Mechanisms of Oncogenesis by Mir-155

Mechanisms of Oncogenesis by miR-155

The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters

Citation Witten, Lisa Walker. 2019. Mechanisms of Oncogenesis by miR-155. Doctoral dissertation, Harvard University, Graduate School of Arts & Sciences.

Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:42029676

Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of- use#LAA

Mechanisms of oncogenesis by miR-155

A dissertation presented

Lisa Walker Witten

The Division of Medical Sciences in partial fulfillment of the requirements

for the degree of

Doctor of Philosophy

in the subject of

Biological and Biomedical Sciences

Harvard University

Cambridge, Massachusetts

March 2019

Dissertation Advisor: Frank Slack Lisa Walker Witten

Mechanisms of oncogenesis by miR-155

Abstract

Tumors develop when abnormalities in a cell accumulate to promote growth and prevent death. These abnormalities can include mutations or changes in gene expression. One type of factor that can contribute to tumor development is microRNAs

(miRNAs), which are small non-coding RNAs that repress expression of target genes. miR-155 is an inflammation-associated miRNA that acts as an oncogene in a number of hematological malignancies. Here, we examine the molecular mechanisms by which miR-155 can promote tumor development and progression.

We used an inducible transgenic mouse model that overexpresses miR-155 and develops hematological malignancy to understand how cooperation between multiple miR-155 targets drives oncogenesis. One category of miR-155 targets includes DNA repair factors, so we performed whole exome sequencing on tumors from our mouse model and identified a single oncogene, c-Kit, which frequently (>93%) bore activating mutations. Tumor growth was dependent on this c-Kit activity, as treatment with an inhibitor resulted in rapid tumor regression. However, the mutation is not sufficient for tumor growth when miR-155 is not overexpressed, because deactivating the miR-155 transgene also results in tumor regression. We showed that c-Kit expression is dependent on miR-155 expression, indicating that targets of miR-155 that affect c-Kit

iii expression must cooperate miR-155-driven DNA repair deficiency in order for tumors to form and progress.

Preliminary data from further studies suggest that this mechanism of oncogenesis is available only under certain conditions. For instance, c-Kit may be epigenetically silenced in adult mice, preventing regulation of its expression by miR-155.

This mechanism is also not available when selection pressure is applied by extended treatment with small molecule inhibitors, and under these conditions, other mechanisms seem to be available for miR-155 to promote oncogenesis. Future research could focus on identifying and understanding these additional mechanisms and the contexts in which they are utilized.

We present here a new model for multi-step oncogenesis by a miRNA, in which multiple targets cooperate to drive oncogenesis. This model may be applicable to other miRNAs and could inform the use of miRNAs as biomarkers and therapeutics.

iv Table of Contents

Abstract ...... iii Table of Contents ...... v Acknowledgements ...... vi Chapter 1: Introduction ...... 1 Overview ...... 2 Tumor development and growth ...... 2 MicroRNA ...... 7 miRNA in Cancer ...... 10 miR-155 ...... 13 Hematopoiesis and Hematological Malignancies ...... 19 c-Kit ...... 21 Objectives ...... 23 Chapter 2: miR-155 drives oncogenesis by promoting and cooperating with mutations in the c-Kit oncogene...... 24 Abstract ...... 25 Introduction ...... 26 Results ...... 28 Discussion ...... 49 Chapter 3: Unpublished Results ...... 52 Introduction ...... 53 Results ...... 53 Discussion ...... 73 Chapter 4: Methods...... 76 Chapter 5: Discussion ...... 89 Cooperation between miRNA targets in oncogenesis ...... 90 miR-155 and c-Kit in cancer ...... 92 Transfection of hematological cells ...... 95 miRNAs and drug resistance ...... 96 Conclusions ...... 99 References ...... 100

v Acknowledgements

I have been incredibly lucky to be surrounded by such an incredible community of scientists during my PhD. Thank you for being my mentors, friends, and inspiration.

To my advisor, Dr. Frank Slack, I am so grateful to have been part of your lab. Your mentorship, support, and guidance were clear, consistent, encouraging, and practical.

You fostered an environment where I felt welcomed, confident, supported, independent, and intellectually engaged. To the members of the Slack lab, I am thankful to have had you to work, talk, joke, commiserate, collaborate, and learn with every day while performing this research, and I’m proud to call myself a Slacker. To the members of my advisory committee, Dr. Matthew Meyerson, Dr. Karen Cichowski, and Dr. Judy

Lieberman, thank you for your interest in my project and in my success, and for the time and guidance you provided toward my accomplishments. To my family, I am so blessed to have been raised in such an inquisitive and curious household, and to have had your infectious enthusiasm for learning encouraging me through this research. Finally, Jacob, you have been everything I needed at every stage of my PhD, and I’m so happy we get to be scientists together.

vi Chapter 1:

Introduction

Overview

The focus of this thesis is the role of a microRNA (miRNA) in cancer, and I will begin with an introduction to the literature on the molecular mechanisms behind tumor growth and on miRNAs. I will highlight the literature about miR-155, an oncogenic miRNA involved in many hematological malignancies, and describe the powerful mouse model of miR-155 driven cancer that I use to perform this research. The second chapter will describe our primary findings of how miR-155 drives oncogenesis in this mouse model, which suggest a multi-step mechanism in which miR-155 promotes both the mutation and expression of an oncogenic receptor tyrosine kinase, c-Kit. The third chapter expands upon these findings by presenting preliminary data to aid in understanding the contexts and pathways in which this mechanism might act, and the alternative mechanisms that might be in play in other contexts. This data, though preliminary, can guide and inform future research. The fourth chapter describes in detail the methods used in the previous chapters, as a reference for readers. Finally, I conclude with a discussion of the findings. I consider their implications for the field’s understanding of oncogenesis by miRNAs and I comment on potential future directions and applications of the research.

Tumor development and growth

Cancer is a diverse and disruptive disease caused by aberrant activity of gene products that normally drive growth and development. This aberrant activity can include too little activity, too much activity, or the correct amount of activity at the wrong time or place. Genes that drive cancer when there is too much activity are oncogenes, while

2 genes that drive cancer when there is too little activity are called tumor suppressors.

Typically, activity of a gene product becomes aberrant either due to a mutation in the gene that alters its function, or due to dysregulated expression such that there is too much or too little of the gene product or its regulators.

Oncogenic mutations and DNA repair

Mutations in oncogenes or tumor suppressors that drive cancer can come from exogenous or endogenous damage. Exogenous sources of DNA damage include UV radiation, cigarette smoke, and carcinogenic chemicals. Damage can also occur more spontaneously, for instance if an incorrect base is inserted during DNA replication, a

“mismatch” occurs (Dexheimer, 2013). If not repaired, the error can be propagated during the next round of DNA replication and the incorrect base is used as the template for transcription.

Cells have DNA repair factors that identify and correct DNA damage. To repair mismatch errors, a combination of “mismatch repair” (MMR) factors including MSH2,

MSH3, MSH6, MLH1, and PMS2 work to identify the mismatch, and resect the damaged section of DNA, so the DNA Polymerase delta can replace it with the correct sequence (Dexheimer, 2013). However, DNA repair mechanisms can fail, either because there is simply too much damage, or because they themselves have deficient activity due to germline or somatic mutation or due to dysregulated expression. When mismatch repair is deficient, it leads to a characteristic profile of mutations in the genome, with transition substitutions (T:A>C:G and C:G>T:A) being the most common

(Greenman et al., 2007). This proclivity for the accumulation of mutations is referred to as a mutator phenotype, and can also be caused by deficiencies in other genome

3 stability mechanisms. A higher frequency of mutations increases the likelihood that a mutation occurs in the DNA coding for an oncogene or tumor suppressor.

When mutations persist in the genome, the altered sequence gets translated into proteins, which are then also mutated. The effect on the protein can either be loss-of- function, where the mutation interferes with the protein’s function, for instance by disrupting proper folding; or it can be gain-of-function, where the mutation causes higher or less well-regulated activity of the protein, for instance by blocking a binding site for an inhibitory regulator.

Regulation of gene expression

Besides mutations in oncogenes and tumor suppressors, changes in the amount, or expression levels, of these genes and their gene products can also contribute to tumor formation and progression. Expression of every gene in a cell is a highly controlled process with many layers of regulation, including transcriptional, post- transcriptional, and protein-level regulation.

For a gene to be expressed, it must first be transcribed into RNA from the DNA template. The amount of RNA transcribed from a gene partially informs the expression levels of the gene product, which is usually a protein. Transcription requires a set of transcription factor proteins to bind the DNA. Different genes require different sets of transcription factors to be transcribed, but there can be overlap in the required transcription factors, especially for genes with related functions. Therefore, the cell regulates transcription in part by regulating the binding of transcription factors, for example by changing levels of the transcription factor expression itself; altering whether the transcription factor is localized to the cytoplasm, where it is inactive, or the nucleus,

4 where it is active; exposing or obscuring binding sites via binding proteins or via epigenetic modifications to the chromatin structure; or introducing molecules that inhibit or activate the transcription factor (Lee and Young, 2013).

However, even after a transcript has been made, the cell can further control gene expression at a post-transcriptional level by controlling the fate of the RNA transcript.

For most genes, RNA must be translated by ribosomes into protein to be functional. If

RNA is degraded or if translation is blocked, protein expression levels from the gene will be lowered. One class of factors that can both inhibit translation and promote RNA degradation is miRNAs, which are small, non-coding RNAs that direct proteins to specific RNAs and repress their expression (Di Leva et al., 2014).

Finally, the half-life and stability of a protein also affects the amount of that protein in the cell. Degradation of specific proteins, or proteolysis, is controlled by the cell using ubiquitination, in which proteins marked with a ubiquitin chain are recognized and degraded by proteases (Lecker et al., 2006).

Hallmarks of cancer and oncogene addiction

Mutation and dysregulation of oncogenes and tumor suppressors are key steps in the development of malignancies. However, cancer is characterized by the accumulation of many abnormalities, and can rarely be attributed to a single alteration.

Often, one mutation or dysregulated gene will lead to more changes in the cell, and the sum of the aberrant activities will be sufficient to drive tumor formation.

In their seminal reviews (Hanahan and Weinberg, 2000, 2011), Hanahan and

Weinberg described a collection of behaviors that are characteristic of tumors, which they dubbed “hallmarks of cancer”. These include sustaining proliferative signaling,

5 evading growth suppressors, avoiding immune destruction, enabling replicative immortality, tumor-promoting inflammation, activating invasion and metastasis, inducing angiogenesis, genome instability and mutation, resisting cell death, and deregulating cellular energetics (Hanahan and Weinberg, 2011). The authors also describe how a cancerous cell achieves these hallmarks by accumulating numerous transformations.

Sometimes a single transformation might facilitate capabilities, and other times multiple transformations might cooperate to drive a capability that neither alone is sufficient to drive.

The multi-step nature of tumor formation does not preclude tumors from being reliant on individual genes, however. In a phenomenon termed oncogene addiction, perturbing the action of one driver of a tumor can be sufficient to induce cell death

(Weinstein, 2002). This can happen, for instance, if the targeted oncogene is at a crucial node in the cancer-specific signaling network (Luo et al., 2009). Many examples of oncogene addiction exist, for instance, transgenic mouse models of sarcoma (Jain et al., 2002), leukemia (Felsher and Bishop, 1999), and pre-malignant skin lesions

(Pelengaris et al., 1999) overexpressing Myc were dependent on continued Myc expression for continued. Similar phenomena were also seen with BCR-ABL, Cyclin-D1,

Ras, and others (Weinstein, 2002). Oncogene addiction is the rationale behind the use of targeted therapies for cancer, which treat the disease by inhibiting the function of a single gene. A tumor is not necessarily addicted to every oncogene that was a driver in its formation, but tumors can be addicted to multiple oncogenes at once (Luo et al.,

2009).

6 MicroRNA

The turn of the 21st century has seen an explosion of research on a newly discovered factor in cells: miRNAs – small, non-coding RNAs that post-transcriptionally repress expression of their target genes. miRNAs differ from canonical genes because they are not translated into proteins, but rather are functional as RNAs molecules (Lee et al., 1993; Wightman et al., 1993). They are a major player in cellular function, and over 2000 miRNAs have been discovered in humans and over 60% of human genes are predicted to be regulated by miRNAs (Friedman et al., 2009).

Discovery of miRNA

miRNA was first described in a pair of seminal papers by Victor Ambros’ and

Gary Ruvkun’s groups in 1993, examining lin-4, a regulator of the lin-14 gene in

Caenorhabditis elegans, a worm model organism (Lee et al., 1993; Wightman et al.,

1993). These papers determined that the functional gene product of lin-4 was not a protein, but rather a short RNA transcript that was complementary to regions of the 3’ untranslated region (UTR) of the lin-14 gene. They hypothesized that binding of lin-4

RNAs to these sites on the lin-14 3’ UTR caused repression through an antisense mechanism.

Seven years later, a second small non-coding RNA, let-7 was identified in C. elegans (Reinhart et al., 2000), and determined to be conserved in humans (Pasquinelli et al., 2000). Like lin-4, this gene impacted worm developmental timing. It produced small (20-21 nucleotides) RNA products, but no protein products were identified. It was complementary to the 3’ UTR of genes downstream in the pathway driving

7 developmental timing, and this complementarity was necessary for let-7 regulation

(Slack et al., 2000). After this finding, a number of studies identified an entire class of short, non-coding RNAs like these abundant in C. elegans, and conserved in other species as well, including humans, and named them miRNAs (Lagos-Quintana et al.,

2001; Lau et al., 2001; Lee and Ambros, 2001).

Biogenesis and Function of miRNAs

miRNAs can exist in the genome as independent genes or within a coding or non-coding host gene. miRNA genes are typically transcribed by RNA Polymerase II and polyadenylated and capped like mRNAs (Lee et al., 2004). The full-length transcript containing the miRNA is referred to as the primary miRNA (pri-miRNA). Within the pri- miRNA, the characteristic stem-loop structure of a miRNA forms when inverted- complementary sequences of ~20-22 nucleotides in length bind each other (Lagos-

Quintana et al., 2001; Lee and Ambros, 2001). For intragenic miRNAs, if the miRNA resides within an intron, the host gene can be spliced as usual, and both the miRNA and host gene can be expressed simultaneously. The nuclear endonuclease

Drosha/DGCR8 processes the pri-miRNA into the next stage, a precursor-miRNA (pre- miRNA), by cleaving at the base of the stem-loop structure, leaving a 2 nucleotide overhang on the 3’ side of the transcript (Lee et al., 2003).

This 2 nucleotide overhang is recognized by Exportin-5, which transfers the pre- miRNA to the cytoplasm (Lee et al., 2002; Yi et al., 2003). Here, the miRNA is further processed by a protein called Dicer, which cleaves the loop off the stem-loop structure, leaving a double stranded molecule (Bernstein et al., 2001; Grishok et al., 2001;

Hutvágner et al., 2001). One side of the stem-loop structure, the “guide” strand, is then

8 loaded into an RNA-induced silencing complex (RISC), which contains an endonuclease, Argonaute (often Ago2, but sometimes Ago1, Ago3, or Ago4, which are catalytically inactive), as well as other proteins which stabilize miRNA binding to the mRNA (Hammond et al., 2000; Khvorova et al.; Meister, 2013; Schwarz et al., 2003).

The miRNA targets the complex to the 3’ UTR of target transcripts through complementary binding (Lee et al., 1993; Reinhart et al., 2000; Slack et al., 2000;

Wightman et al., 1993).

The mechanism of repression of targets by miRNAs and the RISC complex varies based on the level of complementarity between the target and the miRNA.

Complementarity within the first ~8 bases of the miRNA, the “seed” region is highly important, while imperfect complementarity outside the seed region is more tolerated.

One mechanism of repression of targets by RISC is degradation of the target transcript by endolytic cleavage by Ago2. This occurs with high complementarity targets or when siRNAs are loaded into RISC (Hutvagner and Zamore, 2002). On the other hand, miRNAs in RISC more often recruit exolytic destabilizing proteins and drive sequestration into p-bodies which will prevent translation (Liu et al., 2005; Sen and

Blau, 2005).

Often, miRNAs function primarily to fine-tune gene expression in the cell, rather than causing major changes in expression levels. To accomplish this, they often exist in feedback loops, or two targets of a miRNA may have opposite effects on a pathway (Di

Leva et al., 2014).

9 miRNA in Cancer

Because miRNAs can act as master regulators and target many genes at once, it is perhaps not surprising that they can be involved in cancer. miRNAs can act either as oncogenes or as tumor suppressors. One of the first pieces of evidence of miRNA involvement in cancer identified miR-15 and miR-16 as tumor suppressor miRNAs in chronic lymphocytic leukemia (CLL). In this study, loss of expression of miR-15 and miR-16, rather than that of other nearby genes, was implicated in the tumor promoting effects of deleting or translocating a specific locus involved in CLL. (Calin et al., 2002).

Subsequent research identified other dysregulated miRNAs in numerous cancers, including miR-155, miR-21, miR-181, miR-17~92 and others in lymphomas (Fernandez-

Mercado et al., 2015), let-7, miR-34, miR-21, miR-17~92, miR-221/222 and others in lung cancer (Inamura and Ishikawa, 2016), miR-21, miR-155, miR-9, let-7, miR-200, and others in breast cancer (Wang and Luo, 2015), as well as other miRNAs in many additional cancers. Broader studies identified miRNA expression profiles for various types of cancer compared to normal tissue (Volinia et al., 2006). In many cases the type of cancer could be predicted based on its miRNA profile, which often held parallels to the highly tissue-specific expression patterns of normal tissue (Lu et al., 2005).

These links between miRNAs and cancer were bolstered by in vitro studies demonstrating that overexpression or knockdown of these miRNAs could affect the growth of cancer cells in culture. Then, generation of a transgenic mouse overexpressing miR-155 was the first evidence that a miRNA could be sufficient to drive malignancy, in this case a lymphoid leukemia (Costinean et al., 2006). Additional mouse models of cancer followed: overexpression of miR-21 drove hematological malignancy

10 (Medina et al., 2010), miR-17~92 accelerated Myc-induced lymphoma development (He et al., 2005), and independently drove lymphoproliferation of both B and T cells (Xiao et al., 2008), while knockout of these miRNAs decreased tumor growth in mice also bearing other oncogenes (Hatley et al., 2010; Mu et al., 2009). Mouse models also confirmed that miRNAs could act as tumor suppressors: deletion of miR-15a/16-1 led to development of a chronic lymphocytic leukemia (CLL) -like disease in mice (Klein et al.,

2010)

The molecular drivers behind miRNAs’ role in cancer are nearly as diverse as the miRNAs themselves. The dysregulation of miRNA that leads to disease can be caused by anything from chromosomal rearrangements, to epigenetic changes, to overexpression of the host gene (for intergenic miRNAs), to viral insertions, to malfunctioning miRNA processing factors, to abnormal signaling in the cell, or sponging or degradation of the miRNA (Di Leva et al., 2014). Downstream of any miRNA, there are numerous targets that may play an oncogenic or tumor suppressive role. Some miRNAs can have a different impact on tumor growth depending on the tumor type, perhaps based on the expression levels of different targets, dose-dependent effects of the miRNA, or due to varying effects on feedback loops (Narayan et al., 2018). This indicates that simply identifying key targets is not sufficient to understand the role of miRNA in cancer, but rather that identifying how the miRNA and its targets interact with the cellular context is necessary.

Clinical Relevance

Identifying and understanding the role of miRNAs in cancer can have substantial implications in the clinic. miRNAs may be able to serve as biomarkers for diagnostic or

11 prognostic purposes based on the cancer specific miRNA profiles that have been described (Lu et al., 2005; Pogribny, 2018; Volinia et al., 2006). Therapeutics that can target or mimic miRNAs are being extensively explored and have shown some promise

(Rupaimoole and Slack, 2017). Successful development and delivery of a miRNA therapeutic will be particularly exciting clinically because the chemistry and delivery method might be easily translatable other miRNA therapies by simply altering the sequence on the same backbone.

However, there are barriers to the development of miRNA therapeutics. For oligonucleotide therapeutics that inhibit miRNAs by complementary binding, stability and binding strength can be improved by changing the chemistry of the oligonucleotide backbone, for instance, to a locked nucleic acid or 2’-O-methyl group, but these changes must be optimized. For therapeutics that mimic miRNAs, these adjustments are also necessary, but must be made to the passenger strand so that the guide strand can be loaded into the RISC machinery without interference (Stenvang et al., 2012).

Even more difficult is achieving efficient, targeted, and non-immunogenic delivery of the therapeutic to the target site. Gymnotic delivery, in which the oligonucleotide is delivered naked, may result in uptake into cells, but appending additional targeting moieties to the oligonucleotide may help improve targeting and uptake. For instance, our group reported successful treatment of tumors in a transgenic mouse model using a peptide-conjugated anti-miR targeted to low pH environments like tumors (Cheng et al.,

2014). Other approaches have focused on packaging oligonucleotides in nanoparticles, as in a study by our group that used a neutral lipid emulsion to deliver a miRNA mimic delivery to lung cancer in mice (Trang et al., 2011). While more time is still necessary to

12 bring these therapeutics to patients, they highlight the importance of research that identifies and explains the role of miRNAs in cancer. miR-155

One miRNA that has been identified as a key factor in development and disease, and the focus of this research, is miR-155.

Discovery of miR-155

The non-coding RNA host gene of miR-155, BIC (now also known as

MIR155HG), was discovered before miR-155 itself, in a screen of viral insertion sites in avian lymphoma (Clurman and Hayward, 1989). Viral insertion into a site in the genome can perturb expression of key oncogenes, leading to tumor formation. Based on these findings, BIC overexpression was then shown to cooperate with Myc overexpression in formation of avian lymphoma (Tam et al., 2002). The previously uncharacterized BIC locus was hypothesized to function as a non-coding RNA, due to the absence of long, conserved open reading frames and its extensive secondary structure (Tam et al.,

1997). Further research identified homologs of BIC in humans and mice (Tam, 2001).

The miRNA, miR-155, was discovered separately in a screen of short RNAs expressed in mice, and was mapped to the BIC locus within the genome (Lagos-

Quintana et al., 2002). The connection between miR-155 and BIC was confirmed in later studies which showed that ectopic BIC overexpression leads to miR-155 overexpression because BIC gets processed into miR-155 (Eis et al., 2005).

13 Physiological Roles of miR-155

Outside of disease, miR-155 is expressed and functions primarily in the hematopoietic system (Thai et al., 2007), and has roles in immune development and inflammation. miR-155 expression is activated in response to B-cell activation by microbial ligands or cytokines. This signaling occurs through toll-like receptors, which trigger TNF expression, which in turn drives miR-155 expression (O’Connell et al., 2007;

Taganov et al., 2006). Interestingly, miR-155 participates in a feedback loop with TNF in which higher levels of miR-155 then results in even higher levels of TNF (Tili et al.,

2007). Other cytokines, including IFN-gamma are also impacted by miR-155 expression

(O’Connell et al., 2007). This relationship between the oncogenic miR-155 and inflammation has been suggested to be one key factor in the connection between inflammation and cancer (O’Connell et al., 2007; Tili et al., 2007, 2011).

The regulation of cytokine output by miR-155 means that miR-155 is required for proper immune development and function. In two mouse models deficient for miR-155,

B, T, and dendritic cells did not develop properly (Rodriguez et al., 2007; Thai et al.,

2007). In B cells, miR-155 is expressed during the germinal center response required for activation and full maturation of B cells. Without miR-155, the number of germinal- center B cells was reduced, as was antibody output (Thai et al., 2007). In hematopoietic stem cells, miR-155 can promote differentiation of specific lineages; for instance sustained overexpression of miR-155 results in expansion of the myeloid- derived granulocyte/monocyte population (O’Connell et al., 2008).

14 miR-155 in Cancer

miR-155 is considered to be the first miRNA to be identified as overexpressed in tumors. The discovery of BIC and its role in avian lymphoma led to studies of its expression levels and those of miR-155 in clinical samples of lymphomas. In diffuse large B-cell lymphomas (DLBCL), for example, miR-155 and BIC were both more highly expressed than in control B-cells. Among DLBCLs, tumors of the activated B-cell (ABC) subtype, which is associated with worse prognosis, had higher levels of expression of both transcripts than tumors of the germinal center (GC) subtype, which is associated with better prognosis (Eis et al., 2005). These findings of miR-155 overexpression in cancer were then expanded to other lymphomas, including Hodgkin’s lymphoma (van den Berg et al., 2003; Kluiver et al., 2005), to leukemias, including acute myeloid leukemia (AML) (Garzon et al., 2008; O’Connell et al., 2008), and to a number of solid tumors (Volinia et al., 2006).

The causal role of miR-155 in promoting tumor growth was confirmed with a mouse model overexpressing miR-155 from a transgene expressed in hematological tissue which developed B-cell neoplasms (Costinean et al., 2006). This was the first mouse model of overexpression of a miRNA (Di Leva et al., 2014). Another mouse model of miR-155 overexpression-driven B-cell malignancy was developed by our group, and is described in detail below (Babar et al., 2012).

Drivers of miR-155 overexpression in cancer vary. One common mechanism is activation by viral gene products in Epstein Barr virus infected cells (Elton et al., 2013).

In other cases, tumors may develop via miR-155 when it is overexpressed during

15 inflammation or in response to other TNF-alpha activity (van den Berg et al., 2003;

Pedersen et al., 2009).

Understanding of the role of miR-155 in cancer will be valuable clinically, where miR-155 has already been identified as a potential therapeutic target for anti-miRs. Our group previously used miR-155 as a test case to examine new anti-miR delivery methods (Babar et al., 2012; Cheng et al., 2014), and a clinical trial for an anti-miR-155 therapy for various lymphomas reported promising preliminary results in 2018 (Foss et al., 2018).

miR-155 Targets in Cancer

Previous studies have identified a number of tumor suppressors, including

TP53INP1 and SHIP1 as targets of miR-155, and implicated their dysregulation in driving tumor formation. TP53INP1 is a target of the powerful tumor suppressor P53. In pancreatic cancer, it is inactivated, and restoring its expression is sufficient to abolish tumor growth, while treatment of cell lines with anti-miR-155 was sufficient to restore

TP53INP1 expression (Gironella et al., 2007). SHIP1 is a phosphatase that acts as an inhibitor of the PI3K pathway, and several studies showed that it is a direct target of miR-155 in human DLBCL cell lines, transgenic mice, and in vitro and in vivo models of bone marrow macrophages (Costinean et al., 2009; O’Connell et al., 2009; Pedersen et al., 2009). SHIP1 is involved in pro- to pre- B cell maturation, and is correlated with

DLBCL prognosis (Pedersen et al., 2009). Furthermore, siRNA against SHIP1 in bone marrow macrophages caused myeloproliferative phenotypes similar to miR-155 overexpression in the same context (O’Connell et al., 2009).

16 A second class of tumor suppressors targeted by miR-155 are DNA repair factors. miR-155 targets MMR genes MLH1 and MSH6, which leads to an increase in mutation rate, a mutator phenotype (Tili et al., 2011; Valeri et al., 2010a). Additionally, repression of FOXO3a by miR-155 causes dysregulation of the high-fidelity polymerase component POLD1, further promoting mutations in the miR-155 overexpressing environment (Czochor et al., 2016). Interestingly, miR-155 also targets Activation

Induced Cytosine Deaminase (AID), which promotes mutations during antibody maturation in B-cells (Dorsett et al., 2008); in this sense, miR-155 can have tumor suppressive effects. However, in many contexts, including our mouse model described below, the net effect of miR-155 overexpression is an increase in mutation rate, not a decrease (Czochor et al., 2016).

During miR-155 overexpression, many targets of miR-155 are simultaneously dysregulated. The studies described here have examined targets individually, and do not elucidate the interactions that likely exist between miR-155 targets in tumors.

miR-155LsLtTA mouse model

Our lab previously developed one of the mouse models of miR-155 overexpression (Babar et al., 2012), which provides a powerful tool to study interactions and cooperation between miR-155 targets during miR-155-driven oncogenesis. The miR-155LSLtTA mice (referred to here as miR-155 mice) were generated by knocking in a construct containing the pre-miR-155 sequence into the ROSA26 locus. This construct contains a loxP-floxed stop cassette, which must be removed by Cre recombinase to allow transcription of the pre-miR-155. The construct also contains a tTA element, and the pre-miR-155 is downstream from a Tet-off promoter. In the absence of doxycycline,

17 tTA proteins from the tTA element bind the promoter and allow transcription of pre-miR-

155, but in the presence of doxycycline, tTA proteins are inhibited from binding the promoter, and transcription of pre-miR-155 does not occur. Overall, this construct allows for overexpression of miR-155 only in tissues that also express a Cre transgene and only when mice are not exposed to doxycycline. When the mice are crossed with a

Nestin-Cre mouse, they overexpress miR-155 in hematological tissue including spleen, bone marrow, and thymus.

When the transgene is induced from birth by removing doxycycline from their diet, the mice develop an aggressive hematological malignancy within 2-6 months.

Symptoms of malignancy include lymphadenopathy, splenomegaly, rear-limb paresis, hunched posture, and labored breathing. Histopathologic analysis identified tumor cell infiltration in the spleen, thymus, lymph nodes, bone marrow, brain, and kidney.

Analysis of tumor cells isolated from spleen or lymph nodes revealed that they are

B220+, IgM-, and CD43+, which is characteristic of pre-B cells, and that they are clonal, which is typical of human hematological malignancies. These cells are highly tumorigenic: they are transplantable to nude host mice as subcutaneous allografts and could migrate from the subcutaneous site of injection into the spleen and lymph nodes of nude host mice, resulting in splenomegaly and peripheral lymphadenopathy.

Transplantion to syngenic mice was not tested because the mice are from a mixed background due to outcrosses with Nestin-Cre mice and with wild type breeders from different backgrounds, including FVB and B6.

Perhaps the most interesting feature of these tumors is that when miR-155 overexpression is withdrawn by adding doxycycline to the diet of tumor-bearing mice,

18 the tumors undergo rapid tumor regression via apoptosis. This also occurs in subcutaneous allografts when the host mouse is fed doxycycline. Overall, this inducible system provides a powerful tool to examine the drivers of tumor formation and growth triggered by overexpression of miR-155.

Hematopoiesis and Hematological Malignancies

Research on hematological malignancies like the disease that develops in our mouse model has resulted in many therapeutic advances in recent years. In particular, these diseases have been at the forefront of immunotherapy research, thanks in part to the ease of sampling blood tumors, and in part to the inherent association between a hematological malignancy and the immune system. Understanding hematological malignancies is aided by some familiarity with the process of hematopoiesis.

Hematopoiesis

Hematopoiesis is the process by which blood and immune cells develop from a common precursor progenitor cell, the hematopoietic stem cell (HSC). HSCs are characterized by expression of CD34, c-Kit, and CD113, among others (Quesniaux et al., 2005), and have a strong capacity to self-renew and proliferate. Differentiation into various lineages is driven by cytokine signaling and changes in gene expression in the cell. The highest lineage distinction is between myeloid and lymphoid cells. Cells in the myeloid lineage include macrophages, granulocytes, mast cells, and erythrocytes; while cells in the lymphocyte lineage include B-cells, T-cells and natural killer cells. Many of these cell types require multiple stages of maturation, for instance B cells progress from pro-B cells, which are committed B-cell lineage progenitors in the process of undergoing

19 VDJ rearrangement, to pre-B cells, which express the pre-BCR and undergo massive proliferation, to immature B cells which express IgM, to mature B cells, which have undergone selection and editing (Abbas et al., 2015).

Hematological malignancies

Cancers of the hematopoietic system include leukemias, which affect cells in the blood or bone marrow; lymphomas, which affect cells in the lymphatic system; and myeloma, which affects plasma cells. Various types of each disease are defined by the lineage and stage of maturation of the malignant cell type, and each type of hematological malignancy is closely associated with certain genetic abnormalities. For instance, acute myeloid leukemia (AML) often exhibits internal tandem duplications

(ITDs) in the tyrosine kinase FLT-3, or chromosomal translocations that create an

AML1-ETO fusion gene, while frequent BCR-ABL fusions in chronic myelogenous leukemia (CML) meant that it was an early indication for use of the tyrosine kinase inhibitor, imatinib. B-cell lymphomas are associated with Myc and Bcl6 alterations, among others (Taylor et al., 2017). Some of the hematological malignancies most associated with miR-155 overexpression include diffuse large B cell lymphoma

(DLBCL), AML, acute and chronic lymphoblastic leukemia, and mantle cell lymphoma

(Ranganath, 2015).

Mouse models of hematological malignancies

Hematological malignancies and the genes involved can be studied in transgenic mouse models, like our miR-155 mice described above. Symptoms of hematological malignancies in mouse models can include splenomegaly, lymphadenopathy, and rear-

20 limb paresis. However, mouse models of hematological malignancies may not perfectly recapitulate a single human disease. For instance mice bearing a Myc transgene developed either disease resembling T cell lymphomas, or AMLs, or B cell leukemias under slightly varying conditions (Bernardi et al., 2002; Felsher and Bishop, 1999).

Therefore, we primarily focus our analysis on mechanistic rather than pathological implications of our findings, and we consider how they may relate to a variety of hematological malignancies, rather than focusing on a single human disease. c-Kit

The research described here identified mutations in the c-Kit oncogene (Kit,

CD117) as cooperating with miR-155 to drive tumors in the miR-155LsLtTA mice. The c-

Kit gene codes for a receptor tyrosine kinase that has roles in promoting growth and survival in hematopoiesis and development, as well as in cancers including gastrointestinal stromal tumors (GISTs), melanomas, and AML.

Under normal physiological conditions, Kit is expressed in hematopoietic stem cells, melanocytes, germ cells, in addition to other tissues. Expression is regulated at the promoter by transcription factors including Sp1, Ets, and Myb, and post- transcriptionally by miRNAs miR-221 and miR-222 (Lennartsson and Rönnstrand,

2012). Once expressed, Kit’s activity is triggered when its ligand, stem cell factor (SCF) binds and promotes dimerization. Upon dimerization, Kit transphosphorylates its partner on a residue in a region called the juxtamembrane domain. When it is not phosphorylated, this domain sits in the active site of the kinase domain, acting in an autoinhibitory manner, but phosphorylation causes a conformational change that

21 releases this inhibition, freeing the kinase domain to perform further phosphorylation

(Mol et al., 2004). Activated Kit can then activate many downstream pathways including

PI3K/AKT, JAK/STAT, and RAS/EGF (Larizza et al., 2005).

In cancers that depend on Kit, Kit expression levels are usually high, and activity can either be SCF-dependent as in normal cells, or it can become SCF-independent through oncogenic mutations. Specifically, there are two hot-spots for oncogenic mutations, one which corresponds to the juxtamembrane domain (exon 11), and one which corresponds to the kinase domain (exon 17). In both cases, the mutations interfere with the ability of the juxtamembrane domain to bind to the kinase domain and impart its autoinhibitory function (Lennartsson and Rönnstrand, 2012). In GISTs, most tumors have activating mutations in the juxtamembrane domain (Hirota, 1998).

Meanwhile, in AMLs, most tumors express Kit, but only a fraction, usually of a specific subtype, carry Kit mutations, which are most often in the kinase domain (Wang et al.).

The proliferative and pro-survival pathways downstream of Kit then contribute to tumor formation and progression. In AML, one study showed that signaling through STAT3 is of particular importance (Ning et al., 2001).

Targeted therapeutic options are available for patients with Kit-dependent cancers. Tyrosine kinase inhibitors including first-generation inhibitor imatinib, and second-generation inhibitor dasatinib are effective against wild type and mutant Kit.

Imatinib is only effective against juxtamembrane domain mutations, not kinase domain mutations, while dasatinib is effective against both (though a higher concentration is required against kinase domain mutations) (Heinrich et al., 2000; Schittenhelm et al.,

2006). Resistance to these treatments can arise by development of additional

22 mutations, including in the “gatekeeper residue” T670, which is a conserved residue in kinase ATP binding pockets and is a frequent site of mutations that confer drug resistance (Holohan et al., 2013; Tamborini et al., 2004).

Objectives

The aim of the research presented here is to elucidate the complex mechanisms by which miR-155 can drive tumor formation. We hope to expand on the previous research that identifies contributions from individual target genes like SHIP1 and

TP53INP of miR-155 in promoting growth, by using our miR-155 mouse model to test the hypothesis that multiple target genes must cooperate in a multi-step process to drive tumor formation. In particular, we are interested in the DNA repair factor targets of miR-

155, and whether mutations that they form are a crucial step in tumor initiation. The fact that the tumors that develop in our miR-155 mice are clonal and have a variable time-to- onset suggests that secondary mutations may indeed be necessary for tumor formation.

23 Chapter 2: miR-155 drives oncogenesis by promoting and cooperating with mutations in the c-Kit oncogene

This chapter was published in Oncogene in November 2018.

Authors: Lisa W. Witten, Chris J. Cheng, Frank J. Slack

Abstract

MicroRNAs (miRNAs) have emerged as crucial players in the development and maintenance of disease. miR-155 is an inflammation-associated, oncogenic miRNA, frequently overexpressed in hematological malignancies and solid tumors. However, the mechanism of oncogenesis by miR-155 is not well characterized, and research has focused primarily on individual, direct targets, which does not recapitulate the complexities of cancer. Using a powerful, inducible transgenic mouse model that overexpresses miR-155 and develops miR-155-addicted hematological malignancy, we describe here a multi-step process of oncogenesis by miR-155, which involves cooperation between miR-155, its direct targets, and other oncogenes. miR-155 is known to target DNA repair proteins, leading to a mutator phenotype, and we find that over 93% of tumors in our miR-155 overexpressing mice contain activating mutations in a single oncogene, c-Kit. Treating mice with dasatinib or imatinib, which target c-Kit, resulted in complete tumor regression, indicating that c-Kit activity is crucial in the oncogenic process. Interestingly, c-Kit expression is high when miR-155 is overexpressed, indicating further cooperation between miR-155 and c-Kit. Our findings support a multi-step model of oncogenesis by miR-155 in which miR-155 promotes both a mutator phenotype and a cellular environment particularly susceptible to mutations in a given oncogene.

25 Introduction

MicroRNAs (miRNAs) are small, non-coding RNAs that repress their target genes by directing a protein complex to the target through complementary binding. A single miRNA can target many genes, so over- or under- expression of a single miRNA can cause widespread genetic dysregulation, and is exhibited in many cancers, including hematological, breast, and colon cancers (Esquela-Kerscher and Slack,

2006). Many miRNAs can function as oncogenes (oncomiRs) or as tumor suppressors, but the mechanisms for this are not well understood. One such oncomiR is miR-155, which has physiological roles in hematopoiesis and inflammation (Faraoni et al., 2009;

Thai et al., 2007). It is frequently overexpressed in hematopoietic tumors including chronic lymphocytic leukemia (CLL) (Fulci et al., 2007), acute myeloid leukemia (AML)

(O’Connell et al., 2008), and diffuse large B-cell lymphoma (DLBCL) (Eis et al., 2005), as well as in solid tumors such as breast, lung, and colon cancers (Volinia et al., 2006).

miR-155 has been shown to drive hematological malignancy in two independent mouse models (Babar et al., 2012; Costinean et al., 2006). In one transgenic mouse model, developed by our group, miR-155 is overexpressed in an inducible and tissue- specific manner. When induced in hematological tissue, miR-155 overexpression drives aggressive hematological malignancy of pre-B cells in these mice within 2-6 months. By deactivating the inducible miR-155 transgene by adding doxycycline to the diet of these mice, tumor formation can be prevented and existing tumors will regress rapidly (Babar et al., 2012).

This model represents a powerful in vivo tool for studying the mechanism of oncogenesis by miR-155, which is not well understood. Previous studies have

26 implicated repression of some miR-155 target genes such as SHIP1 (Costinean et al.,

2006; O’Connell et al., 2009; Pedersen et al., 2009), TP53INP1 (Gironella et al., 2007), and AID (Dorsett et al., 2008) in oncogenesis by miR-155, but these findings do not capture the complexity of interactions that likely exist between miR-155 targets in tumors. Notably, miR-155 driven tumors in our mice are clonal and occur with highly variable time-to-onset, suggesting that development of these tumors is a multi-step mechanism requiring more than just repression of a few targets. Malignant transformation is widely accepted to occur through the accumulation of cellular abnormalities that cooperate to promote oncogenesis (Hanahan and Weinberg, 2000), so it is likely that tumors driven by miR-155 follow a similar path.

We hypothesized that additional mutations might represent the second step in the development of the miR-155-driven tumors. This is consistent with previous findings that miR-155 targets multiple DNA repair factors including MLH1, MSH6, and drives a mutator phenotype in human cell culture (Tili et al., 2011; Valeri et al., 2010a).

Additionally, in our mouse model, repression of FOXO3a by miR-155 regulates POLD1 expression levels, affecting DNA repair (Czochor et al., 2016). In this study, we demonstrate that mutations in the c-Kit proto-oncogene are selected for in miR-155- overexpressing cells. c-Kit (Kit, CD117, SCF Receptor) is a receptor tyrosine kinase frequently mutated in cancers including gastrointestinal stromal tumors (GIST), mastocytosis, and AML. Specifically, cancer-causing mutations occur in one of two regions of the protein: the juxtamembrane domain, an autoinhibitory loop that loses autoinhibitory function when altered; or the region of the kinase domain that binds the autoinhibitory loop, such that mutations activate kinase activity (Hirota, 1998;

27 Lennartsson and Rönnstrand, 2012; Mol et al., 2004). In the miR-155 overexpressing environment of our mice, c-Kit is highly expressed, and we demonstrate that c-Kit activity promotes growth by showing that inhibiting c-Kit is sufficient to cause tumor regression.

Together, these findings help elucidate the mechanism of oncogenesis by miR-

155, and reveal a multi-step process with cooperation between miR-155, its direct targets, and other oncogenes.

Results

c-Kit is frequently mutated in tumor cells from miR-155LSLtTA mice

First, we sought to identify second-site mutations that might be involved in promoting tumor formation and growth. We performed Whole Exome Sequencing

(WES) on three tumors and three control mice to identify single-nucleotide variants and small insertions or deletions unique to tumors. After performing variant calling, polymorphisms that were present in any of the three controls were filtered out to generate a list of tumor-specific variants. The most common types of mutations across the whole genome were T:A to C:G and C:G to T:A substitutions (Figure 1a). This is qualitatively similar to the pattern determined by Greenman, et al (Greenman et al.,

2007) to be associated with deficient mismatch repair which is one DNA-repair pathway targeted by miR-155 (Tili et al., 2011; Valeri et al., 2010a). We further confirmed that miR-155 targets mismatch repair in our system by testing expression of one mismatch repair gene, MLH1, in tumor cells overexpressing miR-155 compared to regressing tumors in which the transgene had been inactivated for 16 hours. We see near-

Figure 1 a. Whole exome sequencing of three flank tumors from two miR-155 transgenic mice, and three controls revealed a pattern of mutations similar to that associated with mismatch repair deficient tumors (Greenman et al., 2007). Mutation count is the sum of mutations of that class in all three tumors sequenced. “Other” includes mostly insertions and deletions but also multi-base pair substitutions. b. qPCR measurement of MH1 expression in growing subcutaneous tumors overexpressing miR-155 compared to regressing tumors undergoing miR- 155 withdrawal due to doxycycline treatment. Error bars represent SEM across the mean of three experimental replicates with three biological replicates per experiment. c. The number of genes mutated in each tumor. A single gene may have multiple mutations in a single tumor, and be counted once, and a gene with mutations in two or more tumors, even if they are different mutations, is shown in the overlapping region. d. Mutations were detected in the Kit gene of nearly all tumors analyzed, but in no control samples (n=6). Mutations refer to the corresponding human amino acid for easy comparison to published clinical data. All mutations were heterozygous.

Figure 1 (continued)

Figure 2 a. Sanger sequencing traces of genomic DNA from representative tumors with substitutions (Tumor 1 and Tumor 10), deletions (Tumor 2), and insertions (Tumor 4). The effect of the amino acid for each mutation is listed, as well as the type of mutation in the DNA. Tumor numbers correspond to the numbers referenced in Figure 1c. b Sanger sequencing traces of cDNA from the same representative tumors. Only 20% of tumors tested expressed Kit bi- allelically, while the others exhibit a single peak at the mutation site, indicating mono-allelic expression from the mutated allele.

31 significant (p<0.06) de-repression of MLH1 upon miR-155 withdrawal (Figure 1b), as expected. Together, this supports the idea that one way that miR-155 drives oncogenesis is by increasing the frequency of persistent second-site mutations in the genome.

While each individual tumor sequenced had many variants, only three genes, c-

Kit, Akap9, and Arap2, exhibited non-synonymous mutations in all three tumors sequenced (Figure 1c). To validate these initial hits, the WES reads were confirmed with Sanger sequencing, and each gene was screened for mutations by Sanger sequencing in a fourth independent tumor sample. Only c-Kit was mutated in the fourth tumor. In 11 additional tumors, Sanger sequencing revealed heterozygous c-Kit mutations in all but one (Figure 1d, Figure 2). The mouse with no c-Kit mutation also did not respond well to withdrawal of miR-155, and we hypothesize that it may have suffered from a spontaneous lymphoma, unrelated to overexpression of miR-155. In at least 6 normal spleen samples, no mutations were detected across the entire c-Kit transcript. We concluded that c-Kit mutations form de novo in these mice during induction of miR-155 overexpression.

Kit mutations in tumors from miR-155 LSLtTA mice are constitutively activating

By mapping the mutations identified in the c-Kit gene to the protein’s amino acid sequence, we determined that nearly all of the mutations in c-Kit were point mutations or small insertions or deletions (indels) affecting codon 816 or the region spanning codons 559-575 (Figure 3a), which correspond to the kinase and juxtamembrane domains, respectively, frequently mutated in cancer. The absence of frameshift, nonsense, and synonymous mutations, combined with the clustering of mutations

Figure 3 a. Domains of the Kit protein include the extracellular domain, the juxtamembrane domain, and the kinase domain. The juxtamembrane autoinhibitory loop and codon 816 of the kinase domain are commonly associated with cancer. Mutations in our mouse model occurred mainly in the juxtamembrane region and kinase domain. b. Representative western blot of phospho-STAT3 and total STAT3 in healthy B-splenocytes from two transgenic mice on doxycycline (not overexpressing miR-155), and from four independent tumors.

33 around known-activating regions strongly suggests that activating mutations are being selected for.

To confirm whether c-Kit is active, we tested phosphorylation levels of Stat3 in tumors and healthy B cells by western blot. Stat3 is a well-established target of c-Kit’s phosphorylation activity (Ning et al., 2001). In tumors, phospho-Stat3 is high, while it is nearly absent in healthy B cells, despite similar levels of expression of the total protein

(Figure 3b). Together, this supports that when c-Kit is expressed, it is constitutively activated, and that cells with activated c-Kit are selected for during miR-155 overexpression.

c-Kit inhibition is sufficient for tumor regression in miR-155 mice

We hypothesized that activating c-Kit mutations were selected for because they contribute to tumor growth. To test this hypothesis, we tested the effect on tumors of small molecule tyrosine kinase inhibitors dasatinib and imatinib, which block c-Kit activity (Heinrich et al., 2000; Schittenhelm et al., 2006). We generated subcutaneous allografts from tumors from four independent miR-155 transgenic mice and tested each treatment on one allograft from each parental tumor. This allowed both quantitative measurement of tumor size and also direct comparison of drug treatment to vehicle treatment using tumor cells bearing the same c-Kit mutation. Once tumors reached

~150 mm3, we initiated daily treatment with intraperitoneal dasatinib, vehicle, or chow- fed doxycycline to induce miR-155 withdrawal in one experiment; or with intraperitoneal imatinib or vehicle in a second experiment, and measured tumors daily. As expected, tumors in mice treated with the vehicle grew rapidly, and tumors in mice fed doxycycline regressed rapidly (Figure 4a). Both dasatinib (Figure 4b-c) and imatinib (Figure 4d)

Figure 4 a, b, d. Relative tumor size (vs. first day of treatment) of subcutaneous tumors in NSG mice during treatment with doxycycline in the mouse chow, dasatinib (10mg/kg/day via intraperitoneal injection) or imatinib (100 mg/kg/day via intraperitoneal injection). N=4 for each treatment, with tumors derived from four independent transgenic mice. Error bars represent SEM between mice. c. Photos of the same mouse with subcutaneous tumors regressing during treatment with dasatinib. e. Relative tumor size (vs. first day of treatment) of subcutaneous tumors transplanted from dasatinib-resistant tumors during treatment with dasatinib and doxycycline. n=3 for each treatment, with cells derived from three separate tumors that developed dasatinib resistance. Error bars represent SEM between mice. f. Dasatinib treatment is effective at reducing lymphadenopathy in transgenic mice overexpressing miR-155, as well as inducing regression in subcutaneous tumors as in (b, c). g. Relative number of cells with cleaved Caspase 3/7, identified with NucView530, after in vitro treatment with doxycycline or dasatinib for 24 h. Error bars represent SEM between four tumor lines in a representative experiment. h. Representative results from flow cytometry of differentiation markers CD43, IgM, and CD34 in cells from subcutaneous tumors treated with vehicle, doxycycline, or dasatinib for 16 h.

Figure 4 (continued)

36 caused tumor regression. Regression caused by treatment with dasatinib occurred at a similar rate to regression during miR-155 withdrawal in the doxycycline treated mice, while regression with imatinib treatment was slightly slower.

We continued treatment with dasatinib and doxycycline after tumors regressed.

Within 30-40 days, tumors regrew on four out of four mice treated with dasatinib, suggesting that the tumors acquired resistance to dasatinib. None of the mice treated with doxycycline had tumors regrow for the three months they were followed. We performed Sanger sequencing of the c-Kit transcript from cDNA of these tumors and detected no new mutations (data not shown). To determine whether resistance to dasatinib also conferred resistance to miR-155 withdrawal by doxycycline, we collected cells from dasatinib-resistant tumors and injected them as subcutaneous allografts in two mice per resistant tumor. After tumors formed, each mouse was treated with dasatinib or doxycycline. Tumors on mice treated with dasatinib continued to grow, while tumors on mice treated with doxycycline regressed (n=3 per group), demonstrating that dasatinib resistance was not sufficient to eliminate the tumors’ dependence on miR-155 (Figure 4e).

We also tested whether the initial sensitivity to tyrosine kinase inhibitors extends beyond subcutaneous allograft tumors to malignancy in the miR-155 transgenic mice.

We monitored three transgenic mice with induced miR-155 expression for symptoms of malignancy. At the first sign of disease, such as dragging of the hind limbs (paresis), we initiated daily treatment with dasatinib and monitored symptoms. We saw mild recovery of hind limb function, shrinking of lymphadenopathy (Figure 4f), and qualitatively longer health span. Life span could not be quantifiably compared to untreated mice because

37 animals were sacrificed for welfare reasons when symptoms worsened, based on a subjective evaluation of symptoms. Therefore, we concluded that Kit activity, likely from mutational activation, drives tumor growth in miR-155 mice.

Finally, we developed an in vitro system to test whether the effects of dasatinib were cell autonomous, like the effects of miR-155 withdrawal. First, we optimized growth conditions and confirmed that established phenotypes of the cells were not compromised by growing the cells in culture (Figure 5). We confirmed that dasatinib inhibited Kit activity by performing a Western blot for phospho-STAT3 and for phospho-

KIT in cells treated for 8 hours with vehicle and with dasatinib (Figure 6f). Kit auto- phosphorylates, so phospho-KIT, like phospho-STAT3, is an indicator of Kit activity.

Then, we measured apoptosis, which we have previously established is the primary route of tumor regression during miR-155 withdrawal from doxycycline treatment (Babar et al., 2012). In cells treated with dasatinib, apoptosis, measured with NucView530, a cleaved caspase marker, was about 1.5 times higher than in cells treated with vehicle.

This matches the trend with doxycycline treatment (Figure 4g).

In vivo, neither dasatinib nor doxycycline induced differentiation nor de- differentiation of the cells. Differentiation would be indicated by loss of CD43 expression or gain of IgM expression by flow cytometry, while de-differentiation would be indicated by CD34 positive cells (Figure 4h), but the percentage of cells staining positive for each marker was not significantly different from vehicle (p > 0.05) for cells treated with either doxycycline or dasatinib.

Together, this suggests that Kit activity drives tumor growth in miR-155 mice through the same pro-proliferative/anti-apoptotic mechanisms as miR-155, and that both

Figure 5 a. In the growth conditions identified (see Methods), cells actively proliferated for up to 72 or 96 hours, as is visibly apparent in images of cell density. b. Despite the addition of growth factors to the media, cells did not differentiate, and expressed the same cell surface markers (Kit+, CD43+, IgM-, CD34-) in the absence (top panels) and presence (bottom panels) of all growth factors. c. The miR-155 transgene retained sensitivity to doxycycline treatment when cells were grown in culture. Knockdown of miR-155 plateaued at 2ug/mL of doxycycline. d. Growth sensitivity to doxycycline was also retained when cells were grown in culture. Similar to miR-155 expression, effects plateaued at 2ug/mL of doxycycline.

39 effects happen in a cell autonomous manner, indicating that the two oncogenes may exist in the same pathway.

c-Kit expression is high in the miR-155 overexpressing environment

To understand how miR-155 overexpression contributes to the selection of mutations in c-Kit rather than in other oncogenes, we examined gene expression in the miR-155 overexpressing environment. First, we characterized expression of c-Kit in the miR-155 overexpressing environment compared to healthy B-cell splenocytes by qPCR and Western blot. c-Kit expression is nearly undetectable at both the RNA and protein level in healthy B-cell splenocytes, but comparatively high in tumor cells where miR-155 is overexpressed (Figure 6a, b). Interestingly, Sanger sequencing of cDNA from tumors revealed that expression of c-Kit is mono-allelic from the mutated allele in over 80% of tumors (Figure 2), despite copy number remaining unchanged compared to healthy littermates (n=16) (Figure 7). When c-Kit is not expressed, even if there is a constitutively activating mutation, there will be no activity. The absence of protein expression makes it irrelevant whether the protein would be activated or not if expressed. Therefore, miR-155 overexpression is likely involved in the selection of c-Kit mutations. With no miR-155 overexpression, c-Kit would not be expressed, thus there would be no selection pressure for c-Kit activating mutations.

To verify whether the high c-Kit expression can be attributed to the miR-155 overexpressing environment, rather than associated with other signaling that develops during malignant transformation, we compared c-Kit expression in tumor cells overexpressing miR-155 to that in tumor cells undergoing miR-155 withdrawal. Our previous RNA sequencing data (Cheng et al., 2014) compared gene expression in

Figure 6 a. c-Kit expression by qPCR in healthy B-splenocytes (CD19+) and in tumors. n=3 for healthy and n=7 for tumors. Error bars represent SEM. b. Representative Western blot of total c-Kit expression in healthy cells and of subcutaneous tumors treated with vehicle or doxycycline for 16h. Vehicle- and doxycycline-treated tumors are paired subcutaneous tumors derived from the same transgenic mouse. c. Volcano plot of RNAseq data from Cheng et al (Cheng et al., 2014). Fold change is calculated as regressing/growing tumors, so a negative log fold change refers to a gene that is downregulated during miR-155 withdrawal. d. Representative c-Kit expression by qPCR in subcutaneous tumors treated with vehicle or treated with doxycycline for 16 h (in vivo) or in cells treated in culture with vehicle or doxycycline for 10 h (in vitro). Error bars represent SD between three tumor lines for in vivo and between four tumor lines for in vitro. e. Representative histogram of c-Kit (CD-117) expression by flow cytometry in cells grown in culture and treated with doxycycline or vehicle for 6 h. f. Representative Western blot showing expression and phosphorylation of c-Kit and Stat3 in cells treated in culture with vehicle, doxycycline, or dasatinib for 8 h

Figure 6 (continued)

Kit Copy Number 1.5

1.0

0.5 Fold Change

0.0 Tumor Healthy

Figure 7. Relative copy number of the Kit gene in healthy splenocytes (n=3) and in tumors (n=16). The Kit locus does not appear to be amplified in any of the tumors tested.

43 tumors with the miR-155 transgene induced to gene expression in tumors 16 hours after turning off expression of the miR-155 transgene by adding doxycycline to the mice’s chow. From this data (Cheng et al., 2014), we determined that c-Kit had the most statistically significant change in expression of any gene during miR-155 withdrawal: a

19-fold down-regulation during withdrawal with p=9.5×10-28 (Figure 6c). We validated this down-regulation with qPCR from tumor samples undergoing miR-155 withdrawal at

16 hours. c-Kit expression was decreased by 10- fold upon miR-155 withdrawal compared to vehicle treated samples (Figure 6d). Western blot of tumors overexpressing miR-155 and undergoing 16-hour miR-155 withdrawal reveals that these findings hold at the protein level as well (Figure 6b).

We also tested c-Kit expression by qPCR in our in vitro system and saw 13-fold downregulation of c-Kit mRNA after 10 hours of treatment with doxycycline (Figure 6d).

In cells derived from splenocytes of mice negative for the Cre-recombinase transgene, c-Kit expression was not affected by treatment with doxycycline, indicating that repression is mediated by miR-155, not an off-target effect of doxycycline (Figure 8).

Flow cytometry of cells treated in vitro with doxycycline reveals that protein expression of c-Kit is also repressed within 6 hours of treatment (Figure 6e). This is accompanied by a decrease in c-Kit signaling, similar to that seen with dasatinib treatment: a

Western blot of samples harvested 8 hours after treatment with doxycycline shows that p-STAT3 is decreased while total STAT3 stays approximately the same. There is an absence of p-KIT, which is unsurprising due to the absence of total KIT (Figure 6f).

Together, this demonstrates that miR-155 withdrawal causes a decrease in c-Kit expression and that this is likely cell autonomous.

Kit expression in mice without Cre transgene 1.5

1.0

0.5 Fold Change

0.0 Vehicle Doxycycline

Figure 8. qPCR for Kit expression in splenocytes derived from mice bearing the miR-155 transgene, but not the Cre transgene, which is required to activate miR-155 expression in a tissue-specific manner. Cells were grown in culture in the presence of doxycycline or vehicle for 10 hours after being harvested from mice. The difference is not significant, indicating that doxycycline does not have an off-target effect on Kit expression.

45 These findings suggest that miR-155 promotes tumor formation not only by causing a mutator phenotype that results in a high mutation rate, but also by creating a gene expression environment that is particularly susceptible to malignant transformation because expression of an oncogene, c-Kit, is high. This is consistent with our previous finding that these miR-155 tumors are addicted to miR-155 overexpression, and that tumors regress when miR-155 overexpression is withdrawn (Babar et al., 2012), despite the c-Kit mutation remaining unchanged (data not shown). In this model, withdrawal of miR-155 overexpression causes a decrease in expression of c-Kit, thereby eliminating c-Kit activity, which is sufficient to induce tumor regression.

Direct targets of miR-155 include known regulators of Kit

We examined the behavior of genes that are known to be direct targets of miR-

155 during miR-155 withdrawal in order to understand whether they might drive the high expression of c-Kit that occurs in the miR-155 overexpressing environment. We used

Ingenuity Pathway Analysis to identify direct miR-155 targets that also directly regulate c-Kit. (Figure 9a). Based on this analysis, we identified Spi1/PU.1, Socs1, Myb, and

Inpp5d/Ship1 as miR-155 direct targets of interest. We further considered additional criteria to narrow down this list: first, we tested whether expression of these genes was affected by miR-155 in these tumors; second, we considered whether the established mechanism of action of these genes on c-Kit would affect expression or just activity.

We measured expression of these four genes by performing RT-qPCR on RNA from growing and regressing tumors after 16 hours of treatment with doxycycline to induce miR-155 withdrawal. All four genes exhibited approximately two- to four- fold de- repression upon miR-155 withdrawal, with statistically significant (p<0.05) or near-

Figure 9 a. Potential pathways between miR-155 and Kit. SPI1 encodes the PU.1 protein, and INPP5D encodes the SHIP1 protein. b. qPCR of potential mediators of the miR-155-Kit axis reveals that they are de-repressed during treatment of subcutaneous tumors with doxycycline, supporting that they may play a role in regulating Kit. Error bars represent SD of the means across three experimental replicates with three tumor lines per replicate. c. Model for the multi- pronged mechanism of oncogenesis by miR-155.

Figure 9 (continued)

48 significant (p<0.06) changes in all genes except Ship1 (Figure 9b). Of these genes, Myb and Spi1/PU.1 are transcription factors able to regulate expression of their target genes, while Socs1 and Inpp5d/Ship1 inhibit the activity of c-Kit, but not the expression.

Because c-Kit expression is repressed during miR-155 withdrawal, it is more likely that the effects of miR-155 on c-Kit are driven by Myb or PU.1. Unfortunately, we have been unable to functionally validate these relationships due to inability to manipulate the primary murine pre-B tumor cells with transfection or transduction despite attempting multiple various approaches (See Chapter 3).

Discussion

MicroRNAs have recently been identified as potent oncogenes (Esquela-

Kerscher and Slack, 2006), with repression of individual tumor suppressor target genes implicated in the oncogenic process. Here, we present evidence that oncogenesis by miR-155 can occur through a more complex, multi-step process in which multiple miR-

155 targets cooperate to promote malignant transformation. In this model, miR-155 promotes a mutator phenotype, allowing formation of activating mutations in c-Kit.

These mutations are then selected for because they confer a proliferative advantage in the context of the miR-155 overexpressing environment where Kit expression is also high (Figure 9c).

We identified likely-activating mutations in c-Kit in miR-155 driven tumors using a powerful mouse model of inducible miR-155 overexpression. We demonstrated that the tumors are dependent on c-Kit activity because they regress rapidly when treated with tyrosine kinase inhibitors, dasatinib and imatinib. Activating mutations in c-Kit are a well-

49 established driver of oncogenesis in human disease, including in hematological malignancies, most notably AML (Lennartsson and Rönnstrand, 2012). In human AML, miR-155 overexpression and activating c-Kit mutations can co-occur, as exemplified by the well-studied AML cell line, Kasumi-1, which has a constitutively activating c-Kit mutation (Larizza et al., 2005) as well as high miR-155 expression (Klimenko and

Shtilman, 2013).

Our result appears counterintuitive at first, because we have previously shown that the tumors are addicted to miR-155. If the role of miR-155 is only to promote mutations, then once a mutation has formed, miR-155 would not be further involved in tumor growth. To address this apparent discrepancy, we examined how the miR-155 overexpressing environment might cooperate with c-Kit mutations to together promote tumor growth. A recent study in mouse models and clinical samples of AML revealed a cooperative mechanism between miR-155 and another receptor tyrosine kinase, FLT3, supporting that cooperation between miR-155 and other oncogenes may be key to understanding oncogenesis by this miRNA (Wallace et al., 2018).

We determined that expression of c-Kit is absent when miR-155 is not overexpressed, and therefore even an activating mutation cannot drive c-Kit activity in the absence of miR-155 expression. This effect of miR-155 on c-Kit may be mediated by transcription factors Myb or PU.1, which are miR-155 targets known to regulate Kit.

Myb binds directly to the c-Kit promoter at two sites and can have context-specific repressive or activating effects on Kit expression (Ratajczak et al., 1998; Vandenbark,

GR; Chen, Yi; Friday, Ellen; Pavlik, Kevin; Anthony, Bruce; deCastro, Carlos; Kaufman,

1996). PU.1 is also known to cooperate with c-Kit. In a PU.1 overexpressing mouse

50 model of erythroleukemia, 86% of the mice exhibited activating Kit mutations, similar to our model where PU.1 is downregulated (Kosmider et al., 2005). PU.1 affects hematopoiesis and leukemia in a dose- and context- dependent manner. Low levels of

PU.1 drive erythroid differentiation, and in erythroid cells, PU.1 acts as an oncogene. In the myeloid and lymphoid lineages, high levels of PU.1 are necessary for proper differentiation, but PU.1 acts as a tumor suppressor (Kastner and Chan, 2008).

Cooperation between Kit, a proliferation driver, and PU.1, a differentiation factor, is consistent with other findings that leukemias require alteration in both a proliferative factor and in a differentiation factor (Kosmider and Moreau-Gachelin, 2006).

This two-step model of oncogenesis by miR-155 may have implications for oncogenesis by other miRNAs as well. For instance, miR-21 is another highly oncogenic miRNA known to target mismatch repair factors (Valeri et al., 2010b), so it may promote oncogenesis through a similar multi-step mechanism. For miR-155 and other miRNAs, elucidating how they contribute to oncogenesis is crucial to understanding disease and therapeutics. MicroRNAs may present a viable target for therapeutics (Rupaimoole and Slack, 2017), and our findings that inhibition of miR-155 is less likely to drive resistance than kinase inhibitors like imatinib and dasatinib encourages further exploration of this therapeutic modality.

51 Chapter 3:

Unpublished Results

Some experiments in this chapter were performed by Rinni Bhansali and Jihoon Lim, under the supervision of Lisa Witten

Introduction

In the previous chapter, we identified one mechanism by which miR-155 can drive tumor formation, that is, by promoting mutations and simultaneously promoting a gene expression environment in which mutations in one particular oncogene, Kit, are particularly oncogenic. These results introduce some additional questions that, if answered, would enhance our understanding of this complex mechanism, or expand this understanding beyond these particular conditions. Specifically, these questions include: Under what conditions can this mechanism occur, and can human cells meet these conditions? What factors mediate the relationship between miR-155 and Kit expression, and how can we experimentally demonstrate their role? What other mechanisms of oncogenesis by miR-155 might exist?

Here, we describe preliminary data that begins addressing these questions and presents avenues of study for future research in addressing them further.

Results

miR-155-independent silencing of Kit in healthy adult mice

To fully understand the contexts in which the mechanism of oncogenesis by miR-

155 that we described can occur, we first examined the role these mechanisms might play in tumor dynamics in adult mice. In our experience, miR-155LsLtTA mice develop tumors when miR-155 overexpression is induced at birth. But, if the transgene is not 53

Figure 10 a. Expression of miR-155 measured by qRT-PCR from whole spleen of adult transgenic miR-155 mice on or off doxycycline for >1 month. n=3 for doxycycline, n=2 for vehicle, error bars represent SD between mice. b. Expression of c-Kit (x-axis) in CD43 positive or negative cells (y-axis) measured by flow cytometry of cells from whole spleen of the same mice tested in Fig. 1a. Each plot represents one mouse

54 induced until mice are adults, while miR-155 does get overexpressed, mice to not develop disease. To understand how the adult context might prevent our proposed mechanism of oncogenesis by miR-155 from driving tumor formation in adults, we examined adult mice overexpressing miR-155. We harvested spleens from 3 adult mice off doxycycline for at least one month and therefore overexpressing miR-155 (but otherwise healthy), and two control mice on doxycycline. First, we confirmed that miR-

155 was overexpressed in these adult, transgene-induced mice by performing qPCR on

RNA from whole spleen (Figure 10a). We note that whole spleen includes a variety of cell types including pre-B cells that are likely the precursors for tumors in younger mice, but also other cells types such as red pulp cells, which may not overexpress miR-155.

Then, we performed flow cytometry for CD43 and for CD117 (Kit) on these cells. CD43 is a marker of pre-B cells, which is the cell type most similar to tumors from the miR-155 mice. Kit was not up-regulated in either the CD43+ or CD43- populations in miR-155 induced cells compared to non-induced cells (Figure 10b).

This absence of Kit expression even while miR-155 was overexpressed is similar to the wild type allele of Kit in many of our tumors. In most cases, while the activating Kit mutation was heterozygous on the DNA, expression was only from the mutated allele

(Figure 2). Together with our finding that miR-155 cannot induce Kit overexpression in adult mice, we hypothesize that Kit may be epigenetically silenced at some point during development, and that when this occurs, miR-155 can no longer affect its expression levels. Epigenetic silencing of Kit would only not occur when there is a strong growth advantage to select against it. Another alternative is that mutations occur in the promoter region on only one allele that encourage or interfere with regulation of c-Kit

55 expression. Future studies could confirm whether and when Kit gets epigenetically silenced by assaying for DNA methylation or repressive histone marks.

Investigation of miR-155 and c-Kit in human cells

Besides understanding how our proposed mechanism of oncogenesis by miR-

155 behaves in adult mice, we also sought to understand whether this mechanism might be relevant in human cells. To study this, we selected an acute myeloid leukemia cell line, Kasumi-1, that is known to have an activating mutation in Kit (Larizza et al.,

2005), like in the tumors from our miR-155 mice. We sought to understand whether miR-155 regulates Kit expression in these cells, as it does in tumors from our mice.

First, we confirmed that these cells, like our mouse tumor cells, expressed high levels of Kit (Figure 11a) and miR-155 (Figure 11b) compared to CD34+ hematopoietic stem cells. We also confirmed that growth of Kasumi-1 cells was inhibited during treatment with Kit inhibitor dasatinib by showing that cell number was decreased and the proportion of dead cells was increased using cell counting with Trypan Blue exclusion (Figure 11c). Then, we optimized transfection conditions of these cells using electroporation and confirmed that anti-miR-155 reduced the repressive activity of miR-

155 on a co-transfected luciferase reporter, but not on a reporter with a mutated miR-

155 binding site (Figure 11d). Using this optimized protocol, we transfected Kasumi-1 cells with anti-miR-155 and a scrambled anti-miR control and measured gene expression and apoptosis. The anti-miR-155 treatment did not result in any significant changes in cell death, nor did it result in changes in expression of Kit by qPCR (Figure

11e, f). However, it also did not result in changes in expression of a control target gene,

TP53INP (Figure 11f), so we cannot determine with certainty whether the absence of

Figure 11 a-b. Expression of c-Kit (a) or miR-155 (b) measured by qRT-PCR in Kasumi-1 cells or in RNA obtained from CD34+ cells (n=1 per group). c. Cell count of live and dead cells in Kasumi-1 cells 96 hours after treatment with dasatinib in vehicle (n=1 per group). d. Ratio of renilla luciferase activity to firefly luciferase activity in Kasumi-1 cells at 48 hours after treatment with anti-miR-155 or a scrambled control. Luciferase is generated by a psicheck2 reporter plasmid with a miR-155 binding site in the 3’UTR of the renilla luciferase. Error bars represent SD between 3 technical replicates per group. e. Percentage of Kasumi-1 cells positive for cleaved caspase 3/7 after transfection with anti-miR-155 or a scrambled control. Apoptosis is high at day 2 because electroporation induces cell death. Error bars represent SEM between 4 technical replicates per sample. f. Expression of c-Kit and Tp53inp measured by qRT-PCR in Kasumi-1 cells 5 days after treatment with anti-miR-155 or a scrambled control. Error bars represent SD between 4 technical replicates per group.

Figure 11 (continued)

58 phenotypic effect is due to miR-155 not regulating Kit or whether it is due to technical failure.

Investigation of transcriptional regulation of Kit expression by miR-155

In order to further understand the conditions in which miR-155 can regulate Kit expression, we sought to further understand what other factors are involved in this pathway. Our analysis described in Chapter 2 identified four potential mediators of the miR-155-Kit regulation axis. These included two transcription factors and two regulators of Kit activity. We hypothesized that miR-155 is most likely to target a gene that affects

Kit transcription because Kit RNA levels are altered, not just protein levels or activity levels. As a further step toward understanding this pathway, we tested this hypothesis by performing chromatin immunoprecipitation (ChIP) for RNA Polymerase II along the

Kit promoter in two pairs of tumor cells overexpressing or miR-155 or undergoing miR-

155 withdrawal. We found that at multiple points along the Kit promoter, and especially closest to the transcription start site, RNA Pol II levels were higher in tumor cells overexpressing miR-155 than in tumor cells treated with doxycycline that were no longer overexpressing miR-155. This is consistent with RNA Pol II on the promoter of the transgene, which serves as a positive control because doxycycline treatment directly targets transcription of the transgene. However, we also saw lower levels of RNA Pol II occupancy on a control gene, RPL30 (Figure 12a), which could indicate that pulldown efficiency was lower for the doxycycline treated cells. On the other hand, because miR-

155 regulates so many genes and can have such a profound effect on the gene expression environment in the cell, it is also possible that RPL30 is genuinely deregulated during treatment with doxycycline.

Figure 12 a. Occupancy levels of RNA Pol II at various locations on the genome in cells from subcutaneous tumors on nude mice treated with or without doxycycline for 16 hours. Kit promoter late is closest to the transcription start site, while Kit promoter early is upstream from that. Error bars represent SD between two pairs of mice. b. Kit expression measured by qRT- PCR in cells treated for 16 hours in vitro with or without doxycycline and with or without TNF added to the media. n=1.

60 Pursuing the hypothesis that RPL30 is genuinely de-regulated during treatment with doxycycline, and that miR-155 regulates Kit by affecting transcription levels, we sought to perform gain- and loss- of function studies to identify transcription factors that are necessary or sufficient for regulation of Kit. To start, we examined the effects of

TNF, which activates transcription factor NF-kb. TNF did not appear in our initial search of potential mediator genes between miR-155 and Kit (Figure 9a), possibly because

TNF is not a canonical target of miR-155, but rather is activated by miR-155 (Tili et al.,

2007). Nonetheless, TNF is both directly regulated by miR-155 and known to regulate

Kit expression (Khoury et al.), so it is a viable candidate for a mediator gene.

To test whether TNF plays a role in the regulation of Kit by miR-155, we added exogenous recombinant TNF to the media of tumor cells grown in vitro with and without doxycycline and measured Kit expression by qPCR. Our RNAseq data (Figure 6c) showed that TNF expression typically decreases during miR-155 withdrawal, so if TNF is a mediator gene, then addition of exogenous TNF would be expected to rescue the effects of doxycycline treatment on Kit expression, resulting in high Kit expression even with doxycycline treatment. However, Kit expression was substantially lower in doxycycline treated cells than in vehicle treated cells for samples either with or without

TNF (Figure 12b). Therefore, we conclude that TNF is not likely to contribute to regulation of Kit by miR-155.

To examine other potential transcription factor mediators of Kit regulation requires knocking down expression of the miR-155 target gene during doxycycline treatment, which would be expected to rescue the repression of Kit typically seen during doxycycline treatment.

61 Cells from miR-155LsLtTA mouse tumors are resistant to transfection and genetic manipulation

In order to further understand the pathways involved in oncogenesis by miR-155, we attempted to knock down factors such as Myb and PU.1 in the miR-155 tumor cells by transfecting siRNA or transducing shRNA against these factors. Hematological cells are notoriously resistant to transfection by traditional methods, and we found that using an siRNA with transfection reagents including Lipofectamine RNAiMax and FuGeneHD, we were not able to knock down even a control gene, GAPDH, by a sufficient amount to perform loss of function experiments (Figure 13a).

We were able to obtain ~10% transfection of a GFP reporter plasmid using electroporation after optimizing conditions. We used Amaxa Cell Line Nucleofector Kit V with the Amaxa Nucleofector V on program Z001 to obtain the best ratio of transfection rate and cell toxicity. However, even in these optimized conditions, the treatment was too toxic to the cells to perform downstream assays.

Viral transduction has traditionally been more successful in hematological cells, and we were able to deliver a GFP-tagged shRNA lentiviral construct to the cells using the PLL3.7 backbone with psPAX2 and pMD2.G. Unfortunately, we were not able to obtain sufficient knockdown of the target genes using this construct in an experiment designed to measure the effect of shRNAs against Myb or PU.1 on subcutaneous tumor growth and response of tumors to doxycycline. In this experiment, miR-155 tumor cells were treated in vitro with lentivirus containing shRNA against Myb, PU.1, or a scrambled control. Cells were then injected as subcutaneous allografts into the flanks of two nude mice per shRNA and allowed to grow without doxycycline. Once the tumors were

Figure 13 a. Expression of Gapdh measured by qRT-PCR in miR-155 tumor cells transfected with either siRNA against Gapdh or a scrambled siRNA using RNAiMax or FuGeneHD transfection reagents. b. Percentage of cells gated as GFP+ by flow cytometry. shRNA infected miR-155 cells were grown as subcutaneous tumors in nude mice who were fed with or without doxycycline for 16 hours before harvesting tumors. Each bar represents one tumor c. Tumors of shRNA infected cells were measured before treatment with or without doxycycline, and measured again after 16 hours. Fold change is measured relative to the tumor size before treatment. Error bars represent SD between two tumors per group. d. Percentage of cells gated positive for Kit expression from tumors with shRNA against PU.1, Myb or a scramble shRNA, measured by flow cytometry. Error bars represent SD between two tumors per group. e. PU.1 or Myb expression measured by qRT-PCR in tumor cells containing shRNA or a scrambled shRNA, and treated with or without doxycycline. RNA was isolated from the entire tumor population, including GFP+ cells that got infected and GFP- cells that were not infected. Error bars represent SD between two tumors per sample.

Figure 13 (continued)

64 grown, one mouse bearing each shRNA was treated with doxycycline. Tumors were measured before treatment and 16 hours after treatment, at which point tumors were harvested and flow cytometry was performed to measure GFP and Kit expression.

Flow cytometry revealed ~10% to ~80% GFP positive cells (Figure 13b), indicating that the transduction was stable in cells and that cells were capable of proliferating after transduction. If the shRNA affected Kit expression and therefore tumor growth, we would expect the following: 1) tumor regression would be reduced or abolished in tumors bearing targeted shRNAs compared to tumors bearing scrambled shRNAs. We saw similar levels, or even increased of tumor regression in the mice with the targeted shRNAs as with the scrambled shRNA (Figure 13c). 2) For targeted shRNAs, the percentage of GFP positive cells would be higher in the mouse treated with doxycycline (where the shRNA would be protecting only the transduced cells from apoptosis), than in the untreated mice. We saw differences in the percentage of GFP positive cells that were consistent between the scrambled control and the targeted shRNAs (Figure 13b). 3) GFP positive cells containing the shRNA would have lower levels of Kit expression than GFP negative cells in the same tumor, which would not contain the shRNA. In all tumors, we saw similar levels of Kit expression in GFP positive and negative cells, regardless of the tumor was treated with doxycycline. (Figure 13d).

These results suggest that Myb and PU.1 may not be the factors responsible for regulating Kit expression. However, qRT-PCR of the total cell population revealed little to no difference in expression of Myb in cells with shRNA against Myb compared to cells with the scrambled control (Figure 13e). Therefore, it is likely that Myb was not sufficiently knocked down to drive any downstream phenotype.

Figure 14 Western blot showing levels of phospho-Kit and total Kit in dasatinib resistant tumor cells treated in vitro for 8 h with doxycycline, dasatinib, or vehicle, compared to the parental (dasatinib-sensitive) tumor cells. Each line underwent treatment with all conditions, and the different experimental conditions for each line are paired together in adjacent lanes. (Western blot performed by Jihoon Lim)

66 Alternative pathways may be available downstream of miR-155 in dasatinib-resistant tumors

Besides gaining a better understanding of how miR-155 can drive oncogenesis through

Kit, our results in Chapter 2 suggested the question whether miR-155 might be able to drive oncogenesis in a similar manner but through other oncogenes. Specifically, the data in Figure 4e, which reveals that tumors can still be dependent on miR-155 expression even in conditions in which they are no longer dependent on Kit activity, suggests that miR-155 is still driving tumor growth, but must be using another oncogenic pathway. We took preliminary steps to determine whether this might be the case.

First, we further examined the dasatinib-resistant tumor lines generated in

Chapter 2 to rule out the possibility that Kit activity was no longer responding to dasatinib due to mutation or other causes. We performed western blots on the three dasatinib-resistant tumor lines each treated in vitro with vehicle, doxycycline, or dasatinib, along with the single parental tumor line that each resistant line was derived from. We found that phospho-Kit levels were decreased in each tumor during treatment with dasatinib compared to treatment with vehicle. This decrease was of a similar magnitude to the response in the parental line, and was similar to the treatment with doxycycline in all lines after accounting for changes in total Kit expression, which may be due to a feedback loop response to Kit inhibition with dasatinib (Figure 14). This preliminary data suggests that the continued tumor growth that occurs during dasatinib treatment but not during doxycycline treatment may be due to another oncogene downstream of miR-155, rather than to Kit activity that evades inhibition by dasatinib.

Figure 15 a. Copy number analysis of Myc in tumors and healthy cells. n=3 for healthy and n=16 for tumors. Error bar represents SD. b. Average tumor size of four tumors each normalized to their own size on day zero before treatments with JQ1 or vehicle. Mice treated with vehicle in this experiment were also used as a comparison for treatment with imatinib in Figure 4d.

68 As a preliminary indicator of other possible oncogenes or bypass pathways through which miR-155 might be driving growth, we revisited our RNAseq data (Figure

6c) and recognized that Myc is the second-most significantly dysregulated gene during miR-155 withdrawal in these tumors, and therefore has the capacity to act in a similar manner to Kit. We took preliminary steps to understand the behavior of Myc in these tumors. In human cancers, Myc is often overexpressed due to copy number amplifications (Kalkat et al., 2017). We tested whether this was the case in our dasatinib-sensitive tumor cells by measuring copy number in tumors and healthy cells.

We did not detect any amplification of Myc in these tumors compared to healthy cells

(Figure 15a), which is consistent with miR-155 playing a post-transcriptional role in regulating Myc expression.

Furthermore, we tested whether tumors are sensitive to a Brd4 inhibitor molecule, JQ1, capable of inhibiting Myc expression (Delmore et al., 2011). We treated mice bearing subcutaneous miR-155 tumors with JQ1 or vehicle and measured tumor growth over time. Tumors grew rapidly with both JQ1 and vehicle treatment (Figure 15b) and did not regress, indicating that while tumors are addicted to Kit activity, targets of

JQ1 are not required for growth. Future experiments could confirm whether Myc is among targets knocked down by JQ1 in this context, and evaluate whether the role of

Myc in tumor growth changes in dasatinib-resistant tumors.

The miR-155 host gene, BIC, may play a role in oncogenesis by miR-155

Finally, we wanted to explore additional alternative ways in which miR-155 might promote tumor growth in various contexts. The host gene of miR-155 is a long-non- coding RNA called MIR-155HG or BIC. miR-155 is encoded in one of the exons of BIC.

69 Frequently, in disease contexts where miR-155 is overexpressed, like (DLBCL), BIC is also overexpressed (van den Berg et al., 2003; Eis et al., 2005; Kluiver et al., 2005).

Interestingly, unlike many other pri-miRNAs which exist transiently at low levels in the nucleus, BIC is frequently expressed and exported to the cytoplasm, where it is no longer a viable precursor for miR-155, because processing of pri-miRNAs requires

Drosha, which is a nuclear protein (Eis et al., 2005; Lee et al., 2003). Therefore, we hypothesized that BIC might play a role in tumor growth independent of serving as a precursor of miR-155.

To test this, we developed vectors to overexpress full length BIC or BIC with the miR-155 sequence deleted (BICdel). After transfecting these vectors into the human

DLBCL line, U2932, we measured BIC expression, cell growth, miR-155 expression, miR-155 activity, and expression of potential target genes. Potential target genes were selected by identifying genes that correlated (Spearman’s rho) with BIC expression in the TCGA dataset of DLBCL samples, but not with miR-155 expression, then prioritizing by known associations with lymphogenesis.

BIC expression was higher in cells transfected with either the BIC or BICdel plasmids than in cells transfected with an empty vector (Figure 16a), indicating that transfection was successful. Cells with either plasmid also grew faster than cells with the empty vector, after an initial die-off due to toxicity from electroporation (Figure 16b).

To determine whether miR-155 played a role in this accelerated growth, we measured miR-155 expression and activity, using a dual luciferase reporter. In cells transfected with BIC, as expected, miR-155 expression and activity are elevated. In cells transfected with BICdel, we also saw slight increases in miR-155 expression and activity

Figure 16 a. BIC levels measured by qRT-PCR in U2932 cells electroporated with plasmids containing the full BIC sequence (BIC), the BIC with the miR-155 sequence deleted (BICdel), or the empty vector. Error bars represent SD between three independent experiments b. Relative cell number measured by cell counting with Trypan Blue Exclusion every 24 hours after electroporation with BIC, BICdel, or empty vector plasmids. Error bars represent SEM between three independent experiments. c. miR-155 expression measured by qRT-PCR in cells electroporated with BIC, BICdel, or empty vector. d. Relative amounts of miR-155 activity measured as the ratio of renilla luciferase to firefly luciferase in cells electroporated with BIC, BICdel, or empty vector, along with a miR-155 psicheck sensor. Error bars represent SD between three independent experiments e. CFLAR expression in cells electroporated with BIC, BICdel, or empty vector. Error bars represent SD between three independent experiments. (All experiments performed by Rinni Bhansali)

Figure 16 (continued)

72 suggesting a potential feedback loop between BIC expression and miR-155 (Figure

16c,d). Finally, qPCR of a potential target gene, CFLAR revealed down-regulation in cells treated with either BIC or BICdel (Figure 16e). Due to the elevated miR-155 expression and activity in BICdel cells, genes regulated just by miR-155 might be expected to have lower expression levels in both conditions, but to different magnitudes.

CFLAR on the other hand, was repressed to similar levels in both conditions, suggesting BIC might be involved in its regulation.

Discussion

Oncogenesis by miRNAs is a complex and poorly understood process. In

Chapter 2, we presented a model for oncogenesis by miR-155, and here, we share preliminary efforts to further elucidate this model and to understand how it might be applied to a variety of contexts.

First, we examined whether miR-155 regulates Kit expression in healthy adult mice and in human tumor cells. In both cases, our results suggested that the regulation of Kit by miR-155 that we see in tumors in our mice might be restricted to that context.

In the healthy adult mice, we hypothesized that Kit was epigenetically silenced during development and therefore not able to be regulated by miR-155 in more mature mice.

Future experiments comparing this effect to Kit regulation by miR-155 in young mice not bearing disease will be valuable to confirm whether this lack of regulation of Kit by miR-

155 is due to the non-malignant state or the age of the mice. In the human cell line, while our negative results may have been the result of technical complications, it is also possible that different regulatory pathways would exist if miR-155 wasn’t overexpressed

73 during the formation of the tumor, or wasn’t driving the formation of the tumor. Notably, in addition to bearing a Kit mutation, Kasumi-1 cells also have a t(8;21) translocation, which is common for Kit-mutant AMLs, and thought to be a precursor to the development of Kit mutations (Larizza et al., 2005). Therefore, it is unlikely that miR-155 was the initial driver of tumor formation in Kasumi-1. These examples highlight that the role of miR-155 in promoting tumor formation may differ from its roles in promoting tumor progression.

A third role for miR-155, besides promoting tumor formation and tumor progression, may be in promoting drug resistance to traditional targeted therapies, especially if miR-155 operates upstream from the target. We confirmed that resistance to dasatinib described in Chapter 2 was likely not due to reduced sensitivity of Kit activity to dasatinib. We hypothesized that miR-155, upstream of Kit, was able to promote oncogenic signaling through another pathway, like Myc activation. In human tumors even where miR-155 is not initially a driver of tumor formation, if it is overexpressed during drug treatment, it may play a role in driving and defining the mechanisms of resistance by promoting formation of additional mutations in a unique gene expression environment. This has strong implications for the development of therapeutics, and encourages the development of therapeutics that target miRNAs.

Finally, we explored a hypothesis for an alternative mechanism of oncogenesis by miR-155 by examining the role of its long non-coding RNA host gene, BIC in oncogenesis. Our preliminary data suggests that cytoplasmic BIC could act oncogenically, perhaps through independent mechanisms or by participating in a feedback loop with miR-155. Going forward, for a comprehensive understanding of the

74 roles of miR-155 in promoting tumor formation, progression, and drug response, it will be important to further elucidate whether and how its host gene is also contributing and how the effects of the two interact.

75 Chapter 4:

Methods

Portions of this chapter are directly from or adapted from the publication of Chapter 2 in

Oncogene in November 2018

Animal care

All mice were maintained at BIDMC in accordance with Institutional Animal Care and Use Committee guidelines. Generation of the miR-155LSLtTA mice was previously described (Babar et al., 2012). To induce overexpression of miR-155, mice were fed standard chow from birth unless otherwise specified; all other mice were maintained on doxycycline-impregnated food (with 200-mg DOX per kg of chow, Bio-Serv). For flank tumors, 4-6 week old female NSG mice (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ from

Jackson Laboratories) or 4-6 week old female nude mice (J:NU 007850 from Jackson

Laboratories) were used.

Tumor isolation and flank tumor establishment and measurement

Subcutaneous flank miR-155 tumors were established as previously described

(22). Briefly, enlarged spleens from symptomatic miR-155 mice, were dispersed into a single cell suspension in 5% fetal bovine serum (FBS) in PBS, and filtered through a

100um filter. Red blood cells were lysed with ammonium chloride. Cells were pelleted for biochemical analysis (eg. protein, DNA, or RNA analysis), or five million cells were subcutaneously injected into the flank of NSG mice. Tumors could be further passaged by dissecting the flank tumor and processing cells in the same manner. Tumors were palpable within 10 days. Tumor volume was calculated as (length x width2)/2, and tumors were measured by caliper. For blinding, measurements were taken and recorded before checking ear tags and without referring to previous measurements.

77 Drug treatment

Dasatinib, imatinib, and JQ1 were administered intraperitoneally. Dasatinib was administered daily at 10 mg/kg/day. Dasatinib was stored in a 10mM stock solution in

DMSO and diluted in PBS for administration. Imatinib was administered daily at 100 mg/kg/day in PBS. JQ1 was a gift from the Bradner Lab via Karen Cichowski. It was administered at 45 mg/kg/day in 10% 2-hydroxypropyl beta cyclodextran. Doxycycline to induce miR-155 withdrawal was administered in the chow as described above. For tumor regression curves, each group of subcutaneous tumors included four mice bearing tumors derived from four independent miR-155 transgenic mice. Four mice was determined to be sufficient to detect regression phenotypes after a pilot experiment with three mice per group. No mice were excluded, though successful engraftment was a pre-established criterion for inclusion. Treatment groups were assigned randomly by the experimenter.

Whole exome sequencing

Whole exome sequencing was performed on three tumor samples and three healthy spleen samples from close relatives (parents or littermates). Tumor samples were derived from three subcutaneous flank tumors established from two diseased spleens. Genomic DNA (gDNA) was extracted from a single cell suspension using the

DNeasy Blood and Tissue Kit (Qiagen) and following the manufacturer’s protocol.

Library prep was performed by Beckman Coulter Genomics using the SureSelect

Mouse All Exon Target Enrichment Kit (Agilent). Paired-end sequencing was performed to 100X by Beckman Coulter Genomics. Variant calling was performed by the Harvard

78 Chan Bioinformatics Core (Harvard T.H. Chan School of Public Health, Boston, MA) using the bcbio-nextgen Cancer Variant Calling Pipeline.

qPCR

Total RNA was extracted from single cell suspensions from tissue or cell culture using Trizol according to standard procedures or using the Direct-zol RNA Mini-Prep

Plus Kit (Zymo Research) according to the manufacturer’s protocol. Reverse transcription was performed with 400ng of RNA using SuperScriptIV Reverse

Transcriptase (ThermoFisher Scientific) for gene expression, or miScript II Reverse

Transcription Kit (Qiagen) for miRNA expression, each according to manufacturer’s protocols. All qPCR was performed with a Roche LightCycler 480 System. For gene expression, LightCycler 480 SYBR Green I Master Mix (Roche) was used according to the manufacturer’s protocol with primers listed in Table 1. Gene expression was normalized to the geometric mean of 18S and HPRT. For miRNA expression, miScript

SYBR Green PCR Kit (Qiagen) was used with QuantiTect Primer Assays (Qiagen) and normalized to U6. For gene expression in regressing tumors (subcutaneous tumors, or in vitro cultures treated with doxycycline or dasatinib), expression was further normalized to the paired vehicle-treated sample from the same tumor line.

Representative plots are representative of at least three separate experiments. Fold change was calculated as 2^(-ΔΔCt). For statistical analysis, two-tailed T-tests were performed on the ΔCt values, in a paired manner for regressing tumors. For copy number analysis, gDNA was extracted with the DNeasy Blood and Tissue Kit (Qiagen), according to manufacturer protocol, and was analyzed with TaqMan Copy Number

Table 1 Forward and reverse primers used to measure expression or enrichment of each gene

qPCR Primers Forward ATGAAGACTTGCTGGGACG Kit Reverse GGGTTCTCTGGGTTGGG Forward TGTCCTCSSSGCCTTTACCG Myb Reverse TTCACGTSTTTCCGAGCCG Forward CAGGTCAGAGGCAACGCTAA PU.1 Reverse ACCATTTGTTACACCTCTCCAGTCA Forward GAACCATGGCAACATCACCC INPP5D Reverse ACACAATTCCGGAACAGCAC Forward SOCS1 Qiagen Primer Assay Reverse Forward GTTGGGCTTACCTCACTGCT HPRT Reverse TCATCGCTAATCACGACGCT Forward GTAACCCGTTGAACCCCATT 18S Reverse CCATCCAATCGGTAGTAGCG Forward Kit (Human) Qiagen Primer Assay Reverse Forward TP53INP (Human) Qiagen Primer Assay Reverse Forward GAGATGGCTCTAATGGTGGCA BIC Reverse GCAAGCCTTCAGCACTCAGA Forward AGAGTGAGGCGATTTGACCTG CFLAR Reverse GTCCGAAACAAGGTGAGGGTT Forward CCGATGGAGGGAGAGTGCTA Kit Promoter Early Reverse CTGTGCCCTCTAAGACCAGG Forward CAGAGTCTAGCGCAGCCACC Kit Promoter Late Reverse CTTGTCTGCCGCTCTTTCCC Forward ATCCCTTATCCTCTGGCTGC Transgene Promoter Reverse AGGCCAAAACCCCTATCACA

Table 2 Forward and reverse primers used to amplify and sequence each gene or transcript

PCR and sequencing primers Forward CTGCTCATTGGCTTTGTGGT Kit gDNA, juxtamembrane region Reverse TCTTTCTGCTAACCCTGCCA Forward TTCACAGAGATTTGGCAGCC Kit gDNA, kinase region Reverse ARRRCAAACGAAGCGGGGAG Forward GAGGAGCTCAGAGTCTAGCG Kit cDNA 1 Reverse TGCACACCACCGTAAATGTG Forward CGGTACATGGCTGCATTCTG Kit cDNA 2 Reverse TGTACGTCCACTGGTGAGAC Forward GCTTCCGTGACATTCAACGT Kit cDNA 3 Reverse AGCTCAGGACCTTCAGTTCC Forward CCCGACGCAACTTCCTTATG Kit cDNA 4 Reverse AAATGCTCTCTGGTGCCATC Forward TCCTCGCCTCCAAGAATTGT Kit cDNA 5 Reverse ATACAGTTCGGACCTTCGCA Forward TGAGCAGAAACCCAAGTCCA Kit cDNA 6 Reverse ACAAAGGCTGACTAGGGAGG Forward GCTGAGAAGGAACAAGCCAG Kit cDNA 7 Reverse CTTCACTCCGCTAAGCCTCT Forward GTGTGGGTGAGTTGTGTTGG Kit cDNA 8 Reverse GGGACTCATGGGCTCAGG Forward AAAGGCAGCAGGTGTACAAG AKAP9 Reverse ATGAACTCAGGCCTGCAGAT Forward ACAGGGCGATGTCTCAGTAG ARAP2 Reverse ATGAACTCAGGCCTGCAGAT

81 Assays and TaqMan Universal PCR Master Mix (ThermoFisher Scientific), according to the manufacturer’s protocol.

RNA from CD34+ cells for use as a comparison control was purchased from

Miltenyi Biotec.

PCR and Sanger sequencing

PCR was performed according to standard protocols with Phusion High Fidelity

Polymerase (New England Biolabs) on gDNA or cDNA, each obtained as described above. Cycling conditions were as follows: 98° for 30 sec; 30° cycles of 98° for 5 sec,

62° for 20 sec, 72° for 20 sec; 72° for 5 min. Primers are listed in Table 2. PCR products were purified with the PCR Purification Kit (Qiagen), according to manufacturer protocol.

Sanger sequencing was performed by GeneWiz with the forward or reverse PCR primer as the sequencing primer.

Western Blot

Western blotting was performed according to standard protocols. Single cell suspensions of cells were lysed in RIPA buffer with TritonX, then run on an SDS/PAGE gel (Bio-Rad Criterion gels) in Tris-Glycine buffer and transferred to a nitrocellulose membrane. Blocking was performed in 5% BSA in TBS-T. Antibodies used were as follows: Cell Signaling Technologies: Stat3 (124H6) Cat. 9145, p-Stat3 (Y705)

Cat.9139, c-Kit (D13A2) Cat. 3074, p-c-Kit (Y703) Cat. 3073; Santa Cruz

Biotechnologies: beta-actin (sc-47778). Representative Western blots represent at least three samples per condition (samples from separate mice, or cells derived from independent tumor lines) blotted at at least two separate times.

82 Flow Cytometry

Single cell suspensions from tumors or cell culture were stained with antibodies for 30 min at 4 C after 10 min incubation in mouse FcR Blocking Buffer (MACS Miltenyi

Biotec). Flow cytometry was performed on a Gallios Flow Cytometer (Beckman

Coulter). Antibodies were as follows: BD Biosciences: PE Rat Anti-Mouse CD117 (2B8)

Cat. 553355, AlexaFlour 647 Rat anti-mouse CD34 (RAM34) Cat. 560233, FITC Rat

Anti-Mouse IgM (II/41) Cat. 553437, PerCP-Cy 5.5 Rat Anti-mouse CD43 (S7) Cat.

562865. Cytometry was performed on cells from at least four independent tumor lines

(treatment groups paired by tumor line), at separate times. Paired, two-sided T-tests were performed on the percent of cells gated positive for a given marker.

Cell culture

Cells isolated from tumors as described above were cultured in Myelocult H5100 media (Stem Cell Technologies) supplemented with 1uM hydrocortisone (Stem Cell

Technologies), 25ng/mL FLT-3 Ligand, 50ng/mL IL-6, 50ng/mL Thrombopoietin,

10ng/mL IL-7, 50ng/mL SCF, 20ng/mL IL-3. All growth factors are recombinant mouse proteins from R&D Systems. Cells were plated at 150,000-300,000 cells/mL and grew in a loosely adherent manner. Assays and treatments were initiated 0-48 hours after plating cells, and cells were not passaged.

Kasumi-1 was cultured in 20% FBS in RPMI at a density of 500,000 – 2 million cells per milliliter using standard methods. U2932 was cultured in 10% FBS in RPMI at a density of 500,000 – 2 million cells per milliliter using standard methods.

83 Cell Culture Treatments

Doxycycline, dasatinib, and TNF treatments were performed by adding the drug directly to the media. Treatments of miR-155 tumor cells with doxycycline were at

2ug/mL, treatment with dasatinib was at 50nM, and treatment with TNF was at 20 ng/mL unless otherwise specified. Doxycycline and dasatinib were suspended in DMSO and TNF was suspended in 0.1% BSA in PBS. Experiments were excluded if viability precluded attaining adequate sample for assays (eg. too low RNA/protein concentration), or if viability interfered with control phenotypes (eg. miR-155 expression changes with doxycycline treatment).

Electroporation was performed using the Amaxa Nucleofector Kit V with an

Amaxa Nucleofector II electroporator, according to the manufacturer’s protocol. Five million cells were used per reaction in 100uL of nucleofector buffer with 100nM of siRNA against Kit, Myb, or PU.1 (Dharmacon ON-TARGET Plus) or anti-miR-155 (Ambion miRvana) or 2 ug of plasmid DNA. Programs used were as follows: miR-155 cells, Z-

001; Kasumi-1 cells, P-019; U2932, X-001.

As a method trial, miR-155 tumor cells were transfected with GAPDH siRNA using Lipofectamine RNAiMax or FuGeneHD, according to manufacturer’s protocols.

For RNAiMax, for each well of a 12 well plate, 2uL of 10uM siRNA and 6uL of RNAi Max was used. For FuGeneHD, 2uL of 10uM siRNA and 8uL of FuGeneHD was used.

For shRNA transductions, first virus was generated by transfecting HEK293T cells with 10ug of PLL3.7 containing shRNA against Myb, PU.1 or a scrambled control,

5ug PMD2.G, and 5ug of psPAX2 using Fugene6. Virus-containing media was harvested every 24 hours for 72 hours. Virus was concentrated 1:60 through a

84 molecular weight cutoff filter, and miR-155 tumor cells were treated in vitro with concentrated virus at a 1:1 ratio of virus to media with 4ug/mL polybrene overnight before changing media. Cells were then injected subcutaneously into mice 48 hours after infection.

In vitro viability and apoptosis assays

Growth and viability of miR-155 tumor cells in vitro was monitored by counting using an automated cell counter (Nexcelom Cellometer Auto 1000) on cells mixed in a

1:2 ratio with Trypan Blue. Early apoptosis (as Caspase 3/7 activity) was measured 24 hours after doxycycline or dasatinib administration using NucView530 Caspase substrate (Biotium Inc), which fluoresces when cleaved by Caspase 3 or 7. Cells were stained with a 1:500 dilution of substrate and 1:1000 dilution of Hoescht dye for 30 minutes, then imaged and analyzed with a Celigo plate imager and software (Nexcelom) using the Expression Analysis/Target 1 + 2 + Mask application. Assays were performed at least three times with 1-3 biological replicates per experiment for Chapter 2.

Chromatin Immunoprecipitation

ChIP was performed using the SimpleChIP Sonication Chromatin IP Kit with

Magnetic Beads (Cell Signaling Technologies) according to the manufacturer’s protocol.

20 million cells were harvested into a single cell suspension directly from subcutaneous tumors of mice treated with or without doxycycline (n=2 per condition, paired based on parental tumor) for 16 hours then fixed in 1% formaldehyde for 30 minutes in 40 mL of total volume. Sonication was performed for 6.5 minutes with 1 second on, one second off at 50% amplitude. For the pulldown, 3ug of chromatin was used per IP. The lysate

85 was pre-cleared for 20 minutes with magnetic beads, and incubated with 10% Mouse

FcR Blocking Buffer (Miltenyi Biotec) for 10 minutes prior to adding 0.5ug/mL of antibody. Antibodies were as follows: RNA Pol II 8WG16 (SantaCruz); negative control mouse IgG (Cell Signaling Technologies); positive control Histone (included in ChIP

Kit). Antibodies were incubated overnight before adding magnetic beads. Washes were performed as recommended in the manufacturer’s protocol. qPCR was performed as described above and percent input was calculated as 2*2^(Ct2%input – CtIPsample ). Primers are indicated in Table 1. Percent input was calculated as 2*2^(Ct2%input-Ctpulldown).

Luciferase Assays

Dual Luciferase assays were performed with the Promega Dual Luciferase Assay

Kit according to the manufacturer’s protocol on a GloMax Explorer. For suspension cells, pelleted cells were incubated in passive lysis buffer for 5 minutes and 20uL of lysate was added to each well of a 96 well plate in triplicate. 50uL of each luciferase reagent was used per well. The plate reader was set to wait 2 seconds after injection, and to use a 10 second integration for 1 read.

Plasmids

The BIC plasmid was manufactured by Biomatik in the pcDNA3.1(+) backbone by inserting the spliced BIC sequence between the NheI and XbaI restriction sites.

BICdel was generated by site directed mutagenesis using QuikChange Lightning kit

(Agilent). Sequences were confirmed by Sanger Sequencing (GeneWiz).

Table 3 Forward and reverse DNA oligos annealed for cloning into the PLL3.7 vector

Plasmid Inserts Forward TGGCTCCTGATGTCAACAGAGAACTCGAGTTCTCTGTTGACATCAGGAGCTTTTTC Myb shRNA1 Reverse TCGAGAAAAAGCTCCTGATGTCAACAGAGAACTCGAGTTCTCTGTTGACATCAGGAGCCA Forward TGCCATCTTTAGAACTCCAGCTACTCGAGTAGCTGGAGTTCTAAAGATGGTTTTTC Myb shRNA2 Reverse TCGAGAAAAACCATCTTTAGAACTCCAGCTACTCGAGTAGCTGGAGTTCTAAAGATGGCA Forward TGGAGCTATACCAACGTCCAATGCTCGAGCATTGGACGTTGGTATAGCTCTTTTTC PU.1 shRNA1 Reverse TCGAGAAAAAGAGCTATACCAACGTCCAATGCTCGAGCATTGGACGTTGGTATAGCTCCA Forward TGGATGTGCTTCCCTTATCAAACCTCGAGGTTTGATAAGGGAAGCACATCTTTTTC PU.1 shRNA2 Reverse TCGAGAAAAAGATGTGCTTCCCTTATCAAACCTCGAGGTTTGATAAGGGAAGCACATCCA Forward TGCCCTAAGGTTAAGTCGCCCTCGCTCGAGCGAGGGCGACTTAACTTAGGTTTTTC Scramble shRNA Reverse TCGAGAAAAACCTAAGTTAAGTCGCCCTCGCTCGAGCGAGGGCGACTTAACCTTAGGGCA

87 shRNA plasmids were generated by inserting annealed oligos (synthesized by

IDT) into the PLL3.7 backbone between the HpaI and XhoI restriction sites. Inserts were as shown in Table 3.

The miR-155 luciferase reporter plasmid was generated by inserting a miR-155 sequence by annealed oligos (synthesized by IDT) into the psiCheck2 vector backbone between the AsiSI and PmeI restriction sties.

88 Chapter 5:

Discussion

Cooperation between miRNA targets in oncogenesis

Since their discovery in the 1990s and early 2000s, miRNAs have emerged as players in diverse biological processes from development to inflammation to disease. In cancer in particular, miRNAs can be diagnostics, prognostics, drivers of disease, preventers of disease, therapeutics, and therapeutic targets (Pogribny, 2018;

Rupaimoole and Slack, 2017).

Understanding the particular role that any miRNA plays in a tumor can be complex: even something as simple as whether a miRNA promotes or inhibits tumor growth can vary depending on the cellular context, and each miRNA has many targets, on which it may have a small or large effect, that may mediate its role in the disease.

For several well-studied oncogenic miRNAs, identification of strong tumor suppressors as targets of the miRNA has informed some understanding of the mechanism by which it drives oncogenicity.

However, to develop a deeper understanding of the mechanism by which a miRNA can drive cancer, and particularly be sufficient to cause a tumor to form, we hypothesized that cooperation between multiple miRNA targets would be crucial to the process. In support of this hypothesis, we used a transgenic mouse model overexpressing the oncogenic miR-155 to identify that two targets or families of targets,

DNA repair factors and c-Kit expression regulators, cooperate to activate oncogenic signaling from c-Kit. In this model system, signaling from c-Kit requires both mutational activation of the kinase and high levels of c-Kit expression, and neither is sufficient for continued tumor growth without the other. It is particularly interesting to note that one of the key targets of miR-155 in this model system, the DNA repair factors, is something

90 whose effect is irreversible upon miR-155 withdrawal, yet the requirement for cooperation with the c-Kit expression regulators makes the effects of miR-155 reversible.

The results of the studies described here provide a proof of concept for a model of oncogenesis by miRNAs, or indeed by any cellular factor with many diverse targets, in which one target shapes the cellular context (eg. which genes are being expressed), thereby affecting whether or how strongly another target can be oncogenic. This represents an extrapolation of the long-understood concept that multiple alterations in a cell must accumulate for full-blown oncogenesis (Hanahan and Weinberg, 2000, 2011).

In our model, multiple of these alterations are driven by the same oncogene, and each alteration informs or influences the other. Specifically, both mutation of c-Kit and expression of c-Kit are driven by miR-155. Expression of c-Kit dictates whether the c-Kit mutations are oncogenic, and therefore selected for, and conversely the oncogenicity of these c-Kit mutations may select against epigenetic silencing of c-Kit as the immune system matures.

While our findings may be useful to understanding mechanisms of oncogenesis by miR-155, it may also have implications for oncogenesis by other miRNAs. For instance, miR-21 is another miRNA whose targets include DNA repair factors (Valeri et al., 2010b). miR-21 is perhaps one of the most oncogenic miRNAs, and in one study on miRNA profiles in six major cancer types was found to be overexpressed in every cancer type tested (Volinia et al., 2006). Our mouse model of miR-21 overexpression exhibits clonal tumors with a delayed and varied time-to-onset (Medina et al., 2010), suggesting that, like in our miR-155 tumors, a secondary mutation might be necessary

91 for tumor formation. Using our model for the mechanism of tumor formation by miR-155 could guide future studies on the mechanism of oncogenesis by miR-21 and others.

Even for miRNAs that do not inhibit DNA repair or otherwise affect mutation frequency, our model highlights the importance of considering how multiple targets of a miRNA might cooperate in oncogenesis. Understanding this cooperation could have implications for identification of appropriate therapeutic targets downstream of miRNAs and for the likelihood of a tumor to develop drug resistance. miR-155 and c-Kit in cancer

While our model may prove useful in informing mechanisms of oncogenesis by a variety of miRNAs in a variety of cellular contexts, as discussed, it is also valuable to evaluate whether our particular findings about the relationship between miR-155 and c-

Kit can translate to cancers in a more clinical context.

While miR-155 and Kit are each involved in a number of tumor types, one particular tumor type of interest is acute myeloid leukemia (AML), where both miR-155 is commonly overexpressed, and Kit is commonly mutated. AML is a cytogenetically diverse malignancy, characterized by particular recurring mutations or translocations.

Some of the most common genetic alterations include mutations in NPM1 or CEBPA, activating internal tandem duplications (ITDs) in the tyrosine kinase FLT3, and translocations like t(8;21) which creates an AML1-ETO fusion. The type of genetic alteration can be prognostic and can inform treatment decisions. FLT3-ITD AMLs are associated with poor prognosis, and t(8;21) AMLs with Kit mutations have poorer prognosis than those without (Döhner et al., 2015).

92 miR-155 overexpression in AML is particularly common in FLT3-ITD AMLs

(Alemdehy et al., 2016; Faraoni et al., 2012). In fact, FLT3-ITD can induce miR-155 expression through its regulation of STAT5 and NF-kB (Gerloff et al., 2015), and meanwhile miR-155 expression can cooperate with FLT3-ITD to promote malignancy.

One mechanism for this cooperation involves evasion of the anti-proliferative effects of bone marrow interferon signaling. One target of miR-155, CEBPB, is implicated in the interferon response, so when it is repressed by miR-155, anti-proliferative interferon signaling has less effect on the tumor cells (Wallace et al., 2018). In simpler potential mechanisms, miR-155 and FLT3-ITD each contribute to the dual features required in the two-hit model of leukemogenesis (Kosmider and Moreau-Gachelin, 2006): a block in differentiation, and proliferative capacity. miR-155 can promote proliferation by targeting

Ship1 and others, while FLT3-ITD blocks myeloid differentiation by inhibiting CEBPA; or miR-155 can block myeloid differentiation by targeting PU.1, while FLT3-ITD promotes proliferation by activating pathways like MAPK/ERK and PI3K/AKT (Alemdehy et al.,

2016; Gerloff et al., 2015). Understanding the role of miR-155 in AML is additionally complicated by findings that in some contexts, miR-155 can act as a tumor suppressor and promote apoptosis (Schneider et al., 2016). This may be due to a dose-dependent effect of miR-155 (Narayan et al., 2018).

Kit in AML, on the other hand, is rarely associated with FLT3-ITDs. It is notable that these two tyrosine kinases target similar pathways, and therefore it is unsurprising that alterations might be mutually exclusive. Like FLT3-ITD, Kit mutations are not typically sufficient for tumorigenesis and instead co-occur with AML1-ETO fusions in t(8;21) AMLs (Döhner et al., 2015). Up to 50% of t(8;21) AMLs have Kit mutations, but

93 Kit overexpression is also common even in wild-type Kit samples. In these cases, the translocation is likely to occur first, and contribute to a block in differentiation, while Kit mutations occur later and promote proliferation (Wang et al.). Notably, one way that

AML1-ETO inhibits differentiation is by binding to and down-regulating PU.1, which is also a key target of miR-155 (Larizza et al., 2005), supporting that miR-155 and Kit may cooperate.

The studies of miR-155 in FLT3-ITD AMLs described above have examined the role of miR-155 in promoting tumor progression once FLT3-ITD has already developed, in contrast to the current study which assumes that overexpression of miR-155 is the first abnormality in the cell, and therefore examines its role in promoting tumor initiation.

One key difference is that when miR-155 is overexpressed first, the mutations it is most likely to occur with are small point mutations or indels that could be caused by its repression of mismatch repair genes and homologous recombination, rather than larger duplications like the ITDs seen in FLT3.

Aside from their shared role in AML, a relationship between miR-155 and Kit in cancers can also be examined by considering other factors in the cell that they share interactions with and which play a role in hematological malignancies like we see in our mouse model. We identified some of these factors, including PU.1, Myb, Ship1, and

Socs1 in Figure 9a.

It will be interesting to further illuminate the mediator genes and the relationship between miR-155 and Kit expression in our mouse model system in order to understand how these relationships might translate in to human cancers that overexpress miR-155 or have activated Kit. This will have implications for therapeutic selection. For instance,

94 if this relationship holds, then miR-155 levels might serve as a good biomarker for sensitivity to dasatinib. On the other hand, if this relationship does not hold in human cancers, it will highlight the importance of further examination of the cellular context in which miR-155 becomes overexpressed and promotes cancer. In our model system, the development and selection of Kit mutations is highly dependent on the cellular context –

Kit must not be epigenetically silenced, and must be highly expressed via miR-155. In a different cellular context in human tumors that form after the hematopoietic system has developed, perhaps other mutations will be selected for, and different miR-155 targets will cooperate in promoting selection of these mutations, and understanding these contextual factors can lead to identifying these other key mutations and targets.

Transfection of hematological cells

Throughout this thesis we have discussed technical barriers to further identification of the genes that mediate the relationship between miR-155 and Kit expression. While we attempted to manipulate cells using RNAi methods in a unique cell type of transformed, primary, pre-B murine cells, our inability to transfect them is a common barrier in research with hematological cells, especially primary cells (Seiffert et al., 2007).

One of the most common laboratory techniques to genetically manipulate adherent cells, for instance cells from solid tumors, is to use a liposomal transfection. In this technique, DNA or oligonucleotides are packaged in a small lipid-based particle, which the cell absorbs by endocytosis, resulting in delivery of the nucleic acid to the cell

(Felgner et al., 1987). This technique is ineffective on cells that grow in suspension like

95 hematological cells, likely due to lower levels of association between the cell and the liposomal particle (Basiouni et al., 2012).

One of the techniques that has been most effective for genetic manipulation of hematological cells is to transduce them using a viral delivery vector like a lentivirus or retrovirus. Engineering virus pseudotypes to be specific to the cell type being transduced can be crucial in facilitating successful transduction, for instance, measles virus glycoproteins may be more efficient for transduction of B cells than vesicular- stomatitis-virus-G (Lévy et al., 2010). In contrast to transduction of adherent cells, transduction of suspension cells often requires higher titers of virus and/or spinoculation protocols.

Electroporation, in which an electric pulse is applied to the cells to cause a disruption to their membrane structure and allow uptake of extracellular DNA or oligonucleotides, has also been effective in some suspension cells (Seiffert et al., 2007).

However, electroporation is often highly toxic to cells.

It will be prudent for the advancement of the field to continue the development of new technologies that allow simple, rapid, and cost-effective genetic manipulation of suspension cells. miRNAs and drug resistance

In this thesis, we described development of drug resistance to a traditional targeted therapy, dasatinib, in a miRNA driven tumor. Targeted therapies are designed to activate or inhibit a particular tumor suppressor or oncogene that the cancer cell will be sensitive to. This reduces the toxicity of therapies compared to older

96 chemotherapeutic reagents by minimizing the effect of the therapy on healthy cells that do not have aberrant expression or activity of the tumor suppressor or oncogene.

However, by targeting a single factor within the cancer cell, targeted therapeutics are susceptible to development of resistance.

Drug resistance to a targeted therapy can develop either because the target of the drug is no longer sensitive to the drug (eg. the drug cannot bind) or because the tumor is no longer sensitive to the activity of the target. The latter can occur either because downstream factors in the target’s pathway are activated independently, or because the tumor started using other, “bypass,” signaling pathways to promote growth

(Pagliarini et al., 2015). In either case, this process can be facilitated by the fast growth of tumors and the fact that most tumors required genomic instability or a mutator phenotype to develop in the first place. Together, this results in a rapid accumulation of changes that can advance the attainment of resistance-causing alterations.

We showed that resistance to dasatinib developed more easily than resistance to doxycycline, which induced miR-155 withdrawal. This is consistent with the idea that targeting multiple pathways at once is effective at preventing drug resistance (Pagliarini et al., 2015). When multiple pathways are targeted, the cell would need to simultaneously develop multiple alterations to overcome these pathways, and each alteration will not be selected for in the absence of the other. In the case of targeting miR-155 by inducing withdrawal, it is similar to simultaneously inhibiting all pathways downstream of miR-155, not just a single pathway. Therefore, it is unsurprising that resistance might develop less readily than when targeting a single downstream pathway.

97 Interestingly, though, our data showed that the tumors were initially only dependent on a single downstream pathway (ie. the Kit signaling pathways), because dasatinib was effective in causing tumor regression. Therefore, we hypothesize that the mutator phenotype induced by miR-155 allowed additional mutations to form during treatment, which represented a reprogramming sufficient to allow tumor growth in the absence of Kit activity. Like in our model for oncogenesis by miR-155 through Kit, the bypass pathway must be dependent on miR-155 expression, because even dasatinib- resistant tumors retained sensitivity to miR-155 withdrawal by doxycycline (Figure 4e).

We suggest testing this hypothesis as a potential future direction for these studies on mechanisms of oncogenesis by miR-155. Specifically, performing whole exome sequencing on dasatinib-resistant tumors could reveal recurrent mutations in a new pathway. One pathway that we hypothesize might be a hit in a whole exome screen of dasatinib resistant tumors is Myc. In our RNAseq results, Myc was highly dysregulated during miR-155 withdrawal, similar to Kit. We performed preliminary tests that suggested that Myc did not play a role in tumor growth of dasatinib-sensitive lines, but it still represents a viable target for oncogenesis by miR-155 after reprogramming.

If a relationship between miR-155 and other relevant pathways cannot be elucidated, it is possible that the host gene of miR-155, BIC is interacting with these target pathways instead of miR-155. Following up on the studies of BIC described in

Chapter 3 could help shed light on this phenomenon. For instance, comparing the effect of BIC to a miR-155 mimic could confirm whether BIC has a role independent of miR-

155, and identifying interacting partners of BIC through pulldowns could help identify a mechanism.

98 Identifying a new pathway downstream of miR-155 or BIC that is sufficient for driving tumor growth would represent an alternative mechanism of oncogenesis by miR-

155. The therapeutic implications of this could include identifying drug combinations that would be effective in preventing resistance in miR-155 overexpressing cancers.

Conclusions

We examined the molecular mechanisms by which miR-155 promoted tumor formation and progression in a powerful mouse model. Our findings support a model of oncogenesis in which multiple miR-155 targets cooperate to drive activation of a strong oncogenic pathway. This represents a proof of principle for a model of cooperation between miRNA targets and for a model of the gene expression environment in a cell informing the selection of oncogenic mutations. This model is widely applicable to other miRNAs, other tumor types, and even other sources of dysregulated gene expression and mutator phenotypes. It highlights the importance of considering cooperation between oncogenic pathways when seeking to understand how tumors form, grow, and regress. This deeper understanding will advance our ability to develop new technologies to fight cancer.

99 References

Abbas, A.K., Lichtman, A.H., and Pillai, S. (2015). Lymphocyte Development and Antigen Receptor Gene Rearrangement. In Cellular and Molecular Immunology, (Elsevier), p.171-198

Alemdehy, M.F., de Looper, H.W.J., Kavelaars, F.G., Sanders, M.A., Hoogenboezem, R., Löwenberg, B., Valk, P.J.M., Touw, I.P., and Erkeland, S.J. (2016). MicroRNA-155 induces AML in combination with the loss of C/EBPA in mice. Leukemia 30, 2238–2241.

Babar, I. a., Cheng, C.J., Booth, C.J., Liang, X., Weidhaas, J.B., Saltzman, W.M., and Slack, F.J. (2012). Nanoparticle-based therapy in an in vivo microRNA-155 (miR-155)- dependent mouse model of lymphoma. Proc. Natl. Acad. Sci. 109, E1695–E1704.

Basiouni, S., Fuhrmann, H., and Schumann, J. (2012). High-efficiency transfection of suspension cell lines. BioTechniques 53, 1–4. van den Berg, A., Kroesen, B.-J., Kooistra, K., de Jong, D., Briggs, J., Blokzijl, T., Jacobs, S., Kluiver, J., Diepstra, A., Maggio, E., et al. (2003). High expression of B-cell receptor inducible gene BIC in all subtypes of Hodgkin lymphoma. Genes. Chromosomes Cancer 37, 20–28.

Bernardi, R., Grisendi, S., and Pandolfi, P.P. (2002). Modelling haematopoietic malignancies in the mouse and therapeutical implications. Oncogene 21, 3445–3458.

Bernstein, E., Caudy, A.A., Hammond, S.M., and Hannon, G.J. (2001). Role for a bidentate ribonuclease in the initiation step of RNA interference. Nature 409, 363–366.

Calin, G.A., Dumitru, C.D., Shimizu, M., Bichi, R., Zupo, S., Noch, E., Aldler, H., Rattan, S., Keating, M., Rai, K., et al. (2002). Frequent deletions and down-regulation of micro- RNA genes miR15 and miR16 at 13q14 in chronic lymphocytic leukemia. Proc. Natl. Acad. Sci. 99, 6.

Cheng, C.J., Bahal, R., Babar, I. a, Pincus, Z., Barrera, F., Liu, C., Svoronos, A., Braddock, D.T., Glazer, P.M., Engelman, D.M., et al. (2014). MicroRNA silencing for cancer therapy targeted to the tumour microenvironment. Nature 518, 107–110.

Clurman, B.E., and Hayward, W.S. (1989). Multiple proto-oncogene activations in avian leukosis virus-induced lymphomas: evidence for stage-specific events. Mol. Cell. Biol. 9, 2657–2664.

Costinean, S., Zanesi, N., Pekarsky, Y., Tili, E., Volinia, S., Heerema, N., and Croce, C.M. (2006). Pre-B cell proliferation and lymphoblastic leukemia/high-grade lymphoma in E -miR155 transgenic mice. Proc. Natl. Acad. Sci. 103, 7024–7029.

100 Costinean, S., Sandhu, S.K., Pedersen, I.M., Tili, E., Trotta, R., Perrotti, D., Ciarlariello, D., Neviani, P., Harb, J., Kauffman, L.R., et al. (2009). Src homology 2 domain- containing inositol-5-phosphatase and CCAAT enhancer-binding protein are targeted by miR-155 in B cells of E -MiR-155 transgenic mice. Blood 114, 1374–1382.

Czochor, J.R., Sulkowski, P., and Glazer, P.M. (2016). miR-155 Overexpression Promotes Genomic Instability by Reducing High-fidelity Polymerase Delta Expression and Activating Error-Prone DSB Repair. Mol. Cancer Res. 14, 363–373.

Delmore, J.E., Issa, G.C., Lemieux, M.E., Rahl, P.B., Shi, J., Jacobs, H.M., Kastritis, E., Gilpatrick, T., Paranal, R.M., Qi, J., et al. (2011). BET Bromodomain Inhibition as a Therapeutic Strategy to Target c-Myc. Cell 146, 904–917.

Dexheimer, T.S. (2013). DNA Repair Pathways and Mechanisms. In DNA Repair of Cancer Stem Cells, L.A. Mathews, S.M. Cabarcas, and E.M. Hurt, eds. (Dordrecht: Springer Netherlands), pp. 19–32.

Di Leva, G., Garofalo, M., and Croce, C.M. (2014). MicroRNAs in Cancer. Annu. Rev. Pathol. Mech. Dis. 9, 287–314.

Döhner, H., Weisdorf, D.J., and Bloomfield, C.D. (2015). Acute Myeloid Leukemia. N. Engl. J. Med. 373, 1136–1152.

Dorsett, Y., McBride, K.M., Jankovic, M., Gazumyan, A., Thai, T.-H., Robbiani, D.F., Di Virgilio, M., San-Martin, B.R., Heidkamp, G., Schwickert, T.A., et al. (2008). MicroRNA- 155 Suppresses Activation-Induced Cytidine Deaminase-Mediated Myc-Igh Translocation. Immunity 28, 630–638.

Eis, P.S., Tam, W., Sun, L., Chadburn, A., Li, Z., Gomez, M.F., Lund, E., and Dahlberg, J.E. (2005). Accumulation of miR-155 and BIC RNA in human B cell lymphomas. Proc. Natl. Acad. Sci. 102, 6.

Elton, T.S., Selemon, H., Elton, S.M., and Parinandi, N.L. (2013). Regulation of the MIR155 host gene in physiological and pathological processes. Gene 532, 1–12.

Esquela-Kerscher, A., and Slack, F.J. (2006). Oncomirs — microRNAs with a role in cancer. Nat. Rev. Cancer 6, 259–269.

Faraoni, I., Antonetti, F.R., Cardone, J., and Bonmassar, E. (2009). miR-155 gene: A typical multifunctional microRNA. Biochim. Biophys. Acta BBA - Mol. Basis Dis. 1792, 497–505.

Faraoni, I., Laterza, S., Ardiri, D., Ciardi, C., Fazi, F., and Lo-Coco, F. (2012). MiR-424 and miR-155 deregulated expression in cytogenetically normal acute myeloid leukaemia: correlation with NPM1 and FLT3 mutation status. J. Hematol. Oncol.J Hematol Oncol 5.

101 Felgner, P.L., Gadek, T.R., Holm, M., Roman, R., Chan, H.W., Wenz, M., Northrop, J.P., Ringold, G.M., and Danielsen, M. (1987). Lipofection: a highly efficient, lipid- mediated DNA-transfection procedure. Proc. Natl. Acad. Sci. 84, 7413–7417.

Felsher, D.W., and Bishop, J.M. (1999). Reversible Tumorigenesis by MYC in Hematopoietic Lineages. Mol. Cell 4, 199–207.

Fernandez-Mercado, M., Manterola, L., and Lawrie, C.H. (2015). MicroRNAs in Lymphoma: Regulatory Role and Biomarker Potential. Curr. Genomics 16, 349–358.

Foss, F.M., Querfeld, C., Kim, Y.H., Pinter-Brown, L.C., William, B.M., Porcu, P., Pacheco, T., Haverkos, B., DeSimone, J., Guitart, J., et al. (2018). Ph 1 study of MRG- 106, an inhibitor of miR-155, in CTCL. J. Clin. Oncol. 36, 2511–2511.

Friedman, R.C., Farh, K.K.-H., Burge, C.B., and Bartel, D.P. (2009). Most mammalian mRNAs are conserved targets of microRNAs. Genome Res. 19, 92–105.

Fulci, V., Chiaretti, S., Goldoni, M., Azzalin, G., Carucci, N., Tavolaro, S., Castellano, L., Magrelli, A., Citarella, F., Messina, M., et al. (2007). Quantitative technologies establish a novel microRNA profile of chronic lymphocytic leukemia. Blood 109, 4944–4952.

Garzon, R., Volinia, S., Liu, C.-G., Fernandez-Cymering, C., Palumbo, T., Pichiorri, F., Fabbri, M., Coombes, K., Alder, H., Nakamura, T., et al. (2008). MicroRNA signatures associated with cytogenetics and prognosis in acute myeloid leukemia. Blood 111, 3183–3189.

Gerloff, D., Grundler, R., Wurm, A.A., Bräuer-Hartmann, D., Katzerke, C., Hartmann, J.- U., Madan, V., Müller-Tidow, C., Duyster, J., Tenen, D.G., et al. (2015). NF- κB/STAT5/miR-155 network targets PU.1 in FLT3-ITD-driven acute myeloid leukemia. Leukemia 29, 535–547.

Gironella, M., Seux, M., Xie, M.-J., Cano, C., Tomasini, R., Gommeaux, J., Garcia, S., Nowak, J., Yeung, M.L., Jeang, K.-T., et al. (2007). Tumor protein 53-induced nuclear protein 1 expression is repressed by miR-155, and its restoration inhibits pancreatic tumor development. Proc. Natl. Acad. Sci. 104, 16170–16175.

Greenman, C., Stephens, P., Smith, R., Dalgliesh, G.L., Hunter, C., Bignell, G., Davies, H., Teague, J., Butler, A., Stevens, C., et al. (2007). Patterns of somatic mutation in human cancer genomes. Nature 446, 153–158.

Grishok, A., Pasquinelli, A.E., Conte, D., Li, N., Parrish, S., Ha, I., Baillie, D.L., Fire, A., Ruvkun, G., and Mello, C.C. (2001). Genes and Mechanisms Related to RNA Interference Regulate Expression of the Small Temporal RNAs that Control C. elegans Developmental Timing. Cell 106, 23–34.

Hammond, S.M., Bernstein, E., Beach, D., and Hannon, G.J. (2000). An RNA-directed nuclease mediates post-transcriptional gene silencing in Drosophila cells. Nature 404, 293–296.

102 Hanahan, D., and Weinberg, R.A. (2000). The Hallmarks of Cancer. Cell 100, 57–70.

Hanahan, D., and Weinberg, R.A. (2011). Hallmarks of Cancer: The Next Generation. Cell 144, 646–674.

Hatley, M.E., Patrick, D.M., Garcia, M.R., Richardson, J.A., Bassel-Duby, R., van Rooij, E., and Olson, E.N. (2010). Modulation of K-Ras-Dependent Lung Tumorigenesis by MicroRNA-21. Cancer Cell 18, 282–293.

He, L., Thomson, J.M., Hemann, M.T., Hernando-Monge, E., Mu, D., Goodson, S., Powers, S., Cordon-Cardo, C., Lowe, S.W., Hannon, G.J., et al. (2005). A microRNA polycistron as a potential human oncogene. Nature 435, 828–833.

Heinrich, M.C., Griffith, D.J., Druker, B.J., Wait, C.L., Ott, K.A., and Zigler, A.J. (2000). Inhibition of c-kit receptor tyrosine kinase activity by STI 571, a selective tyrosine kinase inhibitor. 96, 925–932.

Hirota, S. (1998). Gain-of-Function Mutations of c-kit in Human Gastrointestinal Stromal Tumors. Science 279, 577–580.

Holohan, C., Van Schaeybroeck, S., Longley, D.B., and Johnston, P.G. (2013). Cancer drug resistance: an evolving paradigm. Nat. Rev. Cancer 13, 714–726.

Hutvagner, G., and Zamore, P.D. (2002). A microRNA in a Multiple-Turnover RNAi Enzyme Complex. Science 297, 2056–2060.

Hutvágner, G., McLachlan, J., Pasquinelli, A.E., Bálint, É., Tuschl, T., and Zamore, P.D. (2001). A Cellular Function for the RNA-Interference Enzyme Dicer in the Maturation of the let-7 Small Temporal RNA. Science 293, 834–838.

Inamura, K., and Ishikawa, Y. (2016). MicroRNA In Lung Cancer: Novel Biomarkers and Potential Tools for Treatment. J. Clin. Med. 5.

Jain, M., Arvanitis, C., Chu, K., Dewey, W., Leonhardt, E., Trinh, M., Sundberg, C.D., Bishop, J.M., and Felsher, D.W. (2002). Sustained Loss of a Neoplastic Phenotype by Brief Inactivation of MYC. Science 297, 102–104.

Kalkat, M., De Melo, J., Hickman, K.A., Lourenco, C., Redel, C., Resetca, D., Tamachi, A., Tu, W.B., and Penn, L.Z. (2017). MYC Deregulation in Primary Human Cancers. Genes 8.

Kastner, P., and Chan, S. (2008). PU . 1 : A crucial and versatile player in hematopoiesis and leukemia. 40, 22–27.

Khoury, E., Andre, C., Pontvert-Delucq, S., Drenou, B., Baillou, C., Guigon, M., Najman, A., and Lemoine, F.M. Tumor Necrosis Factor a (TNFa) Downregulates c-kit Proto- Oncogene Product Expression in Normal and Acute Myeloid Leukemia CD34’ Cells Via p55 TNFa Receptors. Blood 84, 2506–2514.

103 Khvorova, A., Reynolds, A., and Jayasena, S.D. Functional siRNAs and miRNAs Exhibit Strand Bias. Cell 115, 209–216.

Klein, U., Lia, M., Crespo, M., Siegel, R., Shen, Q., Mo, T., Ambesi-Impiombato, A., Califano, A., Migliazza, A., Bhagat, G., et al. (2010). The DLEU2/miR-15a/16-1 Cluster Controls B Cell Proliferation and Its Deletion Leads to Chronic Lymphocytic Leukemia. Cancer Cell 17, 28–40.

Klimenko, O. V, and Shtilman, M.I. (2013). Transfection of Kasumi-1 cells with a new type of polymer carriers loaded with miR-155 and antago-miR-155. Cancer Gene Ther. 20, 237–241.

Kluiver, J., Poppema, S., de Jong, D., Blokzijl, T., Harms, G., Jacobs, S., Kroesen, B.- J., and van den Berg, A. (2005). BIC and miR-155 are highly expressed in Hodgkin, primary mediastinal and diffuse large B cell lymphomas. J. Pathol. 207, 243–249.

Kosmider, O., and Moreau-Gachelin, F. (2006). From Mice to Human: The “Two-Hit Model” of Leukemogenesis. Cell Cycle 5, 569–570.

Kosmider, O., Denis, N., Lacout, C., Vainchenker, W., Dubreuil, P., and Moreau- Gachelin, F. (2005). Kit-activating mutations cooperate with Spi-1/PU.1 overexpression to promote tumorigenic progression during erythroleukemia in mice. Cancer Cell 8, 467–478.

Lagos-Quintana, M., Rauhut, R., Lendeckel, W., and Tuschl, T. (2001). Identification of Novel Genes Coding for Small Expressed RNAs. Science 294, 853–858.

Lagos-Quintana, M., Rauhut, R., Yalcin, A., Meyer, J., Lendeckel, W., and Tuschl, T. (2002). Identification of Tissue-Specific MicroRNAs from Mouse. Curr. Biol. 12, 735– 739.

Larizza, L., Magnani, I., and Beghini, A. (2005). The Kasumi-1 cell line: a t(8;21)-kit mutant model for acute myeloid leukemia. Leuk. Lymphoma 46, 247–255.

Lau, N.C., Lim, L.P., Weinstein, E.G., and Bartel, D.P. (2001). An Abundant Class of Tiny RNAs with Probable Regulatory Roles in Caenorhabditis elegans. Science 294, 858–862.

Lecker, S.H., Goldberg, A.L., and Mitch, W.E. (2006). Protein Degradation by the Ubiquitin–Proteasome Pathway in Normal and Disease States. J. Am. Soc. Nephrol. 17, 1807–1819.

Lee, R.C., and Ambros, V. (2001). An Extensive Class of Small RNAs in Caenorhabditis elegans. Science 294, 862–864.

Lee, T.I., and Young, R.A. (2013). Transcriptional Regulation and Its Misregulation in Disease. Cell 152, 1237–1251.

104 Lee, R.C., Feinbaum, R.L., and Ambros, V. (1993). The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell 75, 843–854.

Lee, Y., Jeon, K., Lee, J.-T., Kim, S., and Kim, V.N. (2002). MicroRNA maturation: stepwise processing and subcellular localization. EMBO J. 21, 4663–4670.

Lee, Y., Ahn, C., Han, J., Choi, H., Kim, J., Yim, J., Lee, J., Provost, P., Rådmark, O., Kim, S., et al. (2003). The nuclear RNase III Drosha initiates microRNA processing. Nature 425, 415–419.

Lee, Y., Kim, M., Han, J., Yeom, K.-H., Lee, S., Baek, S.H., and Kim, V.N. (2004). MicroRNA genes are transcribed by RNA polymerase II. EMBO J. 23, 4051–4060.

Lennartsson, J., and Rönnstrand, L. (2012). Stem Cell Factor Receptor/c-Kit: From Basic Science to Clinical Implications. Physiol. Rev. 92, 1619–1649.

Lévy, C., Frecha, C., Costa, C., Rachinel, N., Salles, G., Cosset, F.-L., and Verhoeyen, E. (2010). Lentiviral vectors and transduction of human cancer B cells. Blood 116, 498– 500.

Liu, J., Valencia-Sanchez, M.A., Hannon, G.J., and Parker, R. (2005). MicroRNA- dependent localization of targeted mRNAs to mammalian P-bodies. Nat. Cell Biol. 7, 719–723.

Lu, J., Getz, G., Miska, E.A., Alvarez-Saavedra, E., Lamb, J., Peck, D., Sweet-Cordero, A., Ebert, B.L., Mak, R.H., Ferrando, A.A., et al. (2005). MicroRNA expression profiles classify human cancers. Nature 435, 834–838.

Luo, J., Solimini, N.L., and Elledge, S.J. (2009). Principles of Cancer Therapy: Oncogene and Non-oncogene Addiction. Cell 136, 823–837.

Medina, P.P., Nolde, M., and Slack, F.J. (2010). OncomiR addiction in an in vivo model of microRNA-21-induced pre-B-cell lymphoma. Nature 467, 86–90.

Meister, G. (2013). Argonaute proteins: functional insights and emerging roles. Nat. Rev. Genet. 14, 447–459.

Mol, C.D., Dougan, D.R., Schneider, T.R., Skene, R.J., Kraus, M.L., Scheibe, D.N., Snell, G.P., Zou, H., Sang, B.-C., and Wilson, K.P. (2004). Structural Basis for the Autoinhibition and STI-571 Inhibition of c-Kit Tyrosine Kinase. J. Biol. Chem. 279, 31655–31663.

Mu, P., Han, Y.-C., Betel, D., Yao, E., Squatrito, M., Ogrodowski, P., Stanchina, E. de, D’Andrea, A., Sander, C., and Ventura, A. (2009). Genetic dissection of the miR-17∼92 cluster of microRNAs in Myc-induced B-cell lymphomas. Genes Dev. 23, 2806–2811.

Narayan, N., Bracken, C.P., and Ekert, P.G. (2018). MicroRNA-155 expression and function in AML: An evolving paradigm. Exp. Hematol. 62, 1–6.

105 Ning, Z., Li, J., and Arceci, R.J. (2001). Signal transducer and activator of transcription 3 activation is required for Asp 816 mutant c-Kit – mediated cytokine-independent survival and proliferation in human leukemia cells. Blood 97, 3559–3568.

O’Connell, R.M., Taganov, K.D., Boldin, M.P., Cheng, G., and Baltimore, D. (2007). MicroRNA-155 is induced during the macrophage inflammatory response. Proc. Natl. Acad. Sci. 104, 1604–1609.

O’Connell, R.M., Rao, D.S., Chaudhuri, A.A., Boldin, M.P., Taganov, K.D., Nicoll, J., Paquette, R.L., and Baltimore, D. (2008). Sustained expression of microRNA-155 in hematopoietic stem cells causes a myeloproliferative disorder. J. Exp. Med. 205, 585– 594.

O’Connell, R.M., Chaudhuri, A.A., Rao, D.S., and Baltimore, D. (2009). Inositol phosphatase SHIP1 is a primary target of miR-155. Proc. Natl. Acad. Sci. 106, 7113– 7118.

Pagliarini, R., Shao, W., and Sellers, W.R. (2015). Oncogene addiction: pathways of therapeutic response, resistance, and road maps toward a cure. EMBO Rep. 16, 280– 296.

Pasquinelli, A.E., Reinhart, B.J., Slack, F., Martindale, M.Q., Kuroda, M.I., Maller, B., Hayward, D.C., Ball, E.E., Degnan, B., Müller, P., et al. (2000). Conservation of the sequence and temporal expression of let-7 heterochronic regulatory RNA. Nature 408, 86–89.

Pedersen, I.M., Otero, D., Kao, E., Miletic, A.V., Hother, C., Ralfkiaer, E., Rickert, R.C., Gronbaek, K., and David, M. (2009). Onco-miR-155 targets SHIP1 to promote TNFα- dependent growth of B cell lymphomas: TNFα regulates onco-miR-155 in DLBCL. EMBO Mol. Med. 1, 288–295.

Pelengaris, S., Littlewood, T., Khan, M., Elia, G., and Evan, G. (1999). Reversible Activation of c-Myc in Skin: Induction of a Complex Neoplastic Phenotype by a Single Oncogenic Lesion. Mol. Cell 3, 565–577.

Pogribny, I.P. (2018). MicroRNAs as biomarkers for clinical studies. Exp. Biol. Med. 243, 283–290.

Quesniaux, V.F.J., Down, J.D., and Schuurman, H.-J. (2005). Hematopoiesis, including lymphocyte development and maturation. In Principles of Immunopharmacology, F.P. Nijkamp, and M.J. Parnham, eds. (Basel: Birkhäuser-Verlag), pp. 3–17.

Ranganath, P. (2015). MicroRNA-155 and Its Role in Malignant Hematopoiesis. Biomark. Insights 10, 95–102.

Ratajczak, M.Z., Perrotti, D., Melotti, P., Powzaniuk, M., Calabretta, B., Onodera, K., Kregenow, D.A., Machalinski, B., and Gewirtz, A.M. (1998). Myb and ets proteins are

106 candidate regulators of c-kit expression in human hematopoietic cells. Blood 91, 1934– 1946.

Reinhart, B.J., Slack, F.J., Basson, M., Pasquinelli, A.E., Bettinger, J.C., Rougvie, A.E., Horvitz, H.R., and Ruvkun, G. (2000). The 21-nucleotide let-7 RNA regulates developmental timing in Caenorhabditis elegans. Nature 403, 901–906.

Rodriguez, A., Vigorito, E., Clare, S., Warren, M.V., Couttet, P., Soond, D.R., Dongen, S. van, Grocock, R.J., Das, P.P., Miska, E.A., et al. (2007). Requirement of bic/microRNA-155 for Normal Immune Function. Science 316, 608–611.

Rupaimoole, R., and Slack, F.J. (2017). MicroRNA therapeutics: towards a new era for the management of cancer and other diseases. Nat. Rev. Drug Discov. 16, 203–222.

Schittenhelm, M.M., Shiraga, S., Schroeder, A., Corbin, A.S., Griffith, D., Lee, F.Y., Bokemeyer, C., Deininger, M.W.N., Druker, B.J., and Heinrich, M.C. (2006). Dasatinib (BMS-354825), a Dual SRC/ABL Kinase Inhibitor, Inhibits the Kinase Activity of Wild- Type, Juxtamembrane, and Activation Loop Mutant KIT Isoforms Associated with Human Malignancies. Cancer Res. 66, 473–481.

Schneider, E., Staffas, A., Röhner, L., Krowiorz, K., Heuser, M., Döhner, K., Bullinger, L., Döhner, H., Fogelstrand, L., Rouhi, A., et al. (2016). MicroRNA-155 is upregulated in MLL-rearranged AML but its absence does not affect leukemia development. Exp. Hematol. 44, 1166–1171.

Schwarz, D.S., Hutvágner, G., Du, T., Xu, Z., Aronin, N., and Zamore, P.D. (2003). Asymmetry in the Assembly of the RNAi Enzyme Complex. Cell 115, 199–208.

Seiffert, M., Stilgenbauer, S., Döhner, H., and Lichter, P. (2007). Efficient nucleofection of primary human B cells and B-CLL cells induces apoptosis, which depends on the microenvironment and on the structure of transfected nucleic acids. Leukemia 21, 1977–1983.

Sen, G.L., and Blau, H.M. (2005). Argonaute 2/RISC resides in sites of mammalian mRNA decay known as cytoplasmic bodies. Nat. Cell Biol. 7, 633–636.

Slack, F.J., Basson, M., Liu, Z., Ambros, V., Horvitz, H.R., and Ruvkun, G. (2000). The lin-41 RBCC Gene Acts in the C. elegans Heterochronic Pathway between the let-7 Regulatory RNA and the LIN-29 Transcription Factor. Mol. Cell 5, 659–669.

Stenvang, J., Petri, A., Lindow, M., Obad, S., and Kauppinen, S. (2012). Inhibition of microRNA function by antimiR oligonucleotides. Silence 3, 1.

Taganov, K.D., Boldin, M.P., Chang, K.-J., and Baltimore, D. (2006). NF-κB-dependent induction of microRNA miR-146, an inhibitor targeted to signaling proteins of innate immune responses. Proc. Natl. Acad. Sci. 103, 12481–12486.

107 Tam, W. (2001). Identification and characterization of human BIC, a gene on chromosome 21 that encodes a noncoding RNA. Gene 274, 157–167.

Tam, W., Ben-Yehuda, D., and Hayward, W.S. (1997). bic, a novel gene activated by proviral insertions in avian leukosis virus-induced lymphomas, is likely to function through its noncoding RNA. Mol. Cell. Biol. 17, 1490–1502.

Tam, W., Hughes, S.H., Hayward, W.S., and Besmer, P. (2002). Avian bic, a Gene Isolated from a Common Retroviral Site in Avian Leukosis Virus-Induced Lymphomas That Encodes a Noncoding RNA, Cooperates with c-myc in Lymphomagenesis and Erythroleukemogenesis. J. Virol. 76, 4275–4286.

Tamborini, E., Bonadiman, L., Greco, A., Albertini, V., Negri, T., Gronchi, A., Bertulli, R., Colecchia, M., Casali, P.G., Pierotti, M.A., et al. (2004). A new mutation in the KIT ATP pocket causes acquired resistance to imatinib in a gastrointestinal stromal tumor patient. Gastroenterology 127, 294–299.

Taylor, J., Xiao, W., and Abdel-Wahab, O. (2017). Diagnosis and classification of hematologic malignancies on the basis of genetics. Blood 130, 410–423.

Thai, T.-H., Calado, D.P., Casola, S., Ansel, K.M., Xiao, C., Xue, Y., Murphy, A., Frendewey, D., Valenzuela, D., Kutok, J.L., et al. (2007). Regulation of the Germinal Center Response by MicroRNA-155. Science 316, 604–608.

Tili, E., Michaille, J.-J., Cimino, A., Costinean, S., Dumitru, C.D., Adair, B., Fabbri, M., Alder, H., Liu, C.G., Calin, G.A., et al. (2007). Modulation of miR-155 and miR-125b Levels following Lipopolysaccharide/TNF-α Stimulation and Their Possible Roles in Regulating the Response to Endotoxin Shock. J. Immunol. 179, 5082–5089.

Tili, E., Michaille, J.-J., Wernicke, D., Alder, H., Costinean, S., Volinia, S., and Croce, C.M. (2011). Mutator activity induced by microRNA-155 (miR-155) links inflammation and cancer. Proc. Natl. Acad. Sci. 108, 4908–4913.

Trang, P., Wiggins, J.F., Daige, C.L., Cho, C., Omotola, M., Brown, D., Weidhaas, J.B., Bader, A.G., and Slack, F.J. (2011). Systemic Delivery of Tumor Suppressor microRNA Mimics Using a Neutral Lipid Emulsion Inhibits Lung Tumors in Mice. Mol. Ther. 19, 1116–1122.

Valeri, N., Gasparini, P., Fabbri, M., Braconi, C., Veronese, A., Lovat, F., Adair, B., Vannini, I., Fanini, F., Bottoni, A., et al. (2010a). Modulation of mismatch repair and genomic stability by miR-155. Proc. Natl. Acad. Sci. 107, 6982–6987.

Valeri, N., Gasparini, P., Braconi, C., Paone, A., Lovat, F., Fabbri, M., Sumani, K.M., Alder, H., Amadori, D., Patel, T., et al. (2010b). MicroRNA-21 induces resistance to 5- fluorouracil by down-regulating human DNA MutS homolog 2 (hMSH2). Proc. Natl. Acad. Sci. 107, 21098–21103.

108 Vandenbark, GR; Chen, Yi; Friday, Ellen; Pavlik, Kevin; Anthony, Bruce; deCastro, Carlos; Kaufman, R.E. (1996). Complex Regulation of Human c-kit Transcription by Promoter Repressors, Activators, and Specific myb Elements. 7, 1383–1392.

Volinia, S., Calin, G.A., Liu, C.-G., Ambs, S., Cimmino, A., Petrocca, F., Visone, R., Iorio, M., Roldo, C., Ferracin, M., et al. (2006). A microRNA expression signature of human solid tumors defines cancer gene targets. Proc. Natl. Acad. Sci. 103, 2257– 2261.

Wallace, J.A., Kagele, D.A., Eiring, A.M., Kim, C.N., Hu, R., Runtsch, M.C., Alexander, M., Huffaker, T.B., Lee, S., Patel, A.B., et al. (2018). miR-155 promotes FLT3-ITD – induced myeloproliferative disease through inhibition of the interferon response. 129, 3074–3087.

Wang, W., and Luo, Y. (2015). MicroRNAs in breast cancer: oncogene and tumor suppressors with clinical potential. J. Zhejiang Univ. Sci. B 16, 18–31.

Wang, Y.-Y., Zhou, G.-B., Yin, T., Chen, B., Shi, J.-Y., Liang, W.-X., Jin, X.-L., You, J.- H., Yang, G., Shen, Z.-X., et al. AML1-ETO and C-KIT mutation/overexpression in t(8;21) leukemia: Implication in stepwise leukemogenesis and response to Gleevec. 6.

Weinstein, I.B. (2002). Addiction to Oncogenes--the Achilles Heal of Cancer. Science 297, 63–64.

Wightman, B., Ha, I., and Ruvkun, G. (1993). Posttranscriptional regulation of the heterochronic gene lin-14 by lin-4 mediates temporal pattern formation in C. elegans. Cell 75, 855–862.

Xiao, C., Srinivasan, L., Calado, D.P., Patterson, H.C., Zhang, B., Wang, J., Henderson, J.M., Kutok, J.L., and Rajewsky, K. (2008). Lymphoproliferative disease and autoimmunity in mice with increased miR-17-92 expression in lymphocytes. Nat. Immunol. 9, 405–414.

Yi, R., Qin, Y., Macara, I.G., and Cullen, B.R. (2003). Exportin-5 mediates the nuclear export of pre-microRNAs and short hairpin RNAs. Genes Dev. 17, 3011–3016.

109