EXPLORING THE CONTROVERSIAL ROLE OF MELK IN CANCER
Ian M. McDonald
A dissertation submitted to the faculty at the University of North Carolina at Chapel Hill in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Pharmacology in the School of Medicine.
Chapel Hill 2020
Approved by:
Adrienne D. Cox
Michael J. Emanuele
Jeanette Gowen Cook
Michael B. Major
Lee M. Graves
© 2020 Ian M. McDonald ALL RIGHTS RESERVED
ii
ABSTRACT
Ian M. McDonald: Exploring the Controversial Role of MELK in Cancer (Under the direction of Lee M. Graves)
The kinase MELK (maternal embryonic leucine zipper kinase) is upregulated in numerous cancers, and high expression levels are correlated with tumor grade, aggressiveness, radioresistance, and poor prognosis. Depletion of MELK with RNAi causes slowed proliferation and induces apoptosis in triple-negative breast cancer (TNBC) and other cancers, and these antiproliferative effects can be rescued by exogenous MELK expression, indicating they are due specifically to MELK loss. Recently, it was demonstrated that CRISPR/Cas9-mediated MELK knockout has no effect on the proliferation of TNBC and other cancer cells. These disparate results have called into question the purported requirement for MELK in cancer, and have underscored the need for a deeper understanding of MELK functions in cancer.
In this dissertation, I demonstrate that the antiproliferative effects of MELK silencing extend to pancreatic cancer (PC), and that treatment of PC cells with the chemotherapeutic agent gemcitabine upregulates MELK expression. Multiple novel MELK inhibitors are characterized and determined to impair PC cell viability, but the lead compound, UNC2025, fails to sensitize cells to gemcitabine and exhibits only moderate selectivity for MELK.
Next, I characterize the selectivity of the MELK inhibitors OTSSP167 and NVS-
MELK8a (8a) in TNBC cells, revealing that OTSSP167, the leading MELK inhibitor in the field, is poorly selective, while 8a is highly selective for MELK. I next demonstrate that MELK
iii inhibition with 8a causes delayed mitotic entry in cancer cells, in parallel with delayed activation of CDK1 and Aurora A and B. Cells overcome this delay to enter and complete mitosis after approximately 2h. 8a also causes an induction of the DNA double strand break marker p-H2AX
(S139) and activation of Chk2, suggesting MELK inhibition causes a transient activation of the
G2/M DNA damage checkpoint. I investigate protein-protein interactions using the technique
BioID, and identify CDK1/cyclin B1 and the MCM complex as putative novel MELK interactors. Collectively, this work provides the field with the rationale to use 8a instead of
OTSSP167 in future studies, on the basis of greatly improved selectivity, and establishes that
MELK inhibition causes delayed mitotic entry in cancer cells, illuminating a function of MELK in cell cycle progression.
iv
ACKNOWLEDGEMENTS
As I near the end of my Ph.D. journey, it is clear to me that I have a great number of people to thank for their support and assistance throughout the process. First, I would like to thank my mentor Lee Graves, for providing scientific guidance and fostering a supportive environment in the lab. I would also like to acknowledge the Graves lab members, past and present, who made the experience fun in the best of times, and tolerable in the worst.
Next, I will thank my thesis committee – Adrienne, Mike, Jean, and Ben – for their help during my studies. In particular, I owe a great deal of gratitude to Mike, who for some inexplicable reason seemed to take a genuine interest in my MELK project early on. I truly appreciate how willing you were to meet and discuss my thesis project, seemingly at any time, and the insights you offered. I am not sure I would have navigated through some of the more tumultuous times without those meetings. Additionally, I would like to acknowledge all of the help from collaborators throughout the course of this project, specifically the CICBDD for their help with MELK inhibitor studies, the Emanuele lab for their help with cell cycle and FoxM1 studies, the Cook lab for their help with cell cycle and MCM studies, and the Johnson lab for their help with synergy studies.
Outside of lab, I was also quite fortunate to have a supportive group of friends and family members. To start, I would like to acknowledge my friends from the Graduate Business &
Consulting Club, in particular Jake and Orrin, who served as mentors that helped to guide me in my pursuit of a more business-focused career path.
v To my parents and grandparents, I owe a great deal. Thank you for always being there to push me when I needed it, and to believe in me when I did not exactly believe in myself. Your support and love, both in my academic and athletic endeavors over the years, has propelled me a long way. To my friends from back home in Maryland, Matt, Nate, Rob, and Mike, I’m not sure how we’ve stayed so close throughout all these years. Your constant friendship has been much appreciated, and the occasional weekend trips down to NC were always something I looked forward to. To the close friends I have made during graduate school, thank you for being a source of fun and support outside of lab. In particular, I will never forget the pain of the 2016
UNC loss in the national championship game, and the subsequent redemption of the 2017 national title run, shared in large part with Robin and Kelsey. On that note, I suppose I should also thank Hickory Tavern for providing the environment where I have watched countless games with friends, and Bill, a fixture there. I don’t know your last name Bill, and you will never read this, but it was great getting to know you over the years.
Finally, I would like to give a huge thank you to my partner Allison, who has been with me throughout the majority of this grad school journey. I am unbelievably grateful for your love and support during this process. I’m not sure how you put up with me during some of the rough times in lab, or how you were able to feign interest while I subsequently vented about problems with experiments. We’ve both conquered our respective grad school challenges together, now on to our next chapter…
vi
PREFACE
During the entirety of my thesis studies (from early 2015 to early 2020), my project was focused on the kinase MELK. While the singular protein of study remained constant, the perspective from which I approached these studies certainly did not. To say the least, the opinions of the field on the requirement of MELK in cancer, and thus the validity of targeting
MELK for therapeutic purposes has become quite controversial. Numerous studies published during the course of my thesis project made it necessary for me to drastically change my approach towards studying MELK. I address, in detail, the controversy surrounding the requirement of MELK in cancer in chapter 1. Here, I would like to orient the reader as to when the studies in each chapter took place, with regards to the prevailing attitude of those in the
MELK field at that particular time.
Chapters 2, 3, and 4 contain research results from my thesis project. The experiments shown in chapter 2 were completed primarily between early 2015 and early 2017. At this point in time, MELK was widely considered a relatively new, attractive therapeutic target in multiple cancers. There was substantial interest in developing novel MELK inhibitors for therapeutic purposes, and accordingly, my project centered around identifying and characterizing a new
MELK inhibitor with therapeutic potential. In early 2017, the notion of targeting MELK in cancer was challenged. We subsequently abandoned our drug discovery efforts and turned our focus to elucidating biological functions of MELK, which were, and still are, quite unclear.
vii The results shown in chapters 3 and 4 are from experiments completed after early 2017.
The abrupt change in experimental approach and model system (from pancreatic to triple- negative breast cancer) were a result of the publications challenging the requirement of MELK in breast cancer. We sought to address this controversy by contributing to the understanding of the biological functions of MELK in cancer.
While the arrival of this paradigm-shifting controversy in the middle of my thesis work certainly made for a tumultuous, and at times seemingly hopeless project, I believe it illustrates the dynamic nature of academic research quite well. The MELK story has made it clear that it is exceedingly difficult and rare to establish any absolute truth. The lesson that everything should be challenged, and that assumptions and paradigms can always be upended, both in science and broadly in other disciplines, is a great one to have learned through experience at a young age. I hope that this work illustrates these principles, and contributes towards a more true understanding of the role of MELK in cancer.
viii
TABLE OF CONTENTS
LIST OF TABLES ...... xi
LIST OF FIGURES ...... xii
LIST OF ABBREVIATIONS ...... xiv
CHAPTER 1: INTRODUCTION ...... 1
A Brief Overview of MELK: From Discovery to Controversy ...... 1
Examining the Requirement for MELK in Cancer ...... 3
The MELK Inhibitor Landscape ...... 15
Processes Regulated by MELK in Cancer ...... 19
Regulation of MELK Expression and Activity ...... 26
MELK Substrates ...... 29
Figures...... 33
CHAPTER 2: MELK KNOCKDOWN AND PHARMACOLOGICAL INHIBITION IMPAIR PANCREATIC CANCER CELL VIABILITY ...... 34
Introduction ...... 34
Results ...... 37
Discussion ...... 41
Experimental Procedures ...... 43
Tables and Figures ...... 47
CHAPTER 3: MASS SPECTROMETRY-BASED SELECTIVITY PROFILING IDENTIFIES A HIGHLY SELECTIVE INHIBITOR OF THE KINASE MELK THAT DELAYS MITOTIC ENTRY IN CANCER CELLS ...... 54
ix Introduction ...... 54
Results ...... 57
Discussion ...... 67
Experimental Procedures ...... 71
Figures...... 79
CHAPTER 4: INVESTIGATION OF PUTATIVE MELK INTERACTORS AND COMBINATORIAL STRATEGIES TO INCREASE MELK-8A POTENCY ...... 94
Introduction ...... 94
Results ...... 98
Discussion ...... 109
Experimental Procedures ...... 114
Tables and Figures ...... 123
CHAPTER 5: CONCLUSIONS AND FUTURE DIRECTIONS ...... 138
REFERENCES ...... 143
x LIST OF TABLES
Table 2.1. MELK inhibitor potency in PC cell viability assays...... 47
Table 4.1. FoxM1 phosphosites identified by MS...... 123
Table 4.2. List of all library compounds used in the synergy screen...... 124
xi LIST OF FIGURES
Figure 1.1. Timeline of major MELK milestones in cancer...... 33
Figure 2.1. Treatment of PC cells with gemcitabine causes MELK upregulation...... 48
Figure 2.2. MELK knockdown slows proliferation of PANC-1 cells...... 49
Figure 2.3. Screening of a small-molecule library reveals potent MELK inhibitors...... 50
Figure 2.4. Novel MELK inhibitors impair PC cell proliferation...... 51
Figure 2.5. The kinase selectivity profile of UNC2025 reveals off-target kinase inhibition...... 52
Figure 2.6. MELK inhibition or knockdown do not sensitize PC cells to gemcitabine...... 53
Figure 3.1. Schematic of competition MIB/MS workflow and sample selectivity output data...... 79
Figure 3.2. 8a is a highly selective MELK inhibitor...... 80
Figure 3.3. Treatment with 8a causes decreased viability and decreased phosphorylation of Histone H3 (S10) in TNBC cells...... 82
Figure 3.4. MELK inhibition with 3 μM 8a causes delayed mitotic entry...... 83
Figure 3.5. Delayed mitotic entry caused by 8a is associated with delayed activation of Aurora A, Aurora B, and CDK1...... 84
Figure 3.6. HeLaS3 cells treated with 8a complete mitosis and begin the subsequent cell cycle following delayed mitotic entry...... 85
Figure 3.7. 8a causes a major, dose-dependent increase in G2 length...... 86
Figure S3.1. Full competition MIB/MS selectivity profiles for 1 μM 8a, HTH, and OTS...... 87
Figure S3.2. Full competition MIB/MS selectivity profiles for 10, 100, and 1000 nM 8a...... 88
Figure S3.3. Full competition MIB/MS selectivity profiles for 1, 3, and 10 µM 8a...... 89
Figure S3.4. 8a has no detectable affinity for STK11 or MAP2K4 at 3 µM...... 90
Figure S3.5. Aurora A and B are minimally inhibited by 8a in cell-free kinase assays...... 91
xii Figure S3.6. Treatment with 8a causes decreased viability in HeLaS3 cells...... 92
Figure S3.7. Additional replicates of live-cell imaging experiment measuring G2 and M phase lengths...... 93
Figure 4.1. BioID schematic...... 126
Figure 4.2. Treatment with 8a causes increased phosphorylation of H2A.X (S139) and activation of Chk2...... 127
Figure 4.3. BirA*-MELK fusion protein design...... 128
Figure 4.4. Biotinylation of proximal proteins differs between BirA*-MELK constructs and BirA* control...... 129
Figure 4.5. Analysis of BirA*-MELK MS data using IPA...... 130
Figure 4.6. BioID and IPA analysis identified putative MELK interactors with roles in cell cycle processes...... 131
Figure 4.7. Investigation of a putative MELK-CDK1/cyclin B1 interaction...... 132
Figure 4.8. MELK interacts with FoxM1 to increase FoxM1 transcriptional activity...... 133
Figure 4.9. FoxM1-S481A/D/E mutants may display differential transcriptional activity...... 134
Figure 4.10. Schematic of the screen for compounds that synergize with MELK-8a...... 135
Figure 4.11. Synergy screen top hits were identified as UNC0642, OTX015, and UNC1999...... 136
Figure 4.12. Secondary cell viability assays indicate no synergy between 8a and UNC0642, OTX015, or UNC1999...... 137
xiii LIST OF ABBREVIATIONS
8a NVS-MELK8a
ACN Acetonitrile
AMPK AMP-activated protein kinase
APC/C Anaphase promoting complex/cyclosome
APEX Ascorbate peroxidase
ASK1 Apoptosis signal-regulating kinase 1
ATM Ataxia-telangiectasia mutated
ATP Adenosine triphosphate
ATR Ataxia telangiectasia and Rad3 related
AURKB Aurora kinase B
BioID Proximity-dependent biotin identification
BirA* BirA-R118G
BLBC Basal-like breast cancer
BRET Bioluminescence resonance energy transfer
BSA Bovine serum albumin
CCNB1 Cyclin B1
CCNB2 Cyclin B2
CDC25 Cell division cycle 25
CDK Cyclin-dependent kinase cDNA Complementary DNA
ChIP Chromatin immunoprecipitation
xiv Chk1 Checkpoint kinase 1
Chk2 Checkpoint kinase 2
Co-IP Co-immunoprecipitation
CRISPR Clustered regularly interspaced short palindromic repeats
CRISPRi CRISPR interference
CV Crystal violet
DDR DNA damage response
DMEM Dulbecco’s Modified Eagle Medium
DMSO Dimethyl sulfoxide
DNA Deoxyribonucleic acid
DSB Double-strand break
DTT Dithiothreitol
ECL Enhanced chemiluminescence
EDTA Ethylenediaminetetraacetic acid eIF4b Eukaryotic translation initiation factor 4B
EMT Epithelial-mesenchymal transition
EZH2 Enhancer of zeste homolog 2
FAK Focal adhesion kinase
FBS Fetal bovine serum
FDR False discovery rate
FKBP12 FK506 binding protein 12
FKH Forkhead homology domain
FoxM1 Forkhead box protein M1
xv GFP Green fluorescent protein gRNA Guide RNA
GSC Glioma stem cell
HRP Horseradish peroxidase
HTH HTH-01-091
IKKβ Inhibitor of NF-κB kinase subunit beta
IP Immunoprecipitation
IPA Ingenuity Pathway Analysis
JNK2 c-JUN N-terminal kinase 2
KA1 Kinase associated domain-1
LC Liquid chromatography
LFQ Label-free quantification
M Mitosis
MAPK Mitogen-activated protein kinase
MCL1 Myeloid cell leukemia 1
MCM Minichromosome maintenance
MDM2 Mouse double minute 2
MDMX Mouse double minute 4
MEK Mitogen-activated protein kinase kinase
MELK Maternal embryonic leucine zipper kinase
MIB/MS Multiplexed kinase inhibitor beads/mass spectrometry
MIBs Multiplexed kinase inhibitor beads
MLST8 mTOR associated protein, LST8 homolog
xvi MPF Mitosis-promoting factor
MPK38 Murine protein serine/threonine kinase 38 mRNA Messenger RNA
MS Mass spectrometry mTOR Mechanistic target of rapamycin
NF-κB Nuclear factor kappa B
NRD N-terminal repressor domain
OTS OTSSP167 p53 Tumor protein 53
PAGE Polyacrylamide gel electrophoresis
PARP Poly (ADP-ribose) polymerase
PBS Phosphate buffered saline
PC Pancreatic cancer
PCNA Proliferating cell nuclear antigen
PCR Polymerase chain reaction
PDK1 3-phosphoinositide-dependent protein kinase-1
PI3K Phosphoinositide 3-kinase
PIC Protease inhibitor cocktail
PLA Proximity ligation assay
PLK1 Polo-like kinase 1
PRAS40 Proline-rich Akt substrate of 40 kDa
PVDF Polyvinylidene difluoride
RhoA Ras homolog family member A
xvii RNA Ribonucleic acid
RNAi RNA interference
RNF168 Ring finger protein 168
RPA1 Replication protein A1
RT Reverse transcription
SAC Spindle assembly checkpoint
SDS Sodium dodecyl sulfate
SEER Surveillance, Epidemiology, and End Results shRNA Short hairpin RNA siRNA Small interfering RNA
SQSTM1 Sequestosome 1
STOX1 Storkhead box 1
STRAP Serine/threonine kinase receptor associated protein
TAD Transactivator domain
TAM TYRO3-AXL-MERTK
TCGA The Cancer Genome Atlas
TNBC Triple-negative breast cancer
TOP2A DNA topoisomerase 2-alpha
UBA Ubiquitin-associated domain
ZPR9 Zinc finger-like protein 9
xviii
CHAPTER 1: INTRODUCTION1
A Brief Overview of MELK: From Discovery to Controversy
Throughout this chapter, it will become clear that the essentiality of MELK in cancer has recently come under scrutiny. Some have even questioned whether MELK has any role in cancer. Before deeply analyzing that controversy, I will begin with a brief chronological history of MELK to provide context (see Fig. 1.1 for an accompanying timeline).
In 1997, MELK cDNA was cloned for the first time, in two studies published in rapid succession. The first group sought to identify genes that were important in embryonic genome activation. They ultimately cloned the cDNA of a gene containing a kinase domain sharing significant sequence identity with the Snf1/AMPK family kinases. The cloned kinase also contained a leucine zipper motif, and transcription of this gene was increased in mouse preimplantation embryos. Accordingly, the kinase was named MELK (maternal embryonic leucine zipper kinase) (1). Shortly after this, another group cloned the same gene from mice and named it MPK38 (murine protein serine/threonine kinase 38) (2). Five years later, the MELK homolog was characterized in Xenopus and named pEg3 (3), not to be confused with PEG3
(paternally expressed gene 3) (4). In this study, pEg3 was demonstrated to be phosphorylated in a cell cycle-dependent manner to regulate its activity, which is maximal in mitosis (3). Naming
1A minor portion of this chapter was originally published in the Journal of Biological Chemistry. The original citation is as follows: McDonald, I. M., Grant, G. D., East, M. P., Gilbert, T. S., Wilkerson, E. M., Goldfarb, D., Beri, J., Herring, L. E., Vaziri, C., Cook, J. G., Emanuele, M. J., and Graves, L. M. (2020) Mass spectrometry-based selectivity profiling identifies a highly selective inhibitor of the kinase MELK that delays mitotic entry in cancer cells. J. Biol. Chem. 295, 2359–2374
1 conventions converged on MELK within a few years, though the name MPK38 is still used sporadically in the literature.
In 2005, the first study characterizing MELK as a promising cancer target was published.
Here, it was noted that MELK RNA levels were elevated in a panel of 20 cancer tissues, relative to matched normal samples, and that siRNA-mediated MELK knockdown decreased proliferation of cervical, breast, colorectal, and pancreatic cancer cell lines (5). This publication spawned dozens of subsequent studies over the next 15 years, demonstrating increased MELK expression in numerous cancers, including glioblastoma, breast, prostate, and gastric, and slowed proliferation of these and other cancers as a result of RNAi-mediated MELK depletion (6–9). As the body of literature implicating MELK as a promising therapeutic target grew, the first MELK inhibitor (OTSSP167 (OTS), also called OTS167) was developed in 2012 (10). OTS effectively impairs growth and induces apoptosis of numerous cancer types, including breast cancer, acute myeloid leukemia, and small cell lung cancer, and is currently in four clinical trials (10–12).
Because of its status as the leading MELK inhibitor, OTS has been used in nearly all functional studies of MELK since its development. It has recently been definitively demonstrated that OTS has extremely poor selectivity for MELK (13–16), yet unfortunately, it is still routinely used as a tool compound to investigate MELK function.
The start of the controversy currently surrounding MELK can be traced back to a comprehensive 2014 study, in which MELK was demonstrated to be essential for the proliferation of basal-like breast cancer (BLBC) cells, one of the most aggressive subtypes of breast cancer. This study utilized RNAi-mediated MELK knockdown to demonstrate slowed proliferation and essentiality, and even showed that growth effects in cells and tumors could be rescued with exogenous wild-type MELK expression, but not with kinase-dead MELK (17).
2 Three years later, in 2017, another group used CRISPR/Cas9 genomic knockout to show that
MELK is not required in triple-negative breast cancer (TNBC) and other cancer types (18). This publication was quickly followed by a study from the group that originally showed essentiality, now also claiming that MELK is not required for TNBC cell proliferation, based upon genomic
MELK knockout and other experimental techniques (16). To further complicate the situation, members of this same group published another study in 2018, asserting that TNBC cells actually have a conditional dependency on MELK for proliferation (19). One could forgive the reader for being somewhat confused at this point; the majority of the MELK field likely is as well. We will next more closely examine the evidence on both sides of the controversy concerning the requirement of MELK in cancer cells.
Examining the Requirement for MELK in Cancer
Evidence supporting the requirement for MELK in cancer
Perhaps the most robust evidence implicating MELK as an important mediator of cancer progression comes in the form of expression analyses between cancerous and normal tissues and cells. Numerous studies have used microarray and TCGA analysis or immunoblotting methods to demonstrate that expression of MELK RNA or protein is significantly increased in neoplastic cells. This effect seems to be a characteristic that broadly defines many cancer types, as it has been described in breast (17, 20, 21), brain (glioma (6, 22), astrocytoma (23), and neuroblastoma
(24)), liver (25), prostate (8), bladder (26), and endometrial cancers (27), among others (5).
Further, high levels of MELK expression correlate with high-grade tumors, increased aggressiveness, poor patient outcomes, and radioresistance (17, 20, 21, 24–28). Increased
MELK expression has been linked to concurrent upregulation of genes important or essential for
3 cell cycle progression including CDK1, CCNB1/2, TOP2A, AURKB, PLK1, and BUB1 (8, 23,
29–32), suggesting MELK likely also plays a role in this process.
The effects of RNAi-mediated MELK knockdown have resolutely indicated that MELK expression is vital to the proliferation and survival of cancer cells. In cancers including basal- like breast (17), endometrial (27), glioma (6), acute myeloid leukemia (11), high-risk neuroblastoma (24), and hepatocellular carcinoma (25), MELK depletion has been shown to slow or halt proliferation of cancer cells and tumors. Some studies have additionally demonstrated that knockdown of MELK sensitizes cells to radiation (17, 21, 33, 34) and impairs migration (8, 9, 27, 35). Crucially, a number of studies have rescued the antiproliferative effects of RNAi-mediated MELK knockdown with ectopic MELK expression, indicating that these effects are specifically due to loss of MELK (6, 17, 22, 25, 36, 37). These results were extended by two studies that demonstrated that complementation of knockdown with kinase-dead MELK-
D150A or -T167A does not restore normal growth, indicating that catalytic activity is required for rescue (17, 37). Additionally, the oncogenic potential of MELK has been shown using rodent fibroblasts that expressed a dominant negative form of p53 (Rat1-p53DD), which necessitate only one oncogenic event for neoplastic transformation. When wild-type, but not kinase-dead,
MELK was overexpressed, these cells gained the ability to grow in an anchorage independent manner and form tumors in vivo, evidence of transformation indicating that MELK may function as a driver oncogene (17).
It was recently demonstrated that CRISPR/Cas9-mediated MELK knockout impaired proliferation and induced apoptosis in chronic lymphocytic leukemia cells (35). It should be noted that this is the only study that has shown a growth effect resulting from MELK knockout.
4 Other studies that demonstrate no phenotype in MELK knockout cells will be discussed in the next section.
Multiple MELK inhibitors have been developed as potential cancer therapeutics
(discussed in more detail later in this chapter). The molecules OTS (10), NVS-MELK8a (8a)
(14, 38), HTH-01-091 (16), MELK-T1 (39), IN17 (40), and C1 (22) have all demonstrated preclinical efficacy in slowing cancer cell proliferation, with OTS exhibiting such potent, broad anti-neoplastic effects that it has advanced to phase I/II clinical trials (11, 12, 24, 35, 41–43).
Certainly, all small-molecule inhibitors exhibit some degree of polypharmacology, so the antiproliferative effects of these compounds cannot be solely attributed to MELK inhibition. For the inhibitor 8a, the cell viability of TNBC cells has been shown to be reduced at concentrations at which 8a is highly selective for MELK, suggesting that inhibition of MELK is a major contributor to its antiproliferative effects (14).
Evidence against the requirement for MELK in cancer
On the other side of the controversy are a series of studies that used CRISPR/Cas9 to genetically knockout MELK, revealing no growth phenotype in MELK null cell lines. The first of these studies used both single and double guide RNA (gRNA) strategies to genetically delete
MELK from a panel of TNBC cell lines, with some experiments additionally completed in other cancers. For the single gRNA strategy, seven independent gRNAs were used to generate clonal
MELK null A375, Cal51, and MDA-MB-231 cell lines, and these lines were compared to
Rosa26 (a non-essential, non-coding gene) knockout control lines in growth assays. Both cell proliferation and anchorage independent growth were unaffected by MELK knockout. The double gRNA strategy was used to induce DNA cutting in two places, excising a portion of the
5 MELK gene that encodes for residues essential for ATP binding in two TNBC cells lines. Long- term growth assays were repeated with these MELK knockout cell lines, and again they were found to proliferate normally, indicating that MELK is not required for TNBC proliferation (18).
Reasoning that signaling pathway adaptations may have occurred during the extensive time required for clonal selection, which could restore normal cell proliferation, the authors employed a GFP dropout assay to assess cell proliferation more proximal to MELK deletion.
Briefly, seven Cas9-expressing TNBC cell lines were transduced with seven independent gRNAs targeting MELK or three gRNAs targeting Rosa26, RPA, or PCNA. Transductions were done at a low MOI to ensure that some portion of cells remained untransduced. gRNA plasmids additionally expressed GFP, which allowed for the proportion of GFP+ to GFP- cells to be monitored over five passages to assess the relative fitness of cells that had been transduced with
MELK or control gRNA relative to their untransduced counterparts. Less than 2-fold dropout of
MELK and Rosa26 knockout cells was observed, compared to 5- to 100-fold dropout of the positive control RPA or PCNA knockout cells, again suggesting that MELK is not a requirement for TNBC proliferation. Crucially, the methods for this study indicate that the baseline
GFP+/GFP- measurement was not taken until 3 days post-transduction, with the first reading assessing GFP dropout measured 3-4 days later. While the authors state that this GFP dropout assay was used in an effort to test the effects of MELK knockout immediately following MELK loss, the 3 days that passed between gRNA transduction and baseline readings could certainly still allow ample time for cellular reprogramming that restores normal growth (18).
The next CRISPR study replicated many of the results described above, while also using additional techniques to test MELK dependency (16). Again, MELK null MDA-MB-468 cells were observed to have no growth phenotype relative to control cells. A chemical-induced
6 protein degradation approach was additionally employed to show that MELK loss had no immediate or prolonged effects on BLBC proliferation. The authors state that MELK expression was maintained throughout the process of creating these cell lines, by first stably expressing a
FKBP12-MELK fusion that could be selectively degraded with an engineered degrader molecule, then deleting endogenous MELK with CRISPR. Importantly, it was never demonstrated that this MELK fusion is similarly functional to endogenous MELK (i.e. interactions and substrate phosphorylations are unperturbed by FKBP12 fusion). It therefore cannot be ruled out that this cell line is simply another MELK knockout line, which has had ample time during clonal selection for reprogramming of signaling pathways, that additionally expresses a non-functional MELK fusion.
This study also demonstrated quite convincingly that MELK knockdown with CRISPR interference (CRISPRi) has no effect on BLBC cell proliferation. MDA-MB-468 cells were transduced with KRAB-dCas9, a catalytically dead version of Cas9 that allows for transcriptional repression of the gRNA-targeted gene. Multiple gRNAs targeting the MELK promoter were shown to efficiently decrease MELK transcript and protein levels, and five of these gRNAs were cloned into a doxycycline-inducible vector and stable cell lines were generated. Using this system, MELK knockdown by doxycycline treatment was not found to significantly decrease cell proliferation in any of the five cell lines. It should be noted that, while not statistically significant, all cell lines did exhibit slightly decreased growth (~5-20%) when treated with doxycycline, compared to untreated controls (16).
A follow-on study to the first MELK CRISPR paper was completed by the same group roughly a year later, in which they expanded upon their previous results (44). In stark contrast to previous results by Wang et al. (17), MELK overexpression was found to be insufficient for
7 neoplastic transformation of immortalized cell lines as measured by anchorage-independent growth assays (44). Multiple cell models were used to test transformation potential, including the Rat1-p53DD system used in the study by Wang et al. Similarly, MELK knockout TNBC, melanoma, and colorectal cell lines did not exhibit impaired anchorage-independent growth.
This study additionally found that genetic deletion of MELK did not impair proliferation of cancer cells plated at varying densities in crystal violet assays, and had no effect on cancer cell sensitivity to five common chemotherapeutic agents or to metabolic stresses including hypoxia, glucose limitation or exposure to reactive oxygen species. CRISPR-mediated MELK knockout also had no effect on the proliferation of tumor xenografts in mice (44).
Commentary on the MELK Requirement Controversy
I will first discuss the specific evidence supporting a requirement for MELK in cancer, then the evidence rebutting a MELK requirement, before providing overall commentary on the controversy, including gaps in knowledge that should be addressed to improve our understanding of the role of MELK in cancer.
Some of the results suggesting that MELK is a cancer requirement have been recently refuted or can now be viewed with added perspective. Specifically, neoplastic transformation of immortalized cells was demonstrated with MELK overexpression in one model (17), but these results were later convincingly refuted by another group using the same and additional models
(44). Multiple MELK inhibitors have demonstrated antiproliferative effects, but it is not possible to conclude that any of the observed effects are due solely to MELK inhibition. The possibility of off-target inhibition contributing to observed phenotypes cannot be discounted. Results from dozens of studies have demonstrated that MELK expression is upregulated in cancer, and that
8 higher levels of MELK correlate with tumor grade and poor prognosis. However, these results are purely correlative in nature. It has been suggested that MELK may be upregulated as part of a cell cycle/mitotic cluster of genes in cancer cells simply due to the fact that these cells proliferate more rapidly than their non-neoplastic counterparts (44). None of these results provide strong evidence that MELK is a cancer dependency.
Other results provide more convincing evidence of MELK’s importance in cancer.
Numerous studies have demonstrated that RNAi-mediated MELK knockdown slows proliferation, induces apoptosis, and decreases migration and radioresistance of cancer cells, providing quite compelling evidence that MELK is required in cancer cells. While it has been documented that RNAi approaches often have off-target silencing effects (45), the use of multiple sh- or siRNA in a single study that cause similar phenotypes suggests that depletion of the common target of the sh/siRNA (i.e. MELK) was responsible for the observed phenotypes.
When this is expanded to a large number of studies (~30) completed by separate groups, all showing evidence of similar phenotypes in varied cancers, it provides quite strong evidence that the phenotypes observed are due to MELK loss. Even stronger evidence that the observed proliferative effects are due specifically to MELK loss were provided by the studies showing successful rescue experiments. Six studies, all completed by different groups, have rescued the effects of MELK knockdown with exogenous MELK expression (6, 17, 22, 25, 36, 37), including at least two that demonstrated a failure to rescue with kinase-dead MELK (17, 37).
The only partially contrasting results come from the study that used CRISPRi to show that depletion of MELK transcript had no significant growth effect in TNBC cells (16). This has only been demonstrated in one study, compared to a large body of work demonstrating that
MELK depletion with RNAi has antiproliferative effects. Further, it is important to note that
9 CRISPRi is an orthogonal approach to RNAi, not an identical one. Differences in observed effects could therefore be due to technical differences between the approaches. This CRISPRi result also does not account for or explain the results observed with rescue experiments. The only effort ever made to rebut rescue experiment results appeared in the discussion of the paper that showed no growth phenotype following CRISPRi-mediated MELK depletion. The authors stated that, “Since the previous study was able to rescue the antiproliferative activity observed for shMELK-2 using an shRNA-resistant MELK (Wang et al., 2014), we postulated that the potential off-target of shMELK-2 might only manifest its effect in the presence of MELK knockdown, a so-called ‘synthetic lethal’ interaction.” Absent any data to support this claim, we find this to be an extremely unlikely scenario to have occurred in one study, and a near- impossibility when considering that six independent groups, all using different RNAi knockdown and rescue reagents, have successfully rescued growth phenotypes with exogenous MELK.
On the other side of the controversy, there are three studies that have demonstrated that
CRISPR-mediated MELK knockout has no effect on proliferation of TNBC and other cancer cells (16, 18, 44). Recently, contrasting results have been presented, indicating that MELK knockout induces apoptosis and slows proliferation of chronic lymphocytic leukemia cells (35), though the former three studies were more thorough in their analyses of MELK deletion and subsequent growth assays. It is possible that technical or cell line-dependent differences account for the conflicting results between these studies. Importantly, in the studies that showed no growth phenotype as a result of MELK knockout, the methods used do not preclude the possibility that cellular reprogramming of signaling pathways occurred to circumvent MELK loss and restore normal growth.
10 While these studies showed compelling evidence that MELK is not a cancer cell dependency, efforts should be made to more comprehensively characterize MELK knockout cell lines. There is at least one documented instance of purported knockout cell lines actually still expressing splice variants of the targeted protein. Bub1 was thought to be essential for proper functioning of the spindle assembly checkpoint (SAC), until CRISPR-mediated Bub1 knockout cells were shown to have a functional SAC. Recently, however, RT-PCR and MS-based approaches were used to show that these Bub1 “knockout” cells still expressed low levels of alternatively spliced Bub1 or Bub1 protein with a small deletion. Subsequent experiments demonstrated that Bub1 is indeed essential to the SAC, and that very low levels of Bub1 were sufficient to restore normal SAC functioning (46–48). This level of evasion of CRISPR- mediated gene deletion is likely more of an exception than a rule. Nonetheless, MS-based approaches should be used to exclude the possibility that MELK knockout cell lines express alternatively spliced MELK, or other gene- or protein-level adaptations that evade complete
MELK knockout.
Overall, assuming no technical shortcomings, CRISPR-based results indicate that MELK is not a strict requirement for cancer cell proliferation. The abundant RNAi studies, however, indicate that MELK does play an important, yet poorly defined role in cancer. How does the field rectify this apparent paradox? One possible explanation is that MELK is dispensable under some conditions, but required, or at least highly important, under others. In this vein, members of the group that initially labeled MELK essential for BLBC cell proliferation (17), and later said it was not necessary for proliferation of these cells (16), recently published a study indicating that TNBC cells have a conditional dependency on MELK for growth (19). In this study, TNBC cells were transduced with a gRNA (plus Cas9) that caused efficient reduction of MELK
11 expression, and selected for infected cells with antibiotic. This approach, which utilized a heterogeneous population of cells instead of clonal selection, is more akin to depletion of MELK expression than total knockout. CRISPR-mediated depletion of MELK was shown to have no effect on the proliferation of TNBC cells in short-term growth assays, in which cells were seeded at medium to high density (MDA-MB-231 cells, 50,000 cells/well seeded on a 12-well plate, 3- day assay). However, when MELK was depleted and cells were seeded at low density (MDA-
MB-231 cells, 500 cells/well seeded on a 12-well plate, 10-day assay), they exhibited markedly slower proliferation relative to control cells. These experiments were also completed with gRNAs targeting the essential mitotic genes AURKB and PLK1, and the oncogenes KRAS and
MYC. CRISPR-mediated depletion of AURKB and PLK1 slowed proliferation of cells seeded at high or low densities, indicating that the effects of knockout of truly essential genes occur independently of assay conditions. Depletion of KRAS and MYC caused moderate decreases in proliferation (50%) when cells were seeded at high densities, and more markedly impaired proliferation (>90%) when seeded at lower densities (19). Collectively, these results suggest that
MELK is likely not universally essential in cancer cells, but under specific growth conditions
MELK expression seems to be important. MELK may conditionally function as an oncogene without being a strict essentiality for cancer cells.
Authors of some of the CRISPR studies have suggested that MELK does not play an important role, or may not even play any role, in mammalian biology and the cell cycle progression of cancer cells (18, 44). However, the fact that MELK expression is cell cycle regulated (17, 27, 49, 50), and that the upregulation of MELK in cancer correlates with upregulation of many other important or essential mitotic genes, including CDK1, CCNB1/2,
TOP2A, AURKB, PLK1, and BUB1 (8, 23, 29–32), suggests that MELK plays a role in cancer
12 cell proliferation despite negative results from CRISPR studies. Why would MELK be cell cycle regulated and consistently upregulated in concert with many essential mitotic proteins if it did not contribute to these processes? There are a few possible explanations for the observations that
MELK expression is tightly correlated with the cell cycle, yet MELK knockout does not perturb growth. In line with the conditional dependency hypothesis, it is possible that MELK acts as a functional redundancy for a specific cell cycle pathway, such that MELK is nonessential during normal cell cycling, but becomes required under certain conditions. Conversely, there may be a functional redundancy for MELK, such that total loss of MELK (i.e. CRISPR knockout) allows for compensatory reprogramming of signaling networks, while RNAi-mediated partial MELK depletion does not trigger the reprogramming necessary to shift to this redundant pathway.
The recent controversy concerning the requirement for MELK in cancer has underscored a few important scientific principles. First, it is imperative not to overinterpret negative results, as there are a wide variety of potential explanations for the lack of any observed phenotype, ranging from functional or biological to purely technical or methodological, as the Bub1
CRISPR story illustrates. Second, technical considerations, while often only given cursory attention in publications, are vitally important. The observation that CRISPR-mediated MELK depletion has differing effects in TNBC cells depending upon cell density serves as an illustration of this. Third, when orthogonal approaches do not give identical results, it does not necessarily mean that one result is correct, while the other is incorrect. Rather, a more complex and interesting biological explanation may be underlying the seemingly discordant results.
To expand upon the third principle in the context of the MELK controversy, authors of the CRISPR studies have attempted to conclude that MELK is not important to cancer cell proliferation, and that MELK is not a viable therapeutic target. Neither of these assertions are
13 definitively supported by their results demonstrating no growth phenotype following CRISPR- mediated MELK deletion. While these studies have provided strong evidence that MELK is not essential to cancer cells, they have done nothing to rebut the studies that coupled RNAi depletion with rescue, indicating that MELK is important for proliferation, and the therapeutic viability of inhibiting MELK should not yet be fully discounted, particularly in combination with other chemotherapeutic agents. It is vitally important to recognize that CRISPR-mediated genetic knockout, RNAi-mediated transcript depletion, and pharmacological inhibition are three orthogonal, but fundamentally different approaches. While some view CRISPR technologies as a universal improvement over RNAi-based approaches, this view is not entirely accurate. Off- target effects are less prevalent with CRISPR than RNAi, but both techniques still have their advantages and limitations (51–54). Results obtained with RNAi knockdown and subsequent rescue, long considered the gold standard for functional studies, still hold tremendous value, as this approach mitigates the off-target effects of RNAi. Further, while CRISPR technologies are undeniably powerful, they are also still relatively new, with CRISPR first being used for genome editing in 2013 (55). Some aspects of CRISPR/Cas9 gene editing are still not entirely understood, including central parts of the technology such as prediction of the on- and off-target
DNA cutting efficiency of specific gRNAs (51, 52). The scientific community is still uncovering some of the intricacies of this powerful, yet complex approach. As such, negative results obtained with CRISPR should be continually re-evaluated as new information becomes available. Finally, the advent and widespread implementation of CRISPR for functional studies does not render RNAi-based approaches obsolete, particularly when coupled with exogenous complementation.
14 The field should continue to investigate the controversy concerning the requirement of
MELK in cancer. Future efforts to better understand the functions of MELK in cancer will likely shed light on the mechanism underlying discordant results observed with CRISPR and RNAi, which could be broadly applicable to functional studies of other proteins using these techniques.
Some studies have begun to investigate the specific conditions under which MELK expression seems to be required, but the conditions tested thus far are certainly not comprehensive (19, 44).
Efforts to elucidate the specific conditions under which MELK is important for cancer cell proliferation will be a crucial step towards a more complete functional understanding of this kinase. There are seemingly more questions than answers in the MELK field at the present moment. In subsequent sections, we will discuss the small-molecule tools available to study
MELK and what is known about MELK functions, regulation, and substrates.
The MELK Inhibitor Landscape
The first MELK inhibitor to be developed, and far and away the most widely-used
MELK inhibitor, is OTSSP167 (OTS) (10). This inhibitor has demonstrated potent anti- proliferative effects against numerous cancers including TNBC (17), multiple myeloma (42), acute myeloid leukemia (11), neuroblastoma (28), and small cell lung cancer (12).
Consequently, OTS has advanced to phase I/II clinical trials for patients with advanced TNBC or refractory or relapsed leukemia (NCT01910545, NCT02795520, NCT02768519,
NCT02926690). It has recently been demonstrated that OTS has extremely poor selectivity for
MELK, both in biochemical assays and in cells, and exerts its antiproliferative effects through inhibition of many kinases, some of which are vital mitotic kinases (13–16). Nonetheless, the polypharmacology that is characteristic of OTS is clearly effective at halting cancer cell proliferation and inducing apoptosis, and thus the rationale for its inclusion in clinical trials
15 remains sound, provided that MELK expression levels are not used to set patient inclusion/exclusion criteria or to stratify experimental groups.
OTS should not be used as a tool compound to study the specific effects of MELK inhibition on cellular processes, as the potential to misattribute observed phenotypes to specific
MELK inhibition is high. In the original study that described the development of OTS, this molecule was described as a highly selective MELK inhibitor, despite a complete dearth of data to support this claim (10). Unfortunately, this has caused a multitude of groups to subsequently use OTS for functional MELK studies. This includes multiple studies that have primarily or exclusively used OTS to draw conclusions about MELK biology (28, 43, 56–58). Additionally, many studies continue to claim that OTS is highly selective, also without supporting data, and despite the fact that multiple studies have been published definitively demonstrating the opposite
(13–16). Conclusions about MELK function drawn solely through experiments using OTS must be interpreted with extreme caution, and should be repeated with RNAi or CRISPRi approaches, or other MELK inhibitors, for validation.
Additional MELK inhibitors have been described and characterized since the development of OTS, although none have garnered serious attention as potential therapeutic options for cancer treatment. Cpd1 and Cpd2 exhibit nanomolar potency for MELK with good selectivity against a panel of 50 kinases, and treatment of multiple cancer cell lines with Cpd2 induces PARP cleavage (59). Compound C1 also showed nanomolar potency for MELK in in vitro assays, though it is a noted multi-kinase inhibitor and was originally designed to inhibit the
Aurora kinases. Treatment of glioma stem cells with this inhibitor caused mitotic arrest, induced apoptosis, and sensitized cells to radiation (22, 60). MELK-T1 (61) and MELK-T2 (62) both have demonstrated nanomolar potency for MELK. The former compound is the more selective
16 of the two, with only 6 kinases (including MELK) out of 235 tested using an in vitro assay inhibited greater than 50% at 1 μM. Phenotypic effects of these inhibitors were not explored in these studies. The inhibitors NVS-MELK8a (8a) and NVS-MELK8b exhibited low nanomolar potency and high selectivity for MELK, as determined using an in vitro assay against a panel of
456 kinases. These molecules had antiproliferative effects against TNBC cells, causing increased apoptosis, polyploidy, and G2/M accumulation, but not luminal breast cancer cells
(38). Another low nanomolar inhibitor of MELK, HTH-01-091 (HTH), showed fairly good selectivity, with only 4% of 141 kinases inhibited greater than 90% by 1 μM compound, determined using an in vitro assay. HTH was described as being cell permeable and having moderate antiproliferative effects (in the low μM range) against breast cancer cells (16).
Compound 17, or IN17, was shown to also have nanomolar potency for MELK, but selectivity was only moderate against a panel of 5 closely related kinases. This molecule additionally impaired proliferation of TNBC cells, though potency in growth assays was notably cell type- dependent (40).
A series of low nanomolar MELK inhibitors were published in 2017, with extensive selectivity data presented from in vitro assays (63). The effects of these inhibitors on cancer cell growth were not demonstrated. Another series of inhibitors was published a year later, though the potency of this series for MELK was rather low compared to other published inhibitors. The most potent molecules from this series, compounds 4 and 16, have low micromolar to high nanomolar potency for MELK. No selectivity data were presented, and cancer cell proliferation was unaffected by these compounds (64). In 2018, yet another series of MELK inhibitors was described, with compounds 52 and 29 presented as the most promising for future studies based upon potency and selectivity for MELK, respectively. The effects of these compounds on cancer
17 cell growth were not investigated (65). Finally, the compound 16h was published in 2020. This molecule was found to be a highly potent (low nanomolar), but poorly selective inhibitor of
MELK, that exhibited antiproliferative effects against breast cancer cells (66). The majority of these studies present extensive structure-activity relationship data, and many additionally show inhibitor binding modes based upon crystallography or computer modeling. This wealth of information could be used for further MELK inhibitor optimization for improved potency and selectivity.
Of these MELK inhibitors, only MELK-T1, HTH, and 8a have been used in studies by groups other than the one responsible for development of the inhibitor. MELK-T1 was found to impair proliferation of breast cancer cells by inducing replication fork stalling and DNA double- strand breaks, resulting in a senescence-like phenotype (39). Despite HTH being previously described as a cell permeable molecule, recent results suggest it does not engage and inhibit
MELK in TNBC cells. A chemical proteomics approach called competition MIB/MS (described in chapter 3) was used to show that treatment of TNBC cells with 1 μM HTH did not prevent the subsequent capture of MELK by immobilized pan-kinase inhibitors. In the same study, the competition MIB/MS technique was used to demonstrate that 1 or 3 μM 8a treatment of TNBC cells significantly prevented MELK binding to inhibitor beads, and did not substantially prevent the binding of over 200 other kinases. These results indicated that 8a is a moderately potent, highly selective inhibitor of MELK in cells. 8a was subsequently used to show that MELK inhibition in HeLaS3 or U2OS cells caused a delay in mitotic entry consistent with transient activation of the G2/M checkpoint (14).
In summation, data from biochemical and cellular assays indicate unequivocally that OTS should not be used as a tool compound to investigate MELK function. NVS-MELK8a is a
18 suitable alternative for functional MELK studies, based upon strong biochemical selectivity data and recent selectivity profiling results completed in TNBC cells. There are a number of additional MELK inhibitors that appear to have good potency for MELK and favorable selectivity profiles in biochemical assays, most notably MELK-T1. Future studies should use cellular selectivity profiling approaches to further characterize these inhibitors, with the goal of identifying additional highly selective molecules suitable for interrogating MELK function.
Processes Regulated by MELK in Cancer
MELK has been implicated in various processes, but a comprehensive understanding of its overarching function in cancer cells is still lacking. In this section I will discuss some of the major processes to which MELK has been linked.
Proliferation and cancer aggressiveness
The connection to these processes was discussed in detail previously, in the section examining the requirement of MELK in cancer. In brief, high MELK expression levels have been consistently correlated with increased proliferation, aggressiveness, and tumor grade of cancers (67). It is unclear whether MELK acts as a driver of these processes; there is some speculation that MELK is simply upregulated as part of a mitotic cluster of genes (44).
To expand upon these results mechanistically, recent studies have suggested that the proliferative effects of MELK may be mediated through the NF-κB pathway. In melanoma cells,
MELK knockdown or inhibition with OTS or 8a decreased NF-κB pathway signaling, which decreased cell proliferation. MELK was found to phosphorylate SQSTM1 (also called p62), a known regulator of NF-κB, to drive these effects. Depletion of SQSTM1 also decreased NF-κB signaling. Overexpression of constitutively active IKKβ, a downstream activator of NF-κB,
19 partially rescued the NF-κB signaling and anti-proliferative effects caused by OTS or 8a treatment, indicating that these effects were driven by loss of MELK-SQSTM1 signaling through the NF-κB pathway (37). It has also been reported that MELK phosphorylates and activates
EZH2, which subsequently methylates NF-κB. This methylation caused increased transcription of NF-κB target genes, and loss of MELK/EZH2-mediated methylation impaired glioblastoma cell and tumor proliferation (68). Other studies have reported that the anti-proliferative effects of
MELK knockdown or inhibition with OTS are due to decreased signaling through the mTOR pathway. In endometrial carcinoma and clear cell renal cell carcinoma, MELK may interact with
MLST8 and phosphorylate PRAS40, respectively, to regulate proliferation of cells via activation of the mTOR signaling pathway (27, 69). Additional studies are needed to clarify the mechanism underlying the role of MELK in cancer cell proliferation and aggressiveness.
Cell cycle progression
A substantial number of studies have linked MELK with cell cycle progression, but the specific functions of MELK in this process have not been well-elucidated. There is not a consensus as to which phases of the cell cycle are most affected by MELK perturbation.
The largest body of evidence suggests that MELK regulates G2 and M phases of the cell cycle. MELK expression is partially regulated by the transcription factors E2F1 and FoxM1 (17,
27, 37, 70, 71). Accordingly, MELK expression oscillates during the cell cycle, increasing throughout S phase, reaching maximal levels around the G2/M transition, and declining upon mitotic exit (17, 49, 50, 72). Multiple studies have reported that MELK knockdown causes a
G2/M accumulation of cells, though these studies have frequently used DNA quantification by flow cytometry to draw this conclusion (5, 9, 35, 38, 41). This approach does not allow for
20 differentiation between cells in G2 and M phases. Overexpression of MELK may also induce a
G2/M accumulation of cells (5, 73).
Some studies have used methods that distinguish between G2 and M phenotypes resulting from MELK perturbation. MELK knockdown has been reported to impair mitotic progression by varied mechanisms, including by decreasing activity and expression of the oncogenic mitotic transcription factor FoxM1 (39, 74), by regulating MCL1 synthesis in mitosis (75), by inducing mitotic catastrophe (22), and by causing cytokinetic defects (17, 76, 77). A role for MELK specifically in G2 has also been demonstrated. Overexpression (73) or selective inhibition of
MELK with 8a (14) has been shown to increase the length of G2 phase, consistent with activation of the G2/M DNA damage checkpoint. Collectively, these results suggest that MELK likely functions as a regulator of both the G2/M checkpoint and mitosis, similar to other crucial cell cycle-regulated kinases including CDK1, PLK1, and Aurora A (78).
A few studies have also described a role for MELK in G1 and S phases of the cell cycle.
MELK knockdown has been shown to induce an accumulation of cells in G1 in glioblastoma
(79) (retracted) or bladder cancer (26). Both studies depleted MELK with RNAi for 48h or more before quantifying DNA content by flow cytometry, which leaves open the possibility that G1 accumulation was a secondary effect of MELK loss following mitotic defects. Progression through S phase is impaired by MELK knockdown (39, 79) (latter paper retracted) or inhibition with MELK-T1 (39) or OTS (26). Specifically, these treatments caused replication fork stalling,
DNA double-strand breaks (DSBs), and activation of the DNA damage response (DDR).
21 DNA damage response and therapeutic resistance
Reduction of MELK expression or activity (with MELK-T1) has been reported to cause
DSBs and induce the DDR specifically through activation of the ATM/Chk2 pathway (26, 39).
Induction of p53 expression has been reported following MELK inhibition with OTS, but has not been replicated with more selective inhibitors (80). Another study showed that MELK silencing in glioblastoma cells induced p53 expression, while overexpression had the opposite effect, and apoptotic effects caused by MELK depletion could be partially rescued by p53 knockdown or inhibition (81). An increase in p53 phosphorylation, with no change in expression, has been demonstrated with MELK-T1 treatment (39). In a separate study by this group (which was later retracted for immunoblot image duplication) MELK knockdown was demonstrated to not affect total p53 levels in glioblastoma cells, but rather caused downregulation of MDMX, an inhibitor of p53, and this effect could be rescued by exogenous MELK expression (79). Clearly, while perturbation of MELK activity or expression results in some sort of DDR, the precise mechanisms remain elusive.
MELK has been implicated as a mediator of radio- and chemotherapeutic resistance.
Treatment of colorectal cancer cells with radiation or the DNA damaging agent 5-fluorouracil caused an upregulation of MELK, and pre-treatment with MELK siRNA sensitized cells to 5- fluorouracil (34). Similarly, MELK expression is correlated with radioresistance in breast cancer cell lines. Knockdown or inhibition of MELK (with OTS) in combination with radiation induced the DNA damage marker p-H2AX (S139) in vitro, and markedly sensitized cells to radiation both in vitro and in tumor xenografts (21). While MELK has been linked to DNA damage response pathways by numerous studies, our understanding of its function in the DDR still requires further characterization.
22 Migration and invasion
Knockdown of MELK has been shown to decrease migration of various cancer types, while overexpression has the opposite effect, as measured by wound healing and transwell migration assays (9, 26, 27, 41, 69). The mechanism underlying suppressed migration as a result of MELK knockdown was investigated further in gastric cancer. MELK depletion reduced the number of actin stress fibers and filopodia, while overexpression increased stress fibers, indicating that MELK regulates cytoskeletal organization. The effect of MELK silencing on the activity of RhoA-GTPases, major regulators of cell migration, was subsequently measured.
RhoA activity was decreased, while Rac1 and Cdc42 activity was unaffected. Modulation of
RhoA activity is likely mediated through the FAK/paxillin pathway. Decreased phosphorylation of both FAK and paxillin was observed in response to MELK silencing, and increased migration induced by MELK overexpression was suppressed by treatment with FAK inhibitor I. A direct physical interaction between MELK and FAK or paxillin was not demonstrated, however, so observed differences in phosphorylation may be indirect effects (9). Another study demonstrated that MELK knockdown or inhibition with OTS suppressed invasion both in vitro and in vivo, and that resulting changes in cell morphology were consistent with epithelial-mesenchymal transition
(EMT). MELK depletion altered expression of proteins involved in the EMT process.
Specifically, E-cadherin was upregulated, and N-cadherin, vimentin, and snail were observed to be downregulated by immunoblotting (41).
Apoptosis
While the vast majority of studies indicate that MELK is a pro-survival kinase (see proliferation and cell cycle progression sections, above), results from one group suggest that it
23 may also function in a pro-apoptotic role. This pro-apoptotic function may be redox-dependent, and is purportedly mediated through direct interaction with and phosphorylation of ASK1 (82), p53 (83), STRAP (84), and SMAD2, -3, -4, and -7 (85). The studies implicating MELK in pro- apoptotic and pro-survival functions have been reviewed previously (67, 86). Future studies should aim to clarify the inputs controlling these disparate functions.
Cancer stem cell phenotype
The role of MELK in tumor initiation and maintenance of cancer stem cell state has been studied fairly extensively. MELK expression has been shown to be increased in breast (36) and glioma (72, 81, 87, 88) cancer stem and tumor initiating cells relative to non-stem cancer cells.
In breast cancer, cells with higher levels of MELK expression were found to form increased tumorspheres in vitro and exhibited enhanced tumor initiation in mice. MELK depletion with shRNA decreased tumorsphere formation, which could be rescued completely with exogenous
MELK expression (36). MELK knockdown in multipotent neural progenitor cells impaired neurosphere formation, while overexpression enhanced formation, indicating that MELK functions in tumor initiation (72). In glioma stem cells (GSCs), treatment with siomycin A, an inhibitor of FoxM1, decreased MELK expression and inhibited self-renewal, proliferation, and invasion. Overexpression of MELK partially rescued these effects, suggesting they are mediated at least in part through MELK. Importantly, treatment of non-stem tumor cells or normal neural progenitor cells with siomycin A had little effect on growth (88). Treatment with compound C1, an inhibitor of MELK and other kinases, impaired the proliferation of GSCs moreso than their non-stem counterparts (22).
24 Mechanistically, the function of MELK in cancer stem cells is mediated in part through the oncogenic transcription factors FoxM1 and c-JUN. The expression of FoxM1 has been shown to be important for neural progenitor cell growth and neurosphere formation using knockdown and overexpression studies. In this system, MELK phosphorylates FoxM1 to increase its transcriptional activity in a PLK1-dependent manner, which results in increased mitotic gene expression. Proliferation of GSCs was again shown to be more sensitive to siomycin A than neural progenitor cells. Antiproliferative effects of this compound could be rescued with overexpression of both FoxM1 and MELK, but not with overexpression of either protein alone, suggesting that disruption of the MELK-FoxM1 interaction mediated growth effects (74). MELK-driven activation of FoxM1 in GSCs may contribute to radioresistance via transcriptional upregulation of the histone methyltransferase enhancer of zeste homolog 2
(EZH2) (33). Recently, it was reported that MELK phosphorylates EZH2 to induce methylation and activation of the transcription factor NF-κB. Signaling through this MELK-EZH2-NF-κB axis was observed moreso in GSCs than non-stem tumors, and suppression of this signaling by
EZH2 knockdown or inhibition, or MELK inhibition (with OTS), reduced GSC self-renewal
(68).
In another study, MELK silencing in GSCs impaired self-renewal capacity in vitro and slowed tumor growth in vivo. MELK expression was found to be regulated by JNK2, via the transcription factor c-JUN, specifically in GSCs and not normal progenitors. Further, MELK formed a protein complex with c-JUN only in GSCs, and these cells were dependent upon the
JNK2-MELK/c-JUN signaling axis for their survival and maintenance of stemness (81). These studies collectively suggest that MELK is important for cancer stem cell self-renewal, and has different functions in tumor initiating and cancer stem cells relative to normal tumor cells.
25 Asymmetric cell division and cytokinesis
The involvement of MELK in the control of asymmetric cell division and cytokinesis has been studied primarily in Caenorhabditis elegans (89–93) and Xenopus (76, 94), respectively.
For this reason, I will not cover these processes in depth here. In brief, MELK has been demonstrated to regulate asymmetric cell division in part through interactions with snail (92),
STOX (93), and LKB1 (90) homologs in C. elegans. In Xenopus, MELK overexpression caused failed cytokinesis and impaired accumulation of active RhoA at cell division furrows (76). The importance of proper control of asymmetric cell division and cytokinesis in cancer has been well-documented (95, 96); thus, a future examination of the role of MELK in these processes in cancer may be useful.
Regulation of MELK Expression and Activity
Regulation of MELK transcription has predominantly been reported to be controlled by the transcription factors E2F1 and FoxM1. E2F1 expression levels peak at the G1/S phase transition to control the transcription of genes crucial for DNA replication and cell cycle progression, among other processes (97). MELK was first reported as an E2F target gene in
2005 (70), and has since been identified as being regulated specifically by the E2F1 isoform
(37). In melanoma, control of E2F1-driven MELK transcription seems to be downstream of the
MAPK pathway. Direct binding of E2F1 to the MELK promoter has been observed by chromatin immunoprecipitation (ChIP), and decreased MELK expression as a result of E2F1 knockdown was detected by immunoblot and transcriptional reporter assays, suggesting that
E2F1 is required to maintain MELK expression (37). Similar results were later reported in endometrial cancer as well (27).
26 FoxM1 is an oncogenic transcription factor that controls the transcription of numerous genes important for mitotic entry and progression, including cyclin B, Aurora B, and PLK1 (98,
99). Much like MELK, the overexpression of FoxM1 has been noted in many cancers, and higher expression levels are correlated with increased migration and invasion, and poor prognosis for patients (17, 100–102). Furthermore, the overexpression of FoxM1 and MELK are tightly correlated in breast cancer, with higher expression levels significantly correlated with the more aggressive basal-like subtype. Direct binding of FoxM1 to the MELK promoter has been observed by ChIP, and FoxM1 depletion or inhibition with thiostrepton has been demonstrated to decrease MELK expression (17).
Recently, a study was published linking the regulation of MELK by both E2F1 and
FoxM1 in breast cancer, in a manner dependent upon p53 mutational status. This study reported that MELK expression is increased in p53-mutant breast cancer cell lines (often TNBC) compared to those with wild-type p53 (often non-TNBC). Mechanistically, E2F1 exhibits reduced binding to the FoxM1 promoter in breast cancer cell lines with wild-type p53, resulting in lower expression of FoxM1, which subsequently causes decreased MELK expression. Wild- type p53 also directly bound FoxM1 to block its recruitment to the MELK promoter.
Conversely, in p53-mutant TNBC cell lines, mutant p53 acted as a dominant negative to reverse the wild-type p53-mediated decrease in E2F1 binding to FoxM1 promoters, and to suppress the binding of wild-type p53 to FoxM1 that impaired recruitment to MELK promoters. This resulted in increased E2F1 binding to FoxM1 promoters, driving higher levels of FoxM1, which increased MELK expression levels. This study also reported a novel FoxM1 binding site on the
MELK promoter (71).
27 As discussed above in the cancer stem cell phenotype section, the oncogenic transcription factor c-JUN has also been reported to control MELK expression specifically in GSCs (81). This has not been corroborated by other cancer studies, and may serve as an example of highly cell- type specific regulation.
Regulation of MELK degradation is not well understood. No E3 ubiquitin ligases have been thoroughly validated to regulate MELK turnover, though evidence suggests that the anaphase promoting complex/cyclosome (APC/C) plays a role in this process. MELK expression is cell cycle regulated, with expression rising in S phase, peaking in late G2 and M phases, and decreasing sharply around mitotic exit (17, 49, 50, 72). This expression pattern mirrors that of other APC/C substrates, including cyclin B and Aurora A and B (103). One study reported that MELK is ubiquitinated by the APC/C-Cdh1 complex (104). Collectively, this information suggests that the APC/C regulates MELK degradation, though further studies are needed to definitively validate this interaction. Separately, thioredoxin may recruit the E3 ligase
MDM2 to induce MELK degradation in a redox-dependent manner, but more experimentation is required to determine the relevance of this in cancer (105).
It has been reported that phosphorylation by an unknown kinase, possibly on the C- terminal non-catalytic domain, stabilizes MELK in mitosis (3, 49). Multiple studies have also reported that pharmacological inhibition of MELK induces its degradation. This seems to be a general effect caused by multiple MELK inhibitors, including OTS, MELK-T1, 8a, and HTH, and has been used to conclude that MELK is inhibited in cells in functional experiments (16, 39,
44). The mechanism underlying inhibitor-induced degradation is not at all understood, and could be due to general cell cycle perturbation rather than a MELK-specific effect, among other possible explanations. For this reason, inhibitor-induced MELK degradation should be used
28 cautiously as a cellular readout of MELK activity, and a more reliable and specific marker for
MELK activity in cells must be pursued.
Unlike other AMPK family members, regulation of MELK activity is not controlled by
LKB1 phosphorylation (106). Rather, MELK autophosphorylation of its activation loop at T167 and S171 regulates its kinase activity (107). The C-terminal noncatalytic domain of MELK is regarded as an autoinhibitory domain, is it has been shown to inhibit MELK activity in a dose- dependent manner in vitro (107). The function of this domain in cells has not been elucidated.
Furthermore, a reducing environment is required for MELK catalytic activity in vitro (107), which may be due to an intramolecular disulfide bridge that forms between C154 and C168 of the activation loop under oxidative conditions (108). It is unclear whether this disulfide represents a physiologically relevant mechanism of redox regulation of MELK activation state in cells. An additional regulatory mechanism may involve the redox-dependent binding of ZPR9 and thioredoxin. These proteins have been reported to bind MELK via two disulfide bridges that form under oxidative conditions, thereby positively and negatively regulating MELK activity, respectively (105, 109). Further characterization of these redox-dependent interactions are necessary to determine their functions in cancer.
MELK Substrates
Despite substantial interest in MELK as a potential therapeutic target in cancer since
2005, MELK substrates remain poorly characterized. MELK was originally reported to have broad substrate specificity. Using a peptide array chip, it was determined that MELK strongly phosphorylated multiple peptides with varied sequence composition surrounding the phosphorylated residue (107). More recently, a positional scanning peptide library screen was used to determine the preferred MELK phosphorylation motif to be X-basic-X-X-S/T, with a
29 strong preference for arginine at the -3 position, a strong selection against hydrophobic residues at the -3 position, and a moderate preference for arginine at the -2 and -4 positions (75). It should be noted, however, that multiple reported MELK substrates do not conform to this phosphorylation consensus sequence, including p53 (Ser15, EPPLS*) (83), ASK1 (Thr838,
PCTET*) (82), ZPR9 (Thr252, AIPIT*) (109), and STRAP (Ser188, MSVSS*) (84).
The most well-characterized MELK substrates to date are FoxM1, EZH2, and eIF4b.
The phosphorylation of FoxM1 by MELK causes increased transcriptional activity. This phosphorylation is reported to be essential for proliferation of GSCs, and to be dependent upon phosphorylation by the known FoxM1 kinase PLK1 (74). As noted in the previous section, the regulation of MELK transcription by FoxM1 is well-validated. Therefore, the observation that
MELK activates FoxM1 transcriptional activity suggests that this interaction creates a positive feedback loop, perhaps to enhance the rapidity of mitotic FoxM1 target gene expression in late
G2 phase. Various additional studies have reported decreased transcription of FoxM1 targets in response to MELK knockdown (33, 110) or inhibition with OTS (11, 12, 35, 57) or MELK-T1
(39). Importantly, while studies have observed decreased phosphorylation of FoxM1-S35 (57,
111) and -T600 (110) in response to MELK targeting, the specific FoxM1 phosphorylation sites regulated by MELK remain unclear.
A MELK-EZH2 relationship was first described in 2015, where this signaling axis was shown to contribute to the radioresistance of GSCs. No direct binding or phosphorylation of
EZH2 by MELK was shown, but rather MELK was demonstrated to control EZH2 expression through phosphorylation of FoxM1 (33). MELK was later reported to directly bind and phosphorylate EZH2, regulating its methyltransferase activity, and EZH2 to purportedly methylate MELK in medulloblastoma stem-like cells (112) and GSCs (68). The MELK-EZH2
30 interaction was further elucidated in extranodal natural killer/T-cell lymphoma. Here, MS was used to observe a concurrent increase in phosphorylation of EZH2 (S220) and decrease in ubiquitination of EZH2 (K222) with MELK overexpression. Notably, this phosphorylation site in EZH2 (S220, RKFPS*) conforms with the reported MELK consensus sequence described previously. Knockdown or inhibition of MELK with OTS caused increased ubiquitination and degradation of EZH2, suggesting that MELK regulates the stability of this protein. Knockdown or overexpression of FoxM1 altered EZH2 transcript and protein, but did not affect ubiquitination of K222, suggesting that MELK-mediated regulation of EZH2 stability is not
FoxM1-dependent (111).
Phosphorylation of eukaryotic translation initiation factor 4b (eIF4b) at S406 has been reported to be regulated by MELK by one study. This study demonstrated decreased phosphorylation of eIF4b (S406) in response to OTS treatment or MELK knockdown. In BLBC cells, the phosphorylation of eIF4b (S406) by MELK was required for synthesis of the antiapoptotic protein MCL1 in mitosis, which regulated cancer cell survival, as demonstrated by knockdown and inhibition studies (75). In another study, it was reported that CRISPR-generated
MELK knockout breast cancer cell lines had normal levels of eIF4b (S406) phosphorylation and maintained normal expression of MCL1, suggesting that MELK may not be required for phosphorylation of eIF4b at this site (44).
Numerous additional proteins have been reported to be MELK substrates, including
CDC25B (73, 113), SQSTM1 (37), ASK1 (82), p53 (83), SMAD2, -3, -4, and -7 (85), PDK1
(114), ZPR9 (115), thioredoxin (105), STRAP (84), PRAS40 (69), stathmin (116), DBNL, and
PSMA1 (10). Some of these studies additionally describe specific phosphosites purported to be regulated by MELK, but none have been well-validated to date. Future studies should focus on
31 further validating these putative MELK substrates and identifying new substrates, as the field critically needs a reliable measure of MELK activity and inhibition in cells.
32 Figures
Figure 1.1. Timeline of major MELK milestones in cancer.
33
CHAPTER 2: MELK KNOCKDOWN AND PHARMACOLOGICAL INHIBITION IMPAIR PANCREATIC CANCER CELL VIABILITY
Introduction
Pancreatic cancer (PC) is among the most deadly and difficult-to-treat forms of cancer.
In the US in 2019, an estimated 56,770 new cases were diagnosed, with an estimated 45,750 deaths occurring as a result of this disease. While PC was only the eleventh most frequently occurring type of cancer in the US in 2019, it accounted for the third most cancer deaths according to the most recent data in the SEER (Surveillance, Epidemiology, and End Results) database (117, 118). Further, the 5-year survival rate for PC is only 9% in the US, lowest amongst all cancers (117). Survival rates are anticipated to rise more slowly in PC than other cancers, and accordingly, PC is projected to surpass colorectal cancer as the second leading cause of cancer deaths in the US by 2030 (119).
One of the major contributors to the poor survival rate for PC patients is the limited efficacy of treatment options. While surgical resection of the pancreas is the best therapeutic strategy, only 15-20% of patients are diagnosed early enough in disease progression to be candidates (120). Even in patients that receive surgery to remove the tumor, recurrence rates can be as high as 80% (121). In these cases, and in the numerous instances of non-resectable disease, patients are treated with chemotherapeutics. The current chemotherapeutic standards of care for
PC are FOLFIRINOX (a combination of 5-fluorouracil, leucovorin, irinotecan, and oxaliplatin) and gemcitabine/nab-paclitaxel (122). Both therapeutic options offer increased survival versus
34 gemcitabine alone, the previous standard of care for PC, though FOLFIRINOX has a higher incidence of serious adverse events than gemcitabine/nab-paclitaxel (122–124). Therefore, when considering new treatment combinations to enhance PC survival, it is likely unrealistic to consider adding another treatment to the toxic four-drug cocktail FOLFIRINOX. A more feasible approach, potentially, would be to identify a treatment that increases the efficacy of the gemcitabine/nab-paclitaxel combination.
Gemcitabine, a cytidine analog, was the standard of care for PC treatment for over a decade, despite only conferring a median survival time of approximately 5.6 months for patients
(125). Often, patients initially respond favorably to gemcitabine therapy, only to rapidly develop drug resistance. A number of resistance factors have been identified, including variation of expression of nucleoside transporter-1 and downstream gemcitabine processing enzymes, changes in expression of apoptosis-related genes, and increased activation of numerous kinase signaling pathways, including PI3K/AKT, FAK, c-SRC, and c-MET (126). Given the potential importance of activation of kinase networks in drug resistance, we sought to characterize perturbations of the kinome in response to gemcitabine in an effort to identify putative, novel therapeutic targets for combatting gemcitabine resistance and increasing the effectiveness of PC treatment with gemcitabine/nab-paclitaxel.
The MIB/MS (multiplexed kinase inhibitor beads/mass spectrometry) technology is a useful approach for identifying global changes in the kinome in response to cellular perturbations. Briefly, this approach utilizes pan-kinase inhibitors immobilized on agarose beads to capture kinases from cell lysates, allowing for comparisons of global kinase expression between treatment and control conditions. This technique has previously been used to define the kinome response of breast cancer cells to MEK inhibition and lapatinib treatment, which
35 subsequently allowed for the rational design of strategies to combat and overcome resistance to these therapeutic approaches (127, 128). MIB/MS kinome profiling was also used to identify kinase pathways driving resistance in imatinib-resistant chronic myelogenous leukemia cells
(129). We sought to employ a similar approach to identify kinome changes in PC cells in response to gemcitabine treatment. A slight adaptation of this strategy, termed competition
MIB/MS, combines short-term treatment of cells or drug addition to cell lysates with MIB/MS to identify direct kinase inhibitor binding targets (130).
In this chapter, we used MIB/MS to characterize the kinome response of PC cells to 24h gemcitabine treatment, in an effort to identify putative resistance mechanisms. This analysis and immunoblotting revealed that expression of the kinase MELK is upregulated by gemcitabine.
RNAi-mediated MELK knockdown was found to slow proliferation of PC cells. To identify novel inhibitors of MELK, we screened a small-molecule library using a MELK kinase assay and subsequently identified 18 compounds with similar or greater potency than the leading
MELK inhibitor OTSSP167 (OTS). These hit compounds were further evaluated using MTS and crystal violet assays, to measure their effects on PC cell viability. A lead compound,
UNC2025, was selected for selectivity profiling using competition MIB/MS, which revealed relatively low selectivity for MELK. PC cells were not sensitized to gemcitabine with MELK inhibition using either OTS or UNC2025, or MELK knockdown. Collectively, the results showing that MELK knockdown and inhibition decrease PC cell viability suggest that this kinase is a putative therapeutic target in PC, though upregulation of MELK may not truly act as a gemcitabine resistance mechanism.
36 Results
Gemcitabine causes increased MELK expression in pancreatic cancer cells
PANC-1 cells were treated with 1 μM gemcitabine for 24h, at which point they were harvested for MIB/MS to assess changes in kinase expression. A total of 128 kinases were quantified using MaxQuant (Fig. 2.1A). Five kinases were captured with at least 2-fold decreased abundance (MTOR, ULK3, STK24, MP2K3, and M3K4), and 2 kinases were captured with greater than 2-fold increased abundance (MELK and KCC2G), indicating that gemcitabine treatment had caused a decrease and increase in expression of those kinases, respectively.
Generally, drug resistance factors are induced in response to exposure to a given drug; thus, we were most interested in the two kinases that exhibited an increase in expression. Of those, the kinase with the largest magnitude change was MELK, which was quantified with nearly 5-fold increased abundance by MIB/MS. We confirmed that 24h treatment with 0.5 or 2 μM gemcitabine caused a 5- to 10-fold increase in MELK expression in MIA PaCa-2 and PANC-1 cells (Fig. 2.1B). To test whether PC cells were sensitive to loss of MELK, we next used an shRNA targeting MELK to deplete this kinase. Indeed, we found that MELK depletion caused a decrease in proliferation in PANC-1 cells, measured using a 21-day crystal violet cell viability assay (Fig. 2.2, A and B)
Screening of UNC small-molecule library reveals novel MELK inhibitors
These observations led us to focus subsequent experiments on investigating the approach of targeting MELK in PC cells, particularly in combination with gemcitabine. At the time these studies were conducted, OTS was universally used in MELK studies. Despite numerous studies describing OTS as a highly selective inhibitor of MELK, no data had ever been presented to support this claim (9, 10, 116). Publicly available data (http://www.kinase-
37 screen.mrc.ac.uk/screening-compounds/594372) and preliminary data from our lab (shown in chapter 3) indicated that OTS was actually quite poorly selective for MELK. Further, while other MELK inhibitors had been published, none had been thoroughly validated or used in
MELK functional studies (38, 39, 61). Thus, we initiated a drug discovery effort in search of a selective MELK inhibitor with potential for both therapeutic use and functional studies.
Through a collaboration with Drs. Stephen Frye and Xiaodong Wang (UNC CICBDD), we obtained access to a library of small-molecule inhibitors of the TAM (TYRO3-AXL-
MERTK) family of kinases (131, 132), known to have moderate affinity for MELK (unpublished data). From this library, we tested 53 compounds for their ability to inhibit MELK activity using the ADP-Glo™ (Promega) kinase assay. All compounds, in addition to OTS, were tested at 50 nM, which was previously determined to be the IC50 of OTS for MELK in this assay (data not shown). A total of 18 compounds were found to inhibit MELK with similar or greater potency than OTS (Fig. 2.3). These compounds were thus selected for further evaluation using PC cell viability assays.
Novel MELK inhibitors impair pancreatic cancer cell viability
First, inhibitors were tested at concentrations ranging from 0.3 nM to 10 μM in MIA
PaCa-2 cells using a 72h MTS assay. The IC50 of each inhibitor was determined (Table 2.1).
The top five inhibitors of MIA PaCa-2 cell viability were found to be OTS, UNC3778,
UNC2371, UNC2025, and UNC3875 (Fig. 2.4A). The effects of these compounds, in addition to UNC1170, UNC570, UNC3906, and UNC4200, were additionally tested against PANC-1 cell viability using a long-term (28-day) crystal violet assay. In order to quantify the IC50 of each compound, crystal violet stain was solubilized with Sorenson’s buffer and absorbance measured
38 at 570 nm (Table 2.1). Similar to MTS assay results, the most potent inhibitors of proliferation in crystal violet assays, besides OTS, were UNC3778, UNC2371, and UNC2025 (Fig. 2.4B).
Notably, there appeared to be a moderate correlation between potency in the MELK kinase assay (Fig. 2.3) and potency for inhibiting MIA PaCa-2 (MTS assay) and PANC-1
(crystal violet) cell viability (Table 2.1). This trend does not hold true for OTS, which is known to significantly inhibit a number of kinases more potently than MELK, including vital regulators of proliferation (chapter 3) (13, 15). Despite being the most potent inhibitor of cell viability,
OTS is a less potent MELK inhibitor than many of the UNC compounds tested. These results suggest that inhibition of MELK by the UNC compounds was at least partially responsible for impaired PC cell viability, while providing further evidence that OTS does not impair cancer cell proliferation primarily through inhibition of MELK activity.
MIB/MS kinome profiling reveals low selectivity of UNC2025 for MELK
Amongst the top three inhibitors of PC cell viability, UNC2025 has the most favorable permeability and pharmacokinetic profile ((131) and unpublished data). We therefore decided to use this compound for subsequent experiments rather than UNC3778 or UNC2371. We utilized competition MIB/MS to determine the kinase selectivity of UNC2025. In brief, we incubated
DMSO or 10, 100, or 1000 nM UNC2025 with equal volumes of PANC-1 lysate for 30 min.
Lysates were then applied to chromatography columns containing MIBs, columns were washed, and bound kinases were eluted and quantified by MS. The fundamental principle underlying this assay is that kinases in the lysate which are bound by UNC2025 are subsequently unable to bind to MIBs; thus, these specific kinases are quantified in lower abundance relative to the DMSO control condition. It is important to note that this particular application of MIB/MS is distinct
39 from the technical setup used previously to study kinome changes resulting from 24h treatment with gemcitabine (Fig. 2.1A), as only direct inhibitor binding is assessed with this competition
MIB/MS approach.
Using competition MIB/MS, we determined the selectivity of UNC2025 for 223 total kinases (Fig. 2.5). A total of 31 kinases, including MELK, exhibited decreased binding to MIBs in a dose-dependent manner, with greater than 2-fold (UNC2025 / DMSO <0.5) decreased binding with 1 μM treatment. The additional kinases that met these criteria were LATS1,
PRKCG, RPS6KA4, PIK3C2A, CAMK2B, CAMK2D, CAMK2G, MARK3, NUAK1, SIK1,
SIK2, SIK3, STK17A, CLK1, CLK2, CLK4, DYRK1A, AAK1, BMP2K, ULK3, AXL,
FGFR1/2, MERTK, PTK2B, MAP3K2, MAP4K3, MAP4K5, MINK1, SLK, and STK10. Of note in this list are the TAM family kinases AXL and MERTK (TYRO3 was not detected).
Moreover, MERTK was the only kinase with more than 5-fold decreased binding (UNC2025 /
DMSO <0.2) at the lowest concentration tested (10 nM). These results are in agreement with previous studies that described MERTK as the primary target of UNC2025 (131, 133). While
MELK binding was decreased more than 5-fold at 1000 nM, it was only slightly decreased (less than 2-fold) at 10 and 100 nM UNC2025. Additionally, 21 other kinases in addition to MELK exhibited greater than 5-fold decreased binding to MIBs with 1000 nM UNC2025 treatment.
These results indicated that UNC2025 has relatively low selectivity for MELK, and that the effects of this compound cannot be solely attributed to MELK inhibition.
MELK inhibition or knockdown does not sensitize pancreatic cancer cells to gemcitabine
While neither OTS nor UNC2025 appeared highly selective, these compounds still strongly inhibited MELK. We therefore next tested whether these inhibitors sensitized PC cells
40 to gemcitabine. MIA PaCa-2 cells were treated with 0.3 nM to 10 μM gemcitabine, plus either
DMSO, 10 nM OTS, or 250 nM UNC2025 (the IC25 concentration for each compound), and the resulting cell viability was measured using a 72h MTS assay (Fig. 2.6A). MIA PaCa-2 cells did not exhibit increased sensitivity to gemcitabine with either OTS or UNC2025 treatment, as the
IC50 remained at or above 20 nM for all conditions. Given the prominent off-target kinase inhibition for both compounds, we further tested our hypothesis that decreased MELK activity would increase susceptibility to gemcitabine using RNAi-mediated knockdown. MELK was depleted with shRNA in MIA PaCa-2 cells, and sensitivity to gemcitabine was again measured with a 72h MTS assay (Fig. 2.6B). While MELK depletion itself caused decreased viability, as with inhibitor treatment, it failed to sensitize cells to gemcitabine.
Discussion
Our results are the first to demonstrate that MELK expression is upregulated by gemcitabine treatment (Fig. 2.1). This observation was not entirely surprising, as others have shown that treatment with the nucleoside analog 5-fluorouracil or radiation also cause upregulation of MELK in colorectal cancer (34). These findings have largely been ignored by the MELK field, and along with our results here, may suggest a role for MELK in response to impaired DNA replication or DNA damage. MELK has been linked to the DNA damage response previously by our group (chapter 4) and others (26, 39). Further studies are needed to more clearly define the role of MELK in these processes, as this is currently an understudied area of investigation in the MELK field.
While the drug discovery efforts undertaken in this study did not ultimately unearth a highly selective inhibitor of MELK (Fig. 2.5), UNC2025 has certainly shown promise as an anticancer agent. This inhibitor has been demonstrated to induce cell death and suppress tumor
41 growth in melanoma, glioblastoma, leukemia, and non-small cell lung cancer primarily through inhibition of MERTK (132–135). Nonetheless, it is possible that partial MELK inhibition by
UNC2025 contributes to the favorable polypharmacology displayed by this inhibitor. Further, no results have yet been published showing the efficacy of UNC2025 or related UNC compounds for the treatment of PC. Our results (Fig. 2.4) indicate that these compounds display anticancer properties in PC cells, and may provide a rationale for further investigation of the role of the
TAM family of kinases in PC. While the approach of targeting AXL in PC has been investigated, MERTK has not yet garnered significant attention (136, 137). UNC2025 could thus serve as a useful tool compound for preliminary studies, and may also hold therapeutic potential. Finally, the results of our MELK kinase assay screen of 53 small-molecules (Fig. 2.3) could be used to define the structure-activity relationship for these scaffolds. This information could potentially be used to modify UNC2025, or other inhibitors, in an effort to improve their potency towards MELK.
As discussed in chapter 1, many groups have shown that RNAi-mediated MELK knockdown causes slowed proliferation of various cancers, including TNBC, glioma, leukemia, and others (6, 11, 17, 24, 25). In many studies, this effect has been shown to be due specifically to MELK loss using complementation experiments (6, 17, 22, 25, 36, 37). The role of MELK in
PC, however, has not yet been the focus of any studies. Results from this chapter suggest that
MELK may play a role in the proliferation of PC cells, albeit without complementation to demonstrate specificity in knockdown experiments. Given that the prognosis for PC is extremely poor, and the need for targeted therapies is undeniably dire, MELK may warrant further attention as a potential target.
42 We observed that MELK inhibition with OTS or UNC2025, or MELK knockdown, did not sensitize PC cells to gemcitabine (Fig. 2.6). Here, we used the IC25 concentration of each of these compounds. Given that both OTS and UNC2025 exhibit significant kinase inhibition with similar or greater potency than that for MELK, MELK activity may not have been substantially inhibited at these concentrations. After the completion of the studies shown in this chapter, additional MELK inhibitors were developed, including NVS-MELK8a (8a), IN17, and HTH-01-
091 (16, 38, 40). 8a in particular has excellent selectivity for MELK and inhibits TNBC cell proliferation (chapter 3), making this compound suitable for functional MELK studies. In future studies, it would be interesting to test whether this inhibitor sensitizes pancreatic or other cancers to chemotherapeutics, including gemcitabine. Lastly, given that MELK expression oscillates in the cell cycle (17, 49), it is possible that gemcitabine treatment, which perturbs the cell cycle, causes an acute upregulation of MELK independent of resistance mechanisms. Future efforts to identify drug resistance mechanisms with MIB/MS may be more successful when applied to cells that have been persistently treated (i.e. for weeks) with chemotherapeutic agents.
Experimental Procedures
Reagents and cell culture
Gemcitabine (cat. # S1714) and OTSSP167 (cat. # S7159) were purchased from
SelleckChem. The small-molecule compound library of putative MELK inhibitors was a generous gift from Drs. Stephen Frye and Xiaodong Wang of the UNC CICBDD. PANC-1 and
MIA PaCa-2 cell lines were kindly provided by Dr. Jen Jen Yeh (UNC). Both cell lines were cultured in RPMI 1640 (Gibco) supplemented with 5% fetal bovine serum (Millipore) and 1% antibiotic-antimycotic (Sigma). Cells were maintained in a humidified chamber with 5% CO2.
43 Short hairpin RNA (shRNA) reagents used in these studies were purchased from Dharmacon™
(ID: TRCN0000001642, sequence: 5’-ACAAAGAGACATAGTTAAGAG-3’).
Immunoblotting
Immunoblotting was completed exactly as described previously (14). The following primary antibodies were used in this study: MELK (Cell Signaling Technologies, cat. # 2274,
1:1000) and β-actin (Santa Cruz Biotechnology, cat. # SC47778, 1:1000).
Kinase assays
Active MELK (1-340) was purchased from SignalChem Lifesciences Corporation (cat. #
M50-11G). ZIPtide substrate (KKLNRTLSFAEPG) was purchased from AnaSpec (cat. # AS-
63366). Tris base (cat. # BP152-500), magnesium chloride hexahydrate (cat. # M33-500), and dithiothreitol (DTT) (cat. # R0861) were purchased from Thermo Fisher Scientific. Bovine serum albumin (BSA) (cat. # A2153) was purchased from Sigma-Aldrich.
ADP-Glo™ (Promega) kinase assays were completed in 96-well white, round bottom plates, according to the Promega protocol. Briefly, the kinase reaction was initiated by adding
25 ng active MELK to a well containing 10 μM ATP, 200 μg/mL ZIPtide, 50 nM MELK inhibitor or DMSO, and kinase reaction buffer (40 mM Tris, pH 7.5, 20 mM MgCl2, 0.1 mg/mL
BSA, and 15 mM DTT). This reaction was allowed to proceed for 30 min at room temperature.
ADP-Glo reagent was added to terminate the kinase reaction and deplete remaining ATP in solution. The plate was incubated for 40 min at room temperature. Kinase detection reagent was added to produce luminescence proportional to the amount of kinase activity (MELK phosphorylation) that occurred in the first step. The plate was incubated for 30 min at room
44 temperature. All incubations were completed with the plate protected from light. Luminescence was measured using a PHERAstar (BMG Labtech) plate reader with luminescence module.
Cell viability assays
For MTS assays, MIA PaCa-2 cells were plated on a 96-well plate at a density of 3000 cells/well. Media was aspirated and replaced, after 24h, with media containing compound
(MELK inhibitor or gemcitabine) at concentrations ranging from 0.3 nM to 10 μM, or DMSO.
MELK inhibitor conditions were tested in technical duplicate, and gemcitabine conditions were tested in technical triplicate. Cells were incubated in a humidified chamber for 72h. After 72h,
0.6 mM resazurin dye (Acros Organics, cat. # 62758-13-8) was added, cells were incubated for
2h, and fluorescence was measured using a PHERAstar plate reader with fluorescence module
(FI: 540-20, 590-20). GraphPad Prism was used to generate dose-response curves and calculate
IC50 values.
Crystal violet cell viability assays were completed by plating 300 PANC-1 cells/well on a
6-well cell culture-treated plate. Media was replaced the following day with fresh media containing DMSO or MELK inhibitor at concentrations ranging from 10 to 1000 nM.
Throughout the duration of the assay, media was refreshed at least every 3 days. After 28 days, media was aspirated and cells were stained by incubating 10 min at room temperature with crystal violet solution (0.5% crystal violet (w/v) (Sigma), 20% methanol (Fisher)). Plates were imaged using a Canon LiDE 110 document scanner. For quantification of relative cell number, crystal violet stain was solubilized using Sorenson’s buffer (0.1 M sodium citrate, pH 4.2, 50% ethanol), and the absorbance (570 nm) of each solubilized stain solution was measured using a
45 SpectraMax microplate reader. GraphPad Prism was used to generate dose-response curves and calculate IC50 values.
Kinome analysis and selectivity profiling by MIB/MS
For standard MIB/MS to determine kinome remodeling following gemcitabine exposure,
PANC-1 cells were treated with 1 μM gemcitabine for 24h. Cells were lysed and MIB enrichment was completed exactly as previously described (138), with the only exception being the kinase inhibitor bead mix. In this study, a mix of the following kinase inhibitors, immobilized on beads, was used: CTx-0294885, PP58, Purvalanol B, UNC2147A, VI-16832,
UNC8088A. Following MIB kinase enrichment steps, MS and MS/MS data acquisition with
MALDI TOF/TOF 5800 (AB SCIEX), MS/MS spectra search against a reviewed human Uniprot database, and quantification and data analysis using ProteinPilot™ were completed as previously described (129).
Competition MIB/MS, used for selectivity profiling of UNC2025, was also completed as previously described (14) with one exception. Here, the inhibitor was incubated with PANC-1 cell lysate on ice for 30 min, rather than being added to cells prior to lysis. Following MIB kinase enrichment, the samples were analyzed by LC/MS/MS using a Q Exactive HF mass spectrometer. Data acquisition, search against a reviewed human Uniprot database, quantification using MaxQuant label-free quantification, and data analysis were completed as previously described (14).
46 Tables and Figures
Table 2.1. MELK inhibitor potency in PC cell viability assays. The 18 compounds that inhibited MELK activity with equal or greater potency than OTS in kinase assays (Fig. 2.3) were tested in PC cell viability assays. For MTS assays, MIA PaCa-2 cells were exposed to inhibitor for 72h prior to quantification of viability by addition of resazurin and measurement of fluorescence. For crystal violet (CV) assays, PANC-1 cells were exposed to inhibitor for 28 days prior to staining with CV. N.D. = not determined.
47
Figure 2.1. Treatment of PC cells with gemcitabine causes MELK upregulation. A, PANC- 1 cells were treated with 1 μM gemcitabine for 24h. Cells were harvested and MIB/MS was completed to assess changes in kinase expression. B, MIA PaCa-2 and PANC-1 cells were treated with 0.5 or 2 μM gemcitabine for 24h and MELK expression was analyzed by immunoblot.
48
Figure 2.2. MELK knockdown slows proliferation of PANC-1 cells. A, PANC-1 cells were transfected with empty vector (vehicle) or an shRNA targeting MELK. Cells were stained with crystal violet after 21 days. B, PANC-1 cells were transfected as in A and MELK expression was analyzed by immunoblot after 72h.
49
Figure 2.3. Screening of a small-molecule library reveals potent MELK inhibitors. An in- vitro kinase assay (ADP-Glo™, Promega) was used to measure MELK activity upon exposure to 53 compounds from a targeted small-molecule library. All compounds, plus OTS (indicated by a blue bar), were screened at 50 nM.
50
Figure 2.4. Novel MELK inhibitors impair PC cell proliferation. A, MIA PaCa-2 cells were exposed to inhibitors for 72h, in technical duplicate, prior to quantification of viability by addition of resazurin and measurement of fluorescence. The 18 compounds shown in Table 2.1 were tested, and dose-response curves for the 4 most potent inhibitors of cell viability, measured by MTS assay, are displayed. B, PANC-1 cells were exposed to inhibitors for 28 days, before being stained with crystal violet dye. The 3 most potent inhibitors of cell viability, measured by this method, are shown here.
51
Figure 2.5. The kinase selectivity profile of UNC2025 reveals off-target kinase inhibition. DMSO or 10, 100, or 1000 nM UNC2025 was incubated for 30 min with equal volumes of lysate from PANC-1 cells. MIB/MS was completed as described in Experimental Procedures to assess the selectivity of UNC2025 for 223 kinases. MELK is indicated with an arrow.
52
Figure 2.6. MELK inhibition or knockdown do not sensitize PC cells to gemcitabine. A, MIA PaCa-2 cells were treated with a range of concentrations of gemcitabine (0.3 nM to 10 μM) and DMSO or the previously determined IC25 of OTS or UNC2025 (10 nM and 250 nM, respectively). Cells were exposed to compounds for 72h and viability assessed using MTS assay, as described previously. B, MIA PaCa-2 cells were transfected with empty vector or shMELK. After 72h, cells were treated with a range of concentrations of gemcitabine, as in A, and viability was assessed by MTS assay.
53
CHAPTER 3: MASS SPECTROMETRY-BASED SELECTIVITY PROFILING IDENTIFIES A HIGHLY SELECTIVE INHIBITOR OF THE KINASE MELK THAT DELAYS MITOTIC ENTRY IN CANCER CELLS2
Introduction
The maternal embryonic leucine zipper kinase (MELK), also known as MPK38 or pEg3, is a highly conserved member of the AMPK family of kinases (1, 86). The proposed role of
MELK in regulation of cell growth has been controversial. MELK expression is increased in cancer relative to normal tissue (5), and high levels of MELK are correlated with tumor grade, poor prognosis, radioresistance, and recurrence in multiple cancers (6, 17, 20–22, 24, 25).
RNAi-mediated depletion of MELK has been shown to cause impaired proliferation in a variety of cancers, including basal-like breast (17), glioma (6), acute myeloid leukemia (11), high-risk neuroblastoma (24), and hepatocellular carcinoma (25). In triple-negative breast cancer (TNBC), knockdown of MELK caused shrinking and radiosensitization of xenografts (17, 21).
Significantly, numerous studies have shown that RNAi effects on growth are due specifically to
MELK depletion by demonstrating rescue with ectopic MELK expression (6, 17, 22, 25, 36, 37).
Complicating the interpretation of these findings, the requirement of MELK for cancer cell proliferation has been challenged by studies showing that genomic deletion of MELK with
CRISPR/Cas9 caused no growth effects (16, 18, 44). Other studies contend that genomic MELK
2This chapter was originally published in the Journal of Biological Chemistry. The original citation is as follows: McDonald, I. M., Grant, G. D., East, M. P., Gilbert, T. S., Wilkerson, E. M., Goldfarb, D., Beri, J., Herring, L. E., Vaziri, C., Cook, J. G., Emanuele, M. J., and Graves, L. M. (2020) Mass spectrometry-based selectivity profiling identifies a highly selective inhibitor of the kinase MELK that delays mitotic entry in cancer cells. J. Biol. Chem. 295, 2359–2374
54 deletion does cause impaired proliferation of cancer cells (35), although this effect may only manifest under certain growth conditions (19).
The discordant results between acute MELK knockdown and longer-term genetic inactivation is reminiscent of other key cell cycle proteins whose essential roles are evident in experiments following acute inactivation by RNAi or with small-molecules, but less clear in knockout experiments. Two examples in particular illustrate this phenomenon. First, the cyclin
D-CDK4/6 complex is non-essential for proliferation in knockout cell lines and virtually all mouse tissue (139, 140), but chemical inhibitors halt progression into S phase in cell lines with a functional retinoblastoma protein (141). This example is particularly relevant given the growing clinical use of CDK4/6 inhibitors for the treatment of cancer (142). Second, recent results using
CRISPR/Cas9 gene editing suggest Bub1 is non-essential for mitotic spindle checkpoint function in humans (143, 144), but it was later shown that these “knockout” cells express low levels of a truncated Bub1 protein, accounting for the negative results and lack of a phenotype (46–48).
Together, these results highlight the importance of examining protein function using multiple, orthogonal approaches and for using caution in interpreting negative results from gene knockout experiments.
The recent controversy concerning the requirement of MELK in cancer has underscored the need for a more comprehensive understanding of MELK function. MELK has been reported to function in a multitude of processes including cell cycle progression (5, 17, 74), maintenance of cancer cell stemness (6, 72, 81), protein synthesis (75), and apoptosis, (7, 22, 67, 82). Many of the functional studies of MELK have focused on the role of this kinase in the cell cycle.
MELK mRNA is cell cycle regulated, and accordingly, protein levels of MELK oscillate throughout the cell cycle, with expression reaching a maximum in G2/mitosis (M) phase, before
55 declining upon mitotic exit (17, 49, 50). Depletion of MELK has been shown to cause a G2/M arrest in cells (5, 17, 35, 41). Some studies have attributed this effect to a role for MELK in the
G2/M DNA damage checkpoint, potentially mediated through interaction with CDC25B, a phosphatase that regulates CDK1 activity at the G2/M checkpoint (73). Additionally, loss of
MELK activity may cause activation of the ATM/Chk2 DNA damage response (39). Other studies indicate that MELK knockdown specifically impairs mitotic progression. A circular relationship between MELK and FoxM1 has been described, whereby MELK expression is both regulated by this mitotic transcription factor, and phosphorylation by MELK regulates the activity and expression of FoxM1 (39, 74). MELK may also regulate synthesis of the anti- apoptotic protein MCL1 in M via interaction with eIF4B (75). Overall, while a number of studies have uncovered putative functions of MELK at G2/M, a comprehensive understanding of the role of MELK at these phases of the cell cycle remains elusive.
One of the key challenges in studying the biological functions of MELK has been the lack of a selective inhibitor. The most commonly used MELK inhibitor, OTSSP167 (OTS), also referred to as OTS167, has shown potent antiproliferative and apoptotic effects against multiple cancer types, including TNBC, acute myeloid leukemia, and high-risk neuroblastoma (10, 11, 17,
24, 28). Consequently, OTS is currently in multiple phase I clinical trials for the treatment of patients with advanced TNBC and refractory or relapsed leukemia (NCT01910545,
NCT02795520, NCT02768519, NCT02926690). While an effective inhibitor of cancer cell viability, OTS is poorly selective for MELK. Recent publications have shown that OTS is a broad-spectrum inhibitor that inhibits numerous kinases that are vital for proliferation, including multiple mitotic kinases (13, 15). A number of studies have relied upon OTS as a means to investigate the biological function of MELK; however, these results must be interpreted with
56 caution due to the poor selectivity of this inhibitor. More recently, additional MELK inhibitors have been developed, including MELK-T1 (39), NVS-MELK8a (8a) (38), HTH-01-091 (HTH)
(16), IN17 (40), and others (22, 65). 8a and HTH in particular appear to have more favorable selectivity profiles than OTS based upon kinase activity profiling assays (16, 38).
The key objectives of this study were to evaluate the selectivity profiles of three MELK inhibitors, 8a, HTH, and OTS, using a cell-based assay, in order to identify a highly selective inhibitor to subsequently investigate MELK function. To this end, we utilized a chemical proteomics approach called multiplexed kinase inhibitor beads/mass spectrometry (MIB/MS) to characterize the selectivity of these MELK inhibitors in TNBC cells (127, 130). OTS was observed to be highly non-selective, in agreement with previous studies, and HTH did not effectively inhibit MELK in cells. In contrast, we found the inhibitor 8a to be highly selective for MELK, providing the first reliable alternative to OTS for functional MELK studies. Using
8a as a tool compound to probe the function of MELK, we observed that MELK inhibition perturbed cell cycle progression by delaying entry into M, without an induction of apoptosis.
Collectively, our results provide a rationale to utilize 8a for functional MELK studies as an orthogonal approach to knockdown and knockout techniques, and provide insight into the role of
MELK in cell cycle progression, specifically at the G2/M checkpoint.
Results
Identification of a selective MELK inhibitor by MIB/MS kinome profiling
MIB/MS is a proteomics technology used for enriching kinases from complex samples
(127, 128). This technique can be used for selectivity profiling of inhibitors in a cell-based fashion, an application termed competition MIB/MS (see Fig. 3.1 for detailed description of technology) (130). Selectivity can also be assessed using MIB/MS by adding inhibitors directly
57 to lysates. A similar chemical proteomics approach, using kinobeads and MS, was recently employed by Klaeger et al. to comprehensively define the selectivity of all clinical and FDA- approved kinase inhibitors, validating the use of this approach for measuring inhibitor selectivity in cells (13). We used the competition MIB/MS approach to profile the selectivity of 8a and
HTH in an effort to identify a highly selective MELK inhibitor suitable for functional studies.
We first determined the kinase target landscape of OTS, HTH, and 8a at a single concentration. MDA-MB-468 cells were selected for initial experiments, as many previous studies utilized this and other TNBC cell lines. MDA-MB-468 cells were treated with 1 μM of each putative MELK inhibitor or DMSO for 30 minutes. Lysates from DMSO- or inhibitor- treated cells were flowed over columns packed with MIBs. Kinases bound to MIBs in the absence or presence of inhibitors were eluted and quantified by MS using label-free quantification in MaxQuant with integrated Andromeda search engine. The specificity of each inhibitor was analyzed for a total of 235 protein kinases quantified in this assay (Fig. S3.1).
Examining the 20 protein kinases detected in lowest abundance after OTS, HTH, or 8a treatment of cells (i.e. most prevented from binding to MIBs), relative to DMSO treatment, revealed stark differences in the selectivity and potency of these three compounds (Fig. 3.2A). As previously reported, OTS exhibited broad specificity, with 52 protein kinases including MELK captured by
MIBs with at least 4-fold decreased abundance relative to DMSO (OTS / DMSO <0.25).
Moreover, OTS decreased the binding of a number of kinases to MIBs more effectively than
MELK. This finding is in agreement with previous studies, confirming that OTS has very poor selectivity, despite high potency for MELK (13, 15). HTH was found to have a much narrower selectivity profile, with only 5 protein kinases captured with at least 4-fold decreased abundance relative to DMSO, but surprisingly MELK binding was largely unaffected by this compound.
58 These results suggest that HTH did not effectively inhibit MELK in this cell-based assay, inconsistent with in vitro enzyme assay data (16). By contrast, the target landscape of 8a was observed to be extremely narrow, with MELK being the only protein kinase captured with at least 4-fold decreased abundance relative to DMSO. These results indicated that 8a is the most selective of the three MELK inhibitors profiled using MIB/MS.
Due to the striking specificity and potency differences between 8a and HTH, we sought to further validate these results in biological triplicate, again at a single concentration of 1 μM.
Subsequent competition MIB/MS results are displayed as volcano plots to assess both kinase fold-change magnitude and significance (Fig. 3.2B). With 8a treatment, MELK was the only kinase that displayed statistically significant decreased binding to MIBs, confirming the high selectivity of this compound. As observed previously (Fig. 3.2A), HTH treatment did not significantly decrease MELK capture, whereas binding of GAK, CSNK2A2, CSNK2B,
CSNK2A1, PIP4K2C, and RIPK2 were significantly decreased. These results further confirm that 8a is a highly selective MELK inhibitor in cells, and underscore the importance of determining compound selectivity in a cellular context prior to biological studies.
Competition MIB/MS was next used to profile the selectivity of 8a at 10, 100, and 1000 nM. As expected, we observed a dose-dependent decrease in MELK binding to MIBs, with no major off-target inhibition over this concentration range (Fig. S3.2). While high selectivity was maintained at 1 μM 8a, we did not observe greater than 90% loss of MELK binding to MIBs, suggesting incomplete inhibition. Since the percent inhibition of MELK required to elicit a phenotype is not well-defined, we tested higher concentrations of 8a (3 and 10 μM), with the goal of assessing selectivity at concentrations required for near-total inhibition of binding to
MIBs. Following MIB/MS, 221 protein kinases were quantified by label-free quantification as
59 above (Fig. S3.3), and the data were filtered to remove any protein kinases that did not exhibit reduced binding at all three concentrations, in a dose-dependent manner. The 32 protein kinases that met these criteria are displayed in a heat map (Fig. 3.2C). At 1 μM, 8a was again observed to be highly selective for MELK, with only MELK, STK11, and MAP2K4 captured with 4-fold less abundance, relative to DMSO. Notably, competition MIB/MS results indicated that MELK was not detected at quantifiable levels following 3 μM 8a treatment, suggesting total loss of
MELK binding to MIBs, while maintaining a high level of selectivity at this concentration. At
10 μM, however, the specificity of 8a was decreased, with robust off-target kinase inhibition observed.
While STK11 and MAP2K4 appear to exhibit decreased binding in a somewhat dose- dependent manner, it should be noted that neither exhibited significantly reduced binding with 1
μM 8a treatment in biological triplicate (Fig. 3.2B). We further tested the binding affinity of 8a for these kinases using the Eurofins DiscoverX KINOMEscan™ assay, which quantitatively measures the ability of an inhibitor to compete with a kinase ligand. 8a was found to have no detectable binding affinity for STK11 (Kd > 30 μM) or MAP2K4 (Kd = 17 μM) at 3 μM or lower, while high affinity for MELK was observed (Kd = 14 nM) (Fig. S3.4). Taken together, results from this cell-based selectivity profiling assay indicate that treatment of cells with 8a at 1
μM to 3 μM concentrations are sufficient for moderately strong levels of inhibition to near-total inhibition of MELK, respectively, while maintaining high selectivity for this kinase.
Effects of MELK inhibition on TNBC cell viability
MELK has been reported to play a role in TNBC proliferation and radioresistance (17,
20, 21). In TNBC and other cancers, RNAi-mediated depletion of this kinase impairs growth, an
60 effect that can be reversed with exogenous MELK rescue, indicating MELK may be an attractive therapeutic target (6, 17, 21, 22, 36, 37). Recent results demonstrating that genetic knockout of
MELK may cause no growth phenotype have called into question the approach of inhibiting this kinase as a cancer monotherapy (16, 18). In light of these disparate results, we sought to test whether selective MELK inhibition with 8a impaired the proliferation of two TNBC cell lines.
Treatment of MDA-MB-231 and MDA-MB-468 cells with 8a for 72 hours caused decreased cell viability, as measured by resazurin assay, with an IC50 of 1.7 μM +/- 0.4 and 2.3 μM +/- 0.4, respectively (Fig. 3.3A). We further measured the viability effects of 8a using long-term crystal violet cell staining assays. MDA-MB-231 and MDA-MB-468 cells were treated with 8a for 8 or
14 days, respectively, prior to staining with crystal violet. Similar IC50 values of between 1 and
3 μM were obtained for both cell lines (Fig. 3.3B), in agreement with shorter-term resazurin assay results. Notably, this inhibition of cell viability was observed at the same concentrations found to result in selective, moderately strong to near-total inhibition of MELK (Fig. 3.2C).
We next examined whether reduced cell viability was mediated through the induction of apoptosis or perturbation of cell cycle progression. MDA-MB-468 cells were treated with 0.5, 1, or 3 μM 8a, or DMSO vehicle control, and cells were collected over a 24h time course. Samples were immunoblotted for PARP/cleaved PARP, cyclins E1, A2, and B1, and p-Histone H3 (S10), to measure effects on apoptosis and cell cycle distribution (Fig. 3.3C). PARP cleavage was not observed at any concentration or timepoint, indicating that 8a did not induce apoptosis. 8a treatment also had no effect on the abundance of the measured cyclins, but there was a dose- dependent decrease in p-H3 (S10), a marker of M that is induced in G2 and dephosphorylated upon mitotic exit. This decrease was observed as early as 1h post-treatment and maintained through 24h.
61 Histone H3 is phosphorylated at Ser10 by Aurora B (145). To be certain that the observed decrease in p-H3 (S10) was not due to direct inhibition of the Aurora kinases, we utilized the Thermo Fisher SelectScreen kinase assay to test the effects of 8a on Aurora A and B activity. Concentrations below 10 μM 8a did not inhibit Aurora A. The IC50 for inhibition of
Aurora B was 24 μM, and only 10% and 21% inhibition was observed with 1.2 and 3.7 μM 8a treatment, respectively. The potency of 8a for the Aurora kinases was 1000-fold lower than that observed for MELK (IC50 = 17 nM) (Fig. S3.5), indicating that 8a does not significantly inhibit
Aurora A and B. It should also be noted that these data were from a cell-free enzymatic assay, so inhibition in cells is likely to be considerably weaker at these concentrations. In agreement with this, we did not observe substantial decreases in Aurora A or B binding to MIBs with 3 μM
8a or lower (Figs. S3.1 and S3.3). Collectively, these results show that MELK inhibition with 8a caused decreased viability through impaired cell cycle progression, specifically in G2/M phases.
MELK inhibition results in delayed mitotic entry mediated through delayed activation of Aurora A, Aurora B, and CDK1
Our results prompted further investigation into the effects of MELK inhibition on cell cycle progression. For subsequent experiments, HeLaS3 cells were used because of their robust and well-characterized response to cell cycle synchronization methods. We first confirmed that
HeLaS3 cells exhibit similar sensitivity to 8a (IC50 of 1.9 μM +/- 0.5) as TNBC cell lines (Fig.
S3.6A). We further observed that 8a treatment caused the same dose-dependent decrease in p-
H3 (S10), with no effect on PARP cleavage or abundance of the measured cyclins, in HeLaS3 cells as in MDA-MB-468 cells (Fig. S3.6B). MELK expression levels were found to be similar between HeLaS3, U2OS, and TNBC cell lines as well (Fig. S3.6C).
62 To specifically examine effects on progression through G2 and M, HeLaS3 cells were synchronized with double thymidine block and released into media containing DMSO, 1 μM, or
3 μM 8a, and harvested every 2h over a 12h time course. Samples were immunoblotted for cell cycle markers to determine the effect of 8a treatment on progression through S phase and into M
(Fig. 3.4A). No differences in cyclin E1 or A2 abundance were observed after 8a treatment from
0h to 8h, indicating that S phase progression was unaffected by MELK inhibition. At 8h post- release, p-H3 (S10) abundance was lower in 3 μM 8a-treated cells relative to DMSO- and 1 μM
8a-treated cells. Further, at 10h post-release, cyclins A2 and B1, and p-H3 (S10) abundances were greater with 3 μM 8a treatment, indicating that entry into and progression through M was delayed (Fig. 3.4B). At the 12h timepoint, cells appeared to have completed M, evident by major decreases in signal for cyclins E1, A2, and B1, and p-H3 (S10). Cyclin E1 abundance was decreased by 3 μM 8a at 12h, relative to other conditions, suggesting the effects of mitotic delay may persist after M. Treatment with 1 μM 8a had little effect on cell cycle progression as measured using this thymidine synchronization technique. Therefore, 3 μM 8a was used in subsequent synchronization experiments.
To further investigate this apparent delay in mitotic entry, we completed another double thymidine synchronization experiment with more frequent collections around M (8-12h post- release). Cells were not collected between 0h and 6h post-release, as no differences were observed with 8a over this timeframe in the previous experiment (Fig. 3.4A). Additionally, we sought to investigate the mechanism underlying this mitotic delay further by expanding the number of cell cycle proteins that were measured by immunoblotting. Aurora A and B have critical roles in mitotic entry and progression, and their activity is regulated by autophosphorylation at T288 and T232, respectively (146, 147). CDK1, regarded as the key
63 kinase for promoting M, is activated by cyclin binding and dephosphorylation of Y15 and T14.
These regulatory sites are phosphorylated by Wee1 and Myt1 in interphase, respectively, and are dephosphorylated by CDC25B (148). We immunoblotted for Aurora A and B and their regulatory phosphorylation sites, CDK1 and its regulatory phosphorylation sites, Myt1, Wee1,
CDC25B, and the cell cycle markers used previously (Fig. 3.5).
With more frequent timepoints, the delay in onset of M caused by 3 μM 8a was even more apparent than in the previous experiment. This was evident by delayed degradation of cyclins A2 and B1, delayed accumulation of the mitotic marker p-H3 (S10), delayed phosphorylation and activation of Aurora A and B, and delayed activation of CDK1 by dephosphorylation of Y15 and T14. In control cells, cyclin A2 accumulation peaked at 7h post- release and persisted through 8.5h, after which it was rapidly degraded. In 3 μM 8a-treated cells, cyclin A2 accumulated similarly, but expression persisted through 11h. Cyclin B1 accumulation was also unaffected by 8a, but its degradation upon mitotic exit was delayed by approximately
3h, beginning at 9h in control cells and 12h in 8a-treated cells. Consistent with this, p-H3 (S10) accumulation began at 8h and decreased after 10h in control cells, compared to accumulation beginning at 9h, peaking at 11h, and persisting at high levels through 12h in 8a-treated cells.
The activation of Aurora A and B by autophosphorylation was also delayed, correlating precisely with p-H3 (S10) patterns. CDK1 activation by dephosphorylation of Y15 and T14 began at 8.5h in control cells, compared to 11h in 8a-treated cells, while CDK1 expression was unaffected.
Interestingly, the dynamics of Wee1 expression appeared mostly unchanged between control and
8a conditions, while inactivation of Myt1 by hyperphosphorylation (slower migrating band on gel) mirrors the delayed dephosphorylation pattern observed with CDK1 T14 and Y15. Finally, the accumulation pattern of CDC25B may also be unaffected by MELK inhibition, though peak
64 levels of CDC25B may be decreased relative to control. These results indicate that MELK inhibition results in delayed mitotic entry due to a delay in activation of multiple kinases essential for this process, namely Aurora A and B, and CDK1.
MELK inhibition does not impair mitotic completion
After observing that MELK inhibition results in a delay in mitotic entry, we next asked whether it also had an effect on mitotic progression and completion. Following the previously described synchronization/release approach, cells were collected over a 20h time course and immunoblotted for the same cell cycle markers (Fig. 3.6). Again, degradation of cyclins A2 and
B1, and induction of p-H3 (S10) were delayed by 3 μM 8a treatment, correlating with delayed activation of Aurora A and B and CDK1. Despite this delay, cells treated with 3 μM 8a appeared to progress through and complete M in a similar amount of time, once M began. This was evident by a return of p-H3 (S10) to steady-state levels at 14h, compared to 12h for DMSO, and dephosphorylation of Myt1, resulting in a faster migrating band relative to hyperphosphorylated mitotic Myt1. MELK expression also decreased rapidly immediately following completion of
M, as described in other studies (17, 49). HeLaS3 cells treated with 3 μM 8a re-entered the subsequent cell cycle, as observed by an induction of cyclin E1 at 14h, compared to 12h for
DMSO. Levels of this G1/S phase cyclin peak at ~16-18h with 3 μM 8a treatment, which is 2h later than control cells, where cyclin E1 peaks at 14-16h. These results indicate that the length of
M appears to be largely unaffected by MELK inhibition, despite mitotic entry being delayed.
These observations, along with delayed activation of Aurora A and B and CDK1 are consistent with activation of the G2/M checkpoint (149). Since M occurs quite rapidly (~30 min), we sought to further examine these results using live-cell imaging for greater temporal resolution.
65 Live-cell imaging quantifies the mitotic entry delay caused by MELK inhibition
To more precisely measure the lengths of G2 and M phases in response to 8a treatment, we conducted live-cell imaging of U2OS osteosarcoma cells stably expressing a fluorescent cell cycle reporter based on PCNA localization (150). U2OS cells are routinely used for cell cycle studies, particularly with imaging methods, and our results indicate that MELK expression is similar between U2OS and TNBC cells (Fig. S3.6C). This live-cell imaging approach employs tracking of individual cells from an unperturbed, asynchronous population. Fluorescent PCNA delineated the start and end of S phase, while nuclear envelope breakdown and cytokinesis delineated the start and end of M. Cells were treated with DMSO, 1 or 3 μM 8a, and G2 length was measured. We observed significant, dose-dependent G2 lengthening in cells treated with either 1 μM or 3 μM 8a, compared to DMSO-treated cells (Fig. 3.7A). In contrast to the lower time resolution immunoblotting experiments with synchronized cells (Fig. 3.4), the increased sensitivity of live-cell imaging revealed that G2 was indeed significantly lengthened by 1 μM 8a treatment (~30% increase, 81 min). Following 3 μM 8a treatment, median G2 length was markedly increased (>250% increase, 426 min) compared to untreated cells (Fig 3.7C).
Live-cell imaging was next used to quantify the length of M phase. While no significant difference in time to complete M was observed between cells treated with DMSO and 1 μM 8a, 3
μM 8a caused detectable lengthening of M in two of three replicates (Figs. 3.7B and S3.7). This increase in mitotic length from 24 minutes in control cells to 36 minutes in 3 μM 8a-treated cells was much smaller in magnitude than the observed increase in G2 (Fig. 3.7C), which likely explains why it was not detected using synchronization methods. An increase in G2 length appears to be the predominant phenotype resulting from MELK inhibition. Taken together, our
66 results indicate that selective MELK inhibition causes a delay in mitotic entry that is consistent with transient G2 arrest.
Discussion
Currently, one of the main questions in the MELK field is how to rectify the seemingly disparate observations that RNAi-mediated MELK depletion and pharmacological inhibition cause a strong growth defect in cancer cells (5, 6, 17, 25, 36, 38, 39, 41, 74, 75), whereas
CRISPR/Cas9 knockout of MELK may have conditional or no effect on proliferation (16, 18, 19,
35, 44). It is vital to recognize that genomic knockout techniques are not identical to RNAi- mediated knockdown or inhibition, but rather that all are valuable orthogonal approaches. Thus, divergent results should be interpreted as evidence that the functional role of the protein being investigated is likely complex, rather than evidence that one set of results is incorrect in its entirety. It is quite possible that the total loss of protein expression and activity, via knockout, could have different effects than partial inhibition of activity with unperturbed expression, via pharmacological inhibition. Further, and because of the inherent differences between these techniques, it is also possible that genomic knockout could result in cellular reprogramming of signaling networks such that compensation for growth defects occurs, whereas partial inhibition or depletion of the same enzyme may not result in the same changes. As described in the introduction, divergent phenotypes have been observed, but not yet fully explained, between cyclin D-CDK4/6 knockout and inhibition, demonstrating that this phenomenon is not constrained solely to MELK (139–141). At this point, these potential explanations are purely speculative, but this will undoubtedly be an important area of investigation for future MELK studies.
67 This controversy has underscored the fact that our understanding of the function of
MELK in cancer is still lacking. As a further complication, functional studies of MELK have long relied upon OTS, an inhibitor with a remarkable lack of selectivity, to complement results obtained using knockdown and other techniques. While RNAi-mediated knockdown and genomic knockout approaches obviously have their merits, the use of a pharmacological inhibitor has distinct advantages as well. First, pharmacological inhibition generally allows for much tighter temporal control than other techniques, such that effects can be measured on a short timescale before significant changes in signaling networks occur. Second, inhibition of kinase activity often does not directly affect expression of the inhibited kinase, which allows for differentiation between activity-dependent and -independent functions. Thus, a selective MELK inhibitor would clearly be a valuable addition to the toolkit available for studying the mechanistic underpinnings of this kinase in cancer.
In an effort to identify a selective MELK inhibitor, we used competition MIB/MS technology to evaluate the selectivity profiles of OTS, HTH, and 8a. Our results (Fig. 3.2A) are in agreement with other studies in demonstrating that OTS is highly non-selective, and therefore is unsuitable for mechanistic studies of MELK biology (13, 15). In contrast to published enzyme assay data demonstrating HTH to be a potent MELK inhibitor, we found that this inhibitor did not significantly affect MELK binding to MIBs (Fig. 3.2, A and B) (16). This illustrates the disparate results that can be observed between enzyme- and cell-based selectivity profiling methods, emphasizing the importance of evaluating compound selectivity in a cellular context.
Our results indicating that 8a exhibits high selectivity for MELK (Fig. 3.2, A-C) are the first to characterize an exquisitely selective MELK inhibitor in cells, and thus rationalize the use of 8a for future functional studies. Since thoroughly validated MELK substrates are lacking, one
68 intriguing possibility is to use MS to compare the phosphoproteome of cells treated with 8a to control cells, in an effort to identify novel MELK substrates. Such studies could serve to establish a much-needed cellular readout for MELK activity, and to further elucidate the role of
MELK in the cell cycle and cancer. To date, MELK phosphoproteomics studies have only been completed using a mutant null allele of the MELK homolog in Caenorhabditis elegans (pig-1)
(93), or with OTS treatment (37).
As previously described, 8a treatment caused decreased viability of TNBC cells (Fig. 3.3,
A and B) at concentration ranges that significantly prevented MELK binding to MIBs (Fig. 3.2C)
(38). We did not observe an induction of apoptosis with 8a (Fig. 3.3C), despite some others describing this effect as a result of MELK knockdown (17, 25, 35, 75). It is possible that the longest time point used in our study (24h) is not sufficient for induction of apoptosis with 8a, and that the observed cell cycle perturbation precedes apoptosis. Both pro-survival and pro-apoptotic roles have been attributed to MELK, and the specific contexts that determine these opposing functions remain to be elucidated (67).
While many studies describe a role for MELK in cell cycle progression, the precise details of this role are not well understood. In response to MELK silencing with RNAi, numerous studies have reported an accumulation of cells in G2/M phases (5, 9, 35, 38, 41). It should be noted that due to the methods used (quantification of cellular DNA by flow cytometry) these particular studies were not able to differentiate between an accumulation in G2 versus M.
Paradoxically, overexpression of MELK may induce a G2/M accumulation of cells as well (5,
73). Other studies have used more specific techniques to differentiate between G2 and M phase effects. MELK knockdown has been reported to impair mitotic progression by varied mechanisms, including by decreasing activity and expression of the oncogenic mitotic
69 transcription factor FoxM1 (39, 74), by regulating MCL1 synthesis in M (75), by inducing mitotic catastrophe (22), and by causing cytokinetic defects (17, 76, 77). A role for MELK in
G2, distinct from M, has been described as well. Davezac et al. observed that ectopic expression of MELK (formerly known as pEg3) caused an accumulation of cells in G2, which could be rescued by co-expression of CDC25B. MELK was found to directly interact with CDC25B in
Xenopus, and phosphorylate this phosphatase in vitro, leading the authors to hypothesize that
MELK negatively regulates the G2/M checkpoint through interaction with CDC25B (73).
As there is no clear consensus as to whether MELK plays a role in G2, M, or both, we used 8a, as an alternative to RNAi knockdown, to further investigate the cell cycle function of
MELK. Time-lapse microscopy experiments (Fig. 3.7) indicated that MELK inhibition caused a major, significant lengthening of G2 phase. Of note, we observed many cells that delayed in G2 phase after exposure to 8a within that same cell cycle, suggesting that the G2 delay phenotype is proximal to MELK inhibition. An additional minor delay in mitotic progression was observed, warranting further investigation in light of previous studies indicating that MELK has mitotic functions. However, we focused on the more predominant delay in mitotic entry. Using thymidine synchronization techniques (Figs. 3.4-3.6), we observed that MELK inhibition caused delayed mitotic entry, likely mediated through delayed activation of CDK1, which was ultimately overcome to allow cells to complete M and begin the subsequent cell cycle. These observations are consistent with a transient activation of the G2/M checkpoint (149). Further studies are required to elucidate the underlying mechanism, though it is possible to speculate based upon the work of others.
One potential explanation for the observed G2 delay phenotype is that MELK positively regulates CDC25B in G2, such that inhibition of MELK causes CDC25B to be unable to
70 dephosphorylate and activate CDK1, thereby delaying mitotic entry. MELK phosphorylates multiple sites on CDC25B to regulate both progression through the G2/M checkpoint, and
CDC25B localization to centrosomes during M (73, 113). Importantly, this putative interaction has never been examined in the context of cancer. A possible interaction with CDC25C, which has a critical role at the G2/M checkpoint and shares high sequence homology with CDC25B, has not been explored either. Additional studies are needed to more clearly define the interaction between MELK and the CDC25 phosphatases in cancer. An alternative hypothesis is that
MELK inhibition causes DNA damage, which subsequently activates the G2/M checkpoint. A study by Beke et al. found that MELK inhibition with MELK-T1 caused replication stress in
MCF-7 cells, manifesting in replication fork stalling and DNA double strand breaks. This resulted in activation of the ATM/Chk2 DNA damage repair pathway, which stimulates the
G2/M checkpoint (39, 151). While this pathway would certainly explain the putative checkpoint activation, one would also expect strong ATM activation to lengthen S phase, which was not observed in our study (Fig. 3.4).
In conclusion, we have validated 8a as a highly selective MELK inhibitor suitable for functional studies, and have shown that MELK inhibition causes a mitotic delay consistent with activation of the G2/M checkpoint.
Experimental Procedures
Antibodies and reagents
MELK-8a hydrochloride was purchased from MedChemExpress (HY-100368A).
OTSSP167 was purchased from Selleck Chemicals (S7159). HTH-01-091 was a generous gift from Dr. Nathanael Gray (Harvard). The following primary antibodies were used in this study.
Aurora A (14475, 1:1000), Aurora B (3094, 1:1000), CDC25B (9525, 1:1000), cyclin A2 (4656,
71 1:2000), cyclin E1 (4129, 1:2500), MELK (2274, 1:1000), Myt1 (4282, 1:1000), PARP (9532,
1:1000), p-Aurora A/B/C (2914, 1:1000), p-CDK1 (T14) (2543, 1:1000), p-CDK1 (Y15) (4539,
1:1000), and p-Histone H3 (S10) (9701, 1:1000) were purchased from Cell Signaling
Technology. β-actin (SC47778, 1:500) was purchased from Santa Cruz Biotechnology. CDK1
(MAB8878, 1:2000) was purchased from Millipore. Cyclin B1 (ab32053, 1:10000) was purchased from Abcam.
Cell culture and viability assays
MDA-MB-468 and MDA-MB-231 cells were kindly provided by Dr. Gary Johnson
(UNC). These cell lines were cultured in RPMI 1640 (Gibco) supplemented with 10% fetal bovine serum (FBS, Millipore) and 1% antibiotic-antimycotic (Sigma). HeLaS3 cells were kindly provided by Dr. Michael Emanuele (UNC), and were cultured in Dulbecco’s Modified
Eagle Medium (DMEM), high glucose (Gibco) supplemented with 10% FBS and 1% antibiotic- antimycotic.
Resazurin assays were completed as described previously (152). Briefly, MDA-MB-231
(1500 cells/well), MDA-MB-468 (4000 cells/well), or HeLaS3 (1000 cells/well) cells were plated on a 96-well plate. After 24h, media was aspirated and replaced with fresh media containing 8a at concentrations ranging from 1 nM to 30 µM, or DMSO (negative control). Each treatment condition was tested in technical quadruplicate. Cells were incubated in media with 8a or DMSO for 72h prior to addition of resazurin dye (0.6 mM, Acros Organics 62758-13-8).
After color change was observed (3h for MDA-MB-231, 2h for MDA-MB-468, 0.75h for
HeLaS3), signaling that measurable reduction of resazurin to resorufin had occurred, fluorescence intensity was measured using a PHERAstar (BMG Labtech) plate reader with
72 fluorescence module (FI: 540-20, 590-20). Dose-response curves were created and IC50 values calculated using GraphPad Prism version 8.
For crystal violet assays, MDA-MB-231 and MDA-MB-468 cells were seeded at a density of 5000 and 10000 cells per well, respectively, on each well of a cell culture-treated 6- well plate (Corning). The following day, media was aspirated from all wells and replaced with media supplemented with either DMSO or 0.1, 0.3, 1, 3, or 10 µM 8a. Every 2 days throughout the course of the assay, media was replaced on all wells with fresh media containing 8a or
DMSO. Cells were observed daily using a bright-field microscope and confluency was noted.
The assay was continued until cells in the DMSO well were observed to be near-confluent (~8d for MDA-MB-231, and ~14d for MDA-MB-468). Media was then aspirated from all wells, and cells were washed gently with ice-cold phosphate buffered saline (PBS, Gibco). Next, 2 mL crystal violet stain (0.5% crystal violet (w/v) (Sigma), 20% methanol (Fisher)) was gently added to each well and plates incubated for 10 minutes at room temperature. Wells were subsequently washed three times with water, allowed to dry, and plates were imaged with a Canon LiDE 110 document scanner.
Double thymidine synchronization
HeLaS3 cells were seeded on 6 cm plates and allowed to adhere overnight. During this and all subsequent incubation steps, cells were maintained in a humidified chamber with 5% CO2 at 37°C. DMEM was aspirated from all plates (except for asynchronous control plates, which were maintained in an incubator until the end of the assay) and replaced with media supplemented with 2 mM thymidine (Sigma). 16h after thymidine block, media was aspirated, plates were washed with warm PBS (1x) and warm DMEM (2x), and fresh media was added for
73 release from block. Plates were incubated for 9h, at which point media was aspirated and again replaced with DMEM supplemented with 2 mM thymidine and incubated for 16h. Media was aspirated and cells washed as above, and released from thymidine block into DMEM supplemented with 8a or DMSO. 0h timepoints were collected at the time of release, and subsequent collections were completed at the timepoints shown in figures. Cells were collected by scraping in media and cell suspension was transferred to a 15 mL conical. Cells were pelleted at 3000 rcf in a centrifuge with swinging bucket rotor, washed with cold PBS, and stored at -
80°C. After all timepoints were collected, cell pellets were lysed and prepared for immunoblotting as described below.
Immunoblotting
HeLaS3 cells were plated in 6-well plates and treated with DMSO or 0.5, 1, or 3 µM 8a for 1h, 2h, 4h, 8h, 12h, or 24h. Cells were collected as described above for double thymidine synchronization protocol. Pellets from asynchronous and synchronized experiments were lysed by addition of RIPA buffer (20 mM Tris pH 7.4, 137 mM NaCl, 10% glycerol, 0.5% sodium deoxycholate, 1% Nonidet P-40 substitute (Fluka), 0.1% sodium dodecyl sulfate, 2 mM EDTA) supplemented with protease and phosphatase inhibitors (2 mM Na(VO3)4, 10 mM NaF, 0.0125
µM calyculin A, and cOmplete EDTA-free Protease Inhibitor Cocktail (Roche). Benzonase was added to each lysate (42 units/sample) and samples were incubated on ice for 10 minutes.
Lysates were clarified by centrifugation at 21000 rcf, 4°C for 15 minutes. Protein concentrations were normalized by Bradford assay (Bio-Rad), and prepared for SDS-PAGE by addition of 4x
Laemmli sample buffer (250 mM Tris pH 6.8, 40% glycerol, 8% sodium dodecyl sulfate, 8% β- mercaptoethanol, 0.4% bromophenol blue). Samples (10-30 µg) were applied to a 10% SDS-
74 PAGE gel for separation of proteins, after which proteins were transferred to a PVDF membrane
(Bio-Rad). Membranes were blocked for 1h with 5% non-fat dry milk or 5% bovine serum albumin (BSA, for phospho-specific antibodies) in TBS-T (20 mM Tris pH 7.6, 137 mM NaCl,
0.05% Tween-20). Membranes were incubated in primary antibody dilutions (described above) in 5% BSA made in TBS-T, at 4°C with gentle shaking. Following overnight incubation, membranes were washed with TBS-T three times and incubated for 1h at room temperature with anti-mouse or -rabbit IgG-HRP conjugated secondary antibody (Promega) dilutions in 5% milk in TBS-T. Membranes were again washed three times in TBS-T, and bands imaged using a
Chemi-Doc Touch Imaging System (Bio-Rad) after addition of Clarity ECL reagent (Bio-Rad).
MIB/MS selectivity profiling sample preparation
The selectivity of MELK inhibitors for kinase targets was profiled using competition
MIB/MS. MDA-MB-468 cells were treated with MELK inhibitor, at concentrations listed in figure legends, or DMSO for 30 minutes. This timepoint is sufficient for kinase engagement and inhibition by compound, but insufficient for major expression-level changes. Cells were then washed twice with ice-cold PBS before being scrape-harvested in PBS. Cell lysis, sample preparation, and MIB kinase enrichment was completed as previously described (138), with the exception of the composition of the kinase inhibitor mix used for enrichment. To enrich for kinases in each sample, a 350 µL volume of the following kinase inhibitors, immobilized on beads, was applied to one Poly-Prep® chromatography column (Bio-Rad) per sample: CTx-
0294885, PP58, Purvalanol B, UNC2147A, VI-16832, UNC8088A.
75 LC-MS/MS Analysis
Samples were analyzed by LC-MS/MS using a Thermo Easy nLC 1200 coupled to an
Orbitrap Q Exactive HF mass spectrometer equipped with an EasySpray nano source. Samples were loaded onto an EasySpray C18 column (75 µm ID X 25cm, 2 µm particle size) and eluted over a 120 min method. The gradient for separation consisted 5-40% B at a 250 nL/min flow rate, where mobile phase A [water, 0.1% formic acid] and mobile phase B [80% acetonitrile,
0.1% formic acid]. The Q Exactive HF was operated in data-dependent mode where the 15 most intense precursors were selected for subsequent fragmentation. Resolution for the precursor scan
(m/z 350–1700) was set to 120,000 with a target value of 3 × 106 ions, 100 ms max IT. MS/MS scans resolution was set to 15,000 with a target value of 1 × 105 ions, 75 ms max IT. The normalized collision energy was set to 27% for HCD. Dynamic exclusion was set to 30 s and precursors with unknown charge or a charge state of 1 and ≥ 8 were excluded.
MS data analysis
Data was processed using the MaxQuant software suite (version 1.6.1.0) (153, 154). The data were searched against a reviewed Uniprot human database (downloaded in February 2018) containing 20,245 sequences. Precursor mass tolerance was set to 4.5 ppm and fragment mass tolerance was set to 20 ppm. A maximum of two missed tryptic cleavages were allowed. The fixed modification specified was carbamidomethylation of cysteine residues. The variable modification specified was oxidation of methionine. A default protein false discovery rate
(FDR) of 1% was used to filter all data within MaxQuant. Proteins were quantified across all samples using MaxLFQ (155). Matching between runs was allowed with the default retention time window. Kinases were parsed from the dataset and those with >1 unique peptide were
76 quantified using label-free quantification (LFQ). Kinases with >50% missing values were removed and LFQ intensities for the missing values were imputed using a constant value
(average lower 10% of all LFQ intensities). The ratio of the LFQ intensity for each kinase in
MELK inhibitor conditions to that kinase in DMSO control was computed. GraphPad Prism version 8 was used to generate bar plots and heat maps. For data used to generate volcano plots, kinases that were not quantified in all three replicates in at least one condition within each analysis were removed. Missing LFQ values were then imputed by randomly sampling from a normal distribution with a mean of 1.8 standard deviations lower than the mean in the original data, and a standard deviation of 0.3 times the standard deviation in the original data. Data were
Log2 transformed and moderated t-tests were computed using the Limma package (156). Briefly, a linear model predicting signal intensity given treatment condition was fit for each kinase, followed by t-statistics calculated by empirical Bayes moderation of standard errors towards the standard error estimated from all kinases. Multiple test correction was performed using the
Benjamini-Hochberg method to control for a 5% FDR (157). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (158) partner repository with the dataset identifier PXD016022.
Live-cell imaging
U2OS cells were a gift from Dr. Michael Whitfield (Dartmouth College) and maintained in DMEM (Sigma) supplemented with 10% FBS (Sigma), 1x penicillin-streptomycin, and 4 mM
L-glutamine and incubated at 5% CO2. Cells were transduced with fluorescent PCNA (pLenti-
Hyg-turq2-PCNA) and histone H2B (pBabe-Puro-mCh-H2B) using standard techniques, as described previously (150). One day prior to imaging, cells were plated on #1.5 glass-bottom
77 plates (Cellvis) in FluoroBrite DMEM (Gibco) supplemented with FBS, penicillin-streptomycin, and L-glutamine, as above. Imaging was performed on a Nikon Ti Eclipse inverted microscope using a 40x (NA 0.95) Plan Apochromat dry objective lens and the Nikon Perfect Focus System.
Still images were captured using an Andor Zyla 4.2 sCMOS detector with 12-bit resolution.
During imaging cells were maintained in a humidified chamber with 5% CO2 at 37°C. Filter sets were CFP - 436/20 nm; 455 nm; 480/40 nm (excitation; beam splitter; emission filter) and mCherry - 560/40 nm; 585 nm; 630/75 nm (Chroma). Images were obtained every 6 minutes using NIS-Elements AR software. No photobleaching or phototoxicity was observed in cells imaged with this protocol.
8a or DMSO (negative control) was added 24 hours after the beginning of the imaging run as described in the figure legends. Cells were manually tracked and scored. The onset and end of S phase were determined visually using the DNA replication associated patterns of PCNA localization in the nucleus. Mitosis length was defined visually by the length of time between nuclear envelope breakdown and cytokinesis. Three individual biological replicates were imaged with 50 cells counted per condition per replicate. Cells that traveled out of the field of view or did not complete a cell cycle (S phase to cytokinesis) were excluded from the analysis.
78 Figures
Figure 3.1. Schematic of competition MIB/MS workflow and sample selectivity output data. MDA-MB-468 cells were treated with DMSO (negative control) or MELK inhibitor for 30 minutes. This timepoint allows sufficient time for inhibitors to penetrate cells and engage kinase targets, but not for significant expression-level changes. After harvest, cell lysates were flowed over columns containing kinase inhibitors immobilized on Sepharose® beads, which bind kinases in the cell lysates (top). Kinases bound to the added MELK inhibitor are unable to bind to the MIBs beads, and thus are not captured. Following wash steps to remove non-specific binders, kinases were subsequently eluted and quantified by MS using label-free quantification. Kinases detected in lower abundance in MELK inhibitor conditions relative to DMSO control (bottom) are prevented from binding to beads due to the inhibitor, and thus are targets of that inhibitor.
79
Figure 3.2. 8a is a highly selective MELK inhibitor. A, MDA-MB-468 cells were treated for 30 minutes with either DMSO or 1 µM 8a, HTH, or OTS. Cells were harvested and competition MIB/MS was completed, as described in Experimental Procedures, to determine the selectivity profile of each MELK inhibitor. The 20 kinases detected in lowest abundance, relative to DMSO control, are displayed for each inhibitor. Arrows indicate the bar corresponding to MELK in each selectivity profile. Results shown are from one experiment. B, MDA-MB-468 cells were treated for 30 minutes with either DMSO, 1 µM 8a, or 1 µM HTH in three biological replicates (distinct from the replicate shown in A). Competition MIB/MS was used to determine selectivity as in A, and results displayed as volcano plots to show both the fold-change and significance of the quantified changes in kinase abundance. Kinases with high significance and log2 fold-change < -1 are colored red and labeled. Kinases with only log2 fold-change < -1 are colored yellow. Kinases with log2 fold-change > -1 and non-significance are colored gray. T- statistics were calculated by empirical Bayes moderation of standard errors towards the standard error estimated from all kinases (156). The Benjamini-Hochberg method was used for multiple
80 test correction with a 5% false discovery rate (157). C, MDA-MB-468 cells were treated with DMSO or 1, 3, or 10 μM 8a for 30 minutes. Cells were harvested and lysates subjected to competition MIB/MS as in A. Data were filtered for kinases that showed decreased binding at all three concentrations in a dose-dependent manner. The 32 kinases that met these criteria are displayed in a heat map, with a double-gradient color scheme ranging from dark blue (near-total loss of binding to MIBs) to red (no loss of binding to MIBs). Results are indicative of one experiment.
81
Figure 3.3. Treatment with 8a causes decreased viability and decreased phosphorylation of Histone H3 (S10) in TNBC cells. A, MDA-MB-231 (left) and MDA-MB-468 (right) cells were treated with DMSO or 8a at concentrations ranging from 1 nM to 30 μM. Cells were exposed to 8a for 72h before viability was measured using a resazurin assay. Three biological replicates were completed for each cell line, and representative dose-response curves are displayed here. B, MDA-MB-231 and MDA-MB-468 cells were treated with DMSO or 8a at indicated concentrations. The medium was changed and fresh inhibitor added every two to three days. Cells were monitored until near-confluent in the DMSO condition (about 8 or 14 days, respectively), at which point they were stained with crystal violet and imaged. Three (MDA- MB-231) or two (MDA-MB-468) biological replicates were completed, and representative images are shown here. C, asynchronous MDA-MB-468 cells were treated with DMSO or 0.5, 1, or 3 μM 8a for 1, 6, or 24h. Cells were harvested and immunoblotted with the indicated antibodies. Data displayed here are representative of three biological replicates.
82
Figure 3.4. MELK inhibition with 3 μM 8a causes delayed mitotic entry. A, HeLaS3 cells were synchronized at G1/S with double thymidine block and released into DMSO, 1 μM, or 3 μM 8a. Cells were collected every 2h over a 12h timecourse. Cells were lysed and immunoblotted with the indicated antibodies. The splits in the image indicate that the samples from DMSO, 1 μM, and 3 μM 8a conditions were immunoblotted separately due to space constraints. Asynchronous HeLaS3 lysate was loaded in the lanes labeled “A” in order to normalize among the blots. The data shown here are representative of three independent experiments. B, the signal intensity of the cyclins E1, A2, and B1, and p-H3 (S10) bands was quantified for the replicate shown in A using Bio-Rad Image Lab™ software. The ratio of the quantified signal intensity at each timepoint relative to the 0h timepoint was computed for each condition. The resulting ratios are plotted as bar charts for each protein, with 1.0 indicating no change relative to the 0h timepoint.
83
Figure 3.5. Delayed mitotic entry caused by 8a is associated with delayed activation of Aurora A, Aurora B, and CDK1. HeLaS3 cells were synchronized at G1/S with double thymidine block and released into DMSO or 3 μM 8a. Cells were collected starting at 6h post- release, with subsequent collections completed every 0.5 or 1h, over a 12h timecourse. Cells were lysed and immunoblotted with the indicated antibodies. The split in the image indicates that the samples from DMSO and 3 μM 8a conditions were blotted separately due to space constraints. Asynchronous HeLaS3 lysate was loaded in the lanes labeled “A” in order to normalize among the blots. The data shown here are representative of two independent experiments.
84
Figure 3.6. HeLaS3 cells treated with 8a complete mitosis and begin the subsequent cell cycle following delayed mitotic entry. HeLaS3 cells were synchronized at G1/S with double thymidine block and released into DMSO or 3 μM 8a. Cells were collected starting at 6h post- release, with subsequent collections completed every 2h over a 20h timecourse. Cells were lysed and immunoblotted with the indicated antibodies. The split in the image indicates that the samples from DMSO and 3 μM 8a conditions were blotted separately due to space constraints. Asynchronous HeLaS3 lysate was loaded in the lanes labeled “A” in order to normalize among the blots. The data shown here are representative of at least two biological replicates.
85
Figure 3.7. 8a causes a major, dose-dependent increase in G2 length. A, asynchronous U2OS cells harboring fluorescent PCNA were treated with DMSO, 1 μM, or 3 μM 8a. Microscopy images were acquired every 6 minutes over the course of 72h. Cells were manually tracked, and the length of G2 was calculated by measuring the amount of time between disappearance of PCNA foci (end of S phase) and nuclear envelope breakdown (start of M). For each of three biological replicates, 50 cells were tracked per condition. Kaplan-Meier curves were generated and significance calculated in GraphPad Prism using the log-rank test for survival curves. Representative data are shown here, and results from additional replicates are shown in Fig. S3.7. B, the length of M was manually calculated in the cells in A by measuring the amount of time between nuclear envelope breakdown (start of mitosis) and cytokinesis (end of mitosis). For each of three biological replicates, 50 cells were tracked per condition. Curves were generated and significance calculated as in A. Data shown here, indicating M to be significantly lengthened by 3 μM 8a, are representative of 2 out of 3 biological replicates. Additional experiments (Fig. S3.7) indicate that M was not significantly lengthened in one replicate. C, median length of G2 and M, in minutes. The displayed cell cycle phase lengths were calculated for the experiment shown in A and B.
86
Figure S3.1. Full competition MIB/MS selectivity profiles for 1 μM 8a, HTH, and OTS. MDA-MB-468 cells were treated with 1μM 8a, HTH, or OTS, and competition MIB/MS was completed as in Fig. 3.2A. All 235 protein kinases that were quantified by MS are displayed. Results are from the experiment shown in Fig. 3.2A.
87
Figure S3.2. Full competition MIB/MS selectivity profiles for 10, 100, and 1000 nM 8a. MDA-MB-468 cells were treated with 10, 100, or 1000 nM 8a, and competition MIB/MS was completed as in Fig. 3.2A. All 235 protein kinases that were quantified by MS are displayed. The results shown here are indicative of one experiment.
88
Figure S3.3. Full competition MIB/MS selectivity profiles for 1, 3, and 10 µM 8a. MDA- MB-468 cells were treated with 10, 100, or 1000 nM 8a, and competition MIB/MS was completed as in Fig. 3.2A. All 221 protein kinases that were quantified by MS are displayed. Results are from the experiment shown in Fig. 3.2C.
89
Figure S3.4. 8a has no detectable affinity for STK11 or MAP2K4 at 3 µM. The Eurofins DiscoverX cell-free assay platform was used to determine the binding affinity of 8a for STK11, MAP2K4, and MELK. Results shown here are indicative of two biological replicates. The binding constant (Kd) displayed in each plot is the average of the two replicates.
90
Figure S3.5. Aurora A and B are minimally inhibited by 8a in cell-free kinase assays. The Thermo SelectScreen kinase assay service was used to determine the percent inhibition of Aurora A and B upon exposure to 8a at concentrations ranging from 5 nM to 100 μM. Results shown are representative of two independent experiments.
91
Figure S3.6. Treatment with 8a causes decreased viability in HeLaS3 cells. A, HeLaS3 cells were treated with DMSO or 8a at concentrations ranging from 1 nM to 30 μM. Cells were exposed to 8a for 72h before viability was measured using a resazurin assay. Two biological replicates were completed, and representative results are displayed here. B, asynchronous HeLaS3 cells were treated with DMSO or 0.5, 1, or 3 μM 8a for 1, 2, 4, 8, 12, or 24h. Cells were harvested and immunoblotted with the indicated antibodies. The split in the center of the image indicates that the 1, 2, and 4h samples were immunoblotted separately from the 8, 12, and 24h samples, due to space constraints. Three biological replicates were completed. Data displayed here are representative results. C, asynchronous HeLaS3, U2OS, MDA-MB-231, and MDA- MB-468 cells were harvested and immunoblotted with the indicated antibodies. The displayed results are representative of three biological replicates.
92
Figure S3.7. Additional replicates of live-cell imaging experiment measuring G2 and M phase lengths. Asynchronous U2OS cells harboring fluorescent PCNA were treated with DMSO, 1 μM, or 3 μM 8a and G2 and M phase lengths were measured as described in Fig 3.7. A and B. Three independent experiments were completed, and representative results are shown in Fig. 3.7. Curves for G2 length (top), M length (middle), and quantified cell cycle phase lengths (bottom) for additional replicates are shown here.
93
CHAPTER 4: INVESTIGATION OF PUTATIVE MELK INTERACTORS AND COMBINATORIAL STRATEGIES TO INCREASE MELK-8A POTENCY3
Introduction
In chapter 3, we demonstrated that MELK inhibition with 8a causes delayed mitotic entry, and that cells ultimately overcome this delay to enter and progress through mitosis. This phenotype is consistent with transient activation of the G2/M DNA damage checkpoint. DNA damage can occur in cells through various means, resulting in activation of the DNA damage response (DDR). Two of the main forms of DNA damage are single-strand and double-strand breaks (DSBs), which canonically result in activation of the ATR/Chk1 and ATM/Chk2 pathways, respectively. Activation of these pathways induces the phosphorylation of a multitude of substrates, including p53 and its stability regulators MDM2 and MDMX, BRCA1 and -2, histone H2A.X, and CDC25 phosphatases, enabling the DDR to slow or halt cell cycle progression and repair the damage (159).
The list of known MELK substrates is short and generally not well-validated, but some putative interactors play a role in the DDR and cell cycle progression, specifically G2/M checkpoint function, including FoxM1, CDC25B, and p53 (67). FoxM1, a transcription factor that regulates key mitotic genes, has been shown to both induce MELK expression and to be phosphorylated by MELK in glioma cells, regulating its transcriptional activity in a PLK1-
3A portion of this chapter, including Fig. 4.1 and descriptions of the BioID technique, was previously published in the Journal of Cellular Biochemistry. The original citation is as follows: Cann, M. L., McDonald, I. M., East, M. P., Johnson, G. L., and Graves, L. M. (2017) Measuring Kinase Activity-A Global Challenge. J. Cell. Biochem. 118, 3595–3606.
94 dependent manner (74). CDC25B is among the first proteins to be recognized as a MELK substrate. This interaction has only been studied in Xenopus, where it was shown that MELK phosphorylates CDC25B to modulate its phosphatase activity, thereby contributing to regulation of mitotic progression through CDK1 activation (73, 113). MELK has been shown to phosphorylate p53, driving increased association with positive regulators and enhanced stability, thus increasing expression of p53 target genes (83). Moreover, it was recently demonstrated that mutant p53 regulates MELK expression by controlling FoxM1 expression, further suggesting that MELK is a regulator of the DDR and cell cycle progression (71). While these putative substrate interactions provide some clues regarding potential mechanisms underlying the observed mitotic delay phenotype, none have been extensively studied, and no unified model of
MELK function exists, either globally or in the specific processes of DDR and G2/M checkpoint function.
It is clear that a more comprehensive understanding of MELK interactors and substrates will be required to elucidate the function of MELK in various processes. Classical affinity-based interaction studies such as immunoprecipitation (IP)/MS are widely used to study protein-protein interactions. However, it can be challenging to identify kinase substrates using these methods due to the transient nature of phosphorylation reactions. For this reason, more recently developed proximity biotinylation approaches such as BioID and APEX have gained popularity for probing the interactome of kinases and protein complexes (160). The BioID (proximity- dependent biotin identification) technique relies on the fusion of the kinase being studied to a mutated, promiscuous form of BirA, a biotin ligase from Escherichia coli (161). The BirA gene used for BioID has a point mutation that results in a ligase (BirA-R118G (BirA*)) with decreased affinity for biotinyl-5’-AMP, the highly reactive intermediate of the biotinylation
95 reaction (162). When a BirA*-kinase fusion protein is expressed in cells grown in media supplemented with biotin, the biotinyl-5’-AMP molecule is released by BirA* and irreversibly biotinylates lysine residues on proximal proteins (within ~10 nm), with no required biotinylation consensus sequence (Fig. 4.1) (163, 164). Biotinylated proteins can then be captured with streptavidin Sepharose and identified by MS, to ultimately identify proximal proteins and candidate kinase substrates.
Another well-established approach for substrate identification and characterization is phosphoproteomics/MS. The coupling of phosphopeptide enrichment strategies to MS allows for the identification and quantification of phosphate modifications on proteins, thereby enabling varied analyses, from examining global phosphoproteome dynamics to mapping phosphorylation sites on a specific protein of interest (165, 166). In the latter scenario, phosphorylation site mapping in combination with either overexpression or selective inhibition of a kinase of interest enables the phosphorylation of specific residues to be attributed to a particular kinase (167).
This can be a powerful approach when probing kinase-substrate interactions, and often serves as a starting point for mechanistic studies of this nature.
The validation of 8a as the first highly selective MELK inhibitor (chapter 3) provides a valuable tool compound for functional studies of MELK. Further, this molecule can also be used to revisit the therapeutic potential of MELK inhibitors for cancer treatment. The leading MELK inhibitor OTS has shown some pre-clinical promise (10–12, 35, 41), spawning multiple clinical trials, but it has also been demonstrated to be non-selective for MELK (13–15, 17). Thus, the clinical fate of this inhibitor will not ultimately settle the dispute concerning whether or not
MELK is an attractive therapeutic target in cancer. In addition to single-agent therapeutic studies with 8a, drug synergy studies could be used to identify efficacious drug combinations.
96 Indeed, the therapeutic use of drug combinations to leverage polypharmacology, particularly involving epigenetic modulation, has proven to be an attractive approach to improve efficacy relative to single-agent therapies (168–170). The elucidation of synergistic relationships has also aided in the investigation of the mechanism of action of drugs and the functional characterization of drug targets (171–173).
In this chapter, we further investigate the mechanism underlying the delay in mitotic entry, observed in chapter 3, as a result of MELK inhibition with 8a. We found that 8a treatment causes dose-dependent induction of the DNA damage marker p-H2A.X (S139) and activation of
Checkpoint kinase 2 (Chk2), suggesting that MELK inhibition delays mitotic entry by activating the G2/M DNA damage checkpoint. We next revisited past results in an effort to shed light on the mechanistic link between MELK inhibition and activation of the DDR. The BioID technology was used to identify putative MELK interactors, which included the MCM complex,
CDK1/cyclin B1, p53, and other proteins involved in cell cycle and DDR-related processes.
Phosphoproteomics/MS techniques were used to probe the MELK-FoxM1 relationship. Our results demonstrated that FoxM1-S481 is a candidate MELK phosphorylation site, and phosphomimetic mutation of this residue was observed to slightly increase FoxM1 transcriptional activity. Finally, we conducted an unbiased screen for kinase inhibitors and epigenetic regulators that synergize with 8a in TNBC cells. The goal of this screen was two- fold; first, we aimed to identify drug combinations that would potentially be clinically actionable, and second, we aimed to gain insight into the function of MELK by examining the targets of synergistic compounds. We identified three epigenetic regulators, namely UNC0642,
OTX015, and UNC1999, that may exhibit favorable combinatorial effects with 8a. While the results presented in this chapter are generally incomplete, they potentially lay the groundwork for
97 future MELK interaction studies and investigations of combinatorial drug strategies involving the MELK inhibitor 8a.
Results
MELK inhibition activates the Chk2 pathway
DNA DSBs are the most deleterious form of DNA damage. ATM kinase is activated in response to DSBs, resulting in activation of Chk2 by phosphorylation of threonine 68 and phosphorylation of serine 139 of the histone variant H2A.X, at the sites of DSBs, which facilitates the recruitment of repair machinery to these sites (151). We treated HeLaS3 cells with
DMSO or 0.5, 1, or 3 μM 8a, for times ranging from 1h to 24h, and immunoblotted for Chk2, p-
Chk2 (T68), and p-H2A.X (S139) (Fig. 4.2A). Treatment with 8a caused a dose-dependent increase in p-H2A.X (S139), a marker of DNA damage, at all measured timepoints. This phosphorylation correlated precisely with increased activation of Chk2, as measured by p-Chk2
(T68), while total Chk2 expression remained largely unchanged. These observations suggest that the delay in mitotic entry caused by MELK inhibition is likely mediated by activation of the
Chk2 DNA damage pathway.
Given these results, we hypothesized that DNA damage occurs in cells following MELK inhibition. To test this hypothesis, we examined whether DNA damage repair-deficient cells were more sensitive to 8a than normal cells. RNF168 is an E3 ubiquitin ligase whose ubiquitination activity is essential for the DDR to DSBs (174). For these experiments, we utilized RNF168 knockout U2OS cells that were generated via CRISPR-mediated genomic knockout by the Cyrus Vaziri lab (UNC). The viability of wild-type and RNF168 knockout
U2OS cells was measured following 72h 8a treatment using an MTS assay (Fig. 4.2B). Both control cells and RNF168 knockout U2OS cells exhibited equal sensitivity to 8a (IC50 = 1 μM),
98 demonstrating that deficiency in the DDR caused by RNF168 loss does not confer increased sensitivity to MELK inhibition.
While the results of these experiments provided some mechanistic insight into the mitotic delay phenotype observed in chapter 3, they did not fully elucidate the steps linking MELK inhibition to Chk2 activation and delayed mitotic entry. In the remainder of this chapter, we revisit previous results in search of a deeper mechanistic understanding of this phenotype.
BioID identifies novel putative MELK interacting proteins
As noted previously throughout this work, the function of MELK in cancer is poorly understood, which is in part due to a dearth of well-validated MELK interactors and substrates
(67). To identify novel putative MELK interacting proteins, we utilized BioID. Because of the small labeling radius inherent to BioID, it is important to carefully consider whether to fuse
BirA* to the N- or C-terminus of the protein of interest, as these two strategies are likely to reveal different proximal proteins (175). Our main goal was to identify novel MELK substrates, so we chose to fuse BirA* to the N-terminus of MELK, where the kinase domain is located (Fig.
4.3A). In an effort to maintain unperturbed substrate binding to MELK, we incorporated a 15- amino acid linker between BirA* and MELK. It is known that linker composition can affect the protein-protein interactions and activity of fusion proteins. To account for this possibility, we designed two different MELK BioID constructs: one bearing a flexible linker (3 x GGGGS), referred to as BirA*-FL-MELK, and one bearing a rigid linker (3 x EAAAK), referred to as
BirA*-RL-MELK (Fig. 4.3B) (176, 177).
We first verified that the BirA*-MELK fusions labeled proteins with biotin in cells.
HEK293T cells were transfected with BirA* (control), BirA*-FL-MELK, or BirA*-RL-MELK,
99 and 3h after transfection, media was supplemented with 50 μM biotin. Cells were incubated in media containing biotin for 16h before being harvested for immunoblot. BirA* and BirA*-
MELK samples were resolved by SDS-PAGE, transferred to PVDF membranes, and detected using streptavidin-HRP (Fig. 4.4). We observed high levels of auto-biotinylation of BirA*,
BirA*-FL-MELK, and BirA*-RL-MELK, as anticipated with this technique, revealing that both
BirA* (37 kDa) and the BirA*-MELK fusions (115 kDa) were of the expected size. Longer exposures revealed apparent differences in proximal protein biotinylation between the BirA* control and BirA*-MELK samples. These differences were observed around 42-60 kDa, though interestingly the biotinylation patterns of BirA*-FL-MELK and BirA*-RL-MELK were not noticeably different from one another. These results indicated that the BirA*-MELK fusion proteins labeled proteins with biotin as expected. We next proceeded to analyze the MELK
BioID samples using MS to identify specific differentially biotinylated proteins.
HEK293T cells were again transfected with control or MELK BioID constructs and exposed to biotin for 16h. Cells were collected and lysed, and biotinylated proteins were captured from cell lysates with streptavidin agarose. On-bead trypsin digestion of proteins and
MS analysis were completed as described in Experimental Procedures. A total of two biological replicates were completed. For both MELK BioID constructs, the ratio of the abundance of each protein quantified in the BirA*-MELK sample to the abundance of that protein in the BirA* control sample was calculated, and this value was averaged between the two biological replicates. Proteins with a ratio of abundance ≤ 1.5 (BirA*-MELK to BirA*) were filtered from the data, and the remaining proteins were analyzed using Ingenuity Pathway Analysis (IPA) software (QIAGEN) (178). In total, 376 and 522 proteins met these cutoffs for BirA*-FL-
MELK and BirA*-RL-MELK, respectively.
100 When we used IPA to analyze the diseases and disorders that most overlapped with the
MELK BioID data, cancer was identified as the top disease (Fig. 4.5A). This result was in agreement with previous studies that have extensively linked MELK to cancer, more frequently than any other disease (67). Next, we used IPA to score the data according to the overlap of identified proteins with canonical pathways. The top three pathways identified for both BirA*-
MELK fusions were eIF2 signaling, regulation of eIF4 and p70S6K signaling, and mTOR signaling (Fig. 4.5B). Notably, the proteins from our data that were assigned to these pathways were exclusively ribosomal proteins and translation initiation factors. These are known contaminants in BioID datasets when using N-terminal BirA* fusions, ostensibly because the functional BirA* domain is synthesized and kept in close proximity to the ribosome while the remainder of the fusion protein is synthesized (179). We therefore omitted these pathways and proteins from further analysis. There was a large degree of overlap between the remaining top canonical pathways identified in the BirA*-FL-MELK and BirA*-RL-MELK data. The two cell cycle pathways, G2/M DNA damage checkpoint regulation and control of chromosomal replication, were of particular note because of the results from chapter 3. We further analyzed the specific proteins from our BioID data that were assigned to these pathways.
The proteins found more highly biotinylated by BirA*-MELK fusions in the G2/M DNA damage checkpoint pathway were TP53 (p53), CDK1, CCNB1/CCNB2 (cyclin B1/B2), and multiple 14-3-3 proteins (Fig. 4.6A). The interaction with p53 is in agreement with previous work that showed MELK directly binds and phosphorylates p53 (83). CDK1, along with its activator cyclin B1, is considered the master regulator of mitotic entry and progression (140).
The 14-3-3 proteins, which bind phosphorylated serine and threonine residues, are known to be
101 involved in hundreds of protein-protein interactions, facilitating a multitude of processes (180–
182).
Within the chromosomal replication pathway, MCM2-7, PCNA, RPA1, and TOP2A were identified as being highly biotinylated by BirA*-MELK fusions (Fig. 4.6B). None of these proteins have previously been identified as MELK interactors. MCM2-7 comprise the minichromosome maintenance (MCM) complex, the core component of the eukaryotic replicative helicase, which is essential for the initiation and elongation of DNA replication (183–
185). PCNA, RPA1, and TOP2A are additional vital parts of the replication machinery. PCNA acts as a DNA clamp, tethering polymerases to DNA and serving as a binding site for numerous proteins, which enables the high processivity inherent to DNA replication (186). RPA1 binds to and stabilizes single-stranded DNA during normal DNA replication and the repair of DNA damage (187). TOP2A is a type 2 topoisomerase that cleaves and unwinds DNA during replication to alleviate tension associated with supercoiling (188).
The interaction with p53 was not pursued further because it has been previously investigated, and due to the anticipated difficulty with linking this interaction to one of the countless functions of p53. A putative interaction with 14-3-3 proteins and PCNA has not yet been pursued because these proteins are very highly abundant, thus increasing the likelihood of these results being false positives when using BioID. Similarly, the probability of RPA1 and
TOP2A being bona fide MELK interactors was considered low because biotinylation of these proteins was only slightly increased in the BirA*-MELK samples relative to BirA* control. The other proteins (Fig. 4.6C) that were quantified in greater abundance in the BirA*-MELK samples, namely SMC2, SUPT16H, MAPRE1, RAN, ARF5, HNRNPH1, VDAC1, and CRK, were not analyzed further at this time because of a less obvious connection to the phenotype
102 observed in chapter 3. We did investigate the putative interaction with the MCM complex, as all six subunits of this heteromeric complex were detected in greater abundance in the BirA*-
MELK samples, and there is a clear connection between faithful replication, which requires the
MCM complex, and activation of the G2/M DNA damage checkpoint. However, results from co-immunoprecipitation (co-IP) experiments were largely negative or inconclusive (data not shown). Finally, we further investigated an interaction between MELK and CDK1/cyclin B1, as the role of CDK1 at the G2/M checkpoint and in mitosis provided a clear link with the mitotic delay phenotype described in chapter 3. The likelihood of CDK1 and cyclin B1 being false positive interactors was also considered low, as both proteins are found in lower abundance relative to many of the other putative interactors identified with BioID.
Investigation of MELK-CDK1/cyclin B1 interaction
Asynchronous HEK293T cells were transfected with FLAG-MELK and harvested after
24h for IP. Mock and FLAG IPs were completed as described in Experimental Procedures.
Samples were analyzed by immunoblot for MELK, CDK1, and cyclin B1 (Fig. 4.7A). IP with
FLAG antibody enriched substantially more MELK than the mock IP. While CDK1 did not seem to co-IP with MELK, more cyclin B1 was detected in the FLAG IP sample than in the mock IP sample. These results suggested that MELK and cyclin B1 may physically interact in cells. However, a second replicate of this experiment did not detect increased cyclin B1 or
CDK1 abundance in the FLAG IP sample (data not shown).
We next tested whether MELK depletion or inhibition affected CDK1 activity in mitosis.
HEK293T cells were either transfected with shMELK42, shMELK46, or an empty vector control
42h prior to harvest, or they were treated with 1 μM dinaciclib, 1 μM flavopiridol, 0.5 μM 8a, or
103 5 μM 8a 20h prior to harvest. 10 μM 8a treatment was done 1.5h before harvest because of the toxicity of 8a at this concentration. Additionally, all conditions were synchronized in mitosis by treatment with 100 ng/mL nocodazole 20h before harvest. Samples were analyzed by immunoblot with MELK, CDK1, CDC25C, cyclin B1, p-H3 (S10), and β-actin antibodies (Fig.
4.7B). The CDC25C antibody is routinely used to measure CDK1 activity, as CDK1 phosphorylates this phosphatase during mitosis, resulting in a slower migration pattern during
SDS-PAGE separation relative to the unphosphorylated form (189). As expected, the CDK1 inhibitors dinaciclib and flavopiridol completely ablated CDK1 activity, as observed by a total loss of CDC25C phosphorylation. Transfection of cells with shMELK42 and -46 resulted in high to moderate levels of MELK depletion, respectively, but had no effect on CDK1 activity, as measured by the proportion of unphosphorylated to total CDC25C. CDK1 and cyclin B1 expression were also unaffected, as was the abundance of p-H3 (S10), a marker for mitosis.
Likewise, MELK inhibition with 0.5, 5, or 10 μM 8a had no effect on CDK1 activity or expression, cyclin B1 expression, or p-H3 (S10) abundance. Collectively, the results from this experiment suggest that MELK depletion or inhibition do not affect CDK1 activity in mitosis.
MELK increases the transcriptional activity of FoxM1
One of the more well-validated MELK substrates is FoxM1, an oncogenic transcription factor that regulates a multitude of mitotic genes (99, 190). While it has been shown previously that MELK interacts with and phosphorylates FoxM1 to regulate its transcriptional activity, the phosphorylation site(s) on FoxM1 that are regulated by MELK are unknown, as is the specific mechanism by which MELK phosphorylation increases FoxM1 activity (74).
104 First, we aimed to replicate results showing that MELK and FoxM1 co-IP. Myc-FoxM1 and FLAG-MELK were overexpressed in HEK293T cells for 24h. Cells were lysed and IPs with or without Myc antibody were completed, as described in Experimental Procedures. We observed that Myc-FoxM1 IP resulted in co-IP of FLAG-MELK, albeit with low efficiency, perhaps suggesting a weak or transient interaction (Fig. 4.8A). Next, a FoxM1 transcriptional reporter assay was completed to test the effects of MELK overexpression on FoxM1 activity.
HEK293T cells were transfected with Myc-FoxM1 and a FoxM1 transcriptional reporter construct (6 x DB), along with varied amounts of the FLAG-MELK construct. After 48h, cells were lysed, and a luciferase assay was completed to quantify the amount of FoxM1 transcriptional activity in each condition. FoxM1 activity was increased by MELK overexpression, and this effect was potentiated by greater amounts of MELK (Fig. 4.8B). Both the co-IP and transcriptional reporter assay results corroborate published work (74).
Identification of FoxM1-S481 as a putative MELK phosphorylation site
Following validation of the MELK-FoxM1 interaction, we aimed to identify specific phosphosites on FoxM1 that are regulated by MELK. We investigated this question with MS, using two separate approaches. Two replicates were completed in which either FoxM1 or
FoxM1 and MELK were overexpressed in HEK293T cells for 24h. A third replicate was completed in which overexpressed FoxM1 was IPed from HEK293T cell lysate, purified, and
FoxM1 was incubated with or without active MELK for 1h. For all replicates, samples were resolved by SDS-PAGE, proteins were labeled with Coomassie stain, FoxM1 bands were excised, and in-gel trypsin digestion and MS analysis were completed as described in
Experimental Procedures. In total, 16 phosphosites were identified by MS analysis, though the
105 abundance of each peptide could not be quantified in all replicates (Table 4.1). FoxM1-S481 was the only phosphosite that was quantified in greater abundance in the FoxM1 + MELK sample than the control sample in all three replicates, suggesting that MELK phosphorylates this residue in FoxM1.
FoxM1 is heavily post-translationally modified, including many known phosphorylation sites, to regulate its transcriptional activity, localization, and stability (191). We used the
PhosphoSitePlus® database to map the phosphorylation sites and relative phosphorylation density on FoxM1 (Fig. 4.9A). Some regions of phosphorylation are well characterized, such as the C-terminal transactivator domain (TAD), which is phosphorylated by CDKs and PLK1 to modulate FoxM1 activity at different stages of the cell cycle (191). However, there is a region of phosphorylation in the central part of FoxM1, including serine 481, that is not well characterized.
The phosphorylation of residues in this region of FoxM1 has been detected using high- throughput techniques, but has not been attributed to any specific kinase, and the effects of this phosphorylation is poorly understood. Serine 481, specifically, has been identified with a phosphate group modification by four studies (165, 192–194). The results of our phosphosite identification experiments suggest that MELK may be one of the kinases responsible for regulating the phosphorylation of serine 481 of FoxM1.
To determine whether phosphorylation of FoxM1 at serine 481 regulates its activity, we used site-directed mutagenesis to create FoxM1 mutants that contain either a non- phosphorylatable alanine, or a phosphomimetic aspartate or glutamate at the 481 position. A luciferase activity assay was conducted to measure the transcriptional activity of wild-type or
FoxM1-S481A, -D, or -E mutants, in the absence of MELK overexpression (Fig. 4.9B). The
FoxM1-S481A mutant exhibited very slightly decreased activity relative to wild-type.
106 Conversely, both of the phosphomimetic mutations resulted in slightly increased FoxM1 activity, with the FoxM1-S481D mutant exhibiting a greater increase in activity (~25%), than the -S481E mutant (~15%). Importantly, the modest increases in transcriptional activity observed with
FoxM1-S481D or -E mutants were much smaller in magnitude than the 250% increase observed with high levels of MELK overexpression (Fig. 4.8B), suggesting that MELK phosphorylates additional residues, perhaps in the same poorly characterized region of the protein, to further activate FoxM1.
Unbiased screen for compounds that synergize with MELK-8a in TNBC cells
We conducted an unbiased screen for kinase inhibitors and epigenetic regulators that synergize with 8a in TNBC cells. This screen, henceforth referred to as the synergy screen, was completed using MDA-MB-231 and MDA-MB-468 cells, two commonly used TNBC cell lines.
A schematic of the synergy screen is provided (Fig. 4.10). Briefly, MDA-MB-231 and MDA-
MB-468 cells were each seeded on 384-well cell culture plates and incubated for 24h to allow for cell adherence. Cells were then treated with either DMSO or 12, 37, 111, 333, 1000, or 3000 nM
8a, in combination with DMSO or one of 176 compounds (Table 4.2) at a concentration of 3, 10,
30, 100, 300 or 1000 nM. Cells were incubated for 96h before viability was measured using the
CellTiter-Glo® assay.
Dose-response curves were generated for 8a in combination with each compound in the synergy screen. High-throughput, objective synergy scoring methods, such as Bliss independence, were not calculated because the range of 8a concentrations used in the screen was too low, which resulted in dose-response curves with no lower plateau. Synergy tests using Bliss independence and other models were not informative due to the non-normal nature of these data,
107 and thus the curves for each compound had to be manually inspected for potential synergy. We analyzed all dose-response curves, in search of drug combinations for which cell viability was substantially lower than what would be expected with single-agent treatment at the same concentrations. For example, the top hit in the synergy screen, UNC0642, appeared to exhibit strong synergy with 8a in MDA-MB-468 cells, where 3 μM 8a alone did not decrease viability, 1
μM UNC0642 alone decreased viability to 80%, and the combination of the two resulted in 0% viability (Fig. 4.11A). These putative synergistic effects were also observed with 1 μM
UNC0642 in combination with 0.3 and 1 μM 8a, suggesting this effect was not a false positive limited to a single datapoint. Using these manual inspection criteria, the top hits from the synergy screen were identified as UNC0642, OTX015, and UNC1999 (Fig. 4.11, A-C). All of these compounds are epigenetic modulators, with known targets identified as G9a/GLP,
BRD2/3/4, and EZH1/2, respectively.
To test the results from the synergy screen using a more optimal range of 8a concentrations, we used MTS cell viability assays. Cells were seeded and incubated for 24h to allow for adherence to plates, then exposed to a combination of DMSO or 1 nM to 30 μM 8a, in combination with DMSO or 10 nM to 10 μM UNC0642, or 3 nM to 3 μM OTX015 or
UNC1999. After 72h, viability was measured as previously described. Dose-response curves were again generated to assess putative synergy (Fig. 4.12, A-C). No synergistic effects with 8a were observed with OTX015 (Fig. 4.12B) or UNC1999 (Fig. 4.12C). Similarly, no synergy was observed between 8a and UNC0642 in MDA-MB-231 cells. In MDA-MB-468 cells, we observed a slight left-shift in the 3 μM UNC0642 dose-response curve (IC50 = 1.5 μM) relative to the DMSO and 10 nM to 1 μM curves (IC50 = 3.5 μM) (Fig. 4.12A). This modest 2-fold increase in 8a potency was not observed in a second replicate of this experiment, however (data not
108 shown). Collectively, the results of the synergy screen demonstrated apparent synergistic effects when 8a was combined with UNC0642, OTX015, or UNC1999, but these results were not replicated using MTS assays, where these epigenetic regulators had no effect on 8a potency.
Discussion
The detection of DNA DSBs triggers a series of events that slow or halt cell cycle progression, which is thought to allow time for DNA damage repair. At the site of DSBs, ATM kinase phosphorylates H2A.X at S139, which acts as the site of recruitment for DSB repair machinery and initiates the DDR (195). ATM also phosphorylates multiple substrates to arrest cell cycle progression. When DSBs are detected in G2, ATM activates Chk2 by phosphorylation of T68 and other sites. Chk2 in turn phosphorylates numerous substrates including CDC25C, at
S216, which is subsequently bound by 14-3-3σ and excluded from the nucleus. This nuclear exclusion prevents CDC25C from dephosphorylating CDK1 at Y15, thereby rendering CDK1 inactive and arresting cell cycle progression at the G2/M checkpoint (151, 159).
As an extension of the results shown in chapter 3, we demonstrated that 8a treatment caused activation of Chk2 and induced phosphorylation of H2A.X (S139) (Fig. 4.2A), strongly suggesting that MELK inhibition delays mitotic entry by activating the G2/M DNA damage checkpoint. Interestingly, other groups have shown that ATM and Chk2 are activated in response to MELK knockdown or inhibition with MELK-T1 (26, 39). To comprehensively test activation of the ATM/Chk2 pathway, future studies should measure the activation of ATM (via
S1981 phosphorylation) and phosphorylation of CDC25C S216 by immunoblot. This could initially be completed in asynchronous cells, as in Fig. 4.2A, but should be extended to synchronous experiments like those shown in Figs. 3.4-3.6 to test whether MELK inhibition is directly linked to activation of the G2/M checkpoint via the ATM/Chk2 pathway.
109 It was surprising that both wild-type and RNF168 knockout U2OS cells displayed equal sensitivity to 8a (Fig. 4.2B), given our findings suggesting that 8a treatment activates the
ATM/Chk2 DDR pathway. Our hypothesis could be tested further using alternative cellular models with deficiency in DNA damage repair pathways. Also, while the observation that 8a treatment induces p-H2A.X (S139) implies that DNA damage occurred, it does not directly provide evidence of DNA lesions. Future studies should test whether MELK inhibition induces
DNA damage using comet assays or similar approaches.
Similar asynchronous (Fig. 4.2A) and synchronous (Figs. 3.4-3.6) approaches to those outlined above should also be employed to test for activation of the ATR/Chk1 pathway, which is canonically activated in response to replication stress and exposed single-stranded DNA (159,
195). The potential observation that MELK inhibition causes activation of the ATM/Chk2 pathway independent of ATR/Chk1 activation would be surprising and mechanistically informative, given that it would suggest that DNA DSBs had formed without prior replication stress or single-strand breaks. A previous study demonstrated that MELK knockdown activates the ATM/Chk2 pathway without activating Chk1, though this article was later retracted for image duplication (79). Preliminary results from our lab (data not shown), indicate that Chk1 is not activated in asynchronous HeLaS3 cells following 8a treatment, but further investigation is required for more definitive evidence.
The BioID experiments (Figs. 4.4-4.6) identified two intriguing putative MELK interactors: the MCM complex and CDK1/cyclin B1. If a putative MELK-MCM complex interaction were to positively regulate normal DNA replication, it could potentially explain how
MELK inhibition induces the DNA damage markers p-H2A.X (S139) and p-Chk2 (T68). In this instance, one would expect to observe ATR/Chk1 activation due to replication fork stalling, as
110 well as lengthened S phase, which was seemingly not observed with 8a treatment (Fig. 3.4).
Nonetheless, the MELK-MCM complex relationship should be further investigated with protein- protein interaction studies beyond the previously attempted co-IP methods. Candidate approaches include proximity ligation assays (PLA) or BRET-based assays, as well as classical phosphorylation assays to determine whether MELK can directly phosphorylate the MCM complex. Since 8a has been validated as a highly selective MELK inhibitor, cell-based, quantitative flow cytometry assays could also be utilized to determine whether MELK inhibition impairs MCM loading onto DNA or S phase progression timing (196).
An interaction between MELK and CDK1/cyclin B1 could potentially explain the observation that MELK inhibition transiently activates the G2/M checkpoint, given that CDK1 activity is the ultimate control mechanism for mitotic entry. To date, our experiments have not uncovered any mechanistic insight into this interaction; however, the possibilities are numerous.
MELK may phosphorylate CDK1 directly to regulate its activity. While our results (Fig. 4.7B) suggest this is not the case, the technical aspects of this experiment could be improved, primarily by using HeLaS3 cells instead of HEK293T, and by using double thymidine synchronization plus 6-8h release to study cells in late G2 instead of cells nocodazole-arrested in mitosis. An additional possibility is that MELK phosphorylates cyclin B1 to regulate its localization, perhaps by modulating 14-3-3σ binding, thus indirectly controlling CDK1 activity. It also cannot be ruled out that CDK1 phosphorylates MELK. Indeed, one study has demonstrated that MPF
(mitosis-promoting factor, a complex of CDK1 and cyclin B1) phosphorylates MELK to regulate its activity in mitosis (197). It should be noted that this study is one of the earliest in the MELK field, was completed solely using Xenopus as a model, and the results were never replicated or expanded upon in mammalian cells or in the context of cancer. To investigate all of these
111 possibilities, PLA, BRET, and phosphorylation assays would again be useful. Secondary experiments, supposing positive results, could include microscopy-based cyclin B1 localization studies and binding experiments (i.e. co-IP) combined with cell cycle synchronization, as previous studies were only completed in asynchronous cells.
With the exception of p53, our BioID results did not identify any other purported MELK substrates such as FoxM1, CDC25B, or eIF4B (73–75). While there are countless potential explanations for these negative results, it may be a worthwhile endeavor to repeat the BioID experiments with further technical optimization. First, HeLaS3 cells should likely be used, given that the mitotic delay phenotype has largely been characterized in this cell line. Second, stable cell lines harboring tetracycline-inducible BirA*-MELK fusions should be created and biotin exposure timing should be optimized. Third, the use of BioID2 could be investigated. This upgrade to the original BioID utilizes a BirA protein with additional mutations, which allows for much more rapid biotin labeling of proteins in cells (198, 199). These optimization steps would allow for tight temporal control of fusion protein expression and proximal protein biotinylation, thereby making it possible to combine cell cycle synchronization with BioID to identify putative
MELK interactors at specific cell cycle stages (e.g. the G2/M checkpoint). Optimization of the
BioID system could be completed in parallel with, or prior to, the aforementioned interaction validation studies to refine the list of putative MELK interactors.
Our phosphoproteomic MS results identified S481 of FoxM1 as a putative MELK phosphorylation site (Table 4.1), and we demonstrated slightly increased transcriptional activity of the phosphomimetic FoxM1-S481D and -E mutants (Fig. 4.9B). This increase in activity, however, was substantially less than what was observed with high levels of MELK overexpression (Fig. 4.8B), suggesting that MELK may phosphorylate additional residues to
112 further activate FoxM1, or that the phosphomimetics acted as poor substitutes for actual phosphorylation. To identify additional FoxM1 sites that are phosphorylated by MELK, we could repeat the approach employed for replicate 3 of our experiments, in which FoxM1 was phosphorylated by MELK in vitro. A TiO2-based phosphopeptide enrichment strategy could also be used to potentially improve FoxM1 sequence coverage. If these strategies were to reveal additional MELK phosphorylation sites, results should be validated using FoxM1 mutation studies, as before (Fig. 4.9B). Upon validation, phosphomimetic mutations of newly identified regulatory sites could be made in FoxM1-S481D and -E mutants to determine whether these double (or triple, etc.) mutants exhibit activity approaching the levels observed with high levels of MELK overexpression (Fig. 4.8B). Any additional regulatory sites that are identified should be characterized to test the effects of phosphorylation of these sites on FoxM1 localization, stability, and promoter binding. Collectively, these experiments would serve to improve our understanding of the mechanism by which MELK activates FoxM1 transcriptional activity.
The synergy screen uncovered three compounds that appeared to enhance the potency of
8a (Fig. 4.11), but attempts at validation with MTS assays did not replicate these results (Fig.
4.12). We do not have a definitive explanation for these divergent findings, but limited evidence suggests this effect may in part be due to differences in cell confluency between the synergy screen and MTS assays. One study has demonstrated that MELK knockdown or genomic deletion has virtually no effect on viability when assays are completed with near-confluent cells, but proliferation is impaired in assays completed with cells plated at low-confluency (19). We tested whether the same effect can be observed with MELK inhibition, and indeed observed that the IC50 of 8a in MDA-MB-231 cells varied up to 20-fold in crystal violet cell viability assays with differential seeding densities (data not shown).
113 We previously showed that the IC50 of 8a in MTS assays with TNBC cells is around 2
μM (Fig. 3.3), yet in the synergy screen, when cells were plated at high confluency on a 384-well plate, 3 μM 8a alone had essentially no effect on the viability of MDA-MB-231 or MDA-MB-
468 cells. These results indicate that careful consideration for cell seeding density must be taken in future experiments. The synergy screen could be repeated with optimized cell plating and 8a concentration ranges to improve the chances of identifying truly synergistic relationships with the kinase inhibitors and epigenetic regulators in this screen. However, therapeutic interest in
MELK inhibitors has significantly waned following the publication of studies suggesting that
MELK is not a cancer dependency (16, 18), so it may be wise to delay future synergy studies until the requirement for and functions of MELK in cancer are better characterized.
Experimental Procedures
Antibodies and reagents
MELK-8a hydrochloride was purchased from MedChemExpress (HY-100368A).
Nocodazole was purchased from AdooQ Bioscience (A13219). Dinaciclib was a gift from the
Gary Johnson lab (UNC). Flavopiridol was purchased from MedChemExpress (HY-10005).
UNC0642 (14604), OTX015 (15947), and UNC1999 (14621) were all purchased from Cayman
Chemical. Brilliant Blue G (B1131) was purchased from Sigma. Myc-tag agarose (3400) was purchased from Cell Signaling Technology. The following primary antibodies were also purchased from Cell Signaling Technology: CDC25C (4688, 1:1000), Chk2 (2662, 1:1000),
MELK (2274, 1:1000), p-Chk2 (T68) (2661, 1:1000), p-H2A.X (S139) (9718, 1:1000), p-H3
(S10) (9701, 1:1000). β-actin (sc-47778, 1:500) and FoxM1 (sc-502, 1:2000) antibodies and protein A/G PLUS-agarose (sc-2003) were purchased from Santa Cruz Biotechnology. CDK1
114 (MAB8878, 1:2000) was purchased from Millipore. Cyclin B1 (ab32053, 1:10000) was purchased from Abcam. FLAG antibody (MA1-91878) was purchased from Thermo.
The shRNA reagents used in these studies were purchased from Dharmacon™: shMELK42 – ID: TRCN0000001642, sequence: 5’-ACAAAGAGACATAGTTAAGAG-3’; shMELK46 – ID: TRCN0000001642, sequence: 5’-ATTTCTCTCTTATGCTGGTTC-3’. The pcDNA3 FLAG-MELK construct was a generous gift from the Kevin Dalby lab (UT-Austin).
The pcDNA3.1 mycBioID plasmid was purchased from Addgene (#35700), for generation of
BirA*-MELK constructs. The Myc-FoxM1 and FoxM1 6x-DBE luciferase reporter plasmid were generous gifts from the Mike Emanuele lab (UNC) (200). Myc-FoxM1 mutagenesis primers were as follows:
FoxM1-S481A
Forward: 5’-ACCCATCAAAGTGGAGGCCCCTCCCTTGGAAGAG-3’
Reverse: 5’-CTCTTCCAAGGGAGGGGCCTCCACTTTGATGGGT-3’
FoxM1-S481D
Forward: 5’-GACCCATCAAAGTGGAGGACCCTCCCTTGGAAGAGT-3’
Reverse: 5’-ACTCTTCCAAGGGAGGGTCCTCCACTTTGATGGGTC-3’
FoxM1-S481E
Forward: 5’-AGACCCATCAAAGTGGAGGAGCCTCCCTTGGAAGAGTGG-3’
Reverse: 5’-CCACTCTTCCAAGGGAGGCTCCTCCACTTTGATGGGTCT-3’
Cell culture and viability assays
HEK293T cells were obtained from the ATCC and were cultured in RPMI 1640 (Gibco) supplemented with 5% FBS (Millipore) and 1% antibiotic-antimycotic. Wild-type and RNF168
115 knockout U2OS cells were a generous gift from the Cyrus Vaziri lab (UNC), and were cultured in DMEM, high glucose (Gibco) supplemented with 10% FBS and 1% antibiotic-antimycotic.
MDA-MB-468 and MDA-MB-231 cells were kindly provided by Dr. Gary Johnson (UNC).
These cell lines were cultured in RPMI 1640 supplemented with 10% FBS and 1% antibiotic- antimycotic. All cells were maintained in a humidified chamber with 5% CO2.
MTS cell viability assays were completed exactly as described previously (referred to as resazurin assays) (14).
Immunoblotting
All immunoblotting was completed exactly as previously described (14).
BioID and mass spectrometry
The full-length MELK gene was amplified using PCR from the pcDNA3 FLAG-MELK plasmid, with XhoI (plus flexible or rigid linker) and BamHI overhangs. MELK was inserted into the pcDNA3.1 myc-BioID plasmid using restriction enzyme cloning to generate the BirA*-
FL-MELK and BirA*-RL-MELK constructs.
HEK293T cells on 15 cm plates were transfected with BirA* (pcDNA3.1 myc-BioID),
BirA*-FL-MELK, or BirA*-RL-MELK plasmids using polyethylenimine (Polysciences,
#24765). After 3h, medium was aspirated and replaced with complete medium supplemented with 50 μM biotin (Sigma, #B4639). Cells were returned to the humidified incubator for 16h.
Plates were washed 2 x 10 mL with PBS (Gibco), then cells were harvested on plates using 1 mL room temperature BioID lysis buffer (50 mM Tris, pH 7.4, 500 mM NaCl, 0.4% SDS, 5 mM
EDTA, 1 mM DTT, 1x protease inhibitor cocktail (PIC) (Roche)), to avoid SDS precipitation.
116 Lysate was transferred to 1.5 mL tubes and sonicated 6 x 3 sec at 35% amplitude. Samples were diluted with an equal volume of ice-cold 50 mM Tris, pH 7.4 and 2% Triton X-100 and stored on ice. Samples, now in BioID2 buffer (50 mM Tris, pH 7.4, 250 mM NaCl, 0.2% SDS, 1% Triton
X-100, 2.5 mM EDTA, 0.5 mM DTT, 0.5x PIC) were again sonicated, on ice, 3 x 10 sec.
Lysates were then clarified by centrifugation at 21,000 rcf for 30 min at 4ºC. Clarified lysate was either prepared in Laemmli sample buffer and immunoblotted with HRP-streptavidin
(Thermo, N100, 1:10000), or stored at -80ºC until ready for pulldown with streptavidin agarose.
For each sample, 60 μL of streptavidin agarose (Thermo, 20347) was pre-washed 2x with
BioID2 buffer. Lysates were applied to equilibrated streptavidin agarose, and rotated for 16h at
4ºC to capture biotinylated proteins. After overnight incubation, beads were pelleted by low- speed centrifugation (375 rcf) at 4ºC. Supernatant was aspirated and beads were subsequently washed 2 x 750 μL ice-cold BioID2 buffer, followed by 3 x 750 μL washes with TAP buffer (50 mM HEPES, pH 8.0, 150 mM NaCl, 10% glycerol, 2 mM EDTA, 0.1% NP-40). Beads were resuspended in 750 μL of 50 mM ammonium bicarbonate, pH 8.0, and transferred to siliconized low-retention microfuge tubes. Beads were washed an additional 2 x 750 μL with ammonium bicarbonate, pelleted, and supernatant aspirated. Beads were then resuspended in 100 μL ammonium bicarbonate and 1 μg of sequencing grade trypsin (Promega, V511A) was added.
Samples were incubated at 37 ºC for 16h with constant shaking at 14,000 rpm using an
Eppendorf ThermoMixer®. Following overnight trypsin digestion, an additional 1 μg of trypsin was added to each sample and samples were incubated for 1h at 37 ºC. Agarose beads were pelleted by centrifugation at 375 rcf, and supernatant containing digested peptides was transferred to new low-retention microfuge tubes. The beads were resuspended 2x with 100 μL
MS-grade water, pelleted, and supernatant transferred to microfuge tubes containing the previous
117 supernatant to collect residual peptides. Samples were extracted with ethyl acetate 4x to remove detergents, and peptides were desalted using C-18 spin columns (Pierce, 89870). Samples were subsequently analyzed by LC/MS/MS using a Q Exactive HF mass spectrometer. Data acquisition and search, quantification using MaxQuant label-free quantification, and analysis were completed as previously described for MIB/MS experiments (14).
MS Data Analysis
Data from BioID MS experiments were analyzed using Ingenuity Pathway Analysis
(IPA) software (QIAGEN) (178). For both the BirA*-FL-MELK and BirA*-RL-MELK samples, the ratio of the abundance of each protein quantified in the BirA*-MELK sample to the abundance of that protein in the BirA* control sample was calculated. This value was averaged between the two biological replicates. Proteins with a ratio of abundance ≤ 1.5 (BirA*-MELK to
BirA*) were filtered from the data, and the remaining proteins were submitted to IPA for further analysis. IPA output data were analyzed according to the diseases and disorders associated with the proteins in each dataset, and the canonical pathways to which each protein was assigned.
Ranking was done according to the significance (-log(p-value)) of their overlap with canonical pathways, as calculated by IPA.
Immunoprecipitation
HEK293T cells were transfected using polyethylenimine with FLAG-MELK or Myc-
FoxM1 as indicated, and lysed for IP after 24h with either CHAPS buffer (40 mM HEPES, pH
7.5, 120 mM NaCl, 0.5% CHAPS, 2.5 mM Na3VO4, 10 mM NaF, 1x PIC) for FLAG IP, or
NETN buffer (20 mM Tris, pH 8.0, 100 mM NaCl, 0.5% NP-40, 0.5 mM EDTA, 2.5 mM
118 Na3VO4, 10 mM NaF, 1x PIC) for Myc IP. Lysates were sonicated 3 x 5 sec on ice, clarified by high-speed centrifugation at 4ºC, and protein was quantified by Bradford assay (Bio-Rad). Equal volumes of each sample containing equal amounts of protein (0.5-1 mg) were incubated with either 20 μL pre-equilibrated Myc agarose for 3h, or FLAG Ab for 3h followed by 50 μL protein
A/G beads for 1h. For mock IPs, samples were incubated with protein A/G beads for equivalent times. All incubations were done at 4ºC with rotation. Beads were washed 2-3x with 500 μL of the appropriate lysis buffer, and pelleted by low-speed centrifugation at 4ºC. Bound proteins were eluted from beads by boiling in Laemmli sample buffer and analyzed by immunoblot.
FoxM1 transcriptional reporter assay
HEK293T cells were transfected with a FoxM1 6x-DBE luciferase reporter construct and
Myc-FoxM1, FoxM1-S481 mutants, or FLAG-MELK as indicated. After 48h, cells were collected and lysed, and a luciferase assay (Promega, #E1500) was completed in a 96-well plate according to the manufacturer’s protocol. Luminescence was measured to quantify transcriptional activity.
In-cell and in-vitro phosphorylation of FoxM1 by MELK
For replicates 1 and 2, Myc-FoxM1 (control) or Myc-FoxM1 and FLAG-MELK were overexpressed in HEK293T cells. After 24h, cells were lysed and a Myc pulldown was completed as previously described. For replicate 3, Myc-FoxM1 was overexpressed in
HEK293T cells for 24h, cells were lysed, and pulldown with Myc agarose was completed. Myc-
FoxM1 bound to beads was incubated, with shaking, for 1h at 30ºC in kinase reaction buffer (40 mM Tris, pH 7.5, 20 mM MgCl2, 0.1 mg/mL BSA, 2.5 mM Na3VO4, 10 mM NaF, 20 mM DTT,
119 150 μM ATP) with or without 100 ng of active MELK (1-340) (SignalChem Lifesciences
Corporation, M50-11G). For all replicates, agarose-bound FoxM1 was eluted from beads by boiling in sample buffer. Samples were resolved by SDS-PAGE (120 V, 1h) and gels stained with Coomassie stain (0.1% Brilliant Blue G, 50% methanol (v/v), 10% glacial acetic acid (v/v)).
Gels were destained (10% methanol, 5% glacial acetic acid) and bands corresponding to FoxM1 were excised and transferred to low-retention microfuge tubes.
In-gel trypsin digestion, MS, and data analysis
Gel slices were further destained (50% methanol, 5% glacial acetic acid) by shaking at
37ºC for 20 min to remove all bound Coomassie. Gel slices were dehydrated with 1 mL of 100% acetonitrile (ACN) for 5 min, ACN was aspirated, and gel slices were dried completely (~10 min) using a vacuum concentrator (LabConco CentriVap). The gel slices were rehydrated with
400 μL of 100 mM ammonium bicarbonate, and proteins reduced by addition of 8 μL of 500 mM
DTT. Samples were incubated at room temperature for 30 min, and DTT solution was aspirated.
A 400 μL volume of 55 mM iodoacetamide (in 50 mM ammonium bicarbonate) was added to each tube, and samples were incubated for 30 min at room temperature in the dark.
Iodoacetamide solution was aspirated, gel slices were washed by addition of 1 mL of 50 mM ammonium bicarbonate, and samples were incubated for 10 min at room temperature. The solution was aspirated and gel slices dehydrated with 1 mL ACN for 5 min, rehydrated with 400
μL of 100 mM ammonium bicarbonate for 10 mins, and dehydrated again with 2 x 800 μL ACN.
Gel slices were dried using a vacuum concentrator. A 150 μL volume of ice-cold 50 mM ammonium bicarbonate containing 3 μg trypsin was added to each sample. Tubes were incubated on ice for 1h, followed by incubation at 37ºC for 16h.
120 The trypsin solution, containing trypsinized peptide fragments, was transferred to a new low-retention microfuge tube for each sample. A 200 μL volume of 100 mM ammonium bicarbonate was added to each tube containing gel slices, and tubes were vortexed and incubated
10 min at room temperature. Ammonium bicarbonate was removed from each sample and combined with the previous trypsin solution for that sample. Peptides were extracted from gel slices with 2 x 100 μL of extraction solution (1% trifluoroacetic acid, 50% ACN), tubes incubated for 10 min at room temperature, and solutions pooled with peptide solutions from previous steps. Peptides were dried using a vacuum concentrator (~1h), then rehydrated using
200 μL of 5% ACN, 0.5% trifluoroacetic acid (in MS-grade water) and desalted using C-18 spin columns according to the manufacturer’s protocol. Dried peptide samples were stored at -80ºC until ready for MS analysis.
LC/MS/MS analysis was completed as described previously (14). The phosphoRS node in Proteome Discoverer™ software was used to assign FoxM1 phosphorylation site localization.
MS/MS spectra of all phosphorylation sites were manually validated. The data were normalized by calculating the ratio of abundance (area, in Proteome Discoverer™) of FoxM1 in the MELK sample compared to the control sample. The abundance of each quantified phosphopeptide in the control sample was then multiplied by this factor. Missing values from both samples were not imputed.
Synergy screen
Cell seeding, drug treatments, and viability measurements by CellTiter-Glo® assay were completed as described previously (171). MELK-8a was screened at 12, 37, 111, 333, 1000, and
3000 nM in combination with each of 176 compounds, screened at 3, 10, 30, 100, 300, and 1000
121 nM, from a library consisting of kinase inhibitors and epigenetic regulators. Synergy screens were completed in MDA-MB-231 and MDA-MB-468 cell lines. Dose-response curves were generated and IC50 values calculated using GraphPad Prism.
122 Tables and Figures
Table 4.1. FoxM1 phosphosites identified by MS. In replicates 1 and 2, either FoxM1 or FoxM1 and MELK were overexpressed in HEK293T cells for 24h, prior to collection. In replicate 3, overexpressed FoxM1 was immunoprecipitated from HEK293T lysate, purified, and incubated with or without active MELK for 1h. Samples were prepared for MS and analyzed as described in Experimental Procedures. The ratios of the abundance of each phosphosite quantified in the FoxM1 + MELK sample to the abundance in the control sample are shown. Missing values indicate that the peptide containing that phosphosite was not identified in either sample. N.Q. indicates that the peptide containing that phosphosite was identified, but not quantified in either sample. A – indicates that the peptide containing that phosphosite was quantified only in the control sample. A + indicates that the peptide containing that phosphosite was quantified only in the FoxM1 + MELK sample.
123 Table 4.2. List of all library compounds used in the synergy screen. The 176 kinase and epigenetic inhibitors screened in combination with 8a are displayed, with known targets indicated.
124
125
Figure 4.1. BioID schematic. BioID schematic with BirA*-MELK fusion protein. Interacting and proximal proteins that fall within the ~10 nm labeling radius (green circle) of BirA*, including MELK, will be biotinylated. Direct interactors may not be biotinylated if they are more than 10 nm from BirA*. BirA* = BirA-R118G.
126
Figure 4.2. Treatment with 8a causes increased phosphorylation of H2A.X (S139) and activation of Chk2. A, asynchronous HeLaS3 cells were treated with DMSO or 0.5, 1, or 3 μM 8a for 1, 2, 4, 8, 12, or 24h. Cells were harvested and immunoblotted with the indicated antibodies. B, Wild-type (WT) or RNF168 knockout (KO) U2OS cells were treated with DMSO or 8a at concentrations ranging from 1 nM to 30 μM in quadruplicate. Cells were exposed to 8a for 72h before viability was measured using MTS assay.
127
Figure 4.3. BirA*-MELK fusion protein design. A, domain structure of MELK. UBA = Ubiquitin-associated domain. KA1 = Kinase associated domain-1. B, two BirA*-MELK fusion proteins were designed, with either a 15-amino acid flexible linker (FL) or rigid linker (RL).
128
Figure 4.4. Biotinylation of proximal proteins differs between BirA*-MELK constructs and BirA* control. HEK293T cells were transfected with BirA*, BirA*-FL-MELK, or BirA*-RL- MELK, and 3h after transfection media was supplemented with 50 μM biotin. Following 16h incubation with biotin-supplemented media, cells were harvested for immunoblot. Biotinylation of proteins was detected using streptavidin-HRP. Short (left) and long (right) exposures are displayed. Regions of apparent differential protein biotinylation between BirA* (blue box) and BirA*-FL-MELK or BirA*-RL-MELK (green boxes) are shown.
129
Figure 4.5. Analysis of BirA*-MELK MS data using IPA. The BioID experiments and MS analysis were completed as described in Experimental Procedures. For both MELK BioID constructs, the ratio of the abundance of each protein quantified in the BirA*-MELK sample to the abundance of that protein in the BirA* control sample was calculated. This value was averaged between the two biological replicates, and proteins with a ratio of 1.5 or higher were submitted to IPA for bioinformatics analysis. A, the top 5 diseases and disorders predicted from IPA analysis of all proteins that passed cutoffs in the BirA*-FL-MELK (left) and BirA*-RL- MELK (right) datasets are displayed. B, the proteins in the BirA*-FL-MELK (top) and BirA*- RL-MELK (bottom) datasets were ranked according to the significance (-log(p-value)) of their overlap with canonical pathways, as calculated by IPA analysis. The top 7 pathways are displayed for each dataset.
130
Figure 4.6. BioID and IPA analysis identified putative MELK interactors with roles in cell cycle processes. The ratio of the abundance (BirA*MELK to BirA* control) of each protein assigned to the G2/M DNA damage checkpoint pathway (A) and chromosomal replication pathway (B) in IPA are displayed. C, additional proteins with high ratios of abundance (BirA*MELK to BirA* control) that were assigned to other pathways.
131
Figure 4.7. Investigation of a putative MELK-CDK1/cyclin B1 interaction. A, FLAG- MELK was overexpressed in HEK293T cells. Cells were lysed and equal volumes of lysate were tumbled with either protein A/G agarose beads (mock IP) or FLAG antibody-linked A/G agarose beads (FLAG IP). Equal volumes of pulldown eluate were immunoblotted with MELK, CDK1, and cyclin B1 antibodies. Images for input, mock, and FLAG IP bands were spliced next to one another due to sub-optimal loading conventions. Images for each target are from the same membrane, with identical exposures, except for cyclin B1, input sample (shorter exposure due to abundance). B, HEK293T cells were either transfected with shMELK42, shMELK46, or an empty vector 48h before collection, or treated with one of the indicated compounds 20h before collection, with the exception of 10 μM 8a (1.5h treatment pre-collection). Additionally, cells were treated with 100 ng/mL nocodazole 20h before collection. Cells were harvested and lysates were analyzed by immunoblot with the indicated antibodies. Phosphorylated CDC25C, which migrates more slowly than unphosphorylated CDC25C, was used as a marker for CDK1 activity. The amount of unphosphorylated and phosphorylated CDC25C was quantified by densitometry using Bio-Rad Image Lab™ software, and the ratio of unphosphorylated CDC25C to total CDC25C for each condition is indicated by an arrow.
132
Figure 4.8. MELK interacts with FoxM1 to increase FoxM1 transcriptional activity. A, Myc-FoxM1 and FLAG-MELK were overexpressed in HEK293T cells. Cells were lysed and equal volumes of lysate were tumbled with either protein A/G agarose beads (mock IP) or Myc antibody-linked A/G agarose beads (Myc IP). Equal volumes of pulldown eluate were immunoblotted FoxM1 and MELK antibodies. B, HEK293T cells were transfected with Myc- FoxM1 and a FoxM1 transcriptional reporter construct (6 x DB). Cells were additionally transfected with varying amounts of FLAG-MELK. After 48h, cells were lysed and lysates were incubated with luciferase assay reagent. Luminescence was measured to quantify FoxM1 transcriptional activity. Portions of this figure were produced by the Michael Emanuele lab.
133
Figure 4.9. FoxM1-S481A/D/E mutants may display differential transcriptional activity. A, FoxM1 domain structure. NRD = N-terminal repressor domain. FKH = Forkhead homology domain. TAD = Transactivator domain. Binding sites of the APC/C and phosphorylation sites of putative (MELK) and established (PLK1 and CDKs) FoxM1 kinases are indicated by pink and green arrows/lines, respectively. Blue lines above the domain structure represent approximate phosphorylation density, as collated in the PhosphoSitePlus® database (www.phosphosite.org/proteinAction.action?id=2992&showAllSites=true). B, HEK293T cells were transfected with Myc-FoxM1 WT, -S481A, -S481D, or -S481E, and a FoxM1 transcriptional reporter construct (6 x DB). After 48h, cells were lysed and lysates were incubated with luciferase assay reagent. Luminescence was measured to quantify FoxM1 transcriptional activity. This figure was produced by the Michael Emanuele lab.
134
Figure 4.10. Schematic of the screen for compounds that synergize with MELK-8a. MDA- MB-231 and MDA-MB-468 cells were seeded on 21 384-well plates each. 24h after seeding, cells were treated with either DMSO or one of six concentrations of 8a, in combination with DMSO or one of six concentrations of each of 176 library compounds. Cell viability was assessed 96h after compound treatment using the CellTiter-Glo® assay (Promega).
135
Figure 4.11. Synergy screen top hits were identified as UNC0642, OTX015, and UNC1999. Dose-response curves were generated for each compound in the synergy screen, and top hits were identified by manual inspection of all dose-response curves. The dose-response curves for the top three hits, UNC0642 (A), OTX015 (B), and UNC1999 (C), are displayed for both MDA- MB-468 (left) and MDA-MB-231 (right) cells.
136
Figure 4.12. Secondary cell viability assays indicate no synergy between 8a and UNC0642, OTX015, or UNC1999. Cells were exposed to a combination of 1, 3, 10, 30, 100, 300, 1000, 3000, 10000, or 30000 nM 8a, or DMSO, and the indicated library compound at 3, 10, 30, 100, 300, 1000, or 3000 nM, or DMSO, for 72h. Viability was measured by MTS assay as described in Experimental Procedures. Dose-response curves for UNC0642 (A), OTX015 (B), and UNC1999 (C) are shown, for both MDA-MB-468 (left) and MDA-MB-231 (right) cells.
137
CHAPTER 5: CONCLUSIONS AND FUTURE DIRECTIONS
One of the main strategies used to generate novel cancer treatments is to identify molecular targets that cancer cells rely on for growth and survival, moreso than non-neoplastic cells, and to develop compounds to inhibit these targets. While this is somewhat of a simplified model, the development of many targeted cancer therapies has taken this general approach. The
MELK controversy, on the other hand, demonstrates that the investigation of putative drug targets does not always neatly fit this reductionist model.
MELK was identified as an attractive target in multiple cancers in 2005, in a study that showed an upregulation of MELK in cancer and slowed proliferation following MELK silencing
(5). Over the subsequent decade, many studies focused on repeating and expanding these methods to identify MELK as a target in additional cancer types, without significantly advancing our understanding of the function of MELK. As the list of cancers for which MELK seemed to be a viable target grew, inhibitor development efforts were launched with haste. The first inhibitor to be developed and published (OTSSP167) was touted as highly MELK-selective (10).
Unfortunately, by the time it was first revealed that OTS was actually very poorly selective (13), the field had already adopted it as the primary small-molecule tool compound to use in functional studies. Despite the fact that dozens of MELK inhibitors have since been published with far superior selectivity profiles, the field has been slow to adopt these alternatives, which has further obfuscated our understanding of the functions of MELK in cancer. More recently,
CRISPR-based techniques were used to allege that MELK is dispensable for cancer growth, and
138 thus is not a viable cancer target. Given that there is such a lack of understanding of the fundamental functions of MELK, I would argue that the question we should be focused on at the moment is not ‘is MELK a requirement in cancer?’ but rather ‘what does MELK do in cancer?’
By working towards an answer to the latter question, the field should be able to approach an answer to the former one.
In chapter 1, I analyze the evidence on both sides of the controversy concerning the requirement for MELK in cancer. I also discuss MELK inhibitors, processes thought to be regulated by MELK, mechanisms of regulation of MELK expression and activity, and reported
MELK substrates. While our understanding of the latter three topics is certainly not comprehensive, considerable strides have been made since the first CRISPR publications.
In chapter 2, I show that treatment of pancreatic cancer (PC) cells with gemcitabine induces MELK expression, and that MELK knockdown slows PC cell proliferation. I also detail our efforts to identify novel MELK inhibitors. Multiple lead compounds were identified that impair PC cell viability, but the top hit (UNC2025) had only moderate selectivity for MELK and failed to sensitize cells to gemcitabine. While the CRISPR studies cast doubt on whether MELK is truly a viable target in pancreatic and other cancers, the observation that gemcitabine upregulates MELK expression is compelling. MELK upregulation has also been observed with radiation therapy and treatment with other chemotherapeutics, including 5-fluorouracil. Future studies should further investigate the mechanism underlying this radio- and chemotherapy- induced upregulation of MELK, as it could serve to improve our understanding of MELK function in the DNA damage response.
In chapter 3, I used the competition MIB/MS technology to evaluate the selectivity of the inhibitors OTS, HTH-01-091 (HTH), and NVS-MELK8a (8a) for MELK. Results showed that
139 OTS is very poorly selective, in agreement with other studies, HTH does not inhibit MELK in cells when used at 1 μM, and 8a is highly selective for MELK at 1 and 3 μM. Impaired TNBC cell viability was observed in the 1 to 3 μM 8a range, indicating that antiproliferative effects were likely a result of MELK inhibition. Interestingly, with treatment times up to 24h, an induction of apoptosis was not observed, but rather cell cycle progression was perturbed. I further investigated this cell cycle defect, and determined that HeLaS3 cells treated with 8a exhibited a delay in mitotic entry, associated with delayed activation of Aurora A, Aurora B, and
CDK1. These mitotic kinases were ultimately activated, approximately 2h later than in untreated cells, which facilitated delayed mitotic entry and progression through mitosis. A similar mitotic delay phenotype was observed in U2OS cells, along with a less pronounced impairment of mitotic progression. These results suggest that the G2/M DNA damage checkpoint is transiently activated in response to MELK inhibition with 8a.
In chapter 4, I probed for a deeper mechanistic understanding of this mitotic delay phenotype. I found that 8a treatment caused a dose-dependent induction of the DNA DSB marker p-H2AX (S139) and activation of Chk2, as measured by phosphorylation of T68. Using
BioID to investigate MELK protein-protein interactions, a number of putative interactors that participate in the chromosomal replication (MCM complex, PCNA) and G2/M DNA damage checkpoint (CDK1, cyclin B1, 14-3-3 proteins, p53) processes were identified. Attempts at validation of these interactions have been inconclusive to date. The MELK-FoxM1 relationship was interrogated with the goal of determining how MELK phosphorylation regulates FoxM1 transcriptional activity. I observed a direct interaction between MELK and FoxM1, along with increased FoxM1 activity, in agreement with other studies. Furthermore, a putative FoxM1 phosphorylation site regulated by MELK was uncovered (S481), and mutation of this site to
140 phosphomimetic or non-phosphorylatable residues may modulate FoxM1 activity. Finally, I used a library of kinase inhibitors and epigenetic regulators to complete a synergy screen with the goal of uncovering molecules that synergize with 8a in TNBC cells. Three compounds exhibited apparent synergy with 8a in the screen (UNC0642, G9a/GLP inhibitor; OTX015,
BRD2/3/4 inhibitor; UNC1999, EZH1/2 inhibitor), but subsequent attempts at validation using viability experiments did not replicate these results.
Future studies should focus on elucidating the mechanism underlying the cell cycle phenotypes observed in chapter 3. With the added information from chapter 4 that the Chk2
DDR pathway is activated by MELK inhibition, along with an induction of p-H2AX (S139), we can begin to form more detailed hypotheses about the role of MELK in the cell cycle. A putative interaction with the MCM complex, identified with BioID, suggests that MELK may regulate
DNA replication. Thus, inhibition of MELK may impair faithful replication such that DNA
DSBs form and the Chk2 DNA damage pathway is activated. Experiments to assess the effects of MELK inhibition on MCM complex loading onto DNA and the timing of S phase progression would begin to test this hypothesis. The effects of MELK inhibition on the G2/M checkpoint could potentially be explained by a direct interaction with CDK1 or cyclin B1, thereby modulating CDK1 activity, or by regulation of FoxM1 activity, thereby interfering with the transcription of genes required for mitotic entry. Both hypotheses should be investigated, though a direct interaction with the CDK1/cyclin B1 complex would be more novel, and could perhaps have broader implications for the cell cycle field’s understanding of mitotic entry regulation.
Basic protein-protein interaction studies should first be used to determine whether a direct
MELK-CDK1/cyclin B1 interaction occurs in cells, before attempting to elucidate the specific mechanistic effects of this interaction in greater detail. The observed effects of 8a on mitotic
141 progression match what others in the field have reported, and could also be mediated by MELK-
CDK1/cyclin B1 or -FoxM1 interactions. Precise cell cycle synchronization techniques should be used with similar protein-protein interaction studies to probe these relationships in mitosis.
Collectively, the experiments presented in this dissertation advance our functional understanding of MELK in the cell cycle and improve upon the tools available to investigate this enigmatic kinase. Our results, and the results of others, demonstrate definitively that the inhibitor OTSSP167 is poorly selective for MELK, and thus should not be used in future functional studies. Instead, NVS-MELK8a should be used in all studies of MELK function, as our work has validated the high selectivity of this compound in cells. The use of 8a in functional
MELK studies should greatly improve the rigor and reproducibility of future research in this field. Our collective poor understanding of MELK substrates represents a major gap in the field that must be addressed. Specifically, the validation of a reliable measure of MELK activity in cells is badly needed. The measurement of MELK inhibitor-induced MELK degradation as a measure of cellular activity has not been well-validated, and the underlying mechanism is unclear; thus, this practice has the potential to produce confounding or misleading results.
Beyond this, the field must place an emphasis on characterizing the functions of MELK in cancer. Much evidence, including some presented in this dissertation, point to an important role for MELK in cell cycle progression, particularly in G2 and M phases. By elucidating the functions of MELK in these and other processes, the field will be able to move towards an explanation for the disparate cell viability effects observed with RNAi and CRISPR, which ignited the present controversy concerning the requirement for MELK in cancer.
142 REFERENCES
1. Heyer, B. S., Warsowe, J., Solter, D., Knowles, B. B., and Ackerman, S. L. (1997) New member of the Snf1/AMPK kinase family, Melk, is expressed in the mouse egg and preimplantation embryo. Mol. Reprod. Dev. 47, 148–156
2. Gil, M., Yang, Y., Lee, Y., Choi, I., and Ha, H. (1997) Cloning and expression of a cDNA encoding a novel protein serine/threonine kinase predominantly expressed in hematopoietic cells. Gene. 195, 295–301
3. Blot, J., Chartrain, I., Roghi, C., Philippe, M., and Tassan, J. P. (2002) Cell cycle regulation of pEg3, a new Xenopus protein kinase of the KIN1/PAR-1/MARK family. Dev. Biol. 241, 327–338
4. Li, L., Keverne, E. B., Aparicio, S. A., Ishino, F., Barton, S. C., and Surani, M. A. (1999) Regulation of maternal behavior and offspring growth by paternally expressed Peg3. Science (80-. ). 284, 330–333
5. Gray, D., Jubb, A. M., Hogue, D., Dowd, P., Kljavin, N., Yi, S., Bai, W., Frantz, G., Zhang, Z., Koeppen, H., de Sauvage, F. J., and Davis, D. P. (2005) Maternal embryonic leucine zipper kinase/murine protein serine-threonine kinase 38 is a promising therapeutic target for multiple cancers. Cancer Res. 65, 9751–61
6. Nakano, I., Masterman-Smith, M., Saigusa, K., Paucar, A. A., Horvath, S., Shoemaker, L., Watanabe, M., Negro, A., Bajpai, R., Howes, A., Lelievre, V., Waschek, J. A., Lazareff, J. A., Freije, W. A., Liau, L. M., Gilbertson, R. J., Cloughesy, T. F., Geschwind, D. H., Nelson, S. F., Mischel, P. S., Terskikh, A. V, and Kornblum, H. I. (2008) Maternal embryonic leucine zipper kinase is a key regulator of the proliferation of malignant brain tumors, including brain tumor stem cells. J. Neurosci. Res. 86, 48–60
7. Lin, M.-L., Park, J.-H., Nishidate, T., Nakamura, Y., and Katagiri, T. (2007) Involvement of maternal embryonic leucine zipper kinase (MELK) in mammary carcinogenesis through interaction with Bcl-G, a pro-apoptotic member of the Bcl-2 family. Breast cancer Res. 9, R17
8. Kuner, R., Fälth, M., Chui Pressinotti, N., Brase, J. C., Balaguer Puig, S., Metzger, J., Gade, S., Schäfer, G., Bartsch, G., Steiner, E., Klocker, H., and Sültmann, H. (2013) The maternal embryonic leucine zipper kinase (MELK) is upregulated in high-grade prostate cancer. J Mol Med. 91, 237–248
9. Du, T., Qu, Y., Li, J., Li, H., Su, L., Zhou, Q., Yan, M., Li, C., Zhu, Z., and Liu, B. (2014) Maternal embryonic leucine zipper kinase enhances gastric cancer progression via the FAK/Paxillin pathway. Mol. Cancer. 13, 100
10. Chung, S., Suzuki, H., Miyamoto, T., Takamatsu, N., Tatsuguchi, A., Ueda, K., Kijima, K., Nakamura, Y., and Matsuo, Y. (2012) Development of an orally-administrative MELK-targeting inhibitor that suppresses the growth of various types of human cancer. Oncotarget. 3, 1629–1640
143 11. Alachkar, H., Mutonga, M. B., Metzeler, K. H., Fulton, N., Malnassy, G., Herold, T., Spiekermann, K., Bohlander, S. K., Hiddemann, W., Matsuo, Y., Stock, W., and Nakamura, Y. (2014) Preclinical efficacy of maternal embryonic leucine-zipper kinase (MELK) inhibition in acute myeloid leukemia. Oncotarget. 5, 12371–12382
12. Inoue, H., Kato, T., Olugbile, S., Tamura, K., Chung, S., Miyamoto, T., Matsuo, Y., Salgia, R., Nakamura, Y., and Park, J.-H. (2016) Effective growth-suppressive activity of maternal embryonic leucine-zipper kinase (MELK) inhibitor against small cell lung cancer. Oncotarget. 7, 13621–13633
13. Klaeger, S., Heinzlmeir, S., Wilhelm, M., Polzer, H., Vick, B., Koenig, P.-A., Reinecke, M., Ruprecht, B., Petzoldt, S., Meng, C., Zecha, J., Reiter, K., Qiao, H., Helm, D., Koch, H., Schoof, M., Canevari, G., Casale, E., Depaolini, S. R., Feuchtinger, A., Wu, Z., Schmidt, T., Rueckert, L., Becker, W., Huenges, J., Garz, A.-K., Gohlke, B.-O., Zolg, D. P., Kayser, G., Vooder, T., Preissner, R., Hahne, H., Tõnisson, N., Kramer, K., Götze, K., Bassermann, F., Schlegl, J., Ehrlich, H.-C., Aiche, S., Walch, A., Greif, P. A., Schneider, S., Felder, E. R., Ruland, J., Médard, G., Jeremias, I., Spiekermann, K., and Kuster, B. (2017) The target landscape of clinical kinase drugs. Science (80-. ). 358, eaan4368
14. McDonald, I. M., Grant, G. D., East, M. P., Gilbert, T. S., Wilkerson, E. M., Goldfarb, D., Beri, J., Herring, L. E., Vaziri, C., Cook, J. G., Emanuele, M. J., and Graves, L. M. (2020) Mass spectrometry-based selectivity profiling identifies a highly selective inhibitor of the kinase MELK that delays mitotic entry in cancer cells. J. Biol. Chem. 295, 2359–2374
15. Ji, W., Arnst, C., Tipton, A. R., Bekier II, M. E., Taylor, W. R., Yen, T. J., and Liu, S.-T. (2016) OTSSP167 abrogates mitotic checkpoint through inhibiting multiple mitotic kinases. PLoS One. 11, e0153518
16. Huang, H.-T., Seo, H.-S., Zhang, T., Wang, Y., Jiang, B., Li, Q., Buckley, D. L., Nabet, B., Roberts, J. M., Paulk, J., Dastjerdi, S., Winter, G. E., McLauchlan, H., Moran, J., Bradner, J. E., Eck, M. J., Dhe-Paganon, S., Zhao, J. J., and Gray, N. S. (2017) MELK is not necessary for the proliferation of basal-like breast cancer cells. Elife. 6, e26693
17. Wang, Y., Lee, Y.-M., Baitsch, L., Huang, A., Xiang, Y., Tong, H., Lako, A., Von, T., Choi, C., Lim, E., Min, J., Li, L., Stegmeier, F., Schlegel, R., Eck, M. J., Gray, N. S., Mitchison, T. J., and Zhao, J. J. (2014) MELK is an oncogenic kinase essential for mitotic progression in basal-like breast cancer cells. Elife. 3, e01763
18. Lin, A., Giuliano, C. J., Sayles, N. M., and Sheltzer, J. M. (2017) CRISPR/Cas9 mutagenesis invalidates a putative cancer dependency targeted in on-going clinical trials. Elife. 6, e24179
19. Wang, Y., Li, B. B., Li, J., Roberts, T. M., and Zhao, J. J. (2018) A conditional dependency on MELK for the proliferation of triple-negative breast cancer cells. iScience. 9, 149–160
20. Pickard, M. R., Green, A. R., Ellis, I. O., Caldas, C., Hedge, V. L., Mourtada-Maarabouni, M., and Williams, G. T. (2009) Dysregulated expression of Fau and MELK is associated
144 with poor prognosis in breast cancer. Breast Cancer Res. 10.1186/bcr2350
21. Speers, C., Zhao, S. G., Kothari, V., Santola, A., Liu, M., Wilder-Romans, K., Evans, J., Batra, N., Bartelink, H., Hayes, D. F., Lawrence, T. S., Brown, P. H., Pierce, L. J., and Feng, F. Y. (2016) Maternal embryonic leucine zipper kinase (MELK) as a novel mediator and biomarker of radioresistance in human breast cancer. Clin Cancer Res. 22, 5864–75
22. Minata, M., Gu, C., Joshi, K., Nakano-Okuno, M., Hong, C., Nguyen, C.-H., Kornblum, H. I., Molla, A., and Nakano, I. (2014) Multi-kinase inhibitor C1 triggers mitotic catastrophe of glioma stem cells mainly through MELK kinase inhibition. PLoS One. 9, e92546
23. Marie, S. K. N., Okamoto, O. K., Uno, M., Hasegawa, A. P. G., Oba-Shinjo, S. M., Cohen, T., Camargo, A. A., Kosoy, A., Carlotti, C. G., Toledo, S., Moreira-Filho, C. A., Zago, M. A., Simpson, A. J., and Caballero, O. L. (2008) Maternal embryonic leucine zipper kinase transcript abundance correlates with malignancy grade in human astrocytomas. Int. J. Cancer. 122, 807–815
24. Guan, S., Lu, J., Zhao, Y., Yu, Y., Li, H., Chen, Z., Shi, Z., Liang, H., Wang, M., Guo, K., Chen, X., Sun, W., Bieerkehazhi, S., Xu, X., Sun, S., Agarwal, S., Yang, J., Guan, S., Lu, J., Zhao, Y., Yu, Y., Li, H., Chen, Z., Shi, Z., Liang, H., Wang, M., Guo, K., Chen, X., Sun, W., Bieerkehazhi, S., Xu, X., Sun, S., Agarwal, S., and Yang, J. (2018) MELK is a novel therapeutic target in high-risk neuroblastoma. Oncotarget. 9, 2591–2602
25. Xia, H., Kong, S. N., Chen, J., Shi, M., Sekar, K., Seshachalam, V. P., Rajasekaran, M., Goh, B. K. P., Ooi, L. L., and Hui, K. M. (2016) MELK is an oncogenic kinase essential for early hepatocellular carcinoma recurrence. Cancer Lett. 383, 85–93
26. Chen, S., Zhou, Q., Guo, Z., Wang, Y., Wang, L., Liu, X., Lu, M., Ju, L., Xiao, Y., and Wang, X. (2020) Inhibition of MELK produces potential anti‐tumour effects in bladder cancer by inducing G1/S cell cycle arrest via the ATM/CHK2/p53 pathway. J. Cell. Mol. Med. 24, 1804–1821
27. Xu, Q., Ge, Q., Zhou, Y., Yang, B., Yang, Q., Jiang, S., Jiang, R., Ai, Z., Zhang, Z., and Teng, Y. (2020) MELK promotes endometrial carcinoma progression via activating mTOR signaling pathway. EBioMedicine. 51, 102609
28. Chlenski, A., Park, C., Dobratic, M., Salwen, H. R., Budke, B., Park, J.-H., Miller, R., Applebaum, M. A., Wilkinson, E., Nakamura, Y., Connell, P. P., and Cohn, S. L. (2019) Maternal embryonic leucine zipper kinase (MELK), a potential therapeutic target for neuroblastoma. Mol Cancer Ther. 18, 507–516
29. Agarwal, R., Narayan, J., Bhattacharyya, A., Saraswat, M., and Tomar, A. K. (2017) Gene expression profiling, pathway analysis and subtype classification reveal molecular heterogeneity in hepatocellular carcinoma and suggest subtype specific therapeutic targets. Cancer Genet. 216–217, 37–51
30. Sun, C., Cheng, X., Wang, C., Wang, X., Xia, B., and Zhang, Y. (2019) Gene expression
145 profiles analysis identifies a novel two-gene signature to predict overall survival in diffuse large B-cell lymphoma. Biosci. Rep. 10.1042/BSR20181293
31. Deng, J. L., Xu, Y. H., and Wang, G. (2019) Identification of potential crucial genes and key pathways in breast cancer using bioinformatic analysis. Front. Genet. 10.3389/fgene.2019.00695
32. Zou, Y. F., Meng, L. B., He, Z. K., Hu, C. H., Shan, M. J., Wang, D. Y., and Yu, X. (2019) Screening and authentication of molecular markers in malignant glioblastoma based on gene expression profiles. Oncol. Lett. 18, 4593–4604
33. Kim, S. H., Joshi, K., Ezhilarasan, R., Myers, T. R., Siu, J., Gu, C., Nakano-Okuno, M., Taylor, D., Minata, M., Sulman, E. P., Lee, J., Bhat, K. P. L., Salcini, A. E., and Nakano, I. (2015) EZH2 protects glioma stem cells from radiation-induced cell death in a MELK/FOXM1-dependent manner. Stem Cell Reports. 4, 226–238
34. Choi, S., and Ku, J. L. (2011) Resistance of colorectal cancer cells to radiation and 5-FU is associated with MELK expression. Biochem. Biophys. Res. Commun. 412, 207–213
35. Zhang, Y., Zhou, X., Li, Y., Xu, Y., Lu, K., Li, P., and Wang, X. (2018) Inhibition of maternal embryonic leucine zipper kinase with OTSSP167 displays potent anti-leukemic effects in chronic lymphocytic leukemia. Oncogene. 37, 5520–5533
36. Hebbard, L. W., Maurer, J., Miller, A., Lesperance, J., Hassell, J., Oshima, R. G., and Terskikh, A. V. (2010) Maternal embryonic leucine zipper kinase is upregulated and required in mammary tumor-initiating cells in vivo. Cancer Res. 70, 8863–8873
37. Janostiak, R., Rauniyar, N., Lam, T. K. T., Ou, J., Zhu, L. J., Green, M. R., and Wajapeyee, N. (2017) MELK promotes melanoma growth by stimulating the NF-κB pathway. Cell Rep. 21, 2829–2841
38. Touré, B. B., Giraldes, J., Smith, T., Sprague, E. R., Wang, Y., Mathieu, S., Chen, Z., Mishina, Y., Feng, Y., Yan-Neale, Y., Shakya, S., Chen, D., Meyer, M., Puleo, D., Brazell, J. T., Straub, C., Sage, D., Wright, K., Yuan, Y., Chen, X., Duca, J., Kim, S., Tian, L., Martin, E., Hurov, K., and Shao, W. (2016) Toward the validation of maternal embryonic leucine zipper kinase: Discovery, optimization of highly potent and selective inhibitors, and preliminary biology insight. J. Med. Chem. 59, 4711–4723
39. Beke, L., Kig, C., Linders, J. T. M., Boens, S., Boeckx, A., van Heerde, E., Parade, M., De Bondt, A., Van den Wyngaert, I., Bashir, T., Ogata, S., Meerpoel, L., Van Eynde, A., Johnson, C. N., Beullens, M., Brehmer, D., and Bollen, M. (2015) MELK-T1, a small- molecule inhibitor of protein kinase MELK, decreases DNA-damage tolerance in proliferating cancer cells. Biosci. Rep. 35, e00267
40. Edupuganti, R., Taliaferro, J. M., Wang, Q., Xie, X., Cho, E. J., Vidhu, F., Ren, P., Anslyn, E. V., Bartholomeusz, C., and Dalby, K. N. (2017) Discovery of a potent inhibitor of MELK that inhibits expression of the anti-apoptotic protein Mcl-1 and TNBC cell growth. Bioorg. Med. Chem. 25, 2609–2616
146 41. Li, S., Li, Z., Guo, T., Xing, X.-F., Cheng, X., Du, H., Wen, X.-Z., and Ji, J.-F. (2016) Maternal embryonic leucine zipper kinase serves as a poor prognosis marker and therapeutic target in gastric cancer. Oncotarget. 7, 6266–80
42. Stefka, A. T., Park, J.-H., Matsuo, Y., Chung, S., Nakamura, Y., Jakubowiak, A. J., and Rosebeck, S. (2016) Anti-myeloma activity of MELK inhibitor OTS167: effects on drug- resistant myeloma cells and putative myeloma stem cell replenishment of malignant plasma cells. Blood Cancer J. 6, e460–e460
43. Bolomsky, A., Heusschen, R., Schlangen, K., Stangelberger, K., Muller, J., Schreiner, W., Zojer, N., Caers, J., and Ludwig, H. (2018) Maternal embryonic leucine zipper kinase is a novel target for proliferation-associated high-risk myeloma. Haematologica. 103, 325– 335
44. Giuliano, C. J., Lin, A., Smith, J. C., Palladino, A. C., and Sheltzer, J. M. (2018) MELK expression correlates with tumor mitotic activity but is not required for cancer growth. Elife. 7, e32838
45. Jackson, A. L., Bartz, S. R., Schelter, J., Kobayashi, S. V., Burchard, J., Mao, M., Li, B., Cavet, G., and Linsley, P. S. (2003) Expression profiling reveals off-target gene regulation by RNAi. Nat. Biotechnol. 21, 635–637
46. Rodriguez-Rodriguez, J. A., Lewis, C., McKinley, K. L., Sikirzhytski, V., Corona, J., Maciejowski, J., Khodjakov, A., Cheeseman, I. M., and Jallepalli, P. V. (2018) Distinct Roles of RZZ and Bub1-KNL1 in Mitotic Checkpoint Signaling and Kinetochore Expansion. Curr. Biol. 28, 3422-3429.e5
47. Zhang, G., Kruse, T., Guasch Boldú, C., Garvanska, D. H., Coscia, F., Mann, M., Barisic, M., and Nilsson, J. (2019) Efficient mitotic checkpoint signaling depends on integrated activities of Bub1 and the RZZ complex. EMBO J. 38, 1–18
48. Meraldi, P. (2019) Bub1 - the zombie protein that CRISPR cannot kill. EMBO J. 38, 1–2
49. Badouel, C., Chartrain, I., Blot, J., and Tassan, J.-P. (2010) Maternal embryonic leucine zipper kinase is stabilized in mitosis by phosphorylation and is partially degraded upon mitotic exit. Exp. Cell Res. 316, 2166–2173
50. Grant, G. D., Brooks, L., Zhang, X., Mahoneya, J. M., Martyanov, V., Wood, T. A., Sherlock, G., Cheng, C., and Whitfield, M. L. (2013) Identification of cell cycle-regulated genes periodically expressed in U2OS cells and their regulation by FOXM1 and E2F transcription factors. Mol. Biol. Cell. 24, 3634–3650
51. Komor, A. C., Badran, A. H., and Liu, D. R. (2017) CRISPR-based technologies for the manipulation of eukaryotic genomes. Cell. 168, 20–36
52. Tsai, S. Q., and Joung, J. K. (2016) Defining and improving the genome-wide specificities of CRISPR-Cas9 nucleases. Nat. Rev. Genet. 17, 300–312
147 53. Seok, H., Lee, H., Jang, E. S., and Chi, S. W. (2018) Evaluation and control of miRNA- like off-target repression for RNA interference. Cell. Mol. Life Sci. 75, 797–814
54. Wittrup, A., and Lieberman, J. (2015) Knocking down disease: A progress report on siRNA therapeutics. Nat. Rev. Genet. 16, 543–552
55. Cong, L., Ran, F. A., Cox, D., Lin, S., Barretto, R., Habib, N., Hsu, P. D., Wu, X., Jiang, W., Marraffini, L. A., and Zhang, F. (2013) Multiplex genome engineering using CRISPR/Cas systems. Science (80-. ). 339, 819–823
56. Simon, M., Mesmar, F., Helguero, L., and Williams, C. (2017) Genome-wide effects of MELK-inhibitor in triple-negative breast cancer cells indicate context-dependent response with p53 as a key determinant. PLoS One. 12, 1–22
57. Maes, A., Maes, K., Vlummens, P., De Raeve, H., Devin, J., Szablewski, V., De Veirman, K., Menu, E., Moreaux, J., Vanderkerken, K., and De Bruyne, E. (2019) Maternal embryonic leucine zipper kinase is a novel target for diffuse large B cell lymphoma and mantle cell lymphoma. Blood Cancer J. 10.1038/s41408-019-0249-x
58. Meel, M. H., Guillén Navarro, M., de Gooijer, M. C., Metselaar, D. S., Waranecki, P., Breur, M., Lagerweij, T., Wedekind, L. E., Koster, J., van de Wetering, M. D., Schouten- van Meeteren, N., Aronica, E., van Tellingen, O., Bugiani, M., Phoenix, T. N., Kaspers, G. J. L., and Hulleman, E. (2020) MEK/MELK inhibition and blood-brain barrier- deficiencies in atypical teratoid/rhabdoid tumors. Neuro. Oncol. 22, 58–69
59. Canevari, G., Re Depaolini, S., Cucchi, U., Bertrand, J. A., Casale, E., Perrera, C., Forte, B., Carpinelli, P., and Felder, E. R. (2013) Structural insight into maternal embryonic leucine zipper kinase (MELK) conformation and inhibition toward structure-based drug design. Biochemistry. 52, 6380–7
60. Hoang, T. M.-N., Favier, B., Valette, A., Barette, C., Nguyen, C. H., Lafanechère, L., Grierson, D. S., Dimitrov, S., and Molla, A. (2009) Benzo[e]pyridoindoles, novel inhibitors of the aurora kinases. Cell cycle. 8, 765–72
61. Johnson, C. N., Berdini, V., Beke, L., Bonnet, P., Brehmer, D., Coyle, J. E., Day, P. J., Frederickson, M., Freyne, E. J. E., Gilissen, R. A. H. J., Hamlett, C. C. F., Howard, S., Meerpoel, L., McMenamin, R., Patel, S., Rees, D. C., Sharff, A., Sommen, F., Wu, T., and Linders, J. T. M. (2015) Fragment-Based Discovery of Type I Inhibitors of Maternal Embryonic Leucine Zipper Kinase. ACS Med. Chem. Lett. 6, 25–30
62. Johnson, C. N., Adelinet, C., Berdini, V., Beke, L., Bonnet, P., Brehmer, D., Calo, F., Coyle, J. E., Day, P. J., Frederickson, M., Freyne, E. J. E., Gilissen, R. A. H. J., Hamlett, C. C. F., Howard, S., Meerpoel, L., Mevellec, L., McMenamin, R., Pasquier, E., Patel, S., Rees, D. C., and Linders, J. T. M. (2015) Structure-based design of type II inhibitors applied to maternal embryonic leucine zipper kinase. ACS Med. Chem. Lett. 6, 31–36
63. Chen, X., Giraldes, J., Sprague, E. R., Shakya, S., Chen, Z., Wang, Y., Joud, C., Mathieu, S., Chen, C. H.-T., Straub, C., Duca, J., Hurov, K., Yuan, Y., Shao, W., and Touré, B. B.
148 (2017) “Addition” and “subtraction”: Selectivity design for type II maternal embryonic leucine zipper kinase inhibitors. J. Med. Chem. 60, 2155–2161
64. Zhou, S., Li, G. B., Luo, L., Zhong, L., Chen, K., Li, H., Jiang, X. J., Fu, Q., Long, X., and Bao, J. K. (2018) Structure-based discovery of new maternal embryonic leucine zipper kinase inhibitors. Org. Biomol. Chem. 16, 1489–1495
65. Boutard, N., Sabiniarz, A., Czerwińska, K., Jarosz, M., Cierpich, A., Kolasińska, E., Wiklik, K., Gluza, K., Commandeur, C., Buda, A., Stasiowska, A., Bobowska, A., Galek, M., Fabritius, C., Bugaj, M., Palacz, E., Mazan, A., Zarębski, A., Krawczyńska, K., Żurawska, M., Zawadzki, P., Milik, M., Węgrzyn, P., Dobrzańska, M., Brzózka, K., and Kowalczyk, P. (2019) 5-Keto-3-cyano-2 , 4-diaminothiophenes as selective maternal embryonic leucine zipper kinase inhibitors. Bioorg. Med. Chem. Lett. 29, 607–613
66. Wang, R., Chen, Y., Yang, B., Yu, S., Zhao, X., Zhang, C., Hao, C., Zhao, D., and Cheng, M. (2020) Design, synthesis, biological evaluation and molecular modeling of novel 1H- pyrrolo[2,3-b]pyridine derivatives as potential anti-tumor agents. Bioorg. Chem. 10.1016/j.bioorg.2019.103474
67. Pitner, M. K., Taliaferro, J. M., Dalby, K. N., and Bartholomeusz, C. (2017) MELK: a potential novel therapeutic target for TNBC and other aggressive malignancies. Expert Opin. Ther. Targets. 21, 849–859
68. Liu, H., Sun, Y., Qi, X., Gordon, R. E., O’Brien, J. A., Yuan, H., Zhang, J., Wang, Z., Zhang, M., Song, Y., Yu, C., and Gu, C. (2019) EZH2 phosphorylation promotes self- renewal of glioma stem-like cells through NF-κB methylation. Front. Oncol. 10.3389/fonc.2019.00641
69. Zhang, H., Wei, P., Lv, W., Han, X., Yang, J., Qin, S., and Zhang, Y. (2019) MELK is upregulated in advanced clear cell renal cell carcinoma and promotes disease progression by phosphorylating PRAS40. Cell Transplant. 10.1177/0963689719890860
70. Verlinden, L., Eelen, G., Beullens, I., Van Camp, M., Van Hummelen, P., Engelen, K., Van Hellemont, R., Marchal, K., De Moor, B., Foijer, F., Te Riele, H., Beullens, M., Bollen, M., Mathieu, C., Bouillon, R., and Verstuyf, A. (2005) Characterization of the condensin component Cnap1 and protein kinase Melk as novel E2F target genes down- regulated by 1,25-dihydroxyvitamin D3. J. Biol. Chem. 280, 37319–30
71. Bollu, L. R., Shepherd, J., Zhao, D., Ma, Y., Tahaney, W., Speers, C., Mazumdar, A., Mills, G. B., and Brown, P. H. (2020) Mutant P53 induces MELK expression by release of wild-type P53-dependent suppression of FOXM1. npj Breast Cancer. 6, 2
72. Nakano, I., Paucar, A. A., Bajpai, R., Dougherty, J. D., Zewail, A., Kelly, T. K., Kim, K. J., Ou, J., Groszer, M., Imura, T., Freije, W. A., Nelson, S. F., Sofroniew, M. V., Wu, H., Liu, X., Terskikh, A. V., Geschwind, D. H., and Kornblum, H. I. (2005) Maternal embryonic leucine zipper kinase (MELK) regulates multipotent neural progenitor proliferation. J. Cell Biol. 170, 413–427
149 73. Davezac, N., Baldin, V., Blot, J., Ducommun, B., and Tassan, J.-P. (2002) Human pEg3 kinase associates with and phosphorylates CDC25B phosphatase: a potential role for pEg3 in cell cycle regulation. Oncogene. 21, 7630–41
74. Joshi, K., Banasavadi-Siddegowda, Y., Mo, X., Kim, S. H., Mao, P., Kig, C., Nardini, D., Sobol, R. W., Chow, L. M. L., Kornblum, H. I., Waclaw, R., Beullens, M., and Nakano, I. (2013) MELK-dependent FOXM1 phosphorylation is essential for proliferation of glioma stem cells. Stem Cells. 31, 1051–1063
75. Wang, Y., Begley, M., Li, Q., Huang, H.-T., Lako, A., Eck, M. J., Gray, N. S., Mitchison, T. J., Cantley, L. C., and Zhao, J. J. (2016) Mitotic MELK-eIF4B signaling controls protein synthesis and tumor cell survival. Proc. Natl. Acad. Sci. 113, 9810–9815
76. Le Page, Y., Chartrain, I., Badouel, C., and Tassan, J. P. (2011) A functional analysis of MELK in cell division reveals a transition in the mode of cytokinesis during Xenopus development. J. Cell Sci. 124, 958–968
77. Chartrain, I., Le Page, Y., Hatte, G., Körner, R., Kubiak, J. Z., and Tassan, J. P. (2013) Cell-cycle dependent localization of MELK and its new partner RACK1 in epithelial versus mesenchyme-like cells in Xenopus embryo. Biol. Open. 2, 1037–1048
78. Schmit, T. L., and Ahmad, N. (2007) Regulation of mitosis via mitotic kinases: New opportunities for cancer management. Mol. Cancer Ther. 6, 1920–1931
79. Kig, C., Beullens, M., Beke, L., Van Eynde, A., Linders, J. T., Brehmer, D., and Bollen, M. (2013) Maternal embryonic leucine zipper kinase (MELK) reduces replication stress in glioblastoma cells. J. Biol. Chem. 288, 24200–24212
80. Chung, S., Kijima, K., Kudo, A., Fujisawa, Y., Harada, Y., Taira, A., Takamatsu, N., Miyamoto, T., Matsuo, Y., and Nakamura, Y. (2016) Preclinical evaluation of biomarkers associated with antitumor activity of MELK inhibitor. Oncotarget. 7, 18171–18182
81. Gu, C., Banasavadi-Siddegowda, Y. K., Joshi, K., Nakamura, Y., Kurt, H., Gupta, S., and Nakano, I. (2013) Tumor-specific activation of the C-JUN/MELK pathway regulates glioma stem cell growth in a p53-dependent manner. Stem Cells. 31, 870–881
82. Jung, H., Seong, H.-A., and Ha, H. (2008) Murine protein serine/threonine kinase 38 activates apoptosis signal-regulating kinase 1 via Thr838 phosphorylation. J. Biol. Chem. 283, 34541–53
83. Seong, H.-A., and Ha, H. (2012) Murine protein serine-threonine kinase 38 activates p53 function through Ser15 phosphorylation. J. Biol. Chem. 287, 20797–810
84. Seong, H. A., Manoharan, R., and Ha, H. (2014) A crucial role for the phosphorylation of STRAP at Ser188 by MPK38 in STRAP-dependent cell death through ASK1, TGF-β, p53, and PI3K/PDK1 signaling pathways. Cell Cycle. 13, 3357–3374
85. Seong, H. A., Jung, H., and Ha, H. (2010) Murine protein serine/threonine kinase 38
150 stimulates TGF-β signaling in a kinase-dependent manner via direct phosphorylation of smad proteins. J. Biol. Chem. 285, 30959–30970
86. Ganguly, R., Mohyeldin, A., Thiel, J., Kornblum, H. I., Beullens, M., and Nakano, I. (2015) MELK-a conserved kinase: Functions, signaling, cancer, and controversy. Clin. Transl. Med. 10.1186/s40169-014-0045-y
87. Liu, G., Yuan, X., Zeng, Z., Tunici, P., Ng, H., Abdulkadir, I. R., Lu, L., Irvin, D., Black, K. L., and Yu, J. S. (2006) Analysis of gene expression and chemoresistance of CD133+ cancer stem cells in glioblastoma. Mol. Cancer. 10.1186/1476-4598-5-67
88. Nakano, I., Joshi, K., Visnyei, K., Hu, B., Watanabe, M., Lam, D., Wexler, E., Saigusa, K., Nakamura, Y., Laks, D. R., Mischel, P. S., Viapiano, M., and Kornblum, H. I. (2011) Siomycin A targets brain tumor stem cells partially through a MELK-mediated pathway. Neuro. Oncol. 13, 622–34
89. Cordes, S., Frank, C. A., and Garriga, G. (2006) The C. elegans MELK ortholog PIG-1 regulates cell size asymmetry and daughter cell fate in asymmetric neuroblast divisions. Development. 133, 2747–2756
90. Chien, S.-C., Brinkmann, E.-M., Teuliere, J., and Garriga, G. (2013) Caenorhabditis elegans PIG-1/MELK acts in a conserved PAR-4/LKB1 polarity pathway to promote asymmetric neuroblast divisions. Genetics. 193, 897–909
91. Pacquelet, A., Uhart, P., Tassan, J. P., and Michaux, G. (2015) PAR-4 and anillin regulate myosin to coordinate spindle and furrow position during asymmetric division. J. Cell Biol. 210, 1085–1099
92. Wei, H., Yan, B., Gagneur, J., and Conradt, B. (2017) Caenorhabditis elegans CES-1 Snail represses pig-1 MELK expression to control asymmetric cell division. Genetics. 206, 2069–2084
93. Offenburger, S.-L., Bensaddek, D., Murillo, A. B., Lamond, A. I., and Gartner, A. (2017) Comparative genetic, proteomic and phosphoproteomic analysis of C. elegans embryos with a focus on ham-1/STOX and pig-1/MELK in dopaminergic neuron development. Sci. Rep. 7, 4314
94. Tassan, J.-P. (2011) Cortical localization of Maternal Embryonic Leucine zipper Kinase (MELK) implicated in cytokinesis in early Xenopus embryos. Commun. Integr. Biol. 4, 483–485
95. Bajaj, J., Zimdahl, B., and Reya, T. (2015) Fearful symmetry: Subversion of asymmetric division in cancer development and progression. Cancer Res. 75, 792–797
96. Lens, S. M. A., and Medema, R. H. (2019) Cytokinesis defects and cancer. Nat. Rev. Cancer. 19, 32–45
97. Kent, L. N., and Leone, G. (2019) The broken cycle: E2F dysfunction in cancer. Nat. Rev.
151 Cancer. 19, 326–338
98. Chen, X., Muller, G. A., Quaas, M., Fischer, M., Han, N., Stutchbury, B., Sharrocks, A. D., and Engeland, K. (2013) The forkhead transcription factor FOXM1 controls cell cycle- dependent gene expression through an atypical chromatin binding mechanism. Mol. Cell. Biol. 33, 227–236
99. Wang, I.-C., Chen, Y.-J., Hughes, D., Petrovic, V., Major, M. L., Park, H. J., Tan, Y., Ackerson, T., and Costa, R. H. (2005) Forkhead Box M1 regulates the transcriptional network of genes essential for mitotic progression and genes encoding the SCF (Skp2- Cks1) ubiquitin ligase. Mol. Cell. Biol. 25, 10875–10894
100. Kong, F. F., Qu, Z. Q., Yuan, H. H., Wang, J. Y., Zhao, M., Guo, Y. H., Shi, J., Gong, X. Di, Zhu, Y. L., Liu, F., Zhang, W. Y., and Jiang, B. (2014) Overexpression of FOXM1 is associated with EMT and is a predictor of poor prognosis in non-small cell lung cancer. Oncol. Rep. 31, 2660–2668
101. Wen, N., Wang, Y., Wen, L., Zhao, S. H., Ai, Z. H., Wang, Y., Wu, B., Lu, H. X., Yang, H., Liu, W. C., and Li, Y. (2014) Overexpression of FOXM1 predicts poor prognosis and promotes cancer cell proliferation, migration and invasion in epithelial ovarian cancer. J. Transl. Med. 12, 134
102. Zhao, F., Siu, M. K. Y., Jiang, L. L., Tam, K. F., Ngan, H. Y. S., Le, X. F., Wong, O. G. W., Wong, E. S. Y., Gomes, A. R., Bella, L., Khongkow, P., Lam, E. W. F., and Cheung, A. N. Y. (2014) Overexpression of forkhead box protein M1 (FOXM1) in ovarian cancer correlates with poor patient survival and contributes to paclitaxel resistance. PLoS One. 9, e113478
103. Zhou, Z., He, M., Shah, A. A., and Wan, Y. (2016) Insights into APC/C: From cellular function to diseases and therapeutics. Cell Div. 10.1186/s13008-016-0021-6
104. Zhao, W.-M., Coppinger, J. A., Seki, A., Cheng, X.-L., Yates Iii, J. R., and Fang, G. (2008) RCS1, a substrate of APC/C, controls the metaphase to anaphase transition. Proc. Natl. Acad. Sci. 105, 13415–13420
105. Manoharan, R., Seong, H.-A., and Ha, H. (2013) Thioredoxin inhibits MPK38-induced ASK1, TGF‐β, and p53 function in a phosphorylation-dependent manner. Free Radic. Biol. Med. 63, 313–324
106. Lizcano, J. M., Göransson, O., Toth, R., Deak, M., Morrice, N. A., Boudeau, J., Hawley, S. A., Udd, L., Mäkelä, T. P., Hardie, D. G., and Alessi, D. R. (2004) LKB1 is a master kinase that activates 13 kinases of the AMPK subfamily, including MARK/PAR-1. EMBO J. 23, 833–43
107. Beullens, M., Vancauwenbergh, S., Morrice, N., Derua, R., Ceulemans, H., Waelkens, E., and Bollen, M. (2005) Substrate specificity and activity regulation of protein kinase MELK. J. Biol. Chem. 280, 40003–11
152 108. Cao, L.-S., Wang, J., Chen, Y., Deng, H., Wang, Z.-X., and Wu, J.-W. (2013) Structural basis for the regulation of maternal embryonic leucine zipper kinase. PLoS One. 8, e70031
109. Seong, H.-A., Manoharan, R., and Ha, H. (2017) Zinc finger protein ZPR9 functions as an activator of AMPK-related serine/threonine kinase MPK38/MELK involved in ASK1/TGF-β/p53 signaling pathways. Sci. Rep. 7, 42502
110. Chen, L., Wei, Q., Bi, S., and Xie, S. (2020) Maternal embryonic leucine zipper kinase promotes tumor growth and metastasis via stimulating FOXM1 signaling in esophageal squamous cell carcinoma. Front. Oncol. 10, 10
111. Li, B., Yan, J., Phyu, T., Fan, S., Chung, T.-H., Mustafa, N. B., Lin, B., Wang, L., Eichhorn, P. J. A., Goh, B.-C., Ng, S.-B., Kappei, D., and Chng, W.-J. (2019) MELK mediates the stability of EZH2 through site-specific phosphorylation in extranodal natural killer/T-cell lymphoma. Blood. 134, 2046–2058
112. Liu, H., Sun, Q., Sun, Y., Zhang, J., Yuan, H., Pang, S., Qi, X., Wang, H., Zhang, M., Zhang, H., Yu, C., and Gu, C. (2017) MELK and EZH2 cooperate to regulate medulloblastoma cancer stem-like cell proliferation and differentiation. Mol. Cancer Res. 15, 1275–1286
113. Mirey, G., Chartrain, I., Froment, C., Quaranta, M., Bouché, J.-P., Monsarrat, B., Tassan, J.-P., and Ducommun, B. (2005) CDC25B phosphorylated by pEg3 localizes to the centrosome and the spindle poles at mitosis. Cell Cycle. 4, 806–811
114. Seong, H. A., Jung, H., Manoharan, R., and Ha, H. (2012) PDK1 protein phosphorylation at Thr354 by murine protein serine-threonine kinase 38 contributes to negative regulation of PDK1 protein activity. J. Biol. Chem. 287, 20811–20822
115. Seong, H. A., Gil, M., Kim, K. T., Kim, S. J., and Ha, H. (2002) Phosphorylation of a novel zinc-finger-like protein, ZPR9, by murine protein serine/threonine kinase 38 (MPK38). Biochem. J. 361, 597–604
116. Marie, S. K. N., Oba-Shinjo, S. M., da Silva, R., Gimenez, M., Nunes Reis, G., Tassan, J.- P., Rosa, J. C., and Uno, M. (2016) Stathmin involvement in the maternal embryonic leucine zipper kinase pathway in glioblastoma. Proteome Sci. 14, 6
117. Siegel, R. L., Miller, K. D., and Jemal, A. (2019) Cancer statistics, 2019. CA. Cancer J. Clin. 69, 7–34
118. Cancer stat facts: Pancreatic cancer (2019) Natl. Cancer Inst. Surveillance, Epidemiol. End Results Progr.
119. Rahib, L., Smith, B. D., Aizenberg, R., Rosenzweig, A. B., Fleshman, J. M., and Matrisian, L. M. (2014) Projecting cancer incidence and deaths to 2030: The unexpected burden of thyroid, liver, and pancreas cancers in the United States. Cancer Res. 74, 2913– 2921
153 120. Vincent, A., Herman, J., Schulick, R., Hruban, R. H., and Goggins, M. (2011) Pancreatic cancer. Lancet. 378, 607–620
121. Moletta, L., Serafini, S., Valmasoni, M., Pierobon, E. S., Ponzoni, A., and Sperti, C. (2019) Surgery for recurrent pancreatic cancer: Is it effective? Cancers (Basel). 10.3390/cancers11070991
122. Saung, M. T., and Zheng, L. (2017) Current standards of chemotherapy for pancreatic cancer. Clin. Ther. 39, 2125–2134
123. Conroy, T., Desseigne, F., Ychou, M., Bouché, O., Guimbaud, R., Bécouarn, Y., Adenis, A., Raoul, J.-L., Gourgou-Bourgade, S., de la Fouchardière, C., Bennouna, J., Bachet, J.- B., Khemissa-Akouz, F., Péré-Vergé, D., Delbaldo, C., Assenat, E., Chauffert, B., Michel, P., Montoto-Grillot, C., and Ducreux, M. (2011) FOLFIRINOX versus gemcitabine for metastatic pancreatic cancer. N. Engl. J. Med. 364, 1817–1825
124. Von Hoff, D. D., Ervin, T., Arena, F. P., Chiorean, E. G., Infante, J., Moore, M., Seay, T., Tjulandin, S. A., Ma, W. W., Saleh, M. N., Harris, M., Reni, M., Dowden, S., Laheru, D., Bahary, N., Ramanathan, R. K., Tabernero, J., Hidalgo, M., Goldstein, D., Van Cutsem, E., Wei, X., Iglesias, J., and Renschler, M. F. (2013) Increased survival in pancreatic cancer with nab-paclitaxel plus gemcitabine. N. Engl. J. Med. 369, 1691–1703
125. Burris, H. A., Moore, M. J., Andersen, J., Green, M. R., Rothenberg, M. L., Modiano, M. R., Cripps, M. C., Portenoy, R. K., Storniolo, A. M., Tarassoff, P., Nelson, R., Dorr, F. A., Stephens, C. D., and Von Hoff, D. D. (1997) Improvements in survival and clinical benefit with gemcitabine as first-line therapy for patients with advanced pancreas cancer: A randomized trial. J. Clin. Oncol. 15, 2403–2413
126. Kim, M. P., and Gallick, G. E. (2008) Gemcitabine resistance in pancreatic cancer: Picking the key players. Clin. Cancer Res. 14, 1284–1285
127. Duncan, J. S., Whittle, M. C., Nakamura, K., Abell, A. N., Midland, A. A., Zawistowski, J. S., Johnson, N. L., Granger, D. A., Jordan, N. V., Darr, D. B., Usary, J., Kuan, P.-F., Smalley, D. M., Major, B., He, X., Hoadley, K. A., Zhou, B., Sharpless, N. E., Perou, C. M., Kim, W. Y., Gomez, S. M., Chen, X., Jin, J., Frye, S. V, Earp, H. S., Graves, L. M., and Johnson, G. L. (2012) Dynamic reprogramming of the kinome in response to targeted MEK inhibition in triple-negative breast cancer. Cell. 149, 307–21
128. Stuhlmiller, T. J., Miller, S. M., Zawistowski, J. S., Nakamura, K., Beltran, A. S., Duncan, J. S., Angus, S. P., Collins, K. A. L., Granger, D. A., Reuther, R. A., Graves, L. M., Gomez, S. M., Kuan, P.-F., Parker, J. S., Chen, X., Sciaky, N., Carey, L. A., Earp, H. S., Jin, J., and Johnson, G. L. (2015) Inhibition of lapatinib-induced kinome reprogramming in ERBB2-positive breast cancer by targeting BET family bromodomains. Cell Rep. 11, 390–404
129. Cooper, M. J., Cox, N. J., Zimmerman, E. I., Dewar, B. J., Duncan, J. S., Whittle, M. C., Nguyen, T. A., Jones, L. S., Ghose Roy, S., Smalley, D. M., Kuan, P. F., Richards, K. L., Christopherson, R. I., Jin, J., Frye, S. V., Johnson, G. L., Baldwin, A. S., and Graves, L.
154 M. (2013) Application of multiplexed kinase inhibitor beads to study kinome adaptations in drug-resistant leukemia. PLoS One. 8, e66755
130. Krulikas, L. J., McDonald, I. M., Lee, B., Okumu, D. O., East, M. P., Gilbert, T. S. K., Herring, L. E., Golitz, B. T., Wells, C. I., Axtman, A. D., Zuercher, W. J., Willson, T. M., Kireev, D., Yeh, J. J., Johnson, G. L., Baines, A. T., and Graves, L. M. (2018) Application of integrated drug screening/kinome analysis to identify inhibitors of gemcitabine- resistant pancreatic cancer cell growth. SLAS Discov. 23, 850–861
131. Zhang, W., DeRyckere, D., Hunter, D., Liu, J., Stashko, M. A., Minson, K. A., Cummings, C. T., Lee, M., Glaros, T. G., Newton, D. L., Sather, S., Zhang, D., Kireev, D., Janzen, W. P., Earp, H. S., Graham, D. K., Frye, S. V., and Wang, X. (2014) UNC2025, a potent and orally bioavailable MER/FLT3 dual inhibitor. J. Med. Chem. 57, 7031–7041
132. Cummings, C. T., Zhang, W., Davies, K. D., Kirkpatrick, G. D., Zhang, D., DeRyckere, D., Wang, X., Frye, S. V., Earp, H. S., and Graham, D. K. (2015) Small molecule inhibition of MERTK is efficacious in non-small cell lung cancer models independent of driver oncogene status. Mol. Cancer Ther. 14, 2014–2022
133. DeRyckere, D., Lee-Sherick, A. B., Huey, M. G., Hill, A. A., Tyner, J. W., Jacobsen, K. M., Page, L. S., Kirkpatrick, G. G., Eryildiz, F., Montgomery, S. A., Zhang, W., Wang, X., Frye, S. V., Earp, H. S., and Graham, D. K. (2017) UNC2025, a MERTK small- molecule inhibitor, is therapeutically effective alone and in combination with methotrexate in leukemia models. Clin. Cancer Res. 23, 1481–1492
134. Sinik, L., Minson, K. A., Tentler, J. J., Carrico, J., Bagby, S. M., Robinson, W. A., Kami, R., Burstyn-Cohen, T., Gail Eckhardt, S., Wang, X., Frye, S. V., Shelton Earp, H., DeRyckere, D., and Graham, D. K. (2019) Inhibition of MERTK promotes suppression of tumor growth in BRAF mutant and BRAF wild-type melanoma. Mol. Cancer Ther. 18, 278–288
135. Sufit, A., Lee-Sherick, A. B., Deryckere, D., Rupji, M., Dwivedi, B., Varella-Garcia, M., Pierce, A. M., Kowalski, J., Wang, X., Frye, S. V., Earp, H. S., Keating, A. K., and Graham, D. K. (2016) MERTK inhibition induces polyploidy and promotes cell death and cellular senescence in glioblastoma multiforme. PLoS One. 11, e0165107
136. Shen, Y., Chen, X., He, J., Liao, D., and Zu, X. (2018) Axl inhibitors as novel cancer therapeutic agents. Life Sci. 198, 99–111
137. Du, W., and Brekken, R. A. (2018) Does Axl have potential as a therapeutic target in pancreatic cancer? Expert Opin. Ther. Targets. 22, 955–966
138. Arend, K. C., Lenarcic, E. M., Vincent, H. A., Rashid, N., Lazear, E., McDonald, I. M., Gilbert, T. S. K., East, M. P., Herring, L. E., Johnson, G. L., Graves, L. M., and Moorman, N. J. (2017) Kinome profiling identifies druggable targets for novel human cytomegalovirus (HCMV) antivirals. Mol. Cell. Proteomics. 10.1074/mcp.M116.065375
155 139. Malumbres, M., Sotillo, R., Santamaría, D., Galán, J., Cerezo, A., Ortega, S., Dubus, P., and Barbacid, M. (2004) Mammalian cells cycle without the D-type cyclin-dependent kinases Cdk4 and Cdk6. Cell. 118, 493–504
140. Malumbres, M., and Barbacid, M. (2009) Cell cycle, CDKs and cancer: A changing paradigm. Nat. Rev. Cancer. 9, 153–166
141. Choi, Y. J., Li, X., Hydbring, P., Sanda, T., Stefano, J., Christie, A. L., Signoretti, S., Look, A. T., Kung, A. L., von Boehmer, H., and Sicinski, P. (2012) The requirement for cyclin D function in tumor maintenance. Cancer Cell. 22, 438–451
142. Otto, T., and Sicinski, P. (2017) Cell cycle proteins as promising targets in cancer therapy. Nat. Rev. Cancer. 17, 93–115
143. Currie, C. E., Mora-Santos, M., Smith, C. A., McAinsh, A. D., and Millar, J. B. A. (2018) Bub1 is not essential for the checkpoint response to unattached kinetochores in diploid human cells. Curr. Biol. 28, R929–R930
144. Raaijmakers, J. A., van Heesbeen, R. G. H. P., Blomen, V. A., Janssen, L. M. E., van Diemen, F., Brummelkamp, T. R., and Medema, R. H. (2018) BUB1 is essential for the viability of human cells in which the spindle assembly checkpoint is compromised. Cell Rep. 22, 1424–1438
145. Crosio, C., Fimia, G. M., Loury, R., Kimura, M., Okano, Y., Zhou, H., Sen, S., Allis, C. D., and Sassone-Corsi, P. (2002) Mitotic phosphorylation of histone H3: Spatio-temporal regulation by mammalian aurora kinases. Mol. Cell. Biol. 22, 874–885
146. Hirota, T., Kunitoku, N., Sasayama, T., Marumoto, T., Zhang, D., Nitta, M., Hatakeyama, K., and Saya, H. (2003) Aurora-A and an interacting activator, the LIM protein Ajuba, are required for mitotic commitment in human cells. Cell. 114, 585–598
147. Yasui, Y., Urano, T., Kawajiri, A., Nagata, K. I., Tatsuka, M., Saya, H., Furukawa, K., Takahashi, T., Izawa, I., and Inagaki, M. (2004) Autophosphorylation of a newly identified site of Aurora-B Is indispensable for cytokinesis. J. Biol. Chem. 279, 12997– 13003
148. Lindqvist, A., Rodríguez-Bravo, V., and Medema, R. H. (2009) The decision to enter mitosis: feedback and redundancy in the mitotic entry network. J. Cell Biol. 185, 193–202
149. Shaltiel, I. A., Krenning, L., Bruinsma, W., and Medema, R. H. (2015) The same, only different – DNA damage checkpoints and their reversal throughout the cell cycle. J. Cell Sci. 128, 607–620
150. Grant, G. D., Kedziora, K. M., Limas, J. C., Cook, J. G., and Purvis, J. E. (2018) Accurate delineation of cell cycle phase transitions in living cells with PIP-FUCCI. Cell Cycle. 17, 2496–2516
151. Guleria, A., and Chandna, S. (2016) ATM kinase: Much more than a DNA damage
156 responsive protein. DNA Repair (Amst). 39, 1–20
152. Graves, P. R., Aponte-Collazo, L. J., Fennell, E. M. J., Graves, A. C., Hale, A. E., Dicheva, N., Herring, L. E., Gilbert, T. S. K., East, M. P., McDonald, I. M., Lockett, M. R., Ashamalla, H., Moorman, N. J., Karanewsky, D. S., Iwanowicz, E. J., Holmuhamedov, E., and Graves, L. M. (2019) Mitochondrial protease ClpP is a target for the anticancer compounds ONC201 and related analogues. ACS Chem. Biol. 14, 1020–1029
153. Cox, J., and Mann, M. (2008) MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat. Biotechnol. 26, 1367–1372
154. Cox, J., Michalski, A., and Mann, M. (2011) Software lock mass by two-dimensional minimization of peptide mass errors. J. Am. Soc. Mass Spectrom. 22, 1373–80
155. Cox, J., Hein, M. Y., Luber, C. A., Paron, I., Nagaraj, N., and Mann, M. (2014) Accurate proteome-wide label-free quantification by delayed normalization and maximal peptide ratio extraction, termed MaxLFQ. Mol. Cell. Proteomics. 13, 2513–26
156. Smyth, G. K. (2004) Linear models and empirical Bayes methods for assessing differential expression in microarray experiments. Stat. Appl. Genet. Mol. Biol. 3, 1–25
157. Benjamini, Y., and Hochberg, Y. (1995) Controlling the false discovery rate : A practical and powerful approach to multiple testing. J. R. Stat. Soc. 57, 289–300
158. Perez-Riverol, Y., Csordas, A., Bai, J., Bernal-Llinares, M., Hewapathirana, S., Kundu, D. J., Inuganti, A., Griss, J., Mayer, G., Eisenacher, M., Pérez, E., Uszkoreit, J., Pfeuffer, J., Sachsenberg, T., Yılmaz, Ş., Tiwary, S., Cox, J., Audain, E., Walzer, M., Jarnuczak, A. F., Ternent, T., Brazma, A., and Vizcaíno, J. A. (2019) The PRIDE database and related tools and resources in 2019: improving support for quantification data. Nucleic Acids Res. 47, D442–D450
159. Smith, J., Mun Tho, L., Xu, N., and A. Gillespie, D. (2010) The ATM-Chk2 and ATR- Chk1 pathways in DNA damage signaling and cancer. Adv. Cancer Res. 108, 73–112
160. Gingras, A. C., Abe, K. T., and Raught, B. (2019) Getting to know the neighborhood: using proximity-dependent biotinylation to characterize protein complexes and map organelles. Curr. Opin. Chem. Biol. 48, 44–54
161. Roux, K. J., Kim, D. I., Raida, M., and Burke, B. (2012) A promiscuous biotin ligase fusion protein identifies proximal and interacting proteins in mammalian cells. J. Cell Biol. 196, 801–810
162. Kwon, K., and Beckett, D. (2000) Function of a conserved sequence motif in biotin holoenzyme synthetases. Protein Sci. 9, 1530–1539
163. Choi-rhee, E., Schulman, H., and Cronan, J. E. (2004) Promiscuous protein biotinylation by Escherichia coli biotin protein ligase. Protein Sci. 13, 3043–3050
157 164. Kim, D. I., KC, B., Zhu, W., Motamedchaboki, K., Doye, V., and Roux, K. J. (2014) Probing nuclear pore complex architecture with proximity-dependent biotinylation. Proc. Natl. Acad. Sci. U. S. A. 111, E2453-61
165. Sharma, K., D’Souza, R. C. J., Tyanova, S., Schaab, C., Wiśniewski, J. R., Cox, J., and Mann, M. (2014) Ultradeep human phosphoproteome reveals a distinct regulatory nature of tyr and ser/thr-based signaling. Cell Rep. 8, 1583–1594
166. Dephoure, N., Gould, K. L., Gygi, S. P., and Kellogg, D. R. (2013) Mapping and analysis of phosphorylation sites: A quick guide for cell biologists. Mol. Biol. Cell. 24, 535–542
167. Macek, B., Mann, M., and Olsen, J. V. (2009) Global and site-specific quantitative phosphoproteomics: Principles and applications. Annu. Rev. Pharmacol. Toxicol. 49, 199– 221
168. Tomaselli, D., Lucidi, A., Rotili, D., and Mai, A. (2019) Epigenetic polypharmacology: A new frontier for epi‐drug discovery. Med. Res. Rev. 40, 190–244
169. Al-Lazikani, B., Banerji, U., and Workman, P. (2012) Combinatorial drug therapy for cancer in the post-genomic era. Nat. Biotechnol. 30, 679–692
170. Lee, A., and Djamgoz, M. B. A. (2018) Triple negative breast cancer: Emerging therapeutic modalities and novel combination therapies. Cancer Treat. Rev. 62, 110–122
171. Bevill, S. M., Olivares-Quintero, J. F., Sciaky, N., Golitz, B. T., Singh, D., Beltran, A. S., Rashid, N. U., Stuhlmiller, T. J., Hale, A., Moorman, N. J., Santos, C. M., Angus, S. P., Zawistowski, J. S., and Johnson, G. L. (2019) GSK2801, a BAZ2/BRD9 bromodomain inhibitor, synergizes with BET inhibitors to induce apoptosis in triple-negative breast cancer. Mol. Cancer Res. 17, 1503–1518
172. Liang, S. Q., Bührer, E. D., Berezowska, S., Marti, T. M., Xu, D., Froment, L., Yang, H., Hall, S. R. R., Vassella, E., Yang, Z., Kocher, G. J., Amrein, M. A., Riether, C., Ochsenbein, A. F., Schmid, R. A., and Peng, R. W. (2019) mTOR mediates a mechanism of resistance to chemotherapy and defines a rational combination strategy to treat KRAS- mutant lung cancer. Oncogene. 38, 622–636
173. Antolin, A., Workman, P., Mestres, J., and Al-Lazikani, B. (2016) Polypharmacology in precision oncology: Current applications and future prospects. Curr. Pharm. Des. 22, 6935–6945
174. Mattiroli, F., Vissers, J. H. A., Van Dijk, W. J., Ikpa, P., Citterio, E., Vermeulen, W., Marteijn, J. A., and Sixma, T. K. (2012) RNF168 ubiquitinates K13-15 on H2A/H2AX to drive DNA damage signaling. Cell. 150, 1182–1195
175. Varnaitė, R., and MacNeill, S. A. (2016) Meet the neighbors: Mapping local protein interactomes by proximity-dependent labeling with BioID. Proteomics. 16, 2503–2518
176. Li, G., Huang, Z., Zhang, C., Dong, B., and Guo, R. (2016) Construction of a linker
158 library with widely controllable flexibility for fusion protein design. Appl Microbiol Biotechnol. 100, 215–225
177. Chen, X., Zaro, J. L., and Shen, W. (2013) Fusion protein linkers: Property, design and functionality. Adv. Drug Deliv. Rev. 65, 1357–1369
178. Krämer, A., Green, J., Pollard, J., and Tugendreich, S. (2014) Causal analysis approaches in Ingenuity Pathway Analysis. Bioinformatics. 30, 523–530
179. Van Itallie, C. M., Aponte, A., Tietgens, A. J., Gucek, M., Fredriksson, K., and Anderson, J. M. (2013) The N and C termini of ZO-1 are surrounded by distinct proteins and functional protein networks. J. Biol. Chem. 288, 13775–88
180. Ballone, A., Centorrino, F., and Ottmann, C. (2018) 14-3-3: A case study in PPI modulation. Molecules. 10.3390/molecules23061386
181. Cau, Y., Valensin, D., Mori, M., Draghi, S., and Botta, M. (2017) Structure, function, involvement in diseases and targeting of 14-3-3 proteins: An update. Curr. Med. Chem. 25, 5–21
182. Khorrami, A., Sharif Bagheri, M., Tavallaei, M., and Gharechahi, J. (2017) The functional significance of 14-3-3 proteins in cancer: Focus on lung cancer. Horm. Mol. Biol. Clin. Investig. 10.1515/hmbci-2017-0032
183. Bochman, M. L., and Schwacha, A. (2009) The Mcm complex: Unwinding the mechanism of a replicative helicase. Microbiol. Mol. Biol. Rev. 73, 652–683
184. Bell, S. D., and Botchan, M. R. (2013) The minichromosome maintenance replicative helicase. Cold Spring Harb. Perspect. Biol. 10.1101/cshperspect.a012807
185. Blow, J. J., and Dutta, A. (2005) Preventing re-replication of chromosomal DNA. Nat. Rev. Mol. Cell Biol. 6, 476–486
186. Choe, K. N., and Moldovan, G. L. (2017) Forging ahead through darkness: PCNA, still the principal conductor at the replication fork. Mol. Cell. 65, 380–392
187. Zou, Y., Liu, Y., Wu, X., and Shell, S. M. (2006) Functions of human replication protein A (RPA): From DNA replication to DNA damage and stress responses. J. Cell. Physiol. 208, 267–273
188. Bush, N. G., Evans-Roberts, K., and Maxwell, A. (2015) DNA Topoisomerases. EcoSal Plus. 10.1128/ecosalplus.ESP-0010-2014
189. Petrone, A., Adamo, M. E., Cheng, C., and Kettenbach, A. N. (2016) Identification of candidate cyclin-dependent kinase 1 (Cdk1) substrates in mitosis by quantitative phosphoproteomics. Mol. Cell. Proteomics. 15, 2448–2461
190. Laoukili, J., Kooistra, M. R. H., Brás, A., Kauw, J., Kerkhoven, R. M., Morrison, A.,
159 Clevers, H., and Medema, R. H. (2005) FoxM1 is required for execution of the mitotic programme and chromosome stability. Nat. Cell Biol. 7, 126–136
191. Liao, G. Bin, Li, X. Z., Zeng, S., Liu, C., Yang, S. M., Yang, L., Hu, C. J., and Bai, J. Y. (2018) Regulation of the master regulator FOXM1 in cancer. Cell Commun. Signal. 10.1186/s12964-018-0266-6
192. Mertins, P., Mani, D. R., Ruggles, K. V., Gillette, M. A., Clauser, K. R., Wang, P., Wang, X., Qiao, J. W., Cao, S., Petralia, F., Kawaler, E., Mundt, F., Krug, K., Tu, Z., Lei, J. T., Gatza, M. L., Wilkerson, M., Perou, C. M., Yellapantula, V., Huang, K. L., Lin, C., McLellan, M. D., Yan, P., Davies, S. R., Townsend, R. R., Skates, S. J., Wang, J., Zhang, B., Kinsinger, C. R., Mesri, M., Rodriguez, H., Ding, L., Paulovich, A. G., Fenyö, D., Ellis, M. J., and Carr, S. A. (2016) Proteogenomics connects somatic mutations to signalling in breast cancer. Nature. 534, 55–62
193. Stuart, S. A., Houel, S., Lee, T., Wang, N., Old, W. M., and Ahn, N. G. (2015) A phosphoproteomic comparison of B-RAF V600E and MKK1/2 inhibitors in melanoma cells. Mol. Cell. Proteomics. 14, 1599–1615
194. Beli, P., Lukashchuk, N., Wagner, S. A., Weinert, B. T., Olsen, J. V., Baskcomb, L., Mann, M., Jackson, S. P., and Choudhary, C. (2012) Proteomic investigations reveal a role for RNA processing Factor THRAP3 in the DNA damage response. Mol. Cell. 46, 212– 225
195. Blackford, A. N., and Jackson, S. P. (2017) ATM, ATR, and DNA-PK: The trinity at the heart of the DNA damage response. Mol. Cell. 66, 801–817
196. Matson, J. P., Dumitru, R., Coryell, P., Baxley, R. M., Chen, W., Twaroski, K., Webber, B. R., Tolar, J., Bielinsky, A. K., Purvis, J. E., and Cook, J. G. (2017) Rapid DNA replication origin licensing protects stem cell pluripotency. Elife. 10.7554/eLife.30473
197. Badouel, C., Körner, R., Frank-Vaillant, M., Couturier, A., Nigg, E. A., and Tassan, J. P. (2006) M-phase MELK activity is regulated by MPF and MAPK. Cell Cycle. 5, 883–889
198. Kim, D. I., Jensen, S. C., Noble, K. A., KC, B., Roux, K. H., Motamedchaboki, K., and Roux, K. J. (2016) An improved smaller biotin ligase for BioID proximity labeling. Mol. Biol. Cell. 27, 1188–96
199. Li, P., Li, J., Wang, L., and Di, L. J. (2017) Proximity labeling of interacting proteins: Application of BioID as a discovery tool. Proteomics. 10.1002/pmic.201700002
200. Arceci, A., Bonacci, T., Wang, X., Stewart, K., Damrauer, J. S., Hoadley, K. A., and Emanuele, M. J. (2019) FOXM1 deubiquitination by USP21 regulates cell cycle progression and paclitaxel sensitivity in basal-like breast cancer. Cell Rep. 26, 3076–3086
160