1 ITPKB SCYL3 AK2 CDC42BPB MAP4K4 MAGI3 AK3 TGFBR2 SPHK1 CDC42BPG MAP4K5 ZAK MPP1 SRC AK4 CDC7 OXSR1 ACTL8 NAGK TAF9 AK5 STE CDK1 ADK NEK7 TBK1 AK6 PAK2 CDK12 AGK NEK9 TBRG4 AK9 PAK3 AKAP13 TFG CDK16 PAK4 NIN ATR AKAP9 TJP2 BCKDK CDK2 SLK NME6 CDK3 ATRIP NRBP1 TLK1 BCR STK24 CDK4 AURKA NRBP2 TLK2 BRD2 CMGC STK25 AURKB TRIM24 CDK5 NRP1 MTOR STK26 BMP2K TRIM27 CDK7 NUP62 PDK1 BUB1 TRIM28

CDK9 STK3 PBK Atypical PDPK1 Other STK4 BUB1B TTK RIOK1 CLK1 CARD11 PEAK1 UCK2 TAOK1 RIOK3 CLK3 CAV2 PFKFB2 UCKL1 SMG1 CRKL TAOK2 CCNH PFKFB4 WEE1 TRPM7 DCK TNIK CHUK PFKL WNK1 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 RImatinib/Control RImatinib/Control RImatinib/Control RImatinib/Control RImatinib/Control RImatinib/Control b BRSK2 AKT1 CLK3 STK40 TYK2 PAN3 AKT2 CAMK1 CLK4 TAOK1 YES1 PBK AKT3 CAMK2A TK a CRKL STE TAOK2 ALK4 CAMK2G PFKL ANPRA DCK ABL1 ALK5 CAMK4 PFKM BCL1 DCLK3 BMX ALK7 CAMKK1 PFKP BTF2 DYRK2 BSK ARAF CAMKK2 PGK1 CCNB1 ERCC2 BTK/ATK BRAF CASK PGK2 CCND1 ERK1 BYK CARK CHEK1 DMP4 ERK2 CAD ILK PI4K2A DMPK CHEK2 ERK3 CKB IRAK PI4K2B DMPK2 MAPKAPK1A ERK5 CSK IRAK4 PI4KB

EXOSC10 MAPKAPK1B TKL GSK3A DDR1 LIMK2 PIG31

GRK2 MAPKAPK1C CMGC GSK3B EGFR LRRK1 PIK3C2A GRK5 MAPKAPK2 MAK EPHA1 LRRK2 PIK3CA GRK6 MAPKAPK3 MAPK14 EPHA2 MLKL LATS1 PIK3CD MARK2 MAPK15 EPHA4 RAF1 MAST3 PIP4K2A MARK4 MAPK7 EPHA7 RIPK1 MASTL PIP4K2B MELK PRPF4B EPHA8 CAMK ZAK MSK2 PIP4K2C NIM1K PRPK EPHB2 AAK1 PDPK1 NUAK2 PRPS1 EPHB3 ADK PIP5K1C PKN2 OBSCN PRPS2 EPHB4 ADVCK1 PIP5K3 PRKAA1 PASK PRPSAP2 ERBB2 AGK PKMYT1 PRKAA2 PHKG2 SRPK1 ERBB4 AURKA PLK1 PRKACA AGC PKDCC SRPK2 FAKD2 AURKB PNCK PRKACB PRAK CIT FER AURKC POLR2A PRKACG PSKH2 LOK FGFR1 CHUK PRKAG1 POLR2B SIK1 MAP2K1 FGFR2 CKMM PRKAG2 POLR2E SIK3 MAP2K3 FGFR4 CMPK1 PRKCA PPP4C STK33 MAP2K4 FGR CMPK2 PRKCB PXK TSSK2 MAP2K7 FLT3 CPNE3 PRKCD RBKS TSSK4 MAP3K1 HCK EIF2AK1

PRKCE TK TSSK6 MAP3K11 IGF1R EIF2AK2 Other SBK1 PRKCG TTN MAP3K2 IGF2R EIF2AK3 SCYL2 PRKCH CSNK1A1 MAP3K3 INSRR EIF2AK4 SNFRK PRKCI CSNK1A1L MAP3K4 JAK1 ERN1 SRC PRKCL1 CSNK2A1 MAP3K5 JAK2 GAK SRK PRKD2 CSNK2A2 MAP3K6 KIT GALK1 TBK1

PRKDC CK1 TTBK2 MAP3K7 LCK HK1 TBRG4 ROCK2 VRK1 MAP4K1 LYK/ITK Other IKBKAP RPS6KB1 TEC VRK2 MAP4K2 LYK5 IKBKB RPS6KC1 STE TJP2 CABLES1 MAP4K3 LYN IP6K1 SGK3 TLK1 CDC42 MAP4K5 MATK IP6K1 STK14A CDK10 MASK MLTK IPMK TLK2 STK14B CDK12 MEK5 NTRK2 IPPK TMPK STK38L CDK13 MEK6 NTRK3 ITPK1 TNK1 AK1 CDK14 NRK PDGFRA MOS TRIB1 AK2 CDK17 OXSR1 PTK2 NAGK ATM TRIM24 CDK19 PAK1 PTK7 NEK1 ATR TRIM28 CDK2 PAK2 PTK8 NEK2 BAZ1A TROVE2 CDK20 SLK RAFTK NEK4 BCKDK TROVE2 CMGC CDK7 STK11 RET NEK5 BRD2 UCK2 CDK9 STK24 ROR1 NEK7 BRD4 UHMK1 Atypical CDKL1 STK3 ROS1 CHAK2 NEK8 CDKL3 STK36 SSK1 ULK1 EEF2K NEK9 CLK1 STK38 SYK ULK3 MTOR NME3 CLK2 STK39 TK1 WNK4 SMG1 NTKL 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 RImatinib/Control RImatinib/Control RImatinib/Control RImatinib/Control RImatinib/Control RImatinib/Control ba ADRBK1 BRSK2 DYRK1A ABL1 CLP1 AAK1 ADRBK2 CAMK2G DYRK1B ABL2 CMPK1 PFKM AKT1 CAMKK2 GSK3A BTK CPNE3 PFKP CAD PI4K2B AKT2 CHEK1 GSK3B DLG1 CSK PI4KB CCND3 CHEK2 MAPK1 DTYMK EXOSC10 DDR1 PIK3C2A MAPKAPK2 MAPK12 EPHB4 EIF2AK2 PIK3C2B GRK6 MAPKAPK3 MAPK14 EPHB6 ERCC2 PIK3CB LATS1 MAPKAPK5 MAPK15 FGFR4 ERCC3 PIK3CD MASTL MARK2 IGF2R FASTKD1 PIK3R2 PKN1 MAPK3 MELK TK LYN FASTKD2 PIK3R3 PKN2 MAPK8 MKNK1 PDGFRB FASTKD3 PIK3R4

CK1 MAP3K1

TK AKT3 CAMK2A

CRKL STE TAOK2 ALK4 CAMK2G PFKL ANPRA DCK ABL1 ALK5 CAMK4 PFKM BCL1 DCLK3 BMX ALK7 CAMKK1 PFKP BTF2 DYRK2 BSK ARAF CAMKK2 PGK1 CCNB1 ERCC2 BTK/ATK BRAF CASK PGK2 CCND1 ERK1 BYK CARK CHEK1 DMP4 ERK2 CAD ILK PI4K2A DMPK CHEK2 ERK3 CKB IRAK PI4K2B DMPK2 MAPKAPK1A ERK5 CSK IRAK4 PI4KB

EXOSC10 MAPKAPK1B TKL GSK3A DDR1 LIMK2 PIG31

PRKDC CK1 TTBK2 MAP3K7 LCK HK1 TBRG4 ROCK2 VRK1 MAP4K1 LYK/ITK Other IKBKAP RPS6KB1 TEC VRK2 MAP4K2 LYK5 IKBKB RPS6KC1 STE TJP2 CABLES1 MAP4K3 LYN IP6K1 SGK3 TLK1 CDC42 MAP4K5 MATK IP6K1 STK14A CDK10 MASK MLTK IPMK TLK2 STK14B CDK12 MEK5 NTRK2 IPPK TMPK STK38L CDK13 MEK6 91 NTRK3 TNK1 ITPK1 AK1 CDK14 NRK PDGFRA MOS TRIB1 AK2 CDK17 OXSR1 PTK2 NAGK ATM TRIM24 CDK19 PAK1 PTK7 NEK1 TRIM28 ATR CDK2 PAK2 PTK8 NEK2 BAZ1A TROVE2 CDK20 SLK RAFTK NEK4 BCKDK TROVE2 CMGC CDK7 STK11 RET NEK5 BRD2 UCK2 CDK9 STK24 ROR1 NEK7 BRD4 UHMK1 Atypical CDKL1 STK3 ROS1 CHAK2 NEK8 CDKL3 STK36 SSK1 ULK1 EEF2K NEK9 CLK1 STK38 SYK ULK3 MTOR NME3 CLK2 STK39 TK1 WNK4 SMG1 NTKL 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 RImatinib/Control RImatinib/Control RImatinib/Control RImatinib/Control RImatinib/Control RImatinib/Control Figure 3.4. Western blot analyses for validating the protein expression levels of kinases in

K-562 cells with or without imatinib treatment. (a) Images from Western blot analyses of the expression levels of select kinases in K-562 cells with or without imatinib treatment.

(b) Representative PRM traces for monitoring the expression levels of the same kinases as shown in (a). (c) The quantification results for the ratios of kinase proteins in imatinib- treated cells over untreated cells. Error bars represent standard deviations.

a b AK1 CCND3 SCYL3 1 1 K-562 1 Forward Imatinib (M) 0 1 0.5 0.5 0.5 Imatinib DMSO AK1 0 0 0 110 114 47 49 51 125 128 CCND3 1 1 1 Reverse Imatinib SCYL3 0.5 0.5 0.5 DMSO

Relative AbundanceRelative 0 β-Actin 0 0 113 117 49 50 51 127 129 c Retention Time (min) 14 PRM 12 Western blot 10

/DMSO) 8 6

RelativeRatio

Imatinib ( 2 0 AK1 CCND3 SCYL3 Kinases

Figure 3.5. The alterations in expression levels of kinase proteins in M14 cells after treatment with two small-molecule BRAF inhibitors, dabrafenib (a) and vemurafenib (b).

The cells were treated with 100 nM inhibitor for 24 h. Displayed are the ratios of expression of kinase proteins in BRAF inhibitor-treated over mock-treated M14 cells, where the X- axis was plotted in log10 scale. The data represent the average ratios obtained from two forward and two reverse SILAC labeling experiments for dabrafenib treatment and two forward and one reverse SILAC labeling experiments for vemurafenib treatment. The red and blue bars designate those kinases that were up- and down-regulated, respectively, by at least 1.5-fold upon the inhibitor treatment.

ADRBK2 RIOK1 GSK3A FES DSTYK AAK1 AKT1 RIOK2 GSK3B FGFR4 DTYMK PFKL AKT2 RIOK3 MAPK1 FLT4 EIF2AK2 PFKM AKT3 SMG1 MAPK14 FYN ENG PFKP TRPM7 MAPK15 IGF1R ERCC2 GRK6 Atypical PI4K2B IGF2R LATS1 CAMK2D MAPK3 ERCC3 PIK3C2A CAMK2G ITK MASTL MAPK7 FASTKD1 PIK3CA CAMK4 MAPK8 JAK1 PKN1 FASTKD2 PIK3CB CAMKK2 MAPK9 JAK3 PKN2 FASTKD3 PIK3CD

CAMKV CMGC PRPK KIT PRKAA1 FKBP1A PIK3R2 LYN PRKACA CHEK1 PRPS1 TK FPGT MERTK PIK3R4 PRKACB CHEK2 PRPS2 GAK PDGFRB PIKFYVE PRKACG DCLK1 PRPSAP1 GALK1 PTK2 PIP4K2A PRKAG1 MAPKAPK2 PRPSAP2 GALK2 MAPKAPK3 PTK7 PIP4K2C PRKAG2 SRPK1 GNE MAPKAPK5 ROR1 PIP5K1A PRKCB SRPK2 GOLGA5 MARK2 SYK PIP5K3 PRKCD CIT GTF2H1 MELK TK1 PKMYT1 PRKCE MAP2K1 GTF2H3 AGC OBSCN TK2 PNKP PRKCH MAP2K2 GTF2H4 PHKA1 TNK1 POLR2A PRKD1 MAP2K3 GUK1 TNK2 POLR2B PHKA2 MAP2K6 HK1 PRKD3 YES1 POLR2C CAMK STK17A HK2 PRKDC MAP3K11 STK33 ZAP70 HK3 POLR2D PRKX MAP3K3 TRIO ACVR1 HKDC1 POLR2E ROCK1 MAP3K4 VRK2 ARAF HSPB8 POLR2G ROCK2 MAP3K5 VRK1 BRAF HUS1 POLR2H RPS6KA1 MAP3K6 CSNK2B ILK IKBKAP POLR2L RPS6KA3 MAP4K1 CSNK2A2 IRAK1 IKBKB

MAP4K4 PPIP5K2 Other

RPS6KA4 Other CSNK2A1 IRAK4 IPPK PPP4C RPS6KA5 MAP4K5 CK1 CSNK1E KSR1 IRS1 OXSR1 RP2 RPS6KB1 CSNK1D LIMK1 TKL ITPK1 SCYL1 RPS6KC1 PAK2 CSNK1A1 STE LIMK2 ITPKA SCYL2 SGK3 PAK4 CDC42BPA RAF1 ITPKB SCYL3 STK38L SLK CDC42BPB RIPK1 KCNH4 SPHK1 ADCK4 STK10 TGFBR1 CDC42BPG MAGI3 SRC AK1 STK24 ZAK CDC7 MPP1 TBCK AK2 STK25 ACTL8 CDK1 NAGK TBK1 AK3 CDK16 STK26 ADK NEK4 TBRG4 AK4 CDK2 STK3 AKAP13 NEK6 TFG AK5 CDK3 STK39 AKAP9 NEK7 TGFB2 AK6 CDK4 STK4 ATRIP NIN TJP2 AK9 CDK5 TAOK1 BMP2K NME6 TRIM27 ATM CDK7 TAOK3 CARD11 NME7 TRIM28

BRD2 CDK8 TNIK CAV2 NRBP1 CMGC FASTK CDK9 CAD CCNH NRP1 TWF1 Atypical MTOR CSK CLP1 UCK1 CLK1 Other NRP2 PDK1 CLK3 DDR1 CMPK1 NUP62 UCK2 PDK2 CRKL EPHB2 CPNE3 PBK UCKL1 PDK3 DCK TK EPHB4 CRIM1 PEAK1 ULK1 PDK4 DYRK1A EPHB6 CTTNBP2 PFKFB2 ULK3 PDPK1 DYRK1B ERBB3 DLG1 PFKFB4 WNK1 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 RDabrafenib/Control RDabrafenib/Control RDabrafenib/Control RDabrafenib/Control RDabrafenib/Control RDabrafenib/Control b AURKB a ADRBK1 FASTK CDK7 ABL1 AAK1 BUB1B NRP2 ADRBK2 MTOR CDK9 AXL CAV2 NUP62 AKT1 PDPK1 CLK1 BTK CAD CCNH PBK AKT2 RIOK1 CLK3 CMPK1 CSK PEAK1 AKT3 RIOK3 CRKL CPNE3 Atypical DDR1 PFKFB2 AMPK1 SMG1 DCK CTTNBP2 EGFR PFKFB3 KAPCB CAMK1D DYRK1B DGUOK PFKFB4 CAMK2D EPHB2 DSTYK LATS1 GSK3A PFKL CMGC EPHB4 CAMK2G MAPK1 DTYMK PFKM MAST3 EPHB6 EIF2AK2 CAMKK2 MAPK14 PFKP MASTL ERBB2 EIF2AK4 CAMKV MAPK15 PI4K2B PKN1 FES ENG CHEK1 PI4KB MAPK3 TK FLT4 EXOSC10 PKN2 CHEK2 PIK3CB MAPK7 IGF1R FASTKD2 PRKAA1 DCLK1 PIK3R2 PRPK IGF2R FASTKD3 PIK3R4 PRKACA DCLK2 FKBP1A PRPS1 JAK3 PIKFYVE PRKACB MAPKAPK2 FPGT PRPS2 LCK PIP4K2A

AGC PRKACG MAPKAPK3 GAK PRPSAP1 LYN PIP4K2C GALK1 PRKAG1 MAPKAPK5 PRPSAP2 PIP5K1A MERTK GALK2 PRKAG2 MARK1 PKMYT1 SRPK1 PDGFRB GNE PRKCD CAMK MARK2 PLK1 SRPK2 PTK2 GOLGA5 MELK PNKP PRKCE CIT PTK7 GTF2H1 PASK POLR2A PRKCH MAP2K1 SYK GTF2H4 PHKA1 POLR2B PRKCI MAP2K2 TK1 GUK1 PHKA2 POLR2C PRKCZ MAP2K3 TK2 HK1 PHKG2 POLR2D YES1 HK2 PRKD1 MAP2K6 POLR2E

SIK2 Other

Other HK3 PRKDC MAP2K7 ZAP70 SIK3 POLR2G ACVR1 HSPB8 PRKX MAP3K1 HUS1 POLR2H STK17A ARAF ROCK1 MAP3K11 IKBKAP POLR2L

TRIO STE MAP3K2 BRAF PPIP5K1 ROCK2 CSNK1A1 IPPK ILK PPIP5K2 RPS6KA1 CSNK1D MAP3K3 IRS1 IRAK1 ITPK1 PPP4C RPS6KA3 CSNK1E MAP3K4 IRAK4 ITPKA RP2 CSNK2A1 MAP3K5 RPS6KC1 KSR1 ITPKB SCYL1 CK1 MAP4K4 STK38 CSNK2A2 LIMK1 ITPKC SCYL2

MAP4K5 TKL STK38L CSNK2B LIMK2 LGMN SPHK1 MINK1 AK1 VRK1 RAF1 MAGI3 TBK1 OXSR1 TBRG4 AK2 VRK2 RIPK1 MNAT1 CDC42BPA PAK2 MPP1 TFG AK3 RIPK3 PAK4 TGFB2 CDC42BPG RIPK4 MPP2 AK4 SLK MYT1 TJP2 CDC7 TGFBR1 AK5 NAGK TRIM28 CDK1 STK10 TGFBR2 AK6 NEK4 TWF1 Atypical CDK14 STK24 ACTL8 NEK6 UCK2 AK9 CDK16 STK26 ADK NEK7 UCKL1 ATM CMGC STK39 CDK2 AGK NIN ULK1 BCR CDK3 STK4 AKAP9 NME6 ULK3 BRD2 CDK4 TAOK2 ATRIP NME7 WNK1 EEF2K CDK5 TAOK3 Other AURKA NRP1 WNK4 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 RVemurafenib/Control RVemurafenib/Control RVemurafenib/Control RVemurafenib/Control RVemurafenib/Control RVemurafenib/Control

ADRBK2 RIOK1 GSK3A FES DSTYK AAK1 b AKT1 RIOK2 GSK3B FGFR4 DTYMK PFKL AKT2 RIOK3 MAPK1 FLT4 EIF2AK2 PFKM AKT3 SMG1 MAPK14 FYN ENG PFKP TRPM7 MAPK15 IGF1R ERCC2 GRK6 Atypical PI4K2B IGF2R LATS1 CAMK2D MAPK3 ERCC3 PIK3C2A CAMK2G ITK MASTL MAPK7 FASTKD1 PIK3CA CAMK4 MAPK8 JAK1 PKN1 FASTKD2 PIK3CB CAMKK2 MAPK9 JAK3 PKN2 FASTKD3 PIK3CD

MAP4K4 PPIP5K2 Other

BRD2 CDK8 TNIK CAV2 NRBP1 CMGC FASTK CDK9 CAD CCNH NRP1 TWF1 Atypical MTOR CSK CLP1 UCK1 CLK1 Other NRP2 PDK1 CLK3 DDR1 CMPK1 NUP62 UCK2 PDK2 CRKL EPHB2 CPNE3 PBK UCKL1 PDK3 DCK TK EPHB4 CRIM1 PEAK1 ULK1 PDK4 DYRK1A EPHB6 CTTNBP2 PFKFB2 ULK3 PDPK1 DYRK1B ERBB3 DLG1 PFKFB4 WNK1 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 RDabrafenib/Control RDabrafenib/Control RDabrafenib/Control RDabrafenib/Control RDabrafenib/Control RDabrafenib/Control b AURKB ADRBK1 FASTK CDK7 ABL1 AAK1 BUB1B NRP2 ADRBK2 MTOR CDK9 AXL CAV2 NUP62 AKT1 PDPK1 CLK1 BTK CAD CCNH PBK AKT2 RIOK1 CLK3 CMPK1 CSK PEAK1 AKT3 RIOK3 CRKL CPNE3 Atypical DDR1 PFKFB2 AMPK1 SMG1 DCK CTTNBP2 EGFR PFKFB3 KAPCB CAMK1D DYRK1B DGUOK PFKFB4 CAMK2D EPHB2 DSTYK LATS1 GSK3A PFKL CMGC EPHB4 CAMK2G MAPK1 DTYMK PFKM MAST3 EPHB6 EIF2AK2 CAMKK2 MAPK14 PFKP MASTL ERBB2 EIF2AK4 CAMKV MAPK15 PI4K2B PKN1 FES ENG CHEK1 PI4KB MAPK3 TK FLT4 EXOSC10 PKN2 CHEK2 PIK3CB MAPK7 IGF1R FASTKD2 PRKAA1 DCLK1 PIK3R2 PRPK IGF2R FASTKD3 PIK3R4 PRKACA DCLK2 FKBP1A PRPS1 JAK3 PIKFYVE PRKACB MAPKAPK2 FPGT PRPS2 LCK PIP4K2A

SIK2 Other

Other HK3 PRKDC MAP2K7 ZAP70 SIK3 POLR2G ACVR1 HSPB8 PRKX MAP3K1 HUS1 POLR2H STK17A ARAF ROCK1 MAP3K11 IKBKAP POLR2L

MAP4K5 TKL STK38L CSNK2B LIMK2 LGMN SPHK1 MINK1 AK1 VRK1 94 RAF1 MAGI3 TBK1 OXSR1 TBRG4 AK2 VRK2 RIPK1 MNAT1 CDC42BPA PAK2 MPP1 TFG AK3 RIPK3 PAK4 TGFB2 CDC42BPG RIPK4 MPP2 AK4 SLK MYT1 TJP2 CDC7 TGFBR1 AK5 NAGK TRIM28 CDK1 STK10 TGFBR2 AK6 NEK4 TWF1 Atypical CDK14 STK24 ACTL8 NEK6 UCK2 AK9 CDK16 STK26 ADK NEK7 UCKL1 ATM CMGC STK39 CDK2 AGK NIN ULK1 BCR CDK3 STK4 AKAP9 NME6 ULK3 BRD2 CDK4 TAOK2 ATRIP NME7 WNK1 EEF2K CDK5 TAOK3 Other AURKA NRP1 WNK4 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 0.33 3.3 RVemurafenib/Control RVemurafenib/Control RVemurafenib/Control RVemurafenib/Control RVemurafenib/Control RVemurafenib/Control Figure 3.6. Western blot analyses for validating the protein expression levels of kinases in

M14 cells with or without dabrafenib/vemurafenib treatment. (a) Images from Western blot analyses of the expression levels of representative kinases in M14 cells with or without dabrafenib or vemurafenib treatment. The quantification results for the ratios of kinase proteins in dabrafenib- and vemurafenib-treated cells over control untreated cells are shown in (b) and (c), respectively. The data represent the mean ± S. D. of the quantification results (n = 3).

95 a Dabrafenib Vemurafenib Dabrafenib Vemurafenib (nM) (nM) (nM) (nM) 0 100 0 100 0 100 0 100 mTor EGFR AK1 ARAF JAK3 BRAF IGF2R MAP2K1 STK26 MAP3K3 CHK1 MAP3K5 β-Actin β-Actin b 2 PRM

1.5 Western blot /DMSO) 1

0.5

RelativeRatio Dabrafenib

( 0 c 3.5 PRM 3 Western blot

2.5 /DMSO) 2 1.5 1

RelativeRatio 0.5 Vemurafenib

( 0

98 a

Clonogenic survival assay results showing the effect of imatinib on sensitizing MDA-MB-

231 cells toward NCS. Error bars represent standard deviation. *, 0.01 < p < 0.05; **, 0.001

< p < 0.01; #, no significant difference. The p-values were calculated against the control using two-tailed, unpaired Student’s t-test.

a b c MRM PRM K-562 1.2 Imatinib 1 1 ** Imatinib Imatinib (M) 0 1 DMSO 0.8 * DMSO * 0.5 0.5 CHK1 0.4 p-CHK1 S296 # 0 0

75 76 77 17 18 19 β-Actin 0 Relative AbundanceRelative Retention Time (min) RelativeCell Survival 0 10 20 50 100 NCS (ng/mL)

100

Figure 3.10. Vemurafenib binds MAP2K5. (a) MRM traces for tryptic peptides of

MAP2K5 in M14 cells with or without vemurafenib treatment. (b) Western blot for monitoring the expression levels of MAP2K5 and ERK5, and the phosphorylation level of

ERK5. (c) Quantitative comparison of ratios of protein expression and activity of MAP2K5 obtained from Western blot analyses, and ATP-binding affinity obtained from MRM analyses. The data represent the mean ± S. D. of the quantification results (n = 3). (d) The

Venn diagram showing the overlap between the cell-based and in vitro kinome profiling by vemurafenib treatment. (e) A scatter plot displaying the correlation between the ratios

(vemurafenib treat/control) obtained from in cellulo (in M14 cells) and in vitro (M14 cell lysate) experiments, respectively. (f) MRM traces of MAP2K5 and SRC obtained from in vitro kinome profiling assay.

101 a b Vemurafenib Vemurafenib M14 DMSO DMSO 1 1 Vemurafenib (nM) 0 100 MAP2K5 0.5 0.5 ERK5

P-ERK5 0 0

79 81 79 81 β-Actin Relative AbundanceRelative Retention Time (min) c DMSO d In cellulo In vitro Vemurafenib 1.2

0.8 Ratio 97 235 33

0.4 Relative

0 ProteinExpression Activity (p- ATP-binding ActivityERK5) ATPAffinity-Biding Expression (p-ERK5) Affinity e f DMSO Vemurafenib 1 MAP2K5 SRC 1 1 0 ARAF -1 ZAK 0.5 0.5 In vitro In -2

MAP2K5 (Vemurafenib/DMSO), (Vemurafenib/DMSO), 2 -3 0 0 -4 -3 -2 -1 0 1 2

Relative AbundanceRelative 76 78 109 110 111 Log Log2(Vemurafenib/DMSO), Retention Time (min) In cellulo

102

Figure 3.11. Relative growth of M14 cells upon a 24-hr treatment with the indicated concentrations of BIX-02188, an inhibitor for MAP2K5.

103

Cell lysis Removal of endogenous nucleotides with NAP-5

Vemurafenib DMSO (2 h) (100 nM, 2 h)

Label with Label with Light ATP Heavy ATP Affinity Probe Affinity Probe Combine, Mix at 1:1 ratio

Tryptic Digestion, Avidin Enrichment

LC-MS/MS (MRM)

Data Analysis

104

Figure 3.13. Differential ATP binding affinities of kinase proteins in lysates of M14 cells with and without a 2-hr pre-treatment with 100 nM vemurafenib. The data represent the means of the results obtained from two forward and one reverse ATP probe labeling results.

105

Chapter 4

Targeted Quantitative Kinome Analysis Identifies PRPS2 as a

Promoter for Colorectal Cancer Metastasis

Introduction

Colorectal cancer (CRC) is a major cause of morbidity and mortality throughout the world. It accounts for over 9% of all cancer incidence, ranks as the 3rd most frequently diagnosed cancer worldwide, and the 4th among the cancer-related deaths (1). Metastases are the main cause of cancer-related mortality in CRC patients (2), where approximately half of CRC patients develop metastatic disease (3) and the 5-year survival rate for metastatic CRC (stage IV) is less than 10% (4).

Aberrant kinase expression is known to be associated with CRC metastasis. For example,

CRC patients with elevated EGFR expression exhibit poor prognosis (5), and overexpression of AXL promotes the migration and invasion in CRC (6). Thus, a comprehensive analysis of kinase protein expression during CRC metastasis may allow for a systematic understanding about the implications of kinases in modulating the metastatic transformation of CRC.

Several quantitative proteomic methods were recently reported for the interrogation of the whole human kinome. For instance, ATP-acyl nucleotide affinity probes could be employed for the enrichment and quantification of kinases by liquid chromatography- multiple-reaction monitoring (LC-MRM) analysis (7-9). In this approach, the labeling of a

106 kinase by the ATP affinity probe can be modulated by both the protein expression level and the ATP-binding affinity of the kinase. Likewise, enrichment of kinases relying on the use of affinity resin immobilized with multiple kinase inhibitors may also be affected by changes in kinase activity (10-13). We recently developed a parallel-reaction monitoring

(PRM)-based targeted proteomic method to analyze the kinase protein expression at the entire proteome scale, and we also applied successfully the method for assessing the reprogramming of the human kinome upon treatment with a kinase inhibitor (14).

In this study, we employed the LC-PRM method to profile the differential protein expression of kinases in a pair of primary/metastatic CRC cells initiated from the same patient. We were able to quantify the expression of 299 unique kinases, and identify PRPS2 as a driver for CRC metastasis.

Materials and Methods

Cell culture

SW480 (primary) and SW620 (metastatic) (ATCC) cells were cultured in Dulbecco's modified eagle medium (DMEM). Culture media were supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA) and penicillin (100 IU/mL). The cells were

7 maintained at 37°C in a humidified atmosphere containing 5% CO2. Approximately 2×10 cells were harvested, washed with cold PBS for three times, and lysed by incubating on ice for 30 min with CelLytic M cell lysis reagent (Sigma) containing 1% protease inhibitor cocktail. The cell lysates were centrifuged at 9,000g at 4°C for 30 min, and the resulting supernatants collected. For SILAC labeling experiments, the cells were cultured in SILAC

107

13 15 13 medium containing unlabeled lysine and arginine or [ C6, N2]-lysine and [ C6]-arginine for at least five cell doublings.

Plasmids and shRNAs

The cDNA sequence of PRPS2 gene was PCR-amplified from a cDNA library prepared from mRNA isolated from SW620 cells, and the primers were 5'-

ATGCCCAACATCGTGCTGTT-3' and 5'-TAGCGGGACATGGCTGAACA-3'. The cDNA was subcloned into BamHI- and XbaI-linearized pRK7 vector. pRK7-PRPS2 and pRK7-empty vector were transfected into SW480 cell by using Transit-2020 (Mirus). All shRNA targeting sequences, which were designed according to Sigma

(https://www.sigmaaldrich.com/life-science/functional-genomics-and- rnai/shrna/individual-genes.html), were cloned into pLKO.1-Puro (Addgene). The sequences for shPRPS2 were 5'-CCATACGCCCGACAAGATAAA-3' (shPRPS2-1) and

5'-GTCACAAACACAATTCCGCAA-3' (shPRPS2-2); and scramble control sequence: 5'-

TCCTAAGGTTAAGTCGCCCTCG-3'.

Lentiviral particles were packaged using HEK293T cells. Virus-containing supernatants were collected at 48 h following transfection and filtered to eliminate cells. SW620 cells were infected with the lentivirus for 48 h prior to puromycin (1.0 μg/ml) selection.

Tryptic digestion of whole-cell protein lysates

The protein lysates prepared from SW480 and SW620 cells were combined at 1:1 ratio, washed with 8 M urea for protein denaturation, and then treated with dithiothreitol and iodoacetamide for cysteine reduction and alkylation, respectively. The proteins were

108 subsequently digested with modified MS-grade trypsin (Pierce) at an enzyme/substrate ratio of 1:100 in 50 mM NH4HCO3 (pH 8.5) at 37°C overnight. The peptide mixture was subsequently dried in a Speed-vac, desalted with OMIX C18 pipette tips (Agilent

Technologies), and subjected to LC-MS/MS analysis in the PRM mode.

LC-PRM Analysis

All scheduled LC-PRM experiments were carried out on a Q Exactive Plus quadrupole-

Orbitrap mass spectrometer coupled with an EASY-nLC 1200 system, as described recently (15, 16). The linear predictor of empirical RT from iRT (17) for targeted kinase peptides was determined by the linear regression of RT vs iRT of tryptic peptides from

BSA obtained for the chromatography setup prior to the analysis of kinase peptides (9, 15,

16). This RT-iRT linear relationship was re-established between every eight LC-PRM runs by analyzing again the tryptic digestion mixture of BSA. The targeted precursor ions were monitored in eight separate injections for each sample in scheduled PRM mode with an 8- min retention time window.

All raw files were processed using Skyline (version 3.5) for the generation of extracted- ion chromatograms and for peak integration (18). Six most abundant y ions found in

MS/MS acquired from shotgun proteomic analysis were chosen for peptide identification and quantification, where a mass accuracy of 20 ppm or less was imposed for fragment ions during the identification of peptides in the Skyline platform. The targeted peptides were first manually checked to ensure the chromatographic profiles of multiple fragment ions derived from the light and heavy forms of the same peptide could be overlaid. The

109 data were then processed to ensure that the distribution of the relative intensities of multiple transitions for the same precursor ion of kinase peptides is correlated with the theoretical distribution in the kinome MS/MS spectral library entry. The sum of peak areas from all transitions of light peptide or the corresponding heavy form was used for the quantification of the peptide.

TCGA and CCLE data analysis

OncoLnc was employed for the analysis of The Cancer Genome Atlas (TCGA) data for the correlation in mRNA expression between PRPS2 and MYC genes (19). Box plot and scatter plots for PRPS2 mRNA expression in CRC cell lines were generated from the data retrieved from The Cancer Cell Line Encyclopedia (CCLE) database

(https://portals.broadinstitute.org/ccle), which included the gene expression data for more than 1,000 cell lines representing 37 types of cancer (20).

Western blot

SW480 and SW620 cells were cultured in 6-well plates and were lysed at 50-70% confluency. The concentrations of the resulting protein lysates were determined using

Bradford Assay (Bio-Rad). The whole cell lysate for each sample (10 μg) was denatured by boiling in Laemmli loading buffer and resolved by using SDS-PAGE. Subsequently, the proteins were transferred onto nitrocellulose membrane at 4°C overnight. The resulting membrane was blocked with PBS-T (PBS with 0.1% Tween 20) containing 5% milk (Bio-

Rad) at 4°C for 6 h. The membrane was subsequently incubated, at 4°C, with primary antibody overnight and then with secondary antibody at room temperature for 1 h. After

110 thorough washing with PBS-T, the HRP signals were detected using Pierce ECL Western

Blotting Substrate (Thermo).

Antibodies recognizing human AK2 (Santa Cruz Biotechnology, sc-374095, 1:2000 dilution), PRPS1/2/3 (Santa Cruz Biotechnology, sc-376440, 1:2000 dilution), IGF2R

(Santa Cruz Biotechnology, sc-14408, 1:1000 dilution), EGFR (Santa Cruz Biotechnology, sc-03, 1:10000 dilution), and CHEK1 (Cell Signaling, #2360, 1:5000 dilution) were employed as primary antibodies for Western blot analysis. Horseradish peroxidase- conjugated anti-rabbit IgG, IRDye® 680LT Goat anti-Mouse IgG (H+L), and donkey anti- goat IgG-HRP were used as secondary antibodies. Membranes were also probed with anti- actin antibody (Cell Signaling #4967, 1:10000 dilution) to confirm equal loading of protein lysate.

Migration and invasion assay

Migration and invasion assays were performed using a Matrigel Transwell Chamber

(Corning) with 8-μm pore polycarbonate filters (21). In the migration assay, the harvested cells were suspended in 100 μl serum-free medium at a final concentration of 1-

2 × 106 cells/ml and were added to the upper chamber of the transwell system. Medium containing 10% FBS was placed in the lower chamber. After incubation for the indicated periods of time, non-migrated cells and media in the upper chamber were removed by using a cotton swab and the migrated cells on the bottom surface of the insert membrane were fixed by incubating with 75% methanol at room temperature for 15 min. The cells were then stained with 0.2% crystal violet in 10% ethanol for 15 min. After washing and drying,

111 the insert membranes were imaged under a light microscope. Since the same number of cells were initially suspended in each well, cell migration was represented by the number of migrated cells.

In the invasion assay, 200-400 μg/ml matrigel in serum-free media was coated on the upper chamber of the filter and the matrigel-containing medium was removed after incubation at 37°C for 2 h. The cells were then dispersed, stained, and counted in a similar way as described above for the migration assay. Cell invasion was calculated by dividing the number of invaded cells over that of the migrated cells.

Real-time quantitative PCR (RT-qPCR)

Cells were seeded in 6-well plates at 50-70% conﬂuence level. Total RNA was extracted from cells using TRI reagent (Sigma). Approximately 3 μg RNA was reverse transcribed by employing M-MLV reverse transcriptase (Promega) and an oligo(dT)18 primer. After a 60-min incubation at 42°C, the reverse transcriptase was deactivated by heating at 75°C for 5 min. RT-qPCR was performed using iQ SYBR Green Supermix kit

(Bio-Rad) on a Bio-Rad iCycler system (Bio-Rad), and the running conditions were at

95°C for 3 min and 45 cycles at 95°C for 15 sec, 55°C for 30 sec, and 72°C for 45 sec. The comparative cycle threshold (Ct) method (ΔΔCt) was used for the relative quantification of gene expression,(22) and the primer sequences are listed in Table 4.1.

Gelatin zymography assay

Concentrated medium collected from cultured CRC cells was loaded without reduction onto a 10% SDS-PAGE gel with 0.1% gelatin. After electrophoresis, the gels were washed

112 with 2.5% Triton X-100 (Sigma) to remove SDS and to renature MMP-2 and MMP-9. The gels were subsequently incubated in the developing buffer for overnight to induce gelatin lysis by the renatured MMP-2 and MMP-9. The relative amounts of active MMP-2 and

MMP-9 were then quantified based on their band intensities using ImageJ (23).

Results and Discussion

1. Differential Expression of Kinase Proteins in Primary and Metastatic Human CRC

Cells

To explore the potential roles of kinases in CRC metastasis, we employed a PRM-based targeted proteomic method (14). in conjunction with SILAC,(24) to examine the differential expression of kinases in a pair of matched primary/metastatic CRC cells, i.e.

SW480 and SW620 (Figure 4.1a), derived from the same CRC patient, respectively (25).

The results from the PRM-based targeted proteomic method led to the quantification of

299 unique kinases in the paired CRC cells, which included more than 200 protein kinases and covered approximately 50% of the human kinome (Figure 4.2). The same retention time with dot product (dotp) being above 0.7 was displayed for all PRM transitions (4-6) used in kinase peptide quantification (26). In addition, consistent trends were observed for more than 90% of the quantified kinase peptides in forward and reverse SILAC labeling experiments (Figure 4.1b&c, and representative results for AK2 and PRPS2 are shown in

Figure 4.3a). We also validated the relative protein expression levels of five quantified kinases (AK2, CHEK1, EGFR, IGF2R and PRPS2) in SW480 and SW620 cells by using

113

Western blot analyses (Figure 4.3). These results confirm that the PRM method, together with SILAC labeling, provided robust quantification of kinase protein expression.

Among the 299 quantified kinases, 83 and 77 were up- and down-regulated by at least

1.5-fold in the SW620 metastatic CRC cells relative to SW480 CRC cells, respectively.

Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis showed that 12 out of the 83 up-regulated kinases were involved in focal adhesion pathway (Figure 4.4a), which is closely associated with cancer metastasis (27). Furthermore , both SRC and CSK in the SRC subfamily, which were previously shown to promote invasion of CRC cells,(28) were up-regulated in the metastatic SW620 cells (Figure 4.2).

2.PRPS2 Drives CRC Metastasis

KEGG pathway analysis also reveals that 10 out of the 83 up-regulated kinases are involved in purine metabolism, rendering it unique among the up-regulated pathways

(Figure S3a). Moreover, a previous study on metabolic reprograming of CRC also suggested that purine metabolism is activated in late-stage CRC patients.(29) Among these kinases, the mRNA expression of PRPS2 gene is uniquely up-regulated in CRC cell lines among the more than 1,000 cell lines representing 37 different cancer types in the Cancer

Cell Line Encyclopedia (CCLE) database (20). suggesting an important role of PRPS2 in

CRC development (Figure 4.4b).

To further substantiate the above findings, we investigated the potential roles of PRPS2 in CRC metastasis by asking how the migration and invasion capacities of CRC cells are influenced by the expression levels of PRPS2 (21). Our results demonstrated that the

114 migratory and invasive abilities of SW480 primary CRC cells were augmented upon overexpression of PRPS2 (Figures 4.5). Reciprocally, shRNA-mediated stable knock- down of PRPS2 in SW620 metastatic CRC cells suppressed their motility and invasion

(Figure 4.5).

3. PRPS2 Modulates the Expression of MMP-9 and E-cadherin

Matrix metalloproteinases 2 and 9 (MMP-2 and MMP-9, a.k.a. gelatinases A and B) play central roles in degrading extracellular matrix (ECM) proteins and in promoting cancer metastasis (30). Hence, we also asked whether the enzymatic activities of MMP-2 and

MMP-9 are modulated by the expression levels of PRPS2. By employing gelatin zymography assay (23), we showed that the activity of secreted MMP-9 from SW480 and

SW620 cells were positively correlated with the expression level of PRPS2. In this regard, ectopic overexpression of PRPS2 in SW480 primary CRC cells gave rise to increased enzymatic activity of secreted MMP-9 (Figures 4.6a-b). Reciprocally, the activity of secreted MMP-9 was diminished in SW620 metastatic CRC cells upon stable knock-down of PRPS2 (Figures 4.6d-e). We also assessed, by employing RT-qPCR, how the mRNA expression levels of MMP2 and MMP9 genes in CRC cells are regulated by the expression levels of PRPS2. As shown in Figure 4.6c, ectopic overexpression of PRPS2 in SW480 primary CRC cells induced heightened mRNA expression of the MMP9 gene, whereas stable knock-down of PRPS2 in the SW620 metastatic CRC cells suppressed the expression of MMP9 gene (Figure 4.6f). However, the mRNA expression or activity of secreted MMP-2 is not positively modulated by PRPS2.

115

Aside from MMPs, epithelial-mesenchymal transition (EMT), a process through which cohesive epithelial cells are transformed to a migratory mesenchymal phenotype, can promote the invasion and metastasis of many types of cancer cells (31). Along this line,

EMT is accompanied with a loss of E-cadherin and a concomitant gain of N-cadherin (32).

E-cadherin, one of the most important molecules in cell-cell adhesion in epithelial tissues, is known to suppress malignant transformation.(33, 34) Our data suggest that the expression of CDH1 gene, which encodes E-cadherin, is inhibited by PRPS2 in CRC cells.

Overexpression of PRPS2 in SW480 primary CRC cells inhibited the transcription of

CDH1 gene, whereas stable knock-down of PRPS2 in the metastatic SW620 cells stimulated the expression of CDH1 gene (Figure 4.6c and f). Nevertheless, the expression of CDH2 gene, which encodes N-cadherin, is not affected by PRPS2. Together, PRPS2 drives CRC metastasis by inhibiting the transcription of E-cadherin and promoting the expression and activities of MMP-9.

4. PRPS2 and Myc

We next explored the potential mechanism through which the elevated expression of

PRPS2 stimulates CRC metastasis in cells. MYC, one of the most potent proto-oncogenes

(35, 36), can promote tumorigenesis in various organs through altering cellular metabolism.

PRPS2 is a crucial kinase in purine metabolism and protein synthesis (37-39). Moreover,

MYC-overexpressing cells often exhibit increased nucleotide biosynthesis through modulating the expression of PRPS2 (39, 40). In addition, elevated expression of MYC gene is known in CRC patients (41), and the metabolic alteration of CRC is also induced

116 by Myc (29). Hence, we reason that the expression of PRPS2 in CRC cells is modulated by Myc.

To examine the relationship between PRPS2 and Myc in CRC, we analyzed the expression of PRPS2 and MYC in all 56 CRC cell lines included in the CCLE database. It turned out that the expression levels of PRPS2 and MYC are well-correlated in CRC cell lines, including the SW480 and SW620 cells employed in the present study (Figure 4.7a).

In this vein, the expression of MYC gene was previously shown to be elevated in SW620 cells relative to SW480 cells (42), suggesting that the elevated expression of PRPS2 protein in SW620 cells could be modulated by Myc. Furthermore, the expression levels of MYC and PRPS2 genes are also correlated with each other in 435 CRC patients in the TCGA database (Figure 4.7b), indicating that Myc may modulate the expression of PRPS2 in CRC cells and patients.

Discussion

By employing a PRM-based targeted proteomic method, we were able to quantify the relative expression levels of 299 kinases in a pair of primary and metastatic CRC cells derived from the same patient. Importantly, we found that among the differentially expressed kinases, PRPS2 promotes the migration and invasion of cultured CRC cells.

Furthermore, we showed that elevated PRPS2 expression confers attenuated transcription of CDH1 and augmented expression and activities of MMP-9. Moreover, our work supports the role of MYC proto-oncogene in up-regulating the PRPS2 gene in metastatic

CRC. Together, we discovered PRPS2 as a novel promoter for CRC metastasis.

117

References

1. Haggar FA, Boushey RP. Colorectal Cancer Epidemiology: Incidence, Mortality, Survival, and Risk Factors. Clin Colon Rectal Surg. 2009;22:191-7. 2. Vatandoust S, Price TJ, Karapetis CS. Colorectal cancer: Metastases to a single organ. World J Gastroenterol. 2015;21:11767-76. 3. Van Cutsem E, Oliveira J, On behalf of the EGWG. Advanced colorectal cancer: ESMO Clinical Recommendations for diagnosis, treatment and follow-up. Ann Oncol. 2009;20:61-3. 4. Field K, Lipton L. Metastatic colorectal cancer-past, progress and future. World J Gastroenterol. 2007;13:3806-15. 5. Warsinggih W, Yusuf I, Manuaba IBTW, Pusponegoro A. Associations of positive epidermal growth factor receptor expression and K-RAS gene mutations with various clinicopathological parameters and survival of colorectal carcinoma patients. Biomarkers Genomic Med. 2015;7:1-7. 6. Uribe DJ, Mandell EK, Watson A, Martinez JD, Leighton JA, Ghosh S, et al. The receptor tyrosine kinase AXL promotes migration and invasion in colorectal cancer. PLoS One. 2017;12:e0179979. 7. Xiao Y, Guo L, Wang Y. A targeted quantitative proteomics strategy for global kinome profiling of cancer cells and tissues. Mol Cell Proteomics. 2014;13:1065-75. 8. Patricelli MP, Szardenings AK, Liyanage M, Nomanbhoy TK, Wu M, Weissig H, et al. Functional Interrogation of the Kinome Using Nucleotide Acyl Phosphates. Biochemistry. 2007;46:350-8. 9. Miao W, Xiao Y, Guo L, Jiang X, Huang M, Wang Y. A high-throughput targeted proteomic approach for comprehensive profiling of methylglyoxal-induced perturbations of the human kinome. Anal Chem. 2016;88:9773-9. 10. Duncan JS, Whittle MC, Nakamura K, Abell AN, Midland AA, Zawistowski JS, et al. Dynamic reprogramming of the kinome in response to targeted MEK inhibition in triple-negative breast cancer. Cell. 2012;149:307-21. 11. Stuhlmiller TJ, Miller SM, Zawistowski JS, Nakamura K, Beltran AS, Duncan JS, et al. Inhibition of lapatinib-induced kinome reprogramming in ERBB2-positive breast cancer by targeting BET family bromodomains. Cell Rep. 2015;11:390-404. 12. Oppermann FS, Gnad F, Olsen JV, Hornberger R, Greff Z, Kéri G, et al. Large-scale proteomics analysis of the human kinome. Mol Cell Proteomics. 2009;8:1751-64. 13. Urisman A, Levin RS, Gordan JD, Webber JT, Hernandez H, Ishihama Y, et al. An optimized chromatographic strategy for multiplexing In parallel reaction monitoring mass spectrometry: insights from quantitation of activated kinases. Mol Cell Proteomics. 2017;16:265- 77. 14. Miao W, Guo L, Wang Y. Imatinib-induced changes in protein expression and ATP- binding affinities of kinases in chronic myelocytic leukemia cells. Anal Chem. 2019;91:DOI: 10.1021/acs.analchem.9b00289. 15. Miao W, Li L, Wang Y. A targeted proteomic approach for heat shock proteins reveals DNAJB4 as a suppressor for melanoma metastasis. Anal Chem. 2018;90:6835-42. 16. Miao W, Li L, Wang Y. Identification of helicase proteins as clients for HSP90. Anal Chem. 2018. 17. Escher C, Reiter L, MacLean B, Ossola R, Herzog F, Chilton J, et al. Using iRT, a normalized retention time for more targeted measurement of peptides. Proteomics. 2012;12:1111- 21.

118

18. MacLean B, Tomazela DM, Shulman N, Chambers M, Finney GL, Frewen B, et al. Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics. 2010;26:966-8. 19. Anaya J. OncoLnc: linking TCGA survival data to mRNAs, miRNAs, and lncRNAs. PeerJ Comput Sci. 2016;2:e67. 20. Barretina J, Caponigro G, Stransky N, Venkatesan K, Margolin AA, Kim S, et al. The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature. 2012;483:603-7. 21. Albini A, Benelli R. The chemoinvasion assay: a method to assess tumor and endothelial cell invasion and its modulation. Nat Protoc. 2007;2:504-11. 22. K.J. L, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2-DDCt method. Methods. 2001;25(1046-2023 (Print)):402-8. 23. Vandooren J, Geurts N, Martens E, Van den Steen PE, Opdenakker G. Zymography methods for visualizing hydrolytic enzymes. Nat Methods. 2013;10:211-20. 24. Ong S-E, Blagoev B, Kratchmarova I, Kristensen DB, Steen H, Pandey A, et al. Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics. Mol Cell Proteomics. 2002;1:376-86. 25. Leibovitz A, Stinson JC, McCombs WB, McCoy CE, Mazur KC, Mabry ND. Classification of Human Colorectal Adenocarcinoma Cell Lines. Cancer Res. 1976;36:4562. 26. de Graaf EL, Altelaar AF, van Breukelen B, Mohammed S, Heck AJ. Improving SRM assay development: a global comparison between triple quadrupole, ion trap, and higher energy CID peptide fragmentation spectra. J Proteome Res. 2011;10:4334-41. 27. Bijian K, Lougheed C, Su J, Xu B, Yu H, Wu JH, et al. Targeting focal adhesion turnover in invasive breast cancer cells by the purine derivative reversine. Br J Cancer. 2013;109:2810. 28. Sirvent A, Bénistant C, Pannequin J, Veracini L, Simon V, Bourgaux JF, et al. Src family tyrosine kinases-driven colon cancer cell invasion is induced by Csk membrane delocalization. Oncogene. 2009;29:1303. 29. Satoh K, Yachida S, Sugimoto M, Oshima M, Nakagawa T, Akamoto S, et al. Global metabolic reprogramming of colorectal cancer occurs at adenoma stage and is induced by MYC. Proc Natl Acad Sci U S A. 2017;114:E7697-E706. 30. Bauvois B. New facets of matrix metalloproteinases MMP-2 and MMP-9 as cell surface transducers: outside-in signaling and relationship to tumor progression. Biochim Biophys Acta. 2012;1825:29-36. 31. Thiery JP. Epithelial–mesenchymal transitions in tumour progression. Nat Rev Cancer. 2002;2:442. 32. Kalluri R, Weinberg RA. The basics of epithelial-mesenchymal transition. J Clin Invest. 2009;119:1420-8. 33. Pećina-Šlaus N. Tumor suppressor gene E-cadherin and its role in normal and malignant cells. Cancer Cell Int. 2003;3:17-. 34. Gumbiner BM. Cell Adhesion: The Molecular Basis of Tissue Architecture and Morphogenesis. Cell. 1996;84:345-57. 35. Stone J, de Lange T, Ramsay G, Jakobovits E, Bishop JM, Varmus H, et al. Definition of regions in human c-myc that are involved in transformation and nuclear localization. Mol Cell Biol 1987;7:1697. 36. Hemann MT, Bric A, Teruya-Feldstein J, Herbst A, Nilsson JA, Cordon-Cardo C, et al. Evasion of the p53 tumour surveillance network by tumour-derived MYC mutants. Nature. 2005;436:807.

119

37. Miller DM, Thomas SD, Islam A, Muench D, Sedoris K. c-Myc and Cancer Metabolism. Clin Cancer Res. 2012;18:5546-53. 38. Rochlitz CF, Herrmann R, de Kant E. Overexpression and Amplification of c-myc during Progression of Human Colorectal Cancer. Oncology. 1996;53:448-54. 39. Cunningham JT, Moreno MV, Lodi A, Ronen SM, Ruggero D. Protein and nucleotide biosynthesis are coupled by a single rate-limiting enzyme, PRPS2, to drive cancer. Cell. 2014;157:1088-103. 40. Mannava S, Grachtchouk V, Wheeler LJ, Im M, Zhuang D, Slavina EG, et al. Direct role of nucleotide metabolism in C-MYC-dependent proliferation of melanoma cells. Cell Cycle. 2008;7:2392-400. 41. Smith DR, Myint T, Goh HS. Over-expression of the c-myc proto-oncogene in colorectal carcinoma. Br J Cancer. 1993;68:407-13. 42. Grand CL, Powell TJ, Nagle RB, Bearss DJ, Tye D, Gleason-Guzman M, et al. Mutations in the G-quadruplex silencer element and their relationship to c-MYC overexpression, NM23 repression, and therapeutic rescue. Proc Natl Acad Sci U S A. 2005;102:516.

120

SW480/SW620 pair of CRC cells. (c) Correlation between the ratios of kinase protein expression in SW480/SW620 cells obtained from forward and reverse SILAC labeling experiments.

121

Cell Lysis SW480 Mix at 1:1 ratio Tryptic Digestion Light Amino Acids

Scheduled SW620 LC-PRM Heavy Amino Acids

Q1 Q2 Q3 100 100

80 80

60 60 Data Analysis

40 40

Intensity Relative

20 20 Abundance

Precursor Fragmentation Fragment 0 0 Selection Selection Retention Time Peptide Time Window

b c 2 1.5 F R 1

(R) 0.5 10 0

33 230 36 log -0.5 -1 -1.5 R2 = 0.63 -2 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 log (F) 10

122

Figure 4.2. Differential expression of kinase proteins in paired SW480/SW620 CRC cells.

Blue, red, and grey bars represent those kinases with ratios (SW480/SW620) that are <

0.67, > 1.5, and between 0.67 and 1.5, respectively.

BRD2 DCK ADRBK1 EPHA2 CMPK1 AAK1 MTOR ADRBK2 DYRK1A EPHB4 CPNE3 PFKP PDK1 AKT1 GSK3A EPHB6 CRIM1 PI4K2B PDK2 AKT2 GSK3B FLT4 DGUOK PI4KB PDK3 AKT3 MAPK1 FYN DLG1 PIK3C2A RAF1 CCND1 MAPK13 IGF1R DSTYK PIK3CA

Atypical RIOK1 CCND3 MAPK14 IGF2R DTYMK SMG1 PIK3CB EXOSC10 MAPK15 INSR EIF2AK2 CAMK2D PIK3R2 GRK6 MAPK3 ERCC2 CAMK2G ITK PIK3R4 LATS1 MAPK7 JAK1 ERCC3 PIP4K2A

CAMKK2 CMGC MASTL MAPK8 FASTKD1 CASK KIT PIP4K2B PDPK1 MAPK9 FASTKD2 CHEK1 LMTK2 PIP4K2C PKN1 PRKM14 LYN FASTKD3 CHEK2 TK PIP5K3 FASTKD5 PKN2 MAPKAPK2 PRPF4B MET PKMYT1 PRKAA1 MAPKAPK3 PRPK PDGFRB FKBP1A PNKP PRKACA MAPKAPK5 PRPS1 PTK2 FPGT POLR2A PRKACB MARK1 PRPS2 PTK7 GAK POLR2B PRKACG MARK2 PRPSAP1 ROR1 GALK1 POLR2C PRKAG1 PRPSAP2 SYK GALK2 CAMK MELK POLR2D PRKAG2 MKNK1 PRS6KA4 TK1 GOLGA5 POLR2E

AGC PRKCB NUAK1 SRPK1 TNK1 GTF2H1 POLR2G GTF2H3 PRKCD OBSCN SRPK2 YES1 POLR2H ARAF GTF2H4 PRKCE PHKA1 CIT POLR2I ILK GUK1 PRKCI PHKA2 MAP2K2 POLR2L IRAK1 HK1 PRKCZ PHKB MAP2K3 POMK KSR1 HK2 PRKD2 PHKG2 MAP2K7 PPIP5K2 LIMK1 HUS1 PRKD3 SIK2 MAP3K4 PPP4C LIMK2 IKBKAP PRKDC TRIO MAP4K1 Other RP2 MLK4 IKBKB ROCK1 CSNK1A1 MAP4K4

Other SCYL1 CSNK1D MLKL IPPK

ROCK2 MAP4K5 TKL SCYL2 PXK ITPK1 RPS6KA1 CSNK2A1 MINK1 SCYL3 RIOK3 ITPKB RPS6KA3 CSNK2A2 OXSR1 SPHK1

CK1 RIPK1 KCNH4 RPS6KB1 CSNK2B STE PAK1 SRC TTBK2 TGFBR1 MAGI3 RPS6KB2 PAK2 TBK1 VRK1 ZAK MAP2K1 RPS6KC1 PAK3 TBRG4 CDC42BPA ACTL8 MPP1 SGK3 PAK4 TFG CDC42BPB ADK NAGK STK38 SLK TJP2 CDK1 AKAP13 NEK6 STK38L STK10 TLK2 CDK13 AKAP9 NEK7 TRIM24 ADCK4 CDK16 STK24 ATRIP NIN TRIM27 ADCK5 CDK19 STK25 AURKA NME6 AK1 CDK2 STK26 AURKB NRBP1 TRIM28 AK2 CDK3 STK3 BMP2K NUP62 TTK AK3 CDK4 STK39 BUB1 PBK TWF1

AK4 CDK5 STK4 Other BUB1B PEAK1 UCK1 CMGC AK5 CDK7 TNIK CARD11 PFKFB2 UCK2 AK6 CDK8 CAD CAV2 PFKFB3 UCKL1 CCNH WEE1 Atypical AK9 CDK9 CSK PFKFB4 CHUK WNK1 ATM CLK3 TK DDR1 PFKL ATR CRKL EGFR CLP1 PFKM WNK4 0.1 1 10 0.1 1 10 0.1 1 10 0.1 1 10 0.1 1 10 0.1 1 10 RSW480/SW620 RSW480/SW620 RSW480/SW620 RSW480/SW620 RSW480/SW620 RSW480/SW620

123

Figure 4.3. AK2, PRPS2 IGF2R, EGFR and CHEK1 are regulated in metastatic CRC cells.

(a) PRM traces for the quantifications of AK2 and PRPS2 proteins in SW480/SW620 cells.

(a) Western blot for the validation of the expression levels of AK2 and PRPS proteins in the paired CRC cells. (c) Quantitative comparison of the ratios of AK2 obtained from PRM and Western blot analysis. (d) Quantitative comparison of the ratios of PRPS proteins obtained from PRM (for PRPS1 and PRPS2) and Western blot analysis (for PRPS1/2/3).

(e) Western blot for the validation of the expression levels of IGF2R, EGFR and CHEK1 in the paired CRC cells. (f) Quantitative comparison of ratios of IGF2R, EGFR and CHEK1 obtained from PRM and Western blot analysis. The data represent the mean ± S. D. of the quantification results (n=3).

124

a SW480 SW480 b SW620 SW620 1 1 AK2

0.5 0.5 PRPS1/2/3 0 0 50 52 47 49 1 1 AK2 PRPS2 0.5 0.5 Actin 0 0

RelativeProtein Expression 63 65 58 60 Retention Time

c d

5 4 SW480 SW620 SW480 SW620 4 3 3 2 2 1 1 0 0 Western PRM PRPS1 PRPS2 PRPS1/2/3

Relative AK2ExpressionRelative blot PRM PRM WB RelativePRPS Expression a b e f IGF2R EGFR CHEK1 4 30 0.8

0.6 IGF2R 20 EGFR 2 0.4 10

CHEK1 0.2 (SW480/SW620) ACTIN 0 0 0 PRM Western PRM Western PRM Western RelativeProtein Expression blot blot blot

125

126 a Up-regulated in SW620

T cell receptor signaling pathway Fc epsilon RI signaling pathway Pathways in cancer ErbB signaling pathway MAPK signaling pathway Purine metabolism Insulin signaling pathway Neurotrophin signaling pathway Focal adhesion Chemokine signaling pathway 0 2 4 6 8 10 12 14 b

D a ta 1

1 0 ***

(PRPS2mRNA) 2

-5 log

r r e e c c n n a a C C - l n ta a c P re lo o C

127

***, p < 0.001.

128

a b

PRPS1/2/3 PRPS1/2/3

Actin Actin

c ac shControl shPRPS2db -1 shPRPS2-2 ** Control PRPS2 1.5

1 Migration Migration 0.5 Migration 0 Relative Control PRPS2

5 *** 4 3 Invasion Invasion 2 1 Cell Invasion 0 ce Control PRPS2

2 # 2.5 2 1.5 * 1.5 ** Migration 1 * 1 0.5 Cell Invasion 0.5

Relative 0 0 shControl shPRPS2-1 shPRPS2-2 shControl shPRPS2-1 shPRPS2-2

129

MMP-2 and MMP-9 by PRPS2 in SW480 cells. (c) RT-qPCR results showing the changes in mRNA levels of CDH2 (encoding N-cadherin), CDH1 (encoding E-cadherin), MMP2 and MMP9 genes in SW480 cells upon ectopic overexpression of PRPS2. (d) Gelatin zymography assay showing the changes in activities of secreted MMP-2 and MMP-9 after shRNA-mediated stable knock-down of PRPS2 gene in SW620 cells. (e) Modulation of activities of secreted MMP-2 and MMP-9 by PRPS2 in SW620 cells. (f) RT-qPCR results showing the modulation in mRNA levels of CDH2, CDH1, MMP2 and MMP9 genes in

SW620 cells upon siRNA-mediated knockdown of PRPS2. The data represent the mean ±

S. D. of the quantification results (n=3). The p values were calculated based on unpaired, two-tailed Student’s t-test: #, p ≥ 0.05; *, 0.01 ≤ p < 0.05; **, 0.001 ≤ p < 0.01; ***, p <

0.001.

130

a b c 2 Control PRPS2 ** 5 Control PRPS2 *** 1.5 # 100 4 MMP-9 3 1 75 ** * 2 # 0.5

MMP-2 1 SecretedMMPs

50 0 RelativeLevels mRNA

Relative ActivitiesRelativeof 0 MMP-2 MMP-9 CDH2 CDH1 MMP-2 MMP-9

d e f 2.5 shControl shPRPS2-1 2 # shControl 2 ** shPRPS2-2 shPRPS2-1 # shPRPS2-2 # * # 1.5 1.5 # * * 100 MMP-9 1 1 * **

75 0.5 ** 0.5 SecretedMMPs MMP-2 0 RelativeLevels mRNA 0 50 ActivitiesRelativeof MMP-2 MMP-9 CDH2 CDH1 MMP-2 MMP-9

131

Myc and PRPS2 genes obtained from 435 CRC patients in TCGA database.

a SW620 b

) 8 16

)

Myc Myc 6 14

4 12 SW480

2 10

(mRNA (mRNA ofLevel (mRNA (mRNA ofLevel

2 N=56 R=0.66 2 N=435

log R=0.37 0 log 8 0 2 4 6 8 8 10 12 14

log2 (mRNA Level of PRPS2) log2 (mRNA Level of PRPS2)

132

Table 4.1. Sequences for RT-qPCR primers.

Gene Forward Primer Reverse Primer Name MMP-2 5- ATACAGGATCATTGGCTACACACCT-3 5- CCAAAGTTGATCATGATGTCTGCC-3 MMP-9 5- ACCAAGGATACAGTTTGTTCCTCGT-3 5- TAGAGGTGCCGGATGCCATTCACGT-3 CDH2 5-GCTTGTCAGGATCAGGTCTGA-3 5-TACTGCATGTGCCCTCAAAT-3 CDH1 5-ACAGACAATGGTTCTCCAGTT-3 5-TGATAGATTCTTGGGTTGGGT-3 PRPS1 5-GCTTCTCAAATTCAGGGCTT-3 5-CAATGGAGGTCACTCTCTTA-3 PRPS2 5-AGGAGAGAGTCGTGCCCCAAT-3 5-AAGAATCCCTGTATCTGAGAAGCA-3 GAPDH 5-CCATGGAGAAGGCTGGGG-3 5-CAAAGTTGTCATGGATGACC-3

133

Chapter 5

A Targeted Proteomic Approach for Heat Shock Proteins

Reveals DNAJB4 as a Suppressor for Melanoma Metastasis

Introduction

Heat shock proteins (HSPs) are molecular chaperones that function in protein folding/unfolding, cell cycle regulation and protection of cells during stress (1). There are six major HSPs, namely, HSP70, HSP40, HSP60, HSP90, HSP110 and small HSP (2), and some of them were shown to correlate with the progression of multiple types of cancer. For instance, HSP90, a molecular chaperone for protein folding and client protein stabilization

(3), is overexpressed in many types of tumors and has been reported to associate with breast cancer progression (4). Thus, targeting HSP90 may inhibit multiple pro-invasive pathways

(5) and enhance cancer immunotherapy.(6) On the basis of these discoveries, 13 inhibitors for HSP90 have been developed for clinical evaluation in anti-cancer therapy (7). Among them, ganetespib has been under clinical trials for the treatment of non-small cell lung cancer (NSCLC) (8, 9) and breast cancer (10). Likewise, inhibitors for HSP27 and HSP70 have also been developed for clinical studies (9). Hence, comprehensive analysis of heat shock proteins will assist drug discovery and cancer treatment. Current methods for studying heat shock proteins rely on low-throughput Western blot analysis. While mass spectrometry has been widely employed for proteomic analysis (11), no analytical methods have yet been developed for proteome-wide interrogation of heat shock proteins.

134

Compared to the proteomic analysis in the data-dependent acquisition (DDA) mode, targeted proteomic methods, which involve LC-MS/MS analyses in multiple-reaction monitoring (MRM) or parallel-reaction monitoring (PRM) mode, exhibit much better sensitivity towards peptide detection (12), and thus have become extensively employed in quantitative proteomics studies (13). Owing to the high mass accuracy and resolution provided by the Orbitrap or time-of-flight mass analyzer, PRM provides better accuracy and specificity for quantifying analytes in complex sample matrices (14); hence, it has been widely used in bioanalysis, such as PTM detection (15) and metabolite quantification (16).

In this work, we developed a PRM-based targeted quantitative proteomic method to interrogate the human heat shock proteome, and we further applied the method for assessing the perturbations of HSPs during melanoma metastasis. Our results revealed many differentially expressed HSPs in paired primary/metastatic melanoma cells, including DNAJB4 that was uniformly down-regulated in the three metastatic lines of melanoma cells relative to the paired primary melanoma cells. We also demonstrated that

DNAJB4 suppressed melanoma metastasis by modulating the expression levels and activities of matrix metalloproteinases 2 and 9 (MMP-2 and MMP-9).

Materials and Methods

Cell culture

WM-115 and WM-266-4 cells (ATCC) were cultured in Eagle's Minimum Essential

Medium. IGR-39, IGR-37 (obtained from Prof. Peter H. Duesberg) (17), WM-793, 1205Lu

(Wistar Institute) cells were cultured in Dulbecco's Modified Eagle Medium. All culture

135 media were supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA) and penicillin (100 IU/mL). The cells were maintained at 37°C in a humidified atmosphere

7 containing 5% CO2. Approximately 2×10 cells were harvested, washed with cold PBS for three times, and lysed by incubating on ice for 30 min in CelLytic M (Sigma) cell lysis reagent containing 1% protease inhibitor cocktail. The cell lysates were centrifuged at

9,000g at 4°C for 30 min, and the resulting supernatants collected. WM-115 and WM-266-

4 cells were derived from the primary and metastatic sites of the same melanoma patient

(18). IGR-39 and IGR-37 cells were derived from the primary and metastatic sites of another melanoma patient (19). 1205Lu cells were initiated from a lung metastasis of WM-

793 human melanoma cells after subcutaneous injection into an immune-deficient

13 15 mouse.(20) Cells were cultured in SILAC medium containing [ C6, N2]lysine and

13 [ C6]arginine for at least 10 days to promote complete incorporation of isotope-labeled amino acids (21).

Plasmid and siRNAs

The sequences for siDNAJB4 was 5'-AACCCGGAAUGAGGAGAAGAA-3' (22). The coding sequences of DNAJB4 gene was amplified from a cDNA library from M14 cells by

PCR primers 5'-GCTCTAGAGCATTCGAAATGGGGAAA-3' and 5'-

CGGGATCCTCTTCATTCTATGAGGCA-3'. cDNA was subcloned into BamHI- and

XbaI-linearized pRK7 vector. siRNA was transfected using RNAiMAX (Invitrogen) following the manufacturer’s protocol, where non-targeting siRNA (Dharmacon, D-

001210-02-20) was used as control. pRK7-DNAJB4 and pRK7-empty vector were

136 transfected into WM-266-4, IGR-37 and 1205Lu cells by using Lipofectamine 2000 (Life

Technologies).

LC-PRM analysis

To assess the differential expression of heat shock proteins in primary and metastatic melanoma cells, we conducted one forward and one reverse SILAC labeling experiments, where lysates of light-labeled primary melanoma cells and heavy-labeled metastatic melanoma cells were combined at 1:1 ratio in the forward labeling experiments. The reverse labeling experiments were conducted in the opposite way. All LC-PRM experiments were performed on a Q Exactive Plus quadrupole-Orbitrap mass spectrometer.

The mass spectrometer was coupled with an EASY-nLC 1200 system (Thermo Scientific), and the samples were automatically loaded onto a 4-cm trapping column (150 µm i.d.) packed with ReproSil-Pur 120 C18-AQ resin (5 µm in particle size and 120 Å in pore size,

Dr. Maisch GmbH HPLC) at 3 µL/min. The trapping column was coupled to a 20-cm fused silica analytical column (PicoTip Emitter, New Objective, 75 µm i.d.) packed with

ReproSil-Pur 120 C18-AQ resin (3 µm in particle size and 120 Å in pore size, Dr. Maisch

GmbH HPLC). The peptides were then separated using a 140-min linear gradient of 9-38% acetonitrile in 0.1% formic acid and at a flow rate of 300 nL/min. The spray voltage was

1.8 kV. Precursor ions were isolated, at an isolation width of 1.0 m/z unit, and collisionally activated in the HCD cell with a collision energy of 29 to yield MS/MS.

137

PRM data analysis

All raw files were processed using Skyline (version 3.5) (23) for the generation of extracted-ion chromatograms and peak integration. We imposed a mass accuracy of within

20 ppm for fragment ions during the identification of peptides in the Skyline platform. The targeted peptides were manually checked to ensure that the transitions for multiple fragment ions derived from light and heavy forms of the same peptide exhibit the same elution time in the pre-selected retention time window. The data were then processed to ensure that the distribution of the relative intensities of multiple transitions associated with the same precursor ion correlates with the theoretical distribution in the MS/MS spectral library entry, which was acquired from shotgun proteomic analysis. The sum of peak areas from all transitions of light or heavy forms of peptides was used for quantification. The

Skyline PRM library for heat shock proteins and the raw files for LC-PRM analyses of heat shock proteins for paired melanoma cells were deposited into PeptideAtlas with the identifier number of PASS01177 (http://www.peptideatlas.org/PASS/PASS01177).

TCGA data analysis

OncoLnc (24) was employed for Kaplan-Meier survival analysis of melanoma patients using The Cancer Genome Atlas data, where patients were stratified based on the expression levels of DNAJB4 gene being among the top (high group) and bottom (low group) quartiles, respectively. Differences in survival with logrank p-values being less than

0.05 were considered significant.

138

Western blot

Melanoma cells were cultured in a 6-well plate and the cells were lysed at 40-50% confluency following the above-described procedures. The concentrations of proteins in the resulting lysates were determined using Bradford Assay (Bio-Rad). The whole cell lysate for each sample (10 μg) was denatured by boiling in Laemmli loading buffer and subjected to SDS-PAGE separation. The proteins were subsequently transferred to a nitrocellulose membrane at 4°C overnight. The resulting membrane was blocked with

PBS-T (PBS with 0.1% Tween 20) containing 5% milk (Bio-Rad) at 4°C for 6 h. The membrane was then incubated with primary antibody at 4°C overnight and subsequently with secondary antibody at room temperature for 1 h. After thorough washing with PBS-

T buffer, the HRP signals were detected with Pierce ECL Western Blotting Substrate

(Thermo).

Antibodies recognizing human DNAJB4 (Santa Cruz Biotechnology, sc-100711, 1:2000 dilution), DNAJC3 (Santa Cruz Biotechnology, sc-393559, 1:1000 dilution), and HSP40

(Santa Cruz Biotechnology, sc-398766, 1:1000 dilution) were employed as primary antibodies. Horseradish peroxidase-conjugated anti-rabbit IgG, IRDye® 680LT Goat anti-

Mouse IgG (1:10000 dilution) were used as secondary antibodies. Membranes were also probed with anti-actin antibody (Cell Signaling #4967, 1:10000 dilution) to confirm equal protein loading.

139

Migration and invasion assay

Migration and invasion assay was performed using a Matrigel Transwell Chamber

(Corning) with 8 μm pore polycarbonate filters. For migration assay, the harvested cells were suspended in 100 μl serum-free media at a final concentration of 3-8 × 105 cells/ml and were added to the upper chambers of the transwell system. Medium containing 10%

FBS was placed in the lower chamber. After incubation for the indicated periods of time, non-migrated cells and media on the upper chamber were removed by using cotton swab and the migrated cells on the bottom surface of the insert membrane were fixed by incubating with 75% methanol at room temperature for 15 min. The cells were then stained with 0.2% crystal violet in 10% ethanol for 15 min. After washing and drying, the insert membranes were imaged under a light microscope. For invasion assay, 200-400 μg/mL matrigel in serum-free media was coated on the upper-chamber of the filter and the matrigel-containing medium was removed after a 2-h incubation at 37°C. The cells were then dispersed, stained, and counted in a similar way as described for the migration assay.

Since the same number of cells was suspended in each well, cell migration was determined based on the number of migrated cells and cell invasion was calculated by dividing the number of invaded cells with that of the migrated cells.

Gelatin zymography assay

Concentrated medium collected from cultured cells was loaded onto 10% SDS-PAGE gels containing 0.1% gelatin under non-reducing conditions. After electrophoresis, the gels were washed with 2.5% Triton X-100 (Sigma) to remove SDS and to renature the MMP-2

140 and MMP-9 proteins. The gels were subsequently incubated in a developing buffer with 1%

Triton X-100 and 1 µM Zn2+ overnight to induce gelatin lysis by the renatured MMP-2 and

MMP-9 (25). The relative amounts of MMP-2 and MMP-9 were then quantified based on their band intensities using ImageJ. Images were inverted using online image inversion tool

(http://pinetools.com/invert-image-colors).

Real-time PCR

Melanoma cells were seeded in 6-well plates at 50% conﬂuence level. Total RNA was extracted from cells using TRI Reagent (Sigma). Approximately 3 μg RNA was reverse transcribed by employing M-MLV reverse transcriptase (Promega) and an oligo(dT)16 primer. After a 60-min incubation at 42°C, the reverse transcriptase was deactivated by heating at 85°C for 5 min. Quantitative real-time PCR was performed using iQ SYBR

Green Supermix kit (Bio-Rad) on a Bio-Rad iCycler system (Bio-Rad), and the running conditions were at 95°C for 3 min and 45 cycles at 95°C for 15 sec, 55°C for 30 sec, and

72°C for 45 sec. The comparative cycle threshold (Ct) method (ΔΔCt) was used for the relative quantification of gene expression (26). Primers are listed in Table 5.1. The mRNA level of each gene was normalized to that of the internal control (GAPDH).

Results

1. Development of a High-throughput PRM Method for the Quantitative Analysis of

Heat Shock Proteome

Construction of PRM Library for Heat Shock Proteins – The major objective of the present study was to develop a high-throughput analytical method, relying on parallel-reaction

141 monitoring (PRM), for profiling quantitatively the human heat shock proteome. To this end, we first constructed a Skyline PRM library using the retention time, MS and MS/MS of peptides of heat shock proteins retrieved from shotgun proteomic analyses of the tryptic digestion mixtures of 10 unique cell lines derived from different human tissue origins.

These data were obtained from more than 200 LC-MS/MS runs and led to the identification of 11,879 protein groups.

Owing to the high level of similarity in protein sequences from some heat shock proteins, we inspected manually all the identified peptides and incorporated only the unique peptides representing individual heat shock proteins into the library. By including a maximum of five unique peptides for any given heat shock proteins into the library, our current PRM

HSP library encompassed 180 unique peptides from 57 distinct human heat shock proteins

(Figure 5.1a). This covers approximately 70% of human heat shock proteome, which contain a total of 84 proteins (27). Among the 57 heat shock proteins in the library, 33, 9 and 5 belong to HSP40, HSP70 and HSP90 groups, respectively (Figure 5.1b).

Retention time calibration - To achieve high-throughput detection of HSP peptides, we adopted scheduled PRM analysis, where the mass spectrometer was programmed to acquire the MS/MS of the precursor ions for a limited number of peptides in each 10-min retention time (RT) window. To achieve this, we calculated the normalized RT (iRT) value for each peptide on our target list following a previously published method (28, 29). By using 10 tryptic peptides of bovine serum albumin (BSA) as standards, we successfully converted the experimentally determined retention times of the 180 peptides from heat shock proteins into normalized iRT scores. The iRT value represents an intrinsic property

142

(i.e. hydrophobicity) of a peptide. Hence, a substantial deviation of measured RT from that projected from the linear plot of RT over iRT is considered a false-positive detection, which is employed as a criterion to validate the results obtained from the PRM assay. The linear

RT vs. iRT relationship was redefined after every 5-7 LC-PRM runs through the analysis of the tryptic digestion mixture of BSA.

2. Scheduled LC-PRM Analysis Revealed Differential Expression of Heat Shock

Proteins during Melanoma Metastasis

To assess the reprogramming of heat shock proteome during melanoma metastasis, we employed LC-MS/MS in the PRM mode, together with metabolic labeling using SILAC

(21), to examine the differential expression of HSPs in three matched primary/metastatic melanoma cells (i.e. WM-115/WM-266-4, IGR-39/IGR-37 and WM-793/1205Lu. Figure

5.1c). As shown in Figure 5.2a, 48 unique HSPs were quantified in WM-115/WM-266-4 paired melanoma cell lines by our PRM-based targeted proteomics method (Figure 5.2a).

All the quantified peptides for heat shock proteins exhibit an excellent linear fit between the observed retention time and iRT in the library. Additionally, all 4-6 transitions used for quantification of each peptide from heat shock proteins were eluted at the same retention time with a dot product (dotp) of > 0.7 when compared to the same fragment ions found in the MS/MS acquired from shotgun proteomic analysis (30), suggesting that our method is highly reliable for peptide identification. Furthermore, all the quantified heat shock proteins appeared in both forward and reverse SILAC labeling experiments (Figure 5.2b).

The ratios of quantified peptides obtained from forward and reverse SILAC labeling experiments exhibited an excellent linear fit (Figure 5.2c) and displayed a strong

143 correlation, supporting the reproducibility of the analytical method. Moreover, the reproducibility of the method is reflected by the observation that consistent ratios were obtained for different tryptic peptides derived from the same heat shock proteins. In this regard, the average relative standard deviations (RSD) among the different quantified peptides from the same heat shock proteins were 11.1% for the data acquired for the WM-

115/WM-266-4 paired melanoma cell lines.

We also analyzed the SILAC samples from paired WM-115/WM-266-4 melanoma cells by using LC-MS/MS in the DDA mode. After pre-fractionation using a strong cation- exchange (SCX) column (31), 20 fractions were subjected to LC-MS/MS analysis in the

DDA mode. The results from this analysis only led to the quantification of 36 heat shock proteins. While our scheduled PRM method allowed for the quantification of 48 heat shock proteins in two LC-PRM runs without pre-fractionation, demonstrating the superior sensitivity and throughput of the PRM method.

A total of 43 and 44 unique heat shock proteins were quantified in IGR-39/IGR-37 and

WM-793/1205Lu paired melanoma cells, respectively. The reliability and reproducibility of the quantification results were similar to those obtained from the WM-115/WM-266-4 paired melanoma cells. Our quantification results showed that 7, 10 and 20 heat shock proteins were up-, and 18, 8 and 5 were down-regulated in primary (WM-115, IGR-39 and

WM-793 cells, respectively) relative to the corresponding metastatic (WM-266-4, IGR-37 and 1205Lu) melanoma cells (Figure 5.2a, Figure 5.3 and Figure 5.2d). We also assessed the differential expression of DNAJB4 in paired WM-115/WM-266-4 melanoma cells by using Western blot analysis. The ratios obtained from Western blot are in keeping with

144 those obtained from PRM analyses (Figure 5.4a), demonstrating that the PRM method is capable of profiling accurately the differential expression of heat shock proteins. In the meantime, we found that HSPB1 (HSP27), which was previously shown to suppress the invasive ability and the activities of secreted matrix metalloproteinases (MMPs) in A375 malignant melanoma cells (32), is up-regulated in the two primary melanoma cells, WM-

115 and IGR-39 (Figure 5.2d).

3. DNAJB4 is Commonly Down-regulated in Metastatic Melanoma Cells and It

Modulates the Invasive Capabilities of Cultured Melanoma Cells

As noted above, our PRM method has revealed the differential expression of a known suppressor for melanoma metastasis, i.e., HSPB1. We next asked whether any other differentially expressed heat shock proteins may act as drivers or suppressors for melanoma metastasis. In this vein, we found that DNAJB4 was down-regulated in all three metastatic melanoma cells relative to the corresponding primary melanoma cells, which we validated by Western blot analysis (Figure 5.4a-c). In addition, Kaplan-Meier survival analysis of

The Cancer Genome Atlas (TCGA) data (24) showed poorer prognosis for those melanoma patients with lower levels of mRNA expression of the DNAJB4 gene (Figure 5.4d), indicating that DNAJB4 may suppress melanoma metastasis. Along this line, DNAJB4 was previously shown to be a suppressor for lung cancer metastasis (33).

To explore the potential roles of DNAJB4 in melanoma metastasis, we next examined how the migratory and invasive abilities of WM-115 and WM-266-4 cells are modulated by the expression level of DNAJB4. Our results showed that, after knocking down the

145 expression of DNAJB4 using siRNA (Figure 5.5a), the invasive ability of WM-115 cells increased significantly (Figure 5.5b). Reciprocally, ectopic overexpression of DNAJB4 led to diminished invasive ability of WM-266-4 cells (Figure 5.5b). However, no significant alterations in the migratory abilities were observed for WM-115 or WM-266-4 cells upon genetic manipulation of the expression level of DNAJB4 (Figure 5.5c). Likewise, we found that ectopic overexpression of DNAJB4 led to significantly diminutions in invasive abilities of the two other metastatic melanoma lines (i.e. IGR-37 and 1205Lu cells, Figure

5.5b), though no apparent alteration in migratory abilities was observed (Figure 5.5c). In addition, siRNA-mediated knock-down of DNAJB4 in the two other lines of primary melanoma cells (i.e. IGR-39 and WM-793) elicited marked elevations in migration and invasion abilities (Figure 5.5b-c). Cumulatively, the above results suggested that DNAJB4 suppresses melanoma metastasis in cultured cells.

4. DNAJB4 Suppresses Melanoma Cell Invasion by Regulating Matrix

Metalloproteinases (MMPs)

Having demonstrated the role of DNAJB4 in suppressing the invasive capabilities of melanoma cells, we next explored the mechanisms through which the invasive capacities of melanoma cells are regulated by DNAJB4. Matrix metalloproteinases (MMPs) are cancer-associated, secreted, zinc-dependent endopeptidases that assume important roles in degrading extracellular matrix (ECM) proteins and in promoting cancer metastasis (34).

Moreover, metastatic cancer cells often display augmented levels of MMPs relative to primary tumor cells. Among the human MMPs, MMP-2 (gelatinase A) and MMP-9

146

(gelatinase B) are two major proteases responsible for remodeling the ECM environment and facilitating cancer metastasis (35).

To examine the potential involvements of MMP-2 and MMP-9 in DNAJB4-mediated alterations in invasive capabilities of melanoma cells, we assessed, by using gelatin zymography assay, how the activities of secreted MMPs in melanoma cells are affected by the expression level of DNAJB4. Our results demonstrated that the activities of secreted

MMP-2 and MMP-9 decreased (Figure 5.6) in the three lines of metastatic melanoma cells

(WM-266-4, IGR-37 and 1205Lu) with ectopic overexpression of DNAJB4. On the other hand, siRNA-mediated knockdown of DNAJB4 led to marked elevations in the activities of MMP-2 and MMP-9 in the three primary melanoma cells (WM-115, IGR-39 and WM-

793. Figure 5.6), suggesting that DNAJB4 inhibits the activity of secreted MMPs.

We also assessed, by employing qRT-PCR, how the mRNA expression levels of MMP2 and MMP9 genes in melanoma cells are regulated by the expression levels of DNAJB4.

Our results showed that ectopic overexpression of DNAJB4 in the three lines of metastatic melanoma cells (WM-266-4, IGR-37 and 1205Lu) suppressed the expression of MMP2 and MMP9 genes (Figure 5.7). siRNA-mediated knock-down of DNAJB4 in the primary melanoma cells (WM-115, IGR-39 and WM-793), however, induced heightened mRNA expression levels of MMP2 and MMP9 (Figure 5.7). Together, DNAJB4 may suppress melanoma metastasis by inhibiting the transcription of genes encoding MMP-2 and MMP-

9 and by diminishing the activities of secreted MMPs.

147

Discussion

In this study, we developed, for the first time, a PRM-based targeted quantitative proteomic method for the comprehensive analysis of the heat shock proteome in cultured human cells. Our PRM library contained 57 heat shock proteins, which encompassed approximately 70% of the human heat shock proteome. We showed that the method exhibited higher throughput and superior sensitivity than the shotgun proteomic method.

We also applied this method to investigate the reprogramming of heat shock proteome during melanoma metastasis by analyzing the differential expression of heat shock proteins in three matched pairs of primary/metastatic melanoma cell lines, WM-115/WM-266-4,

IGR-39/IGR-37 and WM-793/1205Lu. HSPB1 (HSP27), a previously reported suppressor for melanoma metastasis (32), was found to be down-regulated in two metastatic melanoma cells based on PRM analysis. Moreover, DNAJB4, whose elevated expression confers better survival in melanoma patients, was found to be consistently down-regulated in all three lines of metastatic melanoma cells. We also demonstrated that DNAJB4 suppressed melanoma cell invasion by inhibiting the expression and activity of MMP-2 and MMP-9.

DNAJB4 was found to interact with AP-2α to repress the expression of AP-2α target genes in lung cancer cells (36), and MMP2 (36) and MMP9 (37) genes could be transcriptionally regulated by AP-2α. Our RT-qPCR result showed that the mRNA levels of MMP2 and MMP9 genes could be modulated by the expression levels of DNAJB4 in all three pairs of melanoma cells. Thus, the suppressed expression of the MMP2 and MMP9

148 genes by DNAJB4 could be attributed to the binding between DNAJB4 and AP-2α in melanoma cells.

Secretion of MMP-9 was shown to be regulated by SRC tyrosine kinase (38), which is known to promote the metastatic transformations of different types of cancers (39), including melanoma (40). On the grounds of our observation that diminished expression of DNAJB4 could elicit increased secretion of MMP-9 and the previous finding that

DNAJB4 could act as an endogenous inhibitor for SRC (33), we reason that the increased secretion of MMP-9 in WM-793 cells may arise from the elevated activity of SRC induced by siRNA-mediated knockdown of DNAJB4.

Aside from DNAJB4, several other heat shock proteins were found to be differentially expressed in primary/metastatic melanoma cells. These proteins could potentially play critical roles in melanoma metastasis. For instance, DNAJC3, which is commonly up- regulated in the three lines of metastatic melanoma cells, is known to inhibit the phosphorylation of eIF2α (41). The phosphorylation of eIF2α has been demonstrated to be associated with cancer progression (42); therefore, DNAJC3 could also be a potential driver for melanoma metastasis. Melanoma is one of the most common cancers in the

United States, and metastasis contributes to the mortality of the majority of melanoma patients (43). Our targeted proteomics method led to the discovery of novel potential promoters or suppressors of melanoma metastasis, which provided important new knowledge for understanding the etiology of melanoma progression.

149

Taken together, we developed, for the first time, a high-throughput and robust PRM- based targeted proteomic method for the quantitative analysis of the human heat shock proteome. It can be envisaged that the method can be generally applicable for assessing how cells respond to extracellular stimuli (e.g. upon exposure to environmental toxicants or heat shock protein inhibitors) by altering the expression of heat shock proteins.

150

References

1. Feder ME, Hofmann GE. Heat shock proteins, molecular chaperones, and the stress response: evolutionary and ecological physiology. Annu Rev Physiol. 1999;61:243-82. 2. Wu J, Liu T, Rios Z, Mei Q, Lin X, Cao S. Heat Shock Proteins and Cancer. Trends in Pharmacological Sciences. 2016;38:226-56. 3. Trepel J, Mollapour M, Giaccone G, Neckers L. Targeting the dynamic HSP90 complex in cancer. Nat Rev Cancer. 2010;10:537. 4. Pick E, Kluger Y, Giltnane JM, Moeder C, Camp RL, Rimm DL, et al. High HSP90 expression is associated with decreased survival in breast cancer. Cancer Res. 2007;67:2932. 5. Koga F, Kihara K, Neckers LEN. Inhibition of cancer invasion and metastasis by targeting the molecular chaperone heat-shock protein 90. Anticancer Res. 2009;29:797-807. 6. Mbofung RM, McKenzie JA, Malu S, Zhang M, Peng W, Liu C, et al. HSP90 inhibition enhances cancer immunotherapy by upregulating interferon response genes. Nature Communications. 2017;8:451. 7. Katerina S, Evangelia P. HSP90 inhibitors: current development and potential in cancer therapy. Recent Patents on Anti-Cancer Drug Discovery. 2014;9:1-20. 8. Ramalingam S, Goss G, Rosell R, Schmid-Bindert G, Zaric B, Andric Z, et al. A randomized phase II study of ganetespib, a heat shock protein 90 inhibitor, in combination with docetaxel in second-line therapy of advanced non-small cell lung cancer (GALAXY-1). Annals of Oncology. 2015;26:1741-8. 9. Hendriks LEL, Dingemans A-MC. Heat shock protein antagonists in early stage clinical trials for NSCLC. Expert Opinion on Investigational Drugs. 2017;26:541-50. 10. Jhaveri K, Chandarlapaty S, Lake D, Gilewski T, Robson M, Goldfarb S, et al. A phase II open-Label study of ganetespib, a novel heat shock protein 90 inhibitor for patients with metastatic breast cancer. Clinical Breast Cancer. 2014;14:154-60. 11. Skultety L, Hernychova L, Toman R, Hubalek M, Slaba K, Zechovska J, et al. Coxiella burnetii whole cell lysate protein identification by mass spectrometry and tandem mass spectrometry. Ann N Y Acad Sci. 2005;1063:115-522. 12. Lange V, Picotti P, Domon B, Aebersold R. Selected reaction monitoring for quantitative proteomics: a tutorial. Mol Syst Biol. 2008;4:222. 13. Marx V. Targeted proteomics. Nat Meth. 2013;10:19-22. 14. Peterson AC, Russell JD, Bailey DJ, Westphall MS, Coon JJ. Parallel reaction monitoring for high resolution and high aass accuracy quantitative, targeted Proteomics. Mol Cell Proteomics. 2012;11:1475-88. 15. Bilan V, Selevsek N, Kistemaker HAV, Abplanalp J, Feurer R, Filippov DV, et al. New quantitative mass spectrometry approaches reveal different ADP-ribosylation phases dependent on the levels of oxidative stress. Mol Cell Proteomics. 2017;16:949-58. 16. Zhou J, Liu H, Liu Y, Liu J, Zhao X, Yin Y. Development and evaluation of a parallel reaction monitoring strategy for large-scale targeted metabolomics quantification. Anal Chem. 2016;88:4478-86. 17. Bloomfield M, Duesberg P. Inherent variability of cancer-specific aneuploidy generates metastases. Mol Cytogenet. 2016;9:90. 18. Westermark B, Johnsson A, Paulsson Y, Betsholtz C, Heldin CH, Herlyn M, et al. Human melanoma cell lines of primary and metastatic origin express the genes encoding the chains of platelet-derived growth factor (PDGF) and produce a PDGF-like growth factor. Proc Natl Acad Sci USA. 1986;83:7197-200.

151

19. Martin J-M, Luis J, Marvaldi J, Pichon J, Pic P. A human melanoma-derived cell line (IGR39) with a very high number of vasoactive-intestinal-peptide (VIP) receptors. Eur J Biochem. 1989;180:435-9. 20. Simon H-G, Risse B, Jost M, Oppenheimer S, Kari C, Rodeck U. Identification of differentially expressed messenger RNAs in human melanocytes and melanoma cells. Cancer Res. 1996;56:3112. 21. Ong S-E, Blagoev B, Kratchmarova I, Kristensen DB, Steen H, Pandey A, et al. Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics. Mol Cell Proteomics. 2002;1:376-86. 22. Tsai M-F, Wang C-C, Chang G-C, Chen C-Y, Chen H-Y, Cheng C-L, et al. A new tumor suppressor DnaJ-like heat shock protein, HLJ1, and survival of patients with non small cell lung carcinoma. JNCI: Journal of the National Cancer Institute. 2006;98:825-38. 23. MacLean B, Tomazela DM, Shulman N, Chambers M, Finney GL, Frewen B, et al. Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics. 2010;26:966-8. 24. Anaya J. OncoLnc: linking TCGA survival data to mRNAs, miRNAs, and lncRNAs. PeerJ Comput Sci. 2016;2:e67. 25. Vandooren J, Geurts N, Martens E, Van den Steen PE, Opdenakker G. Zymography methods for visualizing hydrolytic enzymes. Nat Methods. 2013;10:211-20. 26. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2−ΔΔCT method. Methods. 2001;25:402-8. 27. Kampinga HH, Hageman J, Vos MJ, Kubota H, Tanguay RM, Bruford EA, et al. Guidelines for the nomenclature of the human heat shock proteins. Cell Stress & Chaperones. 2009;14:105-11. 28. Escher C, Reiter L, MacLean B, Ossola R, Herzog F, Chilton J, et al. Using iRT, a normalized retention time for more targeted measurement of peptides. Proteomics. 2012;12:1111- 21. 29. Miao W, Xiao Y, Guo L, Jiang X, Huang M, Wang Y. A high-throughput targeted proteomic approach for comprehensive profiling of methylglyoxal-induced perturbations of the human kinome. Anal Chem. 2016;88:9773-9. 30. de Graaf EL, Altelaar AF, van Breukelen B, Mohammed S, Heck AJ. Improving SRM assay development: a global comparison between triple quadrupole, ion trap, and higher energy CID peptide fragmentation spectra. J Proteome Res. 2011;10:4334-41. 31. Marchetti N, Fairchild JN, Guiochon G. Comprehensive off-Line, two-dimensional liquid chromatography. application to the separation of peptide digests. Anal Chem. 2008;80:2756-67. 32. Aldrian S, Trautinger F, Fröhlich I, Berger W, Micksche M, Kindas-Mügge I. Overexpression of Hsp27 affects the metastatic phenotype of human melanoma cells in vitro. Cell Stress & Chaperones. 2002;7:177-85. 33. Chen CH, Chang WH, Su KY, Ku WH, Chang GC, Hong QS, et al. HLJ1 is an endogenous Src inhibitor suppressing cancer progression through dual mechanisms. Oncogene. 2016;35:5674- 85. 34. Bauvois B. New facets of matrix metalloproteinases MMP-2 and MMP-9 as cell surface transducers: outside-in signaling and relationship to tumor progression. Biochim Biophys Acta. 2012;1825:29-36. 35. Niedzwiecki. Patterns of MMP-2 and MMP-9 expression in human cancer cell lines. Oncology Reports. 2009;21:1323-33.

152

36. Chang T-P, Yu S-L, Lin S-Y, Hsiao Y-J, Chang G-C, Yang P-C, et al. Tumor suppressor HLJ1 binds and functionally alters nucleophosmin via activating enhancer binding protein 2α complex formation. Cancer Research. 2010;70:1656-67. 37. Schwartz B, Melnikova VO, Tellez C, Mourad-Zeidan A, Blehm K, Zhao YJ, et al. Loss of AP-2alpha results in deregulation of E-cadherin and MMP-9 and an increase in tumorigenicity of colon cancer cells in vivo. Oncogene. 2007;26:4049-58. 38. Cortes-Reynosa P, Robledo T, Macias-Silva M, Wu SV, Salazar EP. Src kinase regulates metalloproteinase-9 secretion induced by type IV collagen in MCF-7 human breast cancer cells. Matrix Biol. 2008;27:220-31. 39. Homsi J, Cubitt C, Daud A. The Src signaling pathway: a potential target in melanoma and other malignancies. Expert Opin Ther Targets. 2007;11:91-100. 40. Hanna SC, Krishnan B, Bailey ST, Moschos SJ, Kuan PF, Shimamura T, et al. HIF1alpha and HIF2alpha independently activate SRC to promote melanoma metastases. J Clin Invest. 2013;123:2078-93. 41. Gao D, Bambang IF, Putti TC, Lee YK, Richardson DR, Zhang D. ERp29 induces breast cancer cell growth arrest and survival through modulation of activation of p38 and upregulation of ER stress protein p58IPK. Lab Invest. 2011;92:200. 42. Podszywalow-Bartnicka P, Cmoch A, Wolczyk M, Bugajski L, Tkaczyk M, Dadlez M, et al. Increased phosphorylation of eIF2α in chronic myeloid leukemia cells stimulates secretion of matrix modifying enzymes. Oncotarget. 2016;7:79706-21. 43. Zbytek B, Carlson JA, Granese J, Ross J, Mihm MC, Jr., Slominski A. Current concepts of metastasis in melanoma. Expert Rev Dermatol. 2008;3:569-85.

153

Experimental strategy for PRM-based targeted proteomic approach.

154

a b Human Heat Shock Proteins (84) sHSPs Other 2, 20% HSP110 5, 100% 2, 50% HSP90 5, 100% Quantified Heat Shock Proteins (~ 45) HSP70 9, 70% HSP40 HSP60 33, 60% Library Heat Shock Proteins (57) 1, 100%

Cell Lysis Tryptic Digestion Primary Melanoma Cells Mix at 1:1 ratio Light Amino Acids

LC-MS/MS Metastatic Melanoma Cells (Scheduled PRM) Heavy Amino Acids

Q1 Q2 Q3 100 100

80 80

60 60 Data Analysis

40 40 Relative

20 20 Abundance

Precursor Fragmentation Fragment 0 0 Selection Selection Retention Time

155

Figure 5.2. Performances of PRM-based targeted proteomic approach for interrogating the perturbations in expression of heat shock proteins during melanoma metastasis. (a)

Differential expression of heat shock proteins in WM-115/WM-266-4 paired melanoma cells. (b) A Venn diagram displaying the overlap between quantified heat shock proteins from the forward and reverse SILAC labelings of WM-115/WM-266-4 paired melanoma cells. (c) Correlation between the ratios obtained from forward and reverse SILAC labeling experiments. (d) A heatmap showing the differences in expression of heat shock proteins in 3 pairs of primary/metastatic melanoma cell lines. Genes were clustered according to

Euclidean distance. The data in (a) and (d) represent the mean of the results obtained from one forward and one reverse SILAC labeling results.

156 a

266 -

115/WM

- WM

0.1

HSP47

HSPA1 HSPB1 HSPA9 HSPA8 HSPA4 HSPE1 HSPA2

HSPH1 HSPD1

DNAJB4 DNAJB1 DNAJA4 DNAJA1 DNAJA3 DNAJB2 DNAJA2 DNAJB6

HSPA13 HSPA14

DNAJC7 DNAJC2 DNAJC6 DNAJC9 DNAJC5 DNAJC8 DNAJC3 DNAJC1 DNAJC2

HSPBP1

DNAJB14 HSP90B1 DNAJB11 DNAJB12

DNAJC30 DNAJC11 DNAJC19 DNAJC21 DNAJC14 DNAJC10 DNAJC25

HSPC030 HSPC125

HSPA12A

HSP90AB1 HSP90AA1

HSP90AA4P HSP90AB4P Heat Shock Proteins

b d F R

0 48 0

c 1.5

0.5 (R)

10 0

log -0.5

-1 R2 = 0.60 -1.5 -1.5 -1 -0.5 0 0.5 1 1.5

log10(F) log2(Ratio)

157

Figure 5.3. Differential expression of heat shock proteins in paired IGR-39/IGR-37 (a) and

WM793/1205Lu (b) primary/metastatic melanoma cells. The data represent the mean of results obtained from one forward and one reverse SILAC labeling experiments.

37 -

39/IGR

- IGR

0.1

HSP47

HSPB1 HSPA2 HSPA1 HSPA4 HSPA9 HSPA8 HSPE1

HSPH1 HSPD1

DNAJB6 DNAJA4 DNAJB4 DNAJB2 DNAJB1 HSPA14 DNAJA1 DNAJA3 DNAJA2

DNAJC9 DNAJC1 DNAJC7 DNAJC6 DNAJC8 DNAJC2 DNAJC3

HSPBP1

DNAJB14 DNAJB12 DNAJB11

HSP90B1

DNAJC21 DNAJC15 DNAJC14 DNAJC10 DNAJC11 HSPC030 HSPC125 DNAJC19

HSP90AA1 HSP90AB1

HSP90AB4P HSP90AA4P Heat Shock Proteins b

793/1205Lu

- WM

0.1

HSP47

HSPA4 HSPA9 HSPA8 HSPA1 HSPA2 HSPE1 HSPB1

HSPD1 HSPH1

DNAJB4 DNAJB1 DNAJB6 DNAJA3 HSPA14 DNAJA2 DNAJA1 DNAJA4

DNAJC2 DNAJC7 DNAJC5 DNAJC1 DNAJC9 DNAJC3 DNAJC6

HSPBP1

DNAJB11 HSP90B1 DNAJB14

DNAJC25 DNAJC17 DNAJC15 DNAJC30 DNAJC19 DNAJC10 DNAJC21 DNAJC11

HSPC030 HSPC125

HSPA12A

HSP90AB1 HSP90AA1

HSP90AB4P HSP90AA4P Heat Shock Proteins

158

Figure 5.4. DNAJB4 is down-regulated in metastatic melanoma cells. (a) Western blot for the validation of the expression levels of DNAJB4 in the three pairs of melanoma cells.

115, 266-4, 39, 37, 793 and 1205 denote WM-115, WM-266-4, IGR-39, IGR-37, WM-793 and 1205Lu cells. (b) PRM traces for the quantification of DNAJB4 protein in 3 pairs of melanoma cells from forward and reverse SILAC labeling experiments. (c) Quantitative comparison of ratios of DNAJB4 obtained from PRM and Western blot analysis (n=3). (d)

Kaplan-Meier survival analysis showing that higher levels of expression of DNAJB4 gene confers better prognosis of melanoma patients.

159

Forward Reverse a 115 266-4 b Primary Primary Metastatic Metastatic DNAJB4 1 1 WM-115/WM-266-4

Actin 0.5 0.5

39 37 0 0 45 46 45.5 46.5 1 DNAJB4 1 IGR-39/IGR-37 0.5 0.5 Actin 0 0 46 47 45.6 46.6 793 1205 1 1 Relative Abundance Relative WM-793/1205Lu DNAJB4 0.5 0.5

0 0 Actin 45.8 46.8 46.3 47.3 Retention Time (min) c d Survival of Data 1:Survival proportions

4 100 Bottom Quartile Western blot PRM 3.5 80 Top Quartile 3 2.5 60 2 1.5 40 1

0.5 20 Primary/Metastatic 0 survival Percent Logrank p-value=0.0479 0 0 2 4 6 8 10 Years

160

161 a WM-115 IGR-39 WM-793

DNAJB4

Actin ba siControl siControl Empty Control vecter 1.5 siDNAJB4 1.0 DNAJB4 siDNAJB4 DNAJB4 O. E. 0.8 1.0 0.6

0.4

0.5

Cell Invasion Cell InvasionCell Cell Invasion Cell 0.2

0.0 0.0 WM-115 IGR-39 WM-793 WM-266-4 IGR-37 1205Lu Melanoma Cell Melanoma Cells Melanoma Cell cb siControl siControl Empty Control vecter 12 1.2 DNAJB4 siDNAJB4 siDNAJB4 DNAJB4 O. E. 1.0 10 0.8

0.6

1 0.4 0.2

0 RelativeMigration Ability 0.0

Relative Migration Ability Migration Relative WM-115 IGR-39 WM-793 WM-266-4 IGR-37 1205Lu Melanoma Cell Melanoma Cells c Migration Invasion

siControl

siDNAJB4 4

162

Figure 5.6. DNAJB4 modulates the enzymatic activities of MMP-2 and MMP-9. (a)

Representative gelatin zymography assay result showing the changes in activities of secreted MMP-2 and MMP-9 in WM-793 cells upon treatment with control, non-targeting siRNA or siRNA targeting DNAJB4. (b) Quantification results showing the modulation of activities of secreted MMP-2 and MMP-9 in primary melanoma cells upon siRNA-induced knockdown of DNAJB4 (left) or in metastatic melanoma cells upon ectopic overexpression of DNAJB4 (right). The data represent the mean and standard deviation of results obtained from three parallel experiments, and were normalized to the results obtained for primary cells treated with control non-targeting siRNA (left) or metastatic melanoma cells treated with control empty vector (right).

163 a

MW (kDa)

100 MMP-9 75

MMP-2

50 b

MMP-2 MMP-2 MMP- MMP-29 MMP -MMP-29 3.0 MMP-9 1.2 MMP-9

2.5 1.0

2.0 0.8

1.5 0.6

1.0 0.4

Secreted MMPs Secreted 0.5 0.2 Relative Activity of Activity Relative

0.0 0.0

WM-115 IGR-39 WM-793 WM-266-4 IGR-37 1205Lu

RelativeActivity of Secreted MMPs RelativeActivity of Secreted MMPs Melanoma Cell Melanoma Cells Melanoma Cell

164

MMP-2 MMP-2 MMP-9 MMP-9 MMP-2 MMP-2 4.5 MMP-9 1.0 MMP-9 4.0 3.5 0.8 3.0 0.6 2.5 2.0

mRNA Levels mRNA 0.4

1.5 of MMPs of 1.0 0.2 0.5 0.0 0.0

Relative Relative WM-115 IGR-39 WM-793 WM-266-4 IGR-37 1205Lu Relative mRNA Levels of MMPs mRNA Relative of Levels Relative mRNA Levels of MMPs mRNA Relative of Levels Melanoma Cell Melanoma Cells Melanoma Cells

165

Table 5.1. Sequences for RT-qPCR primers.

Gene Forward Primer Reverse Primer Name MMP-2 5’-ATACAGGATCATTGGCTACACACCT-3’ 5’-CCAAAGTTGATCATGATGTCTGCC-3’ MMP-9 5’-ACCAAGGATACAGTTTGTTCCTCGT-3’ 5’-TAGAGGTGCCGGATGCCATTCACGT-3’ GAPDH 5’-CCATGGAGAAGGCTGGGG-3’ 5’-CAAAGTTGTCATGGATGACC-3’

166

Chapter 6

HSP90 Inhibitors Stimulate DNAJB4 Protein Expression

through an Epitranscriptomic Mechanism

Introduction

RNA is known to harbor more than 100 types of covalent modifications (1), and the biological functions for most of these modifications remain poorly understood. Recent studies documented the widespread occurrence of N6-methyladenosine (m6A) in mRNA and the discovery of cellular proteins that are involved in the deposition (2-4), recognition

(5-7), and removal (8-10) of this modified nucleoside in mRNA. Thus, reversible methylations in mRNA may constitute a very important mechanism of gene regulation through modulating the stability and translation efficiency of mRNA (6, 7, 11). In addition, this mechanism may also assume important roles in stress response (12, 13).

Heat shock response is among the best studied cellular stress response pathways, where heat shock proteins enable homeostasis of the proteome by preventing the aggregation and maintaining the native structures of proteins in cells (14). In this vein, the 90-kDa heat shock protein (HSP90) is involved in assisting the folding of a large number of so-called

“client” proteins (15). Moreover, small-molecule inhibitors of HSP90 have been employed in clinical trials for treating various human diseases (16). It, however, remains largely unexplored how treatment with these inhibitors modulates the heat shock proteome in human cells.

167

In the present study, we uncovered that treatment of human cells with HSP90 inhibitors led to substantial reprogramming of the heat shock proteome, where pronounced increases in expression were observed for DNAJB4 and HSPA1. We also found that the elevated expression of DNAJB4 occurs, in part, through an m6A-mediated epitranscriptomic mechanism, and a similar mechanism is at play during heat shock response.

Materials and Methods

Cell culture

M14 (National Cancer Institute), Hela, HEK293T (ATCC), and all CRISPR- engineered cells were cultured in Dulbecco's Modified Eagle Medium with 10% fetal bovine serum (Invitrogen, Carlsbad, CA) and penicillin (100 IU/mL). The cells were maintained at 37°C in a humidified atmosphere containing 5% CO2. All cells were authenticated by ATCC and tested for mycoplasma with LookOut® Mycoplasma PCR

Detection Kit (Sigma). Cells were treated with 100 nM HSP90 inhibitors for the indicated period of time, and, for heat shock treatment, the HEK293T cells were incubated in a

42.0C water bath for 60 min. Cell growth was monitored at 24-hr upon HSP90 inhibitors treatment by cell counting (Figure S10). For SILAC labeling experiments, M14 cells were

13 15 13 cultured in SILAC medium containing [ C6, N2]-lysine and [ C6]-arginine for at least 10 days to promote complete incorporation of the heavy isotope-labeled amino acids (17). The complete heavy-isotope incorporation was verified by mass spectrometry before drug treatment (Figure S11).

168

Approximately 2×107 cells were harvested, washed with cold phosphate-buffered saline (PBS) for three times, and lysed by incubating on ice for 30 min in CelLytic M

(Sigma) cell lysis reagent containing 1% protease inhibitor cocktail. The cell lysates were centrifuged at 9,000g at 4°C for 30 min, and the resultant supernatants collected.

Plasmid and siRNAs

The sequences for siMETTL3-1 and siMETTL3-2 were 5'-

CUGCAAGUAUGUUCACUAUGA-3' and 5'-AGGAGCCAGCCAAGAAAUCAA-3', respectively, as described previously (2). The sequences for siALKBH5-1 and siALKBH5-

2 were 5'-ACAAGUACUUCUUCGGCGA-3' and 5'-GCGCCGUCAUCAACGACUA-3', respectively, as described previously (18). siRNA was transfected using RNAiMAX

(Invitrogen) following the manufacturer’s protocol, where non-targeting siRNA

(Dharmacon, D-001210-02-20) was used as control. The pcDNA3.1-DYK plasmids for ectopic expression of ALKBH5 and FTO were kindly provided by Prof. Chuan He (19, 20).

The pRK7-YTHDF3 plasmid was constructed by amplifying the cDNA of YTHDF3 from pGEX-YTHDF3 provided by Prof. Chuan He (6) using the primers of 5'-

GCTCTAGACCACCAATGTCAGATCCA-3' and 5'-

CGGGATCCTTGTTTGTTTCTATTTCTCTCCCTA-3'. The plasmids were transfected into cells by using Lipofectamine 2000 (Life Technologies), following the manufacturer’s recommended procedures.

169

LC-PRM analysis

To assess the differential expression of heat shock proteins in M14 cells after HSP90 inhibitor treatment, we conducted two forward and two reverse SILAC labeling experiments, where lysates of light-labeled, inhibitor-treated M14 cells and heavy-labeled, mock (DMSO)-treated M14 cells were combined at 1:1 ratio in the forward labeling experiment (Figure 6.1). The reverse labeling experiment was conducted in the opposite way. The equal amount of the lysate combination was confirmed by mass spectrometry analysis in data-dependent mode (Table 6.1). The cells were lysed, and the proteins in the resulting lysates were digested using the previously reported filter-aided sample preparation (FASP) protocol (21). Approximately 50 µg of cell lysates were washed with

8 M urea for protein denaturation using a Microcon centrifugal filter with a molecular weight cutoff of 30 kDa, and the urea buffer was then removed by centrifugation at 10,000 rpm. The ensuing denatured proteins were reduced, alkylated, and digested with modified

MS-grade trypsin (Pierce) at an enzyme/substrate ratio of 1:100 in 50 mM NH4HCO3 (pH

8.5) at 37°C overnight. The peptide mixture was subsequently dried in a Speed-vac, desalted with OMIX C18 pipette tips (Agilent Technologies). Around 500 ng peptides were analyzed by LC-MS/MS on a Q Exactive Plus quadruple-Orbitrap mass spectrometer

(Thermo Fisher Scientiﬁc) in the PRM mode.

All LC-PRM experiments were performed on a Q Exactive Plus quadrupole-Orbitrap mass spectrometer, which was coupled with an EASY-nLC 1200 system (Thermo

Scientific). The LC-MS/MS conditions were recently described (22). Top five abundant

170 unique peptides for each heat shock protein were selected in the isolation list of LC-PRM analysis and both light and heavy version of peptides were monitored.

PRM data processing

All raw files were processed using Skyline (version 3.5) (23) for the generation of extracted-ion chromatograms and peak integration. We imposed a mass accuracy of within

20 ppm for fragment ions during the identification of peptides in the Skyline platform. The targeted peptides were manually checked to ensure that the transitions for multiple fragment ions derived from light and heavy forms of the same peptide exhibit the same elution time in the pre-selected retention time window and display similar distribution as those in the MS/MS acquired from shotgun proteomic analysis, with dot product value being > 0.7 (24). The sum of peak areas from all transitions of light or heavy forms of peptides was used for quantification.

Pulse-chase SILAC labeling and LC-PRM analysis

M14 cells were cultured in SILAC light or heavy DMEM with 10% (v/v) dialyzed FBS, as described above. Cells cultured in heavy and light SILAC media were rinsed with PBS, and switched to light and heavy DMEM media, respectively. Immediately after the exchange of culture media, the cells were treated with 100 nM ganetespib in DMSO or mocked treated (with DMSO). After culturing for 6 hr, the cells were harvested and lysates prepared for LC-PRM analysis, as described above (Figure S5). The 6-24 h pulse-chase cells were treated with 100 nM ganetespib in DMSO or mocked treated (with DMSO) at

0-hr and exchange the culture media at 6-hr of ganetespib treatment. After culturing for

171

24-hr, the cells were harvested and lysates prepared for LC-PRM analysis, as described above (Figure S5). The fold difference in protein turnover was calculated by comparing the relative abundance of the converted proteins with or without ganetespib treatment, following a previously described method (25).

CRISPR/Cas9-mediated genome editing of HEK293T cells

CRISPR targeting was conducted following the previously reported protocols,(26) where the single guide RNAs (sgRNAs) were designed as described

(http://www.broadinstitute.org/rnai/public/analysis-tools/sgrna-design). The guide sequences were GTGGTGAGGTATGGAATCGGAGG for YTHDF1,

TGAACCTTACTTGAGTCCACAGG for YTHDF2, and

ATAAAACACAACATGAATATTGG for YTHDF3, where the letters in bold indicate the PAM motif. Oligodeoxyribonucleotides corresponding to target sequences were obtained from Integrated DNA Technologies and ligated into the hSpCas9 plasmid pX330

(Addgene). The constructed plasmids were then transfected into HEK293T cells using

Lipofectamine 2000 (Invitrogen) in a 6-well plate and individual cells were cultured for further analysis. Genomic DNA was extracted from individual clonal cell lines, and specific DNA regions surrounding the targeted sites were screened by PCR, followed by agarose gel electrophoresis to assess the modification efficiency and by Sanger sequencing to identify the deletion loci (Figure S8). A set of clones with both alleles being cleaved by

Cas9 were isolated, and the successful deletion of YTHDF1 and YTHDF3 was further validated by Western blot analysis (Figure 5a).

172

Western blot

Cells were cultured in a 6-well plate and lysed at 60-80% confluency following the above-described procedures. The concentrations of proteins in the resulting lysates were determined using Bradford Assay (Bio-Rad). The whole cell lysate for each sample (10

μg) was denatured by boiling in Laemmli loading buffer and subjected to SDS-PAGE separation. The proteins were subsequently transferred to a nitrocellulose membrane at 4°C overnight. The resulting membrane was blocked with PBS-T (PBS with 0.1% Tween 20) containing 5% milk (Bio-Rad) at 4°C for 6 h. The membrane was subsequently incubated with primary antibody at 4°C overnight and then with secondary antibody at room temperature for 1 h. After thorough washing with PBS-T, the HRP signals were detected with Pierce ECL Western Blotting Substrate (Thermo).

Antibodies recognizing human ALKBH5 (Proteintech, 16837-1-AP, 1:50000 dilution),

DNAJB4 (Santa Cruz Biotechnology, sc-100711, 1:4000 dilution), FTO (Santa Cruz

Biotechnology, sc-271713, 1:2000 dilution), HSP70 (Stressgen SPA-810, 1:10000 dilution), HSP90 (Santa Cruz Biotechnology, sc-13119, 1:10000 dilution), HSPB1 (Santa

Cruz Biotechnology, sc-13132, 1:10000 dilution), METTL3 (Proteintech, 15073-1-AP,

1:4000 dilution), YHTDF1 (Abcam, ab99080, 1:1000 dilution), YTHDF3 (Santa Cruz

Biotechnology, SC-377119, 1:500 dilution), and Flag epitope tag (Cell Signaling, 2368,

1:20000 dilution) were employed as primary antibodies. Horseradish peroxidase- conjugated anti-rabbit IgG, IRDye® 680LT Goat anti-Mouse IgG (1:10000 dilution) were used as secondary antibodies. Membranes were also probed with anti-actin antibody (Cell

Signaling #4967, 1:10000 dilution) to confirm equal protein loading.

173

Polysome profiling

Sucrose solutions were prepared in a polysome buffer (10 mM HEPES, pH 7.4, 100 mM KCl, 5 mM MgCl2, 100 µg/ml cycloheximide and 2% Triton X-100). Sucrose density gradients (15%–60%, w/v) were freshly prepared in SW 55 ultracentrifuge tubes (Backman) using a Gradient Master (BioComp Instruments). M14 cells were pre-treated with 100

µg/ml cycloheximide for 7 min at 37C followed by washing with ice-cold PBS containing

100 µg/ml cycloheximide. The cells were then lysed in polysome lysis buffer. Cell debris was removed by centrifugation at 14,000 r.p.m. for 10 min at 4C. OD254 was measured by using a Nanodrop (Thermo) and the same amount of RNA was loaded onto sucrose gradients followed by centrifugation at 50,000 r.p.m for 90 min at 4C in an SW 55 rotor.

The resulting sample was eluted out at a flow rate of 0.5 ml/min through an automated fractionation system (ISCO UA-5 UV detector, ISCO Inc) that continuously monitored

OD254 values. Aliquots of polysome fraction were used for real-time PCR analysis. m6A RNA immunoprecipitation (RIP)

Eighty µg of total RNA isolated from M14 cells with or without ganetespib treatment was incubated with 4 μg anti-m6A antibody (Millipore ABE572) in 1 × IP buffer (10 mM

Tris-HCl, pH 7.4, 150 mM NaCl, and 0.1% Igepal CA-630) at 4°C for 2 h. The m6A-IP mixture was then incubated with Protein A beads at 4°C on a rotating wheel for additional

2 h. After washing for three times with the IP buffer, the bound RNA was eluted using a

50-µL elution buffer (6.7 mM N6-methyladenosine 5-monophosphate in 1 × IP buffer),

174 followed by ethanol precipitation. Precipitated RNA was used for reverse transcription and real-time PCR as described below.

Real-time PCR

M14 cells were seeded in 6-well plates at 50% conﬂuence level. Total RNA and polysome RNA were extracted from cells or polysome fraction using TRI Reagent (Sigma).

Approximately 3 μg RNA was reverse transcribed by employing M-MLV reverse transcriptase (Promega) and an oligo(dT)18 primer. After a 60-min incubation at 42°C, the reverse transcriptase was deactivated by heating at 75°C for 5 min. Quantitative real-time

PCR was performed using iQ SYBR Green Supermix kit (Bio-Rad) on a Bio-Rad iCycler system (Bio-Rad), and the running conditions were at 95°C for 3 min and 45 cycles at 95°C for 15 sec, 55°C for 30 sec, and 72°C for 45 sec. The comparative cycle threshold (Ct) method (ΔΔCt) was used for the relative quantification of gene expression (27), and the primers are listed in Table 6.2. The total mRNA level of each gene was normalized to that of the internal control to (GAPDH and HPRT1). The polysome and total mRNA level of

DNAJB4 for polysome occupancy calculation was normalized to that of an internal control,

HPRT1(6) and GAPDH. The m6A RIP for m6A level calculation was normalized to that of an internal control, HPRT1.

In vitro m6A demethylase activity assay

The demethylase activity was measured using an ELISA-based m6A demethylase assay kit (ab233489). The M14 cells were cultured in a 6-well plate at 60-80% confluency and lysed in a buffer containing 0.7% CHAPS, 50 mM HEPES (pH 7.4), 0.5 mM EDTA, 100

175 mM NaCl, and 1% protease inhibitor cocktail. The demethylase activity was measured by monitoring absorbance at 450 nm using a Synergy H1 microplate reader (Biotek, Winooski,

VT), and demethylase activity was calculated following the vendor’s recommended procedures.

Single-base elongation- and ligation-based qPCR amplification method (SELECT)

SELECT method was performed following previously reported procedure (28). Total

RNA was mixed with 40 nM Up Primer, 40 nM Down Primer (Table 6.3, designed from

NM_007034.5) and 5 μM dNTP in 17μl 1×CutSmart buffer. The RNA and primers were annealed by incubating mixture at a temperature gradient: 90°C for 1min, 80°C for 1min,

70°C for 1min, 60°C for 1min, 50°C for 1min, and then 40°C for 6min. Subsequently, a 3

μl of mixture containing 0.01 U Bst 2.0 DNA polymerase, 0.5 U SplintR ligase and 10 nmol ATP was added in the former mixture to the final volume 20 μl. The final reaction mixture was incubated at 40°C for 20 min, denatured at 80°C for 20 min and kept at 4°C.

Afterwards, Quantitative real-time PCR was performed using iQ SYBR Green Supermix kit (Bio-Rad) on a Bio-Rad iCycler system (Bio-Rad). The 20μl qPCR reaction was composed of iQ SYBR Green Supermix 200 nM qPCRF primer, 200 nM qPCRR primer

(Table 6.3), 2μl of the final reaction mixture and ddH2O. qPCR was run at the following condition: 95°C, 5min; (95°C, 10s; 60°C, 35s)×40 cycles; 95°C, 15s. The comparative cycle threshold (Ct) method (ΔΔCt) was used for the relative quantification of template abundance (27).

176

Dual-luciferase reporter assay

The 5′-UTR of DNAJB4 gene was amplified from a cDNA library from M14 cells by

PCR primers of 5'-CATGCCATGGAGGATTGAATACAGAGAC-3' and 5'-

CATGCCATGGTTCGAATGCCTTGAAAT-3'. cDNA was subcloned into NcoI- linearized pGL3-promoter vector (designed from NM_007034.5). The A17C, A41C,

A114C, and A121C mutated pGL3 plasmids were constructed by amplifying from pGL3-

DNAJB4-5′UTR using the primers listed in Table 6.4. All plasmids were sequenced to be right.

pGL3 and pRL plasmids were co-transfected into M14 cells for 24 hours. Cells were treated with 100 nM ganetespib for 6 hours before lysis. Luciferase activity of firefly and

Renilla were measured following the vendor’s recommended procedures (Promega) using a Synergy H1 microplate reader (Biotek, Winooski, VT) with gain set as 225, 10 s integration time and 2000 ms delay time.

Results

1. Treatment with HSP90 inhibitors led to elevated expression of a number of heat shock proteins through a post-transcriptional mechanism

By employing a recently developed targeted quantitative proteomic method (Figure 6.1)

(22), we examined the alterations in expression levels of heat shock proteins in response to treatment with three small-molecule inhibitors of HSP90. These include ganetespib and

AT13387 (a.k.a. onalespib) of the resorcinol chemotype, and 17-

(dimethylaminoethylamino)-17-demethoxygeldanamycin (a.k.a. alvespimycin) of the

177 ansamycin chemotype (16). All three inhibitors interact with the ATP-binding pocket located in the N-terminal domain of HSP90 (16). Strikingly, our results showed that a number of heat shock proteins, particularly HSPA1 and DNAJB4, which belong to the

HSP70 and HSP40 subfamilies, respectively, displayed markedly elevated expression in

M14 melanoma cells upon treatment with the three HSP90 inhibitors (Figure 6.2a&b).

We confirmed the altered expressions of several heat shock proteins, namely HSP90,

HSP70, DNAJB4 and HSPB1, by Western blot analysis (Figure 6.2c&d). It is worth noting that the targeted proteomic method, which quantifies proteins on the basis of specific peptides, allows for the independent assessments of different isoforms of HSP90

(HSP90AA1 and HSP90AB1, a.k.a. HSP90 and HSP90) and HSP70 (HSPA1, HSPA2,

HSPA7 and HSPA8, a.k.a. HSP70-1, HSP70-2, HSP70B and HSC70) (Figure 6.2b&d).

The commercially available antibodies employed in our Western analysis, however, recognize multiple isoforms of HSP90 or HSP70. Thus, the magnitudes of changes in expression of some heat shock proteins measured with Western blot analysis were not as pronounced as those determined from LC-MS/MS analysis. The increased expression of heat shock proteins was also captured in Hela and HEK293T cells after 24-hr ganetespib treatment (Figure 6.3).

Our quantitative proteomic data revealed a ~ 30-fold increase in expression level of

DNAJB4 after a 24-hr treatment with ganetespib; the mRNA level of this gene, however, only exhibited a 6-fold elevation (Figure 6.4a). This result suggests that ganetespib stimulates the expression of DNAJB4 protein partly through a post-transcriptional mechanism. This conclusion is further substantiated by the result obtained from polysome

178 fractionation followed by RT-qPCR analysis, which showed augmented occupancy of

DNAJB4 mRNA in the polysome fraction after ganetespib treatment (Figure 6.4b-c, 6.5a).

In addition, absorbance at 254 nm for the polysome fraction was increased at 6 hr following ganetespib treatment, manifesting an elevated translation efficiency of the genes including

DNAJB4, though it returned to pre-treatment levels at 24-hr following ganetespib treatment (Figure 6.5b). The elevated polysome fraction may be ascribed to commonly elevated protein synthesis which we have validated in heat shock proteins by pulse-chase experiment. The pulse-chase SILAC labeling experiment (Figure 6.6) revealed elevated synthesis of many heat shock proteins in M14 cells after a 6-hr treatment with ganetespib

(Figure 6.7), where the increases in newly synthesized heat shock proteins in M14 cells upon ganetespib treatment are correlated with the elevations in their overall levels of expression (Figure 6.5c).

Our proteomic data also showed that the expression levels of three other heat shock proteins (HSPA1, HSPB1 and HSPH1) were up-regulated by 26-, 2.5- and 3.6-fold at 24 hr following ganetespib exposure. However, the mRNA expression levels of the HSPA1,

HSPB1 and HSPH1 genes were increased by 50-, 5- and 5-fold at the same time point

(Figure 6.4a), suggesting that the HSP90 inhibitor-elicited elevation in expression levels of these three heat shock proteins occurs primarily through transcriptional regulation.

179

2. HSP90 inhibitor-stimulated increase in DNAJB4 protein expression involves an m6A-based epitranscriptomic mechanism

Recent studies showed that, in response to heat shock stress, cells overexpress HSP70 protein through a mechanism involving cap-independent translation enabled by m6A in the

5′-untranslated region (UTR) of the mRNA of HSP70 gene (12, 13, 29). Therefore, we asked whether the similar mechanism contributes to the HSP90 inhibitor-induced augmentation in DNAJB4 protein. In this respect, it was previously observed that heat shock could result in a ~ 2-fold increase in the level of m6A in the 5′-UTR of DNAJB4 mRNA and a nearly 4-fold elevation in the ribosomal occupancy of DNAJB4 mRNA, as revealed by m6A-seq and Ribo-seq analyses (12). Furthermore, our m6A RNA immunoprecipitation (RIP) together with RT-qPCR result revealed an increased level of m6A in DNAJB4 mRNA after a 6-hr treatment with ganetespib (Figure 6.5d). Thus, we monitored the expression levels of DNAJB4, HSP70, ALKBH5, FTO, METTL3 and

YTHDF3 proteins in M14 cells at different time intervals following treatment with 100 nM ganetespib. Our results revealed a marked increase in the levels of DNAJB4 and HSP70, which paralleled the temporal profile of YTHDF3, though no substantial change was observed for the expression level of ALKBH5, FTO, or METTL3 (Figure 6.8).

We next asked how the ganetespib-stimulated increase in DNAJB4 protein level is modulated by the eraser, writer and reader proteins of m6A. We found that ectopic overexpression of ALKBH5 and, to a lesser degree, FTO, significantly attenuated the ganetespib-induced progressive increase in expression levels of DNAJB4 protein (Figure

6.9a&b). In this respect, FTO is mainly involved in demethylating both N6,2-O-

180

6 6 dimethyladenosine (m Am) in the mRNA cap structure (10) and internal m A (30), while

ALKBH5-mediated m6A erasure of nuclear mRNA affects nuclear mRNA export, metabolism and gene expression (9). Reciprocal experiment showed that the siRNA- mediated knock-down of ALKBH5 facilitated the ganetespib-induced heightened level of

DNAJB4 (Figure 6.10). On the other hand, METTL3, which encodes the catalytic component of the major m6A methyltransferase complex (2-4), its depletion abolished the ganetespib-induced elevated expression of DNAJB4 (Figure 6.9c&d). Likewise, CRISPR-

Cas9-mediated individual deletion of the three YTH domain-containing proteins (i.e.

YTHDF1, YTHDF2, and YTHDF3), which are reader proteins for m6A (5-7), also led to marked attenuation in the ganetespib-induced increase in expression of DNAJB4 protein

(Figure 6.11a&b and Figure 6.12). Furthermore, ectopic reconstitution of YTHDF3 in the

YTHDF3-knockout cells rescued the elevated expression of DNAJB4 protein following ganetespib treatment (Figure 6.11c&d).

We also assessed the perturbations of the heat shock proteome induced by ALKBH5 overexpression using the aforementioned targeted proteomic method. It turned out that overexpression of ALKBH5 only led to substantially decreased expression of DNAJB4,

DNAJB12 and HSP47 (Figure 6.13a). Consistent with the proteomic results, Western blot results revealed that ALKBH5 overexpression did not exert any apparent effect on the expression of HSPB1 protein (Figure 6.139b&c). This result underscores that ALKBH5 modulates only a small subset of heat shock proteins. Interestingly, the expression of

HSPA14 was induced with reduced m6A level on HSPA14 mRNA. An m6A site was identified in non-coding last exon in HSPA14 gene with cross-linking-induced mutation

181 sites (CIMSs)-based miCLIP (31). However, the function of m6A modification in non- coding last exon remained uncovered. In this respect, m6A residues in the last exons allowed the potential for 3′-UTR regulation (32), while 3′-UTR controls mRNA translation in an ALKBH5-dependent manner (33).

We further explored the HSP90 inhibitor-elicited increase in DNAJB4 protein synthesis by monitoring the m6A demethylase activity in M14 cells with or without a 6-hr treatment with 100 nM ganetespib. Our results showed that the m6A demethylase activity in the lysate of M14 cells was decreased to ~60% of the level observed in DMSO-treated cells at 6 hr after ganetespib treatment (Figure 6.14a). This result suggests that exposure to ganetespib and the ensuring proteotoxic stress may compromise the activity of m6A demethylase(s) (e.g. ALKBH5). This may result in elevated m6A level in the nuclear mRNA of DNAJB4, promoting nuclear mRNA export (9) and therefore, becoming available for translation.

3. m6A modification at site 114 on 5′-UTR promotes the translation of DNAJB4

m6A modification at 5′-UTR has been reported to promote cap-independent translation

(29), we next explored whether the m6A levels at 5′-UTR of DNAJB4 mRNA was elevated, and if the elevation of m6A drives the translation of DNAJB4 or not.

To measure the m6A level of the specific site, a single-base elongation- and ligation- based qPCR amplification method (termed “SELECT”) was employed (28). With the utilization of Bst DNA polymerase, the single base elongation of the Up Probe was selectively hindered with m6A modifications present in the RNA template. Thus, the lower

182 template abundance obtained from quantitative PCR will correlate to the higher the m6A level of the specific site. We measured a total of nine sites from 5′-UTR and CDS of

DNAJB4 mRNA, which covers all of the motif sites containing GAC from 5′-UTR (four sites) and two from CDS. The rest three are the T/CAA sites from 5′-UTR applied as negative control. As a result, SELECT uncovered an increased m6A levels of the six motif sites we selected after ganetespib treatment compared to the three negative control sites

(Figure 6.14b), indicating ganetespib induced m6A levels on the mRNA of DNAJB4, especially at 5′-UTR region.

With the identification of the elevated m6A sites, we next measured the translational efficiency by dual-luciferase reporter assay. Our results showed that only with the point mutation of site 114 on 5′-UTR (Figure 6.14c), the translation was hindered, suggesting that elevated m6A level at site 114 induced by HSP90 inhibitors promoted the translation of DNAJB4 (Figure 6.14d).

4. Heat shock stress stimulated DNAJB4 protein expression through an m6A-based epitranscriptomic mechanism

We next examined whether a similar mechanism is at work in cells under heat shock stress. It turned out that heat shock (by incubating cells at 42C for 1 hr) again induced a marked elevation in the expression level of DNAJB4 protein, and this increase is substantially diminished in cells upon RNA interferencing (RNAi)-mediated depletion of

METTL3 (Figure 6.15a&b), ectopic overexpression of ALKBH5 (Figure 6.15c-d), or

CRISPR-Cas9-mediated depletion of three YTH domain-containing proteins (i.e.

183

YTHDF1-3) (Figure 6.15e-f). Hence, heat shock induces the elevated expression of

DNAJB4 through a similar mechanism.

Discussion

Our quantitative proteomic method facilitated the assessment of the reprogramming of the heat shock proteome in cultured human cells in response to HSP90 inhibitor treatment en masse. We found that treatment with three HSP90 inhibitors leads to elevated expression of multiple heat shock proteins, with the most pronounced increases being observed for

HSPA1 and DNAJB4. HSPA1 and DNAJB4 belong to the HSP70 and HSP40 subfamilies of heat shock proteins, where HSP40 acts as a co-chaperone for HSP70 by stimulating its

ATPase activity and enhancing its interaction with substrate proteins (34). Our study uncovered a novel regulatory mechanism of DNAJB4, where, in response to proteotoxic stress induced by HSP90 inhibitor treatment, cells stimulate the translation of mRNA of

DNAJB4, which is modulated by the writer, eraser and reader proteins of m6A (Figure

6.16). Our work, together with previous studies by Zhou et al. (12, 13), showed that proteotoxic stress, arising from heat shock or treatment with HSP90 inhibitors, results in elevated translation of both HSP70 and its co-chaperone HSP40 via a common m6A- mediated epitranscriptomic mechanism. To our knowledge, this is the first report to show that small molecules can modulate gene expression through an epitranscriptomic mechanism.

184

References

1. Cantara WA, Crain PF, Rozenski J, McCloskey JA, Harris KA, Zhang X, et al. The RNA modification database, RNAMDB: 2011 update. Nucleic Acids Res. 2011;39:D195-D201. 2. Liu J, Yue Y, Han D, Wang X, Fu Y, Zhang L, et al. A METTL3–METTL14 complex mediates mammalian nuclear RNA N6-adenosine methylation. Nat Chem Biol. 2013;10:93. 3. Ping X-L, Sun B-F, Wang L, Xiao W, Yang X, Wang W-J, et al. Mammalian WTAP is a regulatory subunit of the RNA N6-methyladenosine methyltransferase. Cell Res. 2014;24:177. 4. Schwartz S, Mumbach Maxwell R, Jovanovic M, Wang T, Maciag K, Bushkin GG, et al. Perturbation of m6A writers reveals two distinct classes of mRNA methylation at internal and 5′ sites. Cell Rep. 2014;8:284-96. 5. Li A, Chen Y-S, Ping X-L, Yang X, Xiao W, Yang Y, et al. Cytoplasmic m6A reader YTHDF3 promotes mRNA translation. Cell Res. 2017;27:444. 6. Wang X, Lu Z, Gomez A, Hon GC, Yue Y, Han D, et al. N6-methyladenosine-dependent regulation of messenger RNA stability. Nature. 2013;505:117. 7. Wang X, Zhao BS, Roundtree IA, Lu Z, Han D, Ma H, et al. N(6)-methyladenosine modulates messenger RNA translation efficiency. Cell. 2015;161:1388-99. 8. Jia G, Fu Y, Zhao X, Dai Q, Zheng G, Yang Y, et al. N6-Methyladenosine in nuclear RNA is a major substrate of the obesity-associated FTO. Nat Chem Biol. 2011;7:885. 9. Zheng G, Dahl JA, Niu Y, Fedorcsak P, Huang C-M, Li CJ, et al. ALKBH5 Is a mammalian RNA demethylase that impacts RNA metabolism and mouse fertility. Mol cell. 2013;49:18-29. 10. Mauer J, Luo X, Blanjoie A, Jiao X, Grozhik AV, Patil DP, et al. Reversible methylation of m(6)Am in the 5' cap controls mRNA stability. Nature. 2017;541:371-5. 11. Choi J, Ieong K-W, Demirci H, Chen J, Petrov A, Prabhakar A, et al. N(6)- methyladenosine in mRNA disrupts tRNA selection and translation elongation dynamics. Nat Struct Mol Biol. 2016;23:110-5. 12. Zhou J, Wan J, Gao X, Zhang X, Jaffrey SR, Qian S-B. Dynamic m6A mRNA methylation directs translational control of heat shock response. Nature. 2015;526:591. 13. Zhou J, Wan J, Shu XE, Mao Y, Liu X-M, Yuan X, et al. N6-Methyladenosine Guides mRNA Alternative Translation during Integrated Stress Response. Mol Cell. 2018;69:636-47. 14. Richter K, Haslbeck M, Buchner J. The heat shock response: life on the verge of death. Mol Cell. 2010;40(2):253-66. 15. Taipale M, Jarosz DF, Lindquist S. HSP90 at the hub of protein homeostasis: emerging mechanistic insights. Nat Rev Mol Cell Biol 2010;11:515. 16. Taldone T, Ochiana SO, Patel PD, Chiosis G. Selective targeting of the stress chaperome as a therapeutic strategy. Trends Pharmacol Sci. 2014;35:592-603. 17. Ong S-E, Blagoev B, Kratchmarova I, Kristensen DB, Steen H, Pandey A, et al. Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics. Mol Cell Proteomics. 2002;1:376-86. 18. Cheong A, Low JJA, Lim A, Yen Paul M, Woon ECY. A fluorescent methylation- switchable probe for highly sensitive analysis of FTO N6-methyladenosine demethylase activity in cells. Chem Sci. 2018;9:7174-85. 19. Su R, Dong L, Li C, Nachtergaele S, Wunderlich M, Qing Y, et al. R-2HG exhibits anti- tumor activity by targeting FTO/m6A/MYC/CEBPA signaling. Cell. 2018;172:90-105. 20. Zhang S, Zhao BS, Zhou A, Lin K, Zheng S, Lu Z, et al. m6A demethylase ALKBH5 maintains tumorigenicity of glioblastoma stem-like cells by sustaining FOXM1 expression and cell proliferation program. Cancer Cell. 2017;31:591-606.e6.

185

21. Wiśniewski JR, Zougman A, Nagaraj N, Mann M. Universal sample preparation method for proteome analysis. Nat Methods. 2009;6:359-62. 22. Miao W, Li L, Wang Y. A targeted proteomic approach for heat shock proteins reveals DNAJB4 as a suppressor for melanoma metastasis. Anal Chem. 2018;90:6835-42. 23. MacLean B, Tomazela DM, Shulman N, Chambers M, Finney GL, Frewen B, et al. Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics. 2010;26:966-8. 24. Sherwood CA, Eastham A, Lee LW, Risler J, Vitek O, Martin DB. Correlation between y- type ions observed in ion trap and triple quadrupole mass spectrometers. J Proteome Res. 2009;8:4243-51. 25. Fierro-Monti I, Racle J, Hernandez C, Waridel P, Hatzimanikatis V, Quadroni M. A novel pulse-chase SILAC strategy measures changes in protein decay and synthesis rates induced by perturbation of proteostasis with an Hsp90 inhibitor. PLoS One. 2013;8:e80423. 26. Cong L, Ran FA, Cox D, Lin S, Barretto R, Habib N, et al. Multiplex genome engineering using CRISPR/Cas systems. Science. 2013;15:819-23. 27. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2−ΔΔCT method. Methods. 2001;25:402-8. 28. Xiao Y, Wang Y, Tang Q, Wei L, Zhang X, Jia G. An elongation- and ligation-based qPCR amplification method for the radiolabeling-free detection of locus-specific N(6) -methyladenosine modification. Angew Chem Int Ed Engl. 2018;57:15995-6000. 29. Meyer Kate D, Patil Deepak P, Zhou J, Zinoviev A, Skabkin Maxim A, Elemento O, et al. 5'UTR m6A promotes cap-independent translation. Cell. 2015;163:999-1010. 6 6 1 30. Wei J, Liu F, Lu Z, Fei Q, Ai Y, He PC, et al. Differential m A, m Am, and m A demethylation mediated by FTO in the cell nucleus and cytoplasm. Mol Cell. 2018;71:973-85.e5. 31. Linder B, Grozhik AV, Olarerin-George AO, Meydan C, Mason CE, Jaffrey SR. Single- nucleotide-resolution mapping of m6A and m6Am throughout the transcriptome. Nat Methods. 2015;12:767. 32. Ke S, Alemu EA, Mertens C, Gantman EC, Fak JJ, Mele A, et al. A majority of m6A residues are in the last exons, allowing the potential for 3' UTR regulation. Genes Dev. 2015;29:2037-53. 33. Tang C, Klukovich R, Peng H, Wang Z, Yu T, Zhang Y, et al. ALKBH5-dependent m6A demethylation controls splicing and stability of long 3′-UTR mRNAs in male germ cells. Proc Natl Acad Sci USA. 2018;115:E325. 34. Qiu XB, Shao YM, Miao S, Wang L. The diversity of the DnaJ/Hsp40 family, the crucial partners for Hsp70 chaperones. Cell Mol Life Sci. 2006;63:2560-70.

186

Figure 6.1. A schematic diagram showing the forward SILAC-based proteomic workflow

for examining the alterations in expression levels of heat shock proteins after HSP90

inhibitor treatment. a HSP90i-treated cells DMSO-treated cells

Light Amino Acids Heavy Amino Acids

Combine, mix at 1:1 ratio

Tryptic digestion

PRM data acquisition

Data analysis

Ganetespib 17-DMAG 1

2 2

1 3

187 0

AT13387 Figure 6.2. HSP90 inhibitors induced substantial reprogramming of the heat shock proteome. (a) A heat map showing the alterations in expression levels of heat shock proteins in M14 cells upon a 24-hr treatment with 100 nM of ganetespib, AT13387, or 17-

DMAG. The data represent the mean of results obtained from two forward and two reverse

SILAC labeling experiments. (b) Representative PRM traces for monitoring the relative expression levels of several heat shock proteins with or without HSP90 inhibitor treatment.

(c) Western blot for validating the expression levels of select heat shock proteins after treatment with the three HSP90 inhibitors. (d) Quantification data for the differences in expression levels of heat shock proteins in M14 cells with or without HSP90 inhibitor treatment, as obtained from Western blot (n = 3) and LC-PRM analysis (n = 4, two forward and two reverse SILAC labelings).

188

a b HSP90 inhibitors DMSO

HSP90AA1 HSP90AB1HSPA1 HSPA2 DNAJB4 HSPB1 1 1 1 1 1 1 Ganetespib 0.8 0.6 0.5 0.5 0.5 0.5 0.5 0.4 0.2 0 0 0 0 0 0 66 68 70 17 19 104 106 93 95 97 45 47 92 94 96

1 1 1 1 1 1 AT13387

0.5 0.5 0.5 0.5 0.5 0.5

0 0 0 0 0 0 58 60 62 17 19 95 97 84 86 88 44 46 83 85 87

1 1 1 1 1 1

Relative AbundanceRelative

- DMAG 0.5 0.5 0.5 0.5 0.5 0.5

0 0 0 0 0 0 57 59 61 17 19 94 96 83 85 87 44 46 82 84 86 Retention Time (min) log2(Ratio)

Ganetespib c d 40 AT13387 PRM 17-DMAG 30 20

6 MW (kDa) 4 2 HSP90 90 0

70 HSPA1 HSPA2 DNAJB4 HSPB1 HSP70 HSP90AA1HSP90AB1

25 DNAJB4 40 Western blot (short exp.) 20 DNAJB4 15 6 (long exp.) 40 4 2 HSPB1 27 Relative(Inhibitor/DMSO) Ratio 0

Actin 45 HSP90 HSP70 DNAJB4 HSPB1 Heat Shock Proteins

189

The quantification data in (a) and (b) represent the mean ± S. D. of results from three independent experiments.

a 7 DMSO Ganetespib (nM) 0 100 Ganetespib MW (kDa) 6 HSP70 70 5 4 DNAJB4 40 3 2 Actin 45 1 0

HSP70 DNAJB4 Relative Protein Expression Protein Relative

b 8 DMSO Ganetespib (nM) 0 100 Ganetespib MW (kDa) 7 HSP70 70 6 5

DNAJB4 40 4 3 2 Actin 45 1 0

HSP70 DNAJB4 Relative Protein Expression Protein Relative

190

DNAJB4 in M14 cells exhibited a progressive increase in the polysome fraction following ganetespib treatment. Data were normalized to the mRNA level of GAPDH (b) or HPRT1

(c) gene. The quantification data represent the mean  S. D. of results from three independent experiments. The p values were calculated based on unpaired, two-tailed

Student’s t-test: #, p > 0.05; *, 0.01 ≤ p < 0.05; **, 0.001 ≤ p < 0.01; ***, p < 0.001.

a 0 h b 250 0.5 h 2 h 10 200 6 h *** 24 h 8 150 *** *** 100 6 50 4 *** 6 5 2

4 PolysomeDNAJB4 3 (Normalized GAPDH)to 0 RelativeLevel mRNA 2 (Normalized GAPDH)to 0 2 6 12 24 1 0 Time after Ganetespib Treatment (h) HSPA1A DNAJB4 DNAJB12 HSPB1 HSPH1 200

150 c 2 ** 100 **

50 DNAJB4

t 1.5 6 # # 5

4 /Inpu 3 1

RelativeLevel mRNA 2 (Normalized HPRT1)to 1

0 (Normalized HPRT1)to 0.5 Polysome HSPA1A DNAJB4 DNAJB12 HSPB1 HSPH1 0 2 6 12 24 Heat Shock Proteins Time after Ganetespib Treatment (h)

191

Data were normalized to the mRNA level of HPRT1 gene. (b) Traces for fractionation of polysome isolated from untreated M14 cells or M14 cells at 6 or 24 hr following treatment with 100 nM ganetespib. (c) A scatter plot shows the correlation between the expression change and protein synthesis (fold turnover) of heat shock proteins in M14 cells after ganetespib treatment. (d) m6A RIP-qPCR result shows the increased m6A level in DNAJB4 mRNA in M14 cells after a 6-hr treatment with ganetespib. The data were normalized to the mRNA level of HPRT1 gene. The quantification data represent the mean  S. D. of results from three independent experiments. The p values referred to comparison between

0 h and the indicated time points following HSP90 inhibitor treatment and were calculated based on unpaired, two-tailed Student’s t-test: #, p > 0.05; *, 0.01  p < 0.05; **, 0.001  p < 0.01; ***, p < 0.001.

192 a b 0.8 DMSO 10 Ganetespib 6 h *** 0.7 Ganetespib 24 h 80S 8 *** *** 0.6

6 DNAJB4

254nm 0.5 Polysome

4 OD *** 0.4

Polysome 0.3 (Normalized (Normalized HPRT1) to 0 0 2 6 12 24 0.2 200 250 300 Time after Ganetespib Treatment (h) Time (s)

c d 2 5 HSPA1 1.5 **

/DMSO) DNAJB4 4 1 3

DNAJA1 Level of A 6 0.5

Ganetespib

( 10

0 DNAJB4mRNA

1 log R² = 0.67 Relativem -0.5 -1 -0.5 0 0.5 1 1.5 2 0 DMSO Ganetespib log10(Fold Turnover)

193

0 h 6 h Change to heavy SILAC media Harvest, lyse cells, L-Gan Tryptic digestion and LC-PRM analysis Ganetespib treatment Light M14 cells

Change to heavy SILAC media Harvest, lyse cells, L-DMSO Tryptic digestion and LC-PRM analysis DMSO treatment Light M14 cells

Change to light SILAC media Harvest, lyse cells, H-Gan Ganetespib treatment Tryptic digestion and LC-PRM analysis Heavy M14 cells Change to light SILAC media Harvest, lyse cells, H-DMSO DMSO treatment Tryptic digestion and LC-PRM analysis Heavy M14 cells

194

100

10 /DMSO)

Fold Turnover FoldTurnover

Ganetespib (

0.1

SELK

HSP47

HSPA1 HSPE1 HSPB1 HSPA8 HSPA2 HSPA9 HSPB8

HSPH1 HSPD1

DNAJB4 DNAJA1 HSPA13 DNAJA3 DNAJA2 DNAJA4 HSPA14

DNAJC8 DNAJC9 DNAJC1 DNAJC7 DNAJC2 DNAJC3 DNAJC6 DNAJC5

HSPBP1

DNAJB12

HSP90B1

DNAJC11 DNAJC10 DNAJC12 HSPC125 DNAJC14 DNAJC17 DNAJC19 DNAJC13

HSPA12B

HSP90AA1 HSP90AB1 HSP90AB4P

195

Figure 6.8. Time-dependent changes in expression levels of HSP70, DNAJB4, METTL3,

FTO, ALKBH5, and YTHDF3 in M14 cells following treatment with 100 nM of the indicated HSP90 inhibitors. Western blot images and the quantification data showing the alterations in expression levels of the indicated proteins at different time points following treatment with ganetespib (a), AT13387 (b), or 17-DMAG (c). Shown are the ratios of expression of the indicated proteins over β-actin, and further normalized to the ratios obtained for the control cells without HSP90 inhibitor treatment. The data represent the mean  S. D. of results from three independent experiments. The p values referred to comparison between 0 h and 24 h treatment and were calculated using unpaired, two-tailed

Student’s t-test: #, p > 0.05; *, 0.01  p < 0.05; **, 0.001  p < 0.01; ***, p < 0.001.

196 a Ganetespib Treatment (h) 0 2 6 12 24 MW (kDa) *** HSP70 70

DNAJB4 40 METTL3 65 ***

ALKBH5 45

FTO 60 *** # #

YTHDF3 65 * RelativeProtein Expression Actin 45 b AT13387 *** Treatment (h) 0 2 6 12 24 MW (kDa) 18 HSP70 70 15

DNAJB4 40 12 *** METTL3 65

ALKBH5 45 6 FTO 60 3 *** # YTHDF3 65 ** * RelativeProtein Expression 0 Actin 45 FTO HSP70DNAJB4METTL3ALKBH5 YTHDF3 c 17-DMAG Treatment (h) 0 2 6 12 24 MW (kDa) 6080 0 h *** HSP70 70 50 2 h 60 6 h 12 h DNAJB4 40 40 40 24 h 30 METTL3 65 20 8 *** ALKBH5 45 5 64 FTO 60 *** 43 2 YTHDF3 65 2 # 1 * RelativeProtein Expression ** 0 Actin 45 DNAJB4 Relative Expression FTO HSP70 DNAJB4METTL3ALKBH5 YTHDF3

197

Figure 6.9. The ganetespib-stimulated expression of DNAJB4 protein involves writer and eraser proteins of m6A. (a) Western blot for monitoring the expression levels of DNAJB4,

Flag-tagged FTO and ALKBH5, and endogenous FTO in M14 cells transfected with control plasmid or plasmids for the ectopic expression of Flag-FTO or ALKBH5 at different time intervals following treatment with 100 nM ganetespib. (b) Quantification data based on Western blot analysis in (a). (c) Western blot for monitoring the expression levels of DNAJB4 in M14 cells treated with control non-targeting siRNA (siCtrl), siMETTL3-1 or siMETTL3-2 at different time points following treatment with 100 nM ganetespib. (d) Quantification data based on Western blot analysis in (c). -actin was employed as the loading control in (a) and (c). Shown in (b) and (d) are the ratios of expression of DNAJB4 protein over β-actin, and further normalized to the ratios obtained for the control cells without ganetespib treatment. The quantification data in (b) and (d) represent the mean  S. D. of results from three independent experiments. The p values referred to comparisons between control cells and cells with ectopic overexpression of the indicated genes (b), or between controls and siRNA-mediated knock-down of the indicated genes (d). The p values were calculated using unpaired, two-tailed Student’s t-test: #, p >

0.05; *, 0.01  p < 0.05; **, 0.001  p < 0.01; ***, p < 0.001.

198

a FTO ALKBH5 FTO ALKBH5 ALKBH5 Control Control (1 µg) (1 µg) (1 µg) (2 µg) (3 µg)

Ganetespib (h) 0 12 24 0 12 24 0 12 24 0 12 24 0 12 24 0 12 24 0 12 24 MW (kDa) DNAJB4 40

FTO 60 Anti-Flag ALKBH5 45 FTO 60

β-Actin 45 b 30 0 h 12 h 24 h

25 *** ** 20 # 15 * * * 10 Expression ** **

5 ** ** RelativeDNAJB4 0 Control FTO ALKBH5 (1 µg) ALKBH5 (2 µg) ALKBH5 (3 µg) c siCtrl siMETTL3-1 siCtrl siMETTL3-2 Ganetespib (h) 0 12 24 0 12 24 0 12 24 0 12 24 MW (kDa) METTL3 65

DNAJB4 40

β-Actin 45

d 30 0 h 12 h 24 h 25 20 15

10 Expression 5 RelativeDNAJB4 ** *** *** *** 0 siCtrl siMETTL3-1 siMETTL3-2

199

Figure 6.10. The ganetespib-stimulated expression of DNAJB4 protein involves eraser protein of m6A, ALKBH5. (a) Western blot for monitoring the expression levels of

DNAJB4 protein in M14 cells treated with control non-targeting siRNA (siCtrl), siALKBH5-1 or siALKBH5-2 at different time points following treatment with 100 nM ganetespib. (b) Quantification data based on Western blot analysis in (a). -actin was employed as the loading control in (a). Shown in (b) is the ratios of expression of DNAJB4 protein over actin, and further normalized to the ratios obtained for the control cells without ganetespib treatment. The quantification data in (b) represent the mean  S. D. of results from three independent experiments. The p values were calculated based on unpaired, two- tailed Student’s t-test: #, p > 0.05; *, 0.01 ≤ p < 0.05; **, 0.001 ≤ p < 0.01; ***, p < 0.001.

200 a siCtrl siALKBH5-1 siALKBH5-2

0 12 24 0 12 24 0 12 24 MW (kDa) DNAJB4 40

ALKBH5 40

β-Actin 45 b 80 0 h 12 h 24 h * 60 ** ** ** 40

20 Expression

RelativeDNAJB4 0 siCtrl siALKBH5-1 siALKBH5-2

201

Figure 6.11. The ganetespib-stimulated expression of DNAJB4 protein involves reader proteins of m6A. (a) Western blot for monitoring the expression levels of DNAJB4 in

HEK293T cells or the isogenic cells with the YTHDF1, YTHDF2, and YTHDF3 genes being ablated by the CRISPR-Cas9 genome editing method at different time intervals following exposure with 100 nM ganetespib. (b) Quantification data based on Western blot analysis in (a). (c) Western blot for monitoring the expression levels of DNAJB4 in

YTHDF3-deficient HEK293T cells complemented with an empty pRK7 plasmid (control) or pRK7-YTHDF3 at different time points following exposure to 100 nM ganetespib. (d)

Quantification data based on Western blot analysis in (c). -actin was employed as the loading control in (a) and (c). Shown in (b) and (d) are the ratios of expression of DNAJB4 protein over β-actin, and further normalized to the ratios obtained for the control cells without ganetespib treatment. The quantification data in (b) and (d) represent the mean 

S. D. of results from three independent experiments. The p values referred to comparisons between HEK293T cells and the isogenic cells with YTHDF1/2/3 genes being individually ablated (b), or between complementation with control and YTHDF3 plasmid (d). The p values were calculated using unpaired, two-tailed Student’s t-test: #, p > 0.05; *, 0.01  p

< 0.05; **, 0.001  p < 0.01; ***, p < 0.001.

202

a b HEK293T sgRNA-YTHDF1 MW (kDa) Ganetespib (h) 0 2 6 12 24 0 2 6 12 24 9 0 h 2 h 6 h 12 h 24 h 65 YTHDF1 8 DNAJB4 40 7 β-Actin 45 6 HEK293T sgRNA-YTHDF2 5 * Ganetespib (h) 0 2 6 12 24 0 2 6 12 24 4 * DNAJB4 40 * 3 * * β-Actin 45 * # HEK293T sgRNA-YTHDF3 2 ** * * *** Ganetespib (h) 0 2 6 12 24 0 2 6 12 24 1 **

YTHDF3 65 RelativeDNAJB4 Expression 0 DNAJB4 40 HEK293T sgRNA- sgRNA- sgRNA- β-Actin 45 YTHDF1 YTHDF2 YTHDF3

c d 10 * sgRNA-YTHDF3 0 h 12 h 24 h 8 Control YTHDF3 MW (kDa) Ganetespib (h) 0 12 24 0 12 24 6 ** YTHDF3 65 4 DNAJB4 40 2 β-Actin 45 0

Control YTHDF3 RelativeDNAJB4 Expression sgRNA-YTHDF3

203

YTHDF3 were also confirmed by Western blot analysis (See Figure 5a). We were not able to validate the knockout of YTHDF2 gene by Western blot owing to the lack of highly specific antibody.

Exon 4 WT CCCTCCGATTCCATACCTCACCACCTAC -14bp CCCTC------ACCACCTAC YTHDF1 -14bp CCCTC------ACCACCTAC

Exon 3 WT TTTGGGCAAC CAGGAGCCCTAGGTAGCAC +1bp TTTGGGCAACCCAGGAGCCCTAGGTAGCAC YTHDF2 -10bp TTTGGGC------CCTAGGTAGCAC

Exon 4 WT ATGAAT ATTGGAAATTGGGATG +1bp ATGAATT ATTGGAAATTGGGATG YTHDF3 +2bp ATGAATTTATTGGAAATTGGGATG

204

Western blot for monitoring the expression levels of HSPB1, Flag-tagged FTO and

ALKBH5, and endogenous FTO in M14 cells transfected with control plasmid or plasmids for the ectopic expression of Flag-FTO or ALKBH5 at different time intervals following exposure with 100 nM ganetespid. (c) Quantification data of HSPB1 based on Western blot analysis in (b). The quantification data in (c) represent the mean  S. D. of results from three independent experiments.

205

3.3

1.0

(ALKBH5/Control) Relative AbundanceRelative

0.33

SELK

HSP47

HSPA1 HSPA8 HSPA2 HSPA9 HSPA4 HSPB1 HSPB8 HSPE1

HSPD1 HSPH1

DNAJB1 DNAJA1 DNAJA4 HSPA14 DNAJB4 DNAJA2 DNAJB2

DNAJC8 DNAJC2 DNAJC7 DNAJC6 DNAJC5 DNAJC3 DNAJC9

HSPBP1

DNAJB12 DNAJB11 DNAJB14 HSP90B1

DNAJC25 DNAJC19 DNAJC11 DNAJC14 DNAJC10

HSP90AA1 HSP90AB1

HSP90AB4P HSP90AA4P b c

FTO ALKBH5 ALKBH5 Control 7 0 h 12 h 24 h (1 µg) (2 µg) (3 µg) 6 Ganetespib (h) 0 12 24 0 12 24 0 12 24 0 12 24 MW (kDa) 5 HSPB1 27 4 FTO 60 Anti-Flag 3 ALKBH5 45 2 FTO 60 1 0 β-Actin 45 RelativeHSPB1 Expression Control FTO ALKBH5 ALKBH5 (2 µg) (3 µg)

206

Figure 6.14. Ganetespib treatment resulted in increased m6A levels in the 5-UTR of

DNAJB4 mRNA, and methylation at one of these sites led to elevated translation efficiency of DNAJB4. (a) Displayed is a bar graph showing that the activity of m6A demethylase is decreased in M14 cells after a 6-hr treatment with 100 nM ganetespib (n = 6). (b) A diagram showing the 5-UTR of the human DNAJB4 mRNA and the adenosine sites monitored by the SELECT assay. The adenosine sites marked in red and black are the m6A motif sites and the negative control sites, respectively. (c) Relative template abundances after elongation and ligation measured by SELECT (n = 3). A total of seven sites were chosen, four from m6A motif site (GAC) and three from UAA or CAA sites (as negative control).

‘Neg’ represents negative control. The p-values were calculated versus the mean value of the three negative controls. (d) Relative luciferase activities for wild-type (WT) 5′-UTR of

DNAJB4 gene and the corresponding A→C mutants (n=3). Firefly luciferase activity was normalized to that of Renilla luciferase and further normalized to the wild-type construct.

The data represent the mean  S. D. of results obtained from three separate experiments.

The p values referred to comparison with the wild-type plasmid, and were calculated using unpaired, two-tailed Student’s t-test: #, p > 0.05; *, 0.01  p < 0.05; **, 0.001  p < 0.01;

***, p < 0.001.

207 a 1.4 b 1.2 A17 A41 A114 A121 1 0.8 ** 5′ 3′ 0.6 0.4 A63 A155 A161

0.2 A DemethylaseActivity A 6 0 5′-UTR m DMSO Ganetespib 1-172 c 7 DMSO 6 h Ganetespib 6 h Neg 6 Neg 5 Neg 4 3 *** *** 2 *** 1 ***

0 after ElongationandLigation

Relative Template Abundance RelativeTemplate 17 41 63 114 121 155 161 d m6A WT 5' Luciferase 1.2 # # ) #

A17C 5' Luciferase 0.8 Renilla A41C 5' Luciferase 0.4 ***

A114C 5' Luciferase (Firefly/ 0 A121C 5'

Luciferase RelativeActivity Luciferase

208

(HS, at 42.0C for 60 min), or the same cells after siRNA-mediated knockdown of

METTL3 (siMETTL3-1) (a), ectopic overexpression of ALKBH5 (c), or the isogenic

HEK293T cells with YTHDF1, YTHDF2, or YTHDF3 genes being knocked out by the

CRSIPR-Cas9 genomic editing method (e). The relevant quantification data are presented in (b), (d) and (f). -actin was employed as the loading control in (a), (c) and (e). Shown in (b) and (f) are the ratios of expression of DNAJB4 protein over β-actin, and in (d) these ratios were further normalized to the ratios obtained for the cells at t=0 post heat shock.

The quantification data in (b), (d) and (f) represent the mean  S. D. of results from three independent experiments. The p values referred to comparison between control cells and siRNA-mediated knock-down (b) or ectopic expression (d) of the indicated genes, or

HEK293T cells and the isogenic cells with YTHDF1/2/3 genes being individually ablated

(f). The p values were calculated based on unpaired, two-tailed Student’s t-test: *, 0.01  p < 0.05; **, 0.001  p < 0.01; ***, p < 0.001.

209 a b siCtrl siMETTL3 1.2 0 h 5 h 10 h 24h MW (kDa) Post HS Post HS 1 N 0 5 10 24 N 0 5 10 24 0.8 METTL3 65 0.6 ** DNAJB4 40 0.4 **

Expression 0.2 β-Actin 45 *

RelativeDNAJB4 0 siCtrl siMETTL3 c d 0 h 5 h 10 h 24 h Control ALKBH5 80 Post HS Post HS MW (kDa) 60 N 0 5 10 24 N 0 5 10 24 DNAJB4 40 40

ALKBH5 45

Expression 20 β-Actin * ** 45 RelativeDNAJB4 * 0 Control ALKBH5 e HEK293T sgRNA-YTHDF1 f Post HS Post HS MW (kDa) 1.2 N 0 5 10 24 N 0 5 10 24 0 h 5 h 10 h 24 h

DNAJB4 40 1

β-Actin 45 0.8 HEK293T sgRNA-YTHDF2 Post HS Post HS 0.6 ** * N 0 5 10 24 N 0 5 10 24 ** ** DNAJB4 40 0.4 *** *** *** ** β-Actin 45 ** HEK293T sgRNA YTHDF3 0.2

Post HS Post HS RelativeDNAJB4 Expression N 0 5 10 24 N 0 5 10 24 0 YTHDF3 65

DNAJB4 40 β-Actin 45

210

Figure 6.16. A schematic diagram showing an m6A-mediated epitranscriptomic pathway in modulating the expression level of DNAJB4 protein in cells upon treatment with HSP90 inhibitor or heat shock.

Eraser Eraser 3' 3'

5' 114 A 5' 114 m6A Reader

A m6A

211

Table 6.1. LC-MS and MS/MS in the data-dependent acquisition (DDA) mode for confirming the equi-mass mixing of light- and heavy-labeled lysates in SILAC experiments.

Drug Replicate Total Intensity Total Intensity Ratio (Light Peptides) (Heavy Peptides) (Light/Heavy) Ganetespib Forward 6.52 × 1011 5.97 × 1011 1.09 Ganetespib Reverse 9.08 × 1011 8.77 × 1011 1.04 AT13387 Forward 1.15 × 1012 1.12 × 1012 1.03 AT13387 Reverse 7.39 × 1011 6.87 × 1011 1.08 17-DMAG Forward 7.42 × 1011 6.83 × 1011 1.09 17-DMAG Reverse 8.28 × 1011 9.13 × 1011 0.91

212

Table 6.2. Sequences for RT-qPCR primers.

Gene Name Forward Primer Reverse Primer DNAJB4 5-AAGGGTTGAAAGGAGGAGCA-3 5-TTCAGAATCTCTACCACCACCCA-3 HSPA1A 5-GGAGGCGGAGAAGTACA-3 5-GCTGATGATGGGGTTACA-3 HSPH1 5-GACAGCTGTTGCTTTGAATTACGGA-3 5-GCTGTTCCCAGTACCTTCAA-3 DNAJB12 5-CTGTGAAAAGGGTCAAGCAATGT-3 5-TGCCAATGGCTTTGAAGGCTT-3 HSPB1 5-AGATCACCGGCAAGCACGAG-3 5-TTGGACTGCGTGGCTAGCTT-3 HPRT1 5- TGACACTGGCAAAACAATGCA-3 5-GGTCCTTTTCACCAGCAAGCT-3 GAPDH 5-CCATGGAGAAGGCTGGGG-3 5-CAAAGTTGTCATGGATGACC-3

213

Table 6.3. Primer sequences for SELECT method.

Description Primer A17 - Up Primer 5-tagccagtaccgtagtgcgtgTGTCTTAAGGCAGCAAGCAGACAGCG-3 A17 - Down Primer 5-phos/CTCTGTATTCAATCCTcagaggctgagtcgctgcat-3 A41 - Up Primer 5-tagccagtaccgtagtgcgtgATCAGCAATTCAGCTAGCTG-3 A41 - Down Primer 5-phos/CTTAAGGCAGCAAGCAGACAGCGTCTCcagaggctgagtcgctgcat-3 A63 - Up Primer 5-tagccagtaccgtagtgcgtgAAGCTGGGTATTTTAAAAGT-3 A63 - Down Primer 5-phos/AATCAGCAATTCAGCTAGCTGTCTTAAcagaggctgagtcgctgcat-3 A114 - Up Primer 5-tagccagtaccgtagtgcgtgAAAGAAAACAGCGTCCCCAG-3 A114 - Down Primer 5-phos/CTTAGCAACAGATTCTAAGAAAAATAAcagaggctgagtcgctgcat-3 A121 - Up Primer 5-tagccagtaccgtagtgcgtgCTTTGTAAAAGAAAACAGCG-3 A121 - Down Primer 5-phos/CCCCAGTCTTAGCAACAGATTCTAAGAcagaggctgagtcgctgcat-3 A155 - Up Primer 5-tagccagtaccgtagtgcgtgCATTTCGAATGCCTTGAAAT-3 A155 - Down Primer 5-phos/AACTTAGATTTCCCTTTGTAAAAGAAAcagaggctgagtcgctgcat-3 A161 - Up Primer 5-tagccagtaccgtagtgcgtgTTTCCCCATTTCGAATGCCT-3 A161 - Down Primer 5-phos/GAAATTAACTTAGATTTCCCTTTGTAAcagaggctgagtcgctgcat-3 qPCR forward primer 5-ATGCAGCGACTCAGCCTCTG-3 qPCR reverse primer 5-TAGCCAGTACCGTAGTGCGTG-3

214

Table 6.4. Primer sequences for mutagenesis amplification from pGL3-DNAJB4-5′UTR.

The C and G in bold indicate the mutation sites, where A is mutated to C.

Description Primer A17C - Forward 5-GAATACAGAGCCGCTGTCTGCTTGCTGCCTT-3 A17C - Reverse 5-AAGCAGACAGCGGCTCTGTATTCAATCCT-3 A41C - Forward 5-CTGCCTTAAGCCAGCTAGCTGAATTGCT-3 A41C - Reverse 5-TCAGCTAGCTGGCTTAAGGCAGCAAGCA-3 A114C - Forward 5-TGTTGCTAAGCCTGGGGACGCTGTTTT-3 A114C - Reverse 5-CGTCCCCAGGCTTAGCAACAGATTCTA-3 A121C - Forward 5-TAAGACTGGGGCCGCTGTTTTCTTTTACA-3 A121C - Reverse 5-AGAAAACAGCGGCCCCAGTCTTAGCAACA-3

215

Chapter 7

Identification of Helicase Proteins as Clients for HSP90

Introduction

Nucleic acid unwinding is an important biological process that regulates nearly all aspects of nucleic acid metabolism, including DNA replication, transcription and repair, as well as mRNA translation (1, 2). Helicases are nucleic acid-dependent ATPases that are capable of unwinding duplex and more complex structures of nucleic acids (2). Owing to these unique roles of helicases, their aberrant functions are strongly associated with genetic instability, cancer development and premature aging. For instance, FANCJ, BLM, and

WRN, which are DNA helicases that can help maintain genomic integrity by resolving guanine-quadruplex (G4) DNA structures, are implicated in genetic disorders of Fanconi anemia, Bloom’s syndrome and Werner’s syndrome, respectively (3, 4).

HSP90 is a molecular chaperone that ensures the proper folding of proteins (clients), thereby maintaining the homeostasis of the proteome (5). HSP90 has a large group of client proteins, including protein kinases (6) and steroid hormone receptors (7). In addition, several helicases, such as yeast SSL2 (8) and human MCM4 (9), were shown by in vitro protein binding assay to be client proteins of HSP90. However, there has been no method reported for examining systematically the interactions between HSP90 and the helicase proteome.

Targeted proteomic methods, involving LC-MS/MS analyses in the parallel-reaction monitoring (PRM) or multiple-reaction monitoring (MRM) mode, exhibit much higher

216 sensitivity and reproducibility in peptide detection than proteomic analysis in the data- dependent acquisition (DDA) mode (10, 11). Due to the high mass resolving power and mass accuracy afforded by a time-of-flight or Orbitrap mass analyzer, the PRM method offers better accuracy and specificity than MRM in quantifying analytes present in complex sample matrices (10, 11); hence, it has become increasingly used in bioanalysis (12-14).

In this study, we developed, for the first time, an LC-PRM-based targeted quantitative proteomic method for the high-throughput proteome-wide analysis of helicase proteins in human cells. With this method, we assessed the differential expression of helicase proteins in cultured human cells upon treatment with two HSP90 inhibitors. We also introduced a tandem affinity tag to the C-terminus of endogenous HSP90 protein using CRISPR-Cas9

(15, 16), and examined the interaction between HSP90 and helicases by employing affinity pull-down in conjunction with LC-PRM analysis.

Materials and Methods

Cell culture

M14 and HEK293T cells were cultured in Dulbecco's Modified Eagle Medium. All culture media were supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA) and penicillin (100 IU/mL). The cells were maintained at 37°C in a humidified atmosphere

9,000g at 4°C for 30 min, and the resulting supernatants collected. For quantitative proteomic experiments, the cells were cultured in SILAC medium containing

217

13 15 13 [ C6, N2]lysine and [ C6]arginine for at least 10 days to promote complete incorporation of isotope-labeled amino acids (17). M14 cells were treated with 100 nM HSP90 inhibitors

(i.e. onalespib or alvespimycin) for 24 h before harvesting.

To assess the differential expression of helicases, we conducted one forward and one reverse SILAC labeling experiments, where lysates of light-labeled HSP90 inhibitor- treated cells and heavy-labeled mock (DMSO)-treated cells were combined at 1:1 ratio in the forward labeling experiments. The reverse labeling experiments were conducted in the opposite way.

Tryptic digestion of the whole cell lysates

The above-mentioned cell lysates from the two cell lines were washed with 8 M urea for protein denaturation, and then treated with dithiothreitol and iodoacetamide for cysteine reduction and alkylation, respectively. The proteins were subsequently digested with modified MS-grade trypsin (Pierce) at an enzyme/substrate ratio of 1:100 in 50 mM

NH4HCO3 (pH 8.5) at 37°C overnight. The resulting peptide mixture was dried in a Speed- vac, desalted with OMIX C18 pipette tips (Agilent Technologies), and analyzed by LC-

MS and MS/MS on a Q Exactive Plus quadruple-Orbitrap mass spectrometer (Thermo

Fisher Scientiﬁc) in the PRM mode.

LC-PRM analysis

All LC-PRM experiments were performed on a Q Exactive Plus quadrupole-Orbitrap mass spectrometer. The mass spectrometer was coupled with an EASY-nLC 1200 system

(Thermo Scientific), and the samples were automatically loaded onto a 4-cm trapping

218 column (150 µm i.d.) packed with ReproSil-Pur 120 C18-AQ resin (5 µm in particle size and 120 Å in pore size, Dr. Maisch GmbH HPLC) at 3 µL/min. The trapping column was coupled to a 20-cm fused silica analytical column (PicoTip Emitter, New Objective, 75 µm i.d.) packed with ReproSil-Pur 120 C18-AQ resin (3 µm in particle size and 120 Å in pore size, Dr. Maisch GmbH HPLC). The peptides were then separated with a 140-min linear gradient of 9-38% acetonitrile in 0.1% formic acid and at a flow rate of 300 nL/min. The spray voltage was 1.8 kV. Precursor ions were isolated with an isolation width of 1.0 m/z unit and collisionally activated in the HCD cell at a collision energy of 29 to yield MS/MS.

PRM data analysis

All raw files were processed using Skyline (version 3.5) (18) for the generation of extracted-ion chromatograms and peak integration. To robustly identify peptides in the

Skyline platform, we imposed a mass accuracy of within 20 ppm for fragment ions. The targeted peptides were manually checked to ensure that the transitions for multiple fragment ions derived from light and heavy forms of the same peptide exhibit the same elution time in the pre-selected retention time window. The data were then processed, where the distribution of the relative intensities of multiple transitions associated with the same precursor ion must correlate with the theoretical distribution in the MS/MS spectral library entry (i.e. data acquired from shotgun proteomic analysis). The sum of peak areas from all transitions of light or heavy forms of peptides was used for quantification. The

Skyline PRM library for helicases (including the iRT file) and the raw files for LC-PRM analyses of helicases were deposited into PeptideAtlas with the identifier number of

PASS01191 (http://www.peptideatlas.org/PASS/PASS01191).

219

Targeted integration of the tandem affinity tag using CRISPR-Cas9

Genome editing-based integration of tandem affinity tag (3×FLAG and 2×Strept) to endogenous HSP90 was carried out according to published procedures (16). DNA sequence for the production of sgRNA targeting HSP90AB1 gene, which encodes HSP90, was inserted into the hSpCas9 plasmid pX330 (Addgene, Cambridge, MA, USA). The donor plasmid for tagging HSP90AB1 gene was obtained from IDT and inserted into the pUC19 plasmid. The constructed Cas9 plasmid were transfected together with the donor plasmid into HEK293T cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA), and individual cells were cultured for further screening. Lysates from cultures initiated from individual cells were employed for Western blot analysis with the use of anti-Flag antibody (Cell Signaling) to validate the insertion of the tandem affinity tag. The guide sequence for HSP90AB1 was 5-TGGAAGAAGTCGATTAGGTT-3.

Pull-down and immunoprecipitation assay for mass spectrometry analysis

For pull-down assay, HSP90-tagged HEK293T cells were collected and lysed in

CelLytic M cell lysis reagent (Sigma) with 1× protease inhibitor mixture (Sigma) for 30 min on ice. After centrifugation, the lysate (supernatant) containing 1 mg protein was incubated with 40 µL pre-equilibrated anti-FLAG M2 affinity gel (Sigma) at 4°C for 4 h.

The beads were washed with a buffer containing 50 mM Tris (pH 8.0), 150 mM NaCl, and

0.1% Tween (Sigma) for five times, and then treated with dithiothreitol and iodoacetamide for cysteine reduction and alkylation, respectively. The proteins were subsequently digested with modified MS-grade trypsin (Pierce) at an enzyme/substrate ratio of 1:100 in

220

50 mM NH4HCO3 (pH 8.5) at 37°C overnight. The resulting peptide mixture was dried in a Speed-vac, desalted with OMIX C18 pipette tips (Agilent Technologies), and subjected to LC-MS/MS analysis on a Q Exactive Plus quadruple-Orbitrap mass spectrometer

(Thermo Fisher Scientiﬁc) in the PRM mode, as described above.

Western blot

Cells were cultured in a 6-well plate and lysed at 40-50% confluency following the above-described procedures. The concentrations of proteins in the resulting lysates were determined using Bradford Assay (Bio-Rad). The whole cell lysate for each sample (10 μg) was denatured by boiling in Laemmli loading buffer and subjected to SDS-PAGE separation. The proteins were subsequently transferred to a nitrocellulose membrane at 4°C overnight. The resulting membrane was blocked with PBS-T (PBS with 0.1% Tween 20) containing 5% milk (Bio-Rad) at 4°C for 6 h. The membrane was then incubated with primary antibody at 4°C overnight and subsequently with secondary antibody at room temperature for 1 h. After thorough washing with PBS-T buffer, the HRP signals were detected with Pierce ECL Western Blotting Substrate (Thermo).

Antibodies recognizing human ARAF (Santa Cruz Biotechnology, sc-166771, 1:4000 dilution), DDX11 (Santa Cruz Biotechnology, sc-271711, 1:2000 dilution), DDX34 (Santa

Cruz Biotechnology, sc-514665, 1:2000 dilution), EGFR (Santa Cruz Biotechnology, sc-

03, 1:10000 dilution), HSP90 (Santa Cruz Biotechnology, sc-13119, 1:20000 dilution),

MCM4 (Santa Cruz Biotechnology, sc-28317, 1:2000 dilution), MCM7 (Santa Cruz

Biotechnology, sc-9966, 1:2000 dilution), and Flag epitope tag (Cell Signaling, 2368,

1:20000 dilution) were employed as primary antibodies. Horseradish peroxidase-

221 conjugated anti-rabbit IgG, IRDye® 680LT Goat anti-Mouse IgG (1:10000 dilution) were used as secondary antibodies. Membranes were also probed with anti-actin antibody (Cell

Signaling #4967, 1:10000 dilution) to confirm equal protein loading.

Results

We first constructed, based on data acquired from shotgun proteomic analyses of the tryptic digestion mixtures of protein lysates of 10 unique cell lines derived from different human tissue origins (14), a Skyline (19) PRM library for helicase proteins. The PRM library contains the retention time, MS and MS/MS of tryptic peptides of helicases. In this respect, we inspected all the identified helicase peptides and included, in the library, only those peptides that could be uniquely assigned to individual helicases. By incorporating 3-

6 such signature peptides for each helicase, our helicase PRM library encompasses 411 unique peptides, which represent 121 distinct human helicases (Figure 7.1a). Among them,

20, 60 and 41 were DNA helicases, RNA helicases and other helicases, respectively (Figure

7.1b). Hence, the library covers approximately 84% of the 95 known DNA and RNA helicases (20).

By adopting scheduled PRM analysis, where we set up the mass spectrometer to collect the MS/MS for the precursor ions from a limited number of peptides in each 10- min retention time (RT) window, we were able to achieve high-throughput detection of helicase proteins. In this vein, we calculated the normalized RT (iRT) value for each peptide on our target list according to previously published methods (21, 22). In particular, we employed 10 tryptic peptides of bovine serum albumin (BSA) as standards for establishing the iRT scale and for calibrating the retention time. We subsequently

222 converted the experimentally determined retention times for all the 411 helicase peptides into normalized iRT scores, where the linear relationship of RT vs. iRT was re-determined in every 5-7 LC-PRM runs through the analysis of the tryptic digestion mixture of BSA.

On the grounds that iRT of a tryptic peptide represents an inherent attribute of the peptide, a substantial difference of the measured RT of a tryptic peptide from that predicted from the linear plot of RT vs. iRT is considered false-positive, which is employed together with characteristic fragment ions observed for the peptide as parameters to validate the results obtained from the PRM assay.

Some HSP90 inhibitors can disrupt the binding of HSP90 to ATP and compromise

HSP90’s capability in maintaining the proper folding of client proteins, thereby resulting in their degradation via the ubiquitin-dependent proteasomal pathway (23, 24). Hence, we next applied the established LC-PRM method, together with stable isotope-labeling by amino acids in cell culture (SILAC) (17), to identify helicases as putative client proteins of

HSP90. In particular, we assessed the reprogramming of the human helicase proteome in cells upon treatment with two HSP90 inhibitors, i.e. onalespib and alvespimycin (Figure

7.1c), which are resorcyclic dihydroxybenzamide and bacterial-derived benzoquinone ansamycin derivatives, respectively (24). These two inhibitors occupy the ATP-binding pocket of the N-terminal domain of HSP90 protein and inhibit its chaperone activity (24).

Our LC-PRM analysis results showed that 49 out of the 95 quantified helicases were down- regulated upon onalespib treatment, and 28 out of the 96 quantified helicases were with decreased expression upon alvespimycin exposure (Figure 7.2). We also confirmed the

223 altered expression of four helicases (i.e. DDX11, DDX34, MCM4 and MCM7) by Western blot analysis (Figure 7.4c-f).

All the quantified peptides for helicases exhibit a reasonably good linear fit between the observed retention time and iRT in the library, lending support for the robust detection of helicases. In addition, all 4-6 transitions used for the quantification of each peptide from helicases exhibited the same retention time with a dot product (dotp) of > 0.7 when compared to the same fragment ions found in the MS/MS acquired from shotgun proteomic analysis (Figure 7.3) (25), further confirming the reliability of the method in peptide identification.

Our results showed that all the quantified helicases could be detected in both forward and reverse SILAC labeling experiments (Figure 7.4a). In addition, the ratios for the quantified peptides obtained from forward and reverse SILAC labeling experiments are consistent (Figures 7.4a). Furthermore, the average relative standard deviation (RSD) for the tryptic peptides quantified for the same helicases was 13.1%. These results underscored the reasonably good reproducibility of the analytical method.

We also analyzed the same samples by using LC-MS/MS in the DDA mode, where the peptide mixture was first separated into 20 fractions with a strong cation-exchange (SCX) column, and the resultant fractions were then individually subjected to LC-MS/MS analysis in the DDA mode. The resulting LC-MS/MS data only led to the quantification of 60 helicases. By contrast, our scheduled PRM method allowed for the quantification of ~95 helicases in only 4-5 LC-PRM runs without any pre-fractionation, revealing the superior sensitivity and throughput of the PRM method.

224

Linear regression analysis of the ratios for the commonly quantified helicases in cells treated with the two HSP90 inhibitors yielded an R2 value of 0.50 (Figure 7.4b), suggesting that the two inhibitors led to similar reprogramming of a large number of helicases.

Considering that HSP90 inhibition can lead to improper folding and proteasomal degradation of its client proteins (23), those helicases that were down-regulated upon the inhibitor treatment are considered candidate client proteins for HSP90.

We reason that those helicases that are client proteins of HSP90 should also interact with HSP90. Hence, we embarked on an interactome study by assessing, at the proteome- wide scale, the interactions between helicases and HSP90. To this end, we employed the

CRISPR-Cas9 genome-editing method to introduce a tandem aﬃnity tag (3 × Flag, 2 ×

Strept) to the C-terminus of endogenous HSP90(6) protein in HEK293T cells.(16)

Screening using Western blot with anti-Flag antibody led to the identification of two positive clones (Figure 7.5a, clone 8 and clone 15). The different intensity could be due to heterozygote and homozygote clones. The successful incorporation of the tandem affinity tag was also validated by Western blot analysis with the use of an anti-HSP90 antibody

(Figure 7.5a, 7.6a). As the HSP90 antibody can recognize all the HSP90 isoforms, the migrated upper band is the HSP90 incorporated with the tandem aﬃnity tag. Clone 15 was applied to the following studies mentioned as Flag-HSP90 cell. Affinity purification of HSP90 using anti-Flag M2 beads from the lysate of CRISPR-engineered cells, followed by tryptic digestion and LC-MS/MS analysis in the DDA mode, gave rise to the identification of HSP90 with a 65% sequence coverage (Figure 7.6b).

225

Relative quantification using SILAC labeling, affinity pull-down with anti-Flag beads, and LC-PRM analysis revealed the interactions between HSP90 and a number of helicases (Figure 7.5b-d). In particular, we were able to quantify 66 helicases, among which

40 were enriched by at least 1.5-fold with the lysate of the Flag-HSP90 cells relative to the use of the lysate from parental HEK293T cells (Figure 7.7). The ratios obtained from forward and reverse SILAC experiments are again consistent (Figure 7.5c). Together with the data from the above-mentioned inhibitor experiment, 21 helicases were both enriched from HSP90-tagged cell line over the parental HEK293T cells, and down-regulated upon treatment with at least one of the two HSP90 inhibitors (Figure 7.5d, Figure 7.8). These 21 helicases are considered strong candidates for client proteins of HSP90. Among them, eight (BTAF1, CHTF18, DDX32, DDX49, DDX55, ERCC2, MCM4 and MCM7) were down-regulated upon treatment with both HSP90 inhibitors (Figure 7.5d, Figure 7.9a,

Figure 7.8). In this vein, MCM4 and MCM7 were previously shown to interact with HSP90 through its cochaperone, FKBP51 (26). Moreover, we validated the interaction between

HSP90 and MCM4/MCM7 by immunoprecipitation followed by Western blot analysis

(Figure 7.9b). In this regard, similar affinity pull-down followed by Western blot analysis confirmed the interaction between HSP90 and ARAF kinase, a previously reported client protein of HSP90(6) (Figure 7.6a). On the other hand, we failed to detect EGFR, which is not a client protein of HSP90 (6), in the affinity pull-down mixture. It is worth noting that, among the helicases that are putative clients of HSP90, FANCJ, ERCC2 (a.k.a. XPD) and

DDX11 (a.k.a. CHIR1) are iron-sulfur cluster proteins that are known to be important in

DNA replication and repair (27).

226

In summary, we developed a PRM-based targeted proteomic approach to fulfill large- scale relative quantification of helicase proteins. We applied this method to interrogate quantitatively the interaction between HSP90 and helicases. We found that the expression levels for a large number of helicases were diminished in human cells upon treatment with two small-molecule inhibitors of HSP90. By incorporating a tandem affinity tag to endogenous HSP90 using CRISPR-Cas9, we found that 40 out of the 66 quantified helicases could be enriched by affinity pull-down of HSP90 from the engineered cells, supporting the interactions between helicases and HSP90. Hence, the results from these studies support that helicases are a group of client proteins for HSP90.

227

References

1. Brosh RM. DNA helicases involved in DNA repair and their roles in cancer. Nat Rev Cancer. 2013;13:542-58. 2. Singleton MR, Dillingham MS, Wigley DB. Structure and mechanism of helicases and nucleic acid translocases. Annu Rev Biochem. 2007;76:23-50. 3. Wu Y, Shin-ya K, Brosh RM. FANCJ Helicase Defective in Fanconia Anemia and Breast Cancer Unwinds G-Quadruplex DNA To Defend Genomic Stability. Mol Cell Biol. 2008;28:4116- 28. 4. Mohaghegh P, Karow JK, Brosh RM, Jr., Bohr VA, Hickson ID. The Bloom's and Werner's syndrome proteins are DNA structure-specific helicases. Nucleic Acids Res. 2001;29(13):2843-9. 5. Taipale M, Jarosz DF, Lindquist S. HSP90 at the hub of protein homeostasis: emerging mechanistic insights. Nat Rev Mol Cell Biol. 2010;11(7):515-28. 6. Taipale M, Krykbaeva I, Koeva M, Kayatekin C, Westover KD, Karras GI, et al. Quantitative analysis of HSP90-client interactions reveals principles of substrate recognition. Cell. 2012;150:987-1001. 7. Manninen T, Purmonen S, Ylikomi T. Interaction of nuclear receptors with hsp90 in living cells. J Steroid Biochem Mol Biol. 2005;96:13-8. 8. Flom G, Weekes J, Johnson JL. Novel interaction of the Hsp90 chaperone machine with Ssl2, an essential DNA helicase in Saccharomyces cerevisiae. Curr Genet. 2005;47:368-80. 9. Taipale M, Tucker G, Peng J, Krykbaeva I, Lin ZY, Larsen B, et al. A quantitative chaperone interaction network reveals the architecture of cellular protein homeostasis pathways. Cell. 2014;158:434-48. 10. Shi T, Song E, Nie S, Rodland KD, Liu T, Qian WJ, et al. Advances in targeted proteomics and applications to biomedical research. Proteomics. 2016;16(15-16):2160-82. 11. Peterson AC, Russell JD, Bailey DJ, Westphall MS, Coon JJ. Parallel reaction monitoring for high resolution and high aass accuracy quantitative, targeted Proteomics. Molecular & Cellular Proteomics. 2012;11:1475-88. 12. Zhou J, Liu H, Liu Y, Liu J, Zhao X, Yin Y. Development and evaluation of a parallel reaction monitoring strategy for large-scale targeted metabolomics quantification. Anal Chem. 2016;88:4478-86. 13. Guo L, Wang Q, Weng L, Hauser LA, Strawser CJ, Rocha AG, et al. Liquid chromatography-high resolution mass spectrometry analysis of platelet frataxin as a protein biomarker for the rare disease Friedreich's ataxia. Anal Chem. 2018;90(3):2216-23. 14. Miao W, Li L, Wang Y. A Targeted Proteomic Approach for Heat Shock Proteins Reveals DNAJB4 as a Suppressor for Melanoma Metastasis. Anal Chem. 2018. 15. Cong L, Ran FA, Cox D, Lin S, Barretto R, Habib N, et al. Multiplex genome engineering using CRISPR/Cas systems. Science. 2013;15:819-23. 16. Dalvai M, Loehr J, Jacquet K, Huard CC, Roques C, Herst P, et al. A Scalable Genome- Editing-Based Approach for Mapping Multiprotein Complexes in Human Cells. Cell Rep. 2015;13:621-33. 17. Ong S-E, Blagoev B, Kratchmarova I, Kristensen DB, Steen H, Pandey A, et al. Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics. Mol Cell Proteomics. 2002;1:376-86. 18. MacLean B, Tomazela DM, Shulman N, Chambers M, Finney GL, Frewen B, et al. Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics. 2010;26:966-8.

228

19. MacLean B, Tomazela DM, Shulman N, Chambers M, Finney GL, Frewen B, et al. Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics. 2010;26(7):966-8. 20. Umate P, Tuteja N, Tuteja R. Genome-wide comprehensive analysis of human helicases. Commun Integr Biol. 2011;4:118-37. 21. Escher C, Reiter L, MacLean B, Ossola R, Herzog F, Chilton J, et al. Using iRT, a normalized retention time for more targeted measurement of peptides. Proteomics. 2012;12:1111- 21. 22. Miao W, Xiao Y, Guo L, Jiang X, Huang M, Wang Y. A high-throughput targeted proteomic approach for comprehensive profiling of methylglyoxal-induced perturbations of the human kinome. Anal Chem. 2016;88:9773-9. 23. Butler LM, Ferraldeschi R, Armstrong HK, Centenera MM, Workman P. Maximizing the Therapeutic Potential of HSP90 Inhibitors. Mol Cancer Res. 2015;13:1445. 24. Taldone T, Ochiana SO, Patel PD, Chiosis G. Selective targeting of the stress chaperome as a therapeutic strategy. Trends Pharmacol Sci. 2014;35:592-603. 25. de Graaf EL, Altelaar AF, van Breukelen B, Mohammed S, Heck AJ. Improving SRM assay development: a global comparison between triple quadrupole, ion trap, and higher energy CID peptide fragmentation spectra. J Proteome Res. 2011;10:4334-41. 26. Taipale M, Tucker G, Peng J, Krykbaeva I, Lin ZY, Larsen B, et al. A quantitative chaperone interaction network reveals the architecture of cellular protein homeostasis pathways. Cell. 2014;158(2):434-48. 27. Wu Y, Brosh RM. DNA helicase and helicase–nuclease enzymes with a conserved iron– sulfur cluster. Nucleic Acids Res. 2012;40:4247-60.

229

Figure 7.1. A PRM-based targeted proteomic approach for interrogating the human helicase proteome. (a) A Venn diagram depicting the numbers of helicases included in the

PRM library and those that could be quantified by the PRM method. (b) A bar graph showing the coverages of different groups of helicases in the PRM library. (c) The experimental strategy, involving the use of forward SILAC labeling together with the

PRM-based targeted proteomic analysis, for assessing the alterations in expression of helicase proteins in human cells upon treatment with the two HSP90 inhibitors.

230 a Human Helicases (95 non- b redundant helicases) 70 60 4 50 Quantified Helicases (~ 95 40 total helicases, ~75 30 60 non-redundant 11 20 helicases) 41

Numbers of Numbers Helicases 10 20 0 DNA RNA Other Library Helicases (121 total helicases, Helicases Helicases Helicases 80 non-redundant helicases) In Library Not in Library c

Onalespib Alvespimycin

Q1 Q2 Q3 100 100 Cell Lysis Tryptic Digestion and 80 80 HSP90i-Treated Cells Mix at 1:1 ratio LC-PRM Analysis 60 60

Light Amino Acids 40 40 Relative

20 20 Abundance

Precursor Fragmentation Fragment 0 0 Selection Selection Retention DMSO-Treated Cells Time Heavy Amino Acids

231

Figure 7.2. The alterations in expression levels of helicase proteins in M14 cells after treatment with two small-molecule HSP90 inhibitors, onalespib (a) and alvespimycin (b).

The cells were treated with 100 nM inhibitor for 24 h. Displayed are the ratios of expression of helicase proteins in HSP90 inhibitor-treated over mock-treated M14 cells, where the Y- axis was plotted in log10 scale. The data represent the average ratios obtained from two biological replicates (i.e. one forward and one reverse SILAC labeling experiments). The red and blue bars designate those helicases that were up- and down-regulated, respectively, by at least 1.5-fold upon the inhibitor treatment.

232 a

0.5

0.2

0.1

/DMSO

BLM

BAT1

DDX1 DDX5

CHD5 CHD1

ARIP4

BTAF1

DDX13 DDX29 DDX31 DDX37 DDX58 DDX11 DDX15 DDX17 DDX18 DDX20 DDX21 DDX23 DDX24 DDX27 DDX30 DDX32 DDX33 DDX36 DDX38 DDX39 DDX42 DDX46 DDX47 DDX49 DDX50 DDX51 DDX54 DDX55 DDX59

CETN2

DDX3X ASCC3 DDX2A DDX2B DDX3Y

CHTF18

DDX19A DDX19B DDX39B

5 Onalespib

2 1

0.5

0.2

0.1

WRN

TTF2

IFIH1

HELZ

HELB NAV2 SUV3

DDX6 DDX8 DDX9 DNA2 TIP49

INO80

MCM4 MCM7

HELZ2

FANCJ

DHX16 G3BP1 DHX30 DHX34 DHX57

RENT1

SCAR1

EIF4A2

ERCC3 MOV10 XRCC6 ERCC2 SMBP2 XRCC5

RECQL SRCAP

SHPRH

HELIC1 HELIC2

DICER1

GTF2F2

FBXO18

RAD54B

SKIT2L2

ERCC6L

YTHDC2

SMARCA1 SMARCA2 SMARCA5 SMARCAD1 b

2 1

0.5

/DMSO 0.2

0.1

BLM

BAT1

DDX1 DDX5

CHD3 CHD1 CHD4 CHD5 CHD6

BTAF1

DDX15 DDX20 DDX27 DDX30 DDX35 DDX11 DDX13 DDX16 DDX17 DDX18 DDX21 DDX23 DDX24 DDX29 DDX31 DDX32 DDX33 DDX37 DDX39 DDX40 DDX42 DDX46 DDX47 DDX49 DDX51 DDX52

CETN2 CHD1L

ASCC3 DDX2A DDX2B DDX3X DDX3Y

CHTF18

DDX39A DDX39B

10 Alvespimycin 5

2 1

0.5

0.2

0.1

WRN

TTF2

IFIH1

DDX8 DDX6 DDX9 DNA2 PEO1 TIP49

INO80

MCM6 MCM4 MCM7

HELZ2

HELLS

FANCJ

DHX57 DDX54 DDX55 DDX58 DHX16 DHX30 DHX34 G3BP1

RENT1

SCAR1

EIF4A2

ERCC3 XRCC6 ERCC2 XRCC5

MOV10

RECQL SRCAP

HELIC1 HELIC2

GTF2F2 SUV3L1

FBXO18

RAD54B

ERCC6L

SKIV2L2

YTHDC2

IGHMBP2

SMARCA4 SMARCA1 SMARCA5 SMARCAD1

233

Figure 7.3. Extracted-ion chromatograms for monitoring representative helicases (MCM4,

MCM7, RUVBL1 and EIF4A1) in M14 cells with or without a 24-h treatment with 100 nM onalespib. The peptide sequences and the transitions employed for plotting the ion chromatograms are listed in the figure.

MCM4, ATPAQTPR, dotp=0.82 MCM7, SEDDESGAGELTR, dotp=0.91

RUVBL1, VPFCPMVGSEVYSTEIK, dotp=0.95 EIF4A1, DFTVSAMHGDMDQK, dotp=0.93

234

235 a b 0.8

1 0.4 Forward Reverse /DMSO) 0 0 -1

(Reverse) -0.4

10 -2 0 96 0 DDX11

R2 = 0.68 log Alvespimycin -0.8 -3 ( DDX32 10 DDX34 R2 = 0.50 -3 -2 -1 0 1 -1.2 log10(Forward) log -1.2 -0.8 -0.4 0 0.4 0.8 log10(Onalespib/DMSO) c d Onalespib DMSO

DDX11 DHX34 1 1 DDX11 0.5 0.5 DHX34 0 0 82 84 86 19 20 MCM4 MCM4 MCM7 MCM7 1 1

ARAF Abundance Relative 0.5 0.5 0 0 Actin 16 18 32 34 e f Retention Time

1 1

PRM Western blot /DMSO) PRM Western blot

0.8 0.8 /DMSO) 0.6 0.6

0.4 0.4 Onalespib

0.2 Alvespimycin 0.2

0 0

Ratio ( Ratio Ratio ( Ratio

236

Cas9 method. (b) Experimental strategy for combining forward SILAC labeling with the

LC-PRM-based targeted proteomic approach for the identification of cellular proteins that can interact with HSP90 . (c) Representative PRM traces showing the relative quantification results of MCM4 and MCM7 from the anti-Flag pull-down mixture in

HEK293T cells with or without the integration of tandem affinity tag to the C-terminus of

HSP90 protein from both forward and reverse SILAC labeling experiments. (d) A Venn diagram depicting the number of helicases that could bind with HSP90 , i.e., those helicases that could be enriched from affinity pull-down from Flag-HSP90 cells, and that could be down-regulated upon HSP90 inhibitor treatment.

237 a b ‘Light’ cells with ‘Heavy’ cells 293T HSP90TAPtag

MW (kDa)

100 Anti-Flag Immunoprecipitation with Immunoprecipitation with anti-FLAG M2 affinity resin anti-FLAG M2 affinity resin 100 Anti-HSP90

46 Anti-actin Mix at 1:1 ratio

On beads digestion and LC-PRM c d Forward SILAC Reverse SILAC

293T Taptag Onalespib, Alvespimycin, 293T Taptag ratio<0.67 ratio<0.67 MCM4 1 1 17 14 0.5 0.5 5 0 0 8 77 78 79 80 77 78 79 80 12 MCM7 1 1 1

Relative Abundance Relative 0.5 0.5 19

0 0 25 26 27 28 25 26 27 28 Retention Time Taptag, ratio>1.5

238

Figure 7.6. Validation for the incorporation of Flag-tag in endogenous HSP90β protein. (a)

Immunoprecipitation followed by Western blot analysis for validating the interactions between HSP90β and ARAF, and the lack of interaction between HSP90β and EGFR. (b)

Immunoprecipitation followed by LC-MS/MS analysis led to the identification of HSP90β with a 65% sequence coverage. Amino acid sequences highlighted in red were identified from LC-MS/MS analysis.

239 a Flag-pull down Input

Flag-HSP90β HSP90 HSP90 HSP90 ARAF ARAF

EGFR EGFR

Actin Actin b

MPEEVHHGEEEVETFAFQAEIAQLMSLIINTFYSNKEIFLRELISNASDALDKI RYESLTDPSKLDSGKELKIDIIPNPQERTLTLVDTGIGMTKADLINNLGTIAKS GTKAFMEALQAGADISMIGQFGVGFYSAYLVAEKVVVITKHNDDEQYAWES SAGGSFTVRADHGEPIGRGTKVILHLKEDQTEYLEERRVKEVVKKHSQFIG YPITLYLEKEREKEISDDEAEEEKGEKEEEDKDDEEKPKIEDVGSDEEDDSG KDKKKKTKKIKEKYIDQEELNKTKPIWTRNPDDITQEEYGEFYKSLTNDWE DHLAVKHFSVEGQLEFRALLFIPRRAPFDLFENKKKKNNIKLYVRRVFIMDS CDELIPEYLNFIRGVVDSEDLPLNISREMLQQSKILKVIRKNIVKKCLELFSEL AEDKENYKKFYEAFSKNLKLGIHEDSTNRRRLSELLRYHTSQSGDEMTSLS EYVSRMKETQKSIYYITGESKEQVANSAFVERVRKRGFEVVYMTEPIDEYC VQQLKEFDGKSLVSVTKEGLELPEDEEEKKKMEESKAKFENLCKLMKEIL DKKVEKVTISNRLVSSPCCIVTSTYGWTANMERIMKAQALRDNSTMGYMM AKKHLEINPDHPIVETLRQKAEADKNDKAVKDLVVLLFETALLSSGFSLEDP QTHSNRIYRMIKLGLGIDEDEVAAEEPNAAVPDEIPPLEGDEDASRMEEVD

240

Figure 7.7. Bar graph depicting the ratios of helicase proteins obtained from affinity down from lysates expressing Flag-tagged HSP90 vs. the corresponding pull-down from lysate of parental HEK293T cells. Red bars designate those helicases with at least a 1.5-fold enrichment from lysates of HEK293T cells with a tandem affinity tag being incorporated to the C-terminus of endogenous HSP90 protein over the parental HEK293T cells.

/293T

BAT1

DDX1 DDX5

ARIP4

BTAF1

DDX15 DDX16 DDX18 DDX20 DDX23 DDX27 DDX30 DDX31 DDX33 DDX34 DDX37 DDX39 DDX41 DDX46 DDX47 DDX10 DDX17 DDX21 DDX32 DDX36 DDX42 DDX49

CHD1L

DDX3X ASCC3 DDX2A

CHTF18 DDX39B

10 Taptag

HELZ

DDX8 DDX9 DDX6

INO80

MCM4 MCM7 MCM6

FBX18

FANCJ

DDX54 DDX55 DHX57 G3BP1 DDX50 DDX59 DHX16 DHX30

RENT1

SCAR1

ERCC2 MOV10 SMBP2 XRCC5 XRCC6 ERCC3

RECQ1

HELIC2

GTF2F2

RUVBL1

SKIV2L2 YTHDC2 RECQL5

241

Figure 7.8. Helicases as putative client proteins of HSP90. The inner and middle layers represent those helicases that were both enriched from the lysates of HSP90-tagged cells and down-regulated in M14 cells upon treatment with both and one of the two HSP90 inhibitors, respectively. The outer layer designates those helicases that were enriched from the lysate of the HSP90-tagged cells but whose expression levels in M14 cells were not modulated by either of the two HSP90 inhibitors.

DDX2A

DDX21 BTAF1

HSP90

DDX55 INO80

242

MCM4/MCM7.

a b 1

0.8 Flag-pull down Input

0.6 MCM4 0.4

(Tag/293T) MCM7

10 DDX32 MCM4 MCM4 0.2

log 1.0 2.5 1.0 1.3 0 MCM7 MCM7 1.0 1.9 1.0 1.0 -0.2 Actin Actin -1.5 -1 -0.5 0 0.5 1.0 1.0

log10(Onalespib/DMSO)

243