Targeting MYC in cancer therapy: RNA processing offers new opportunities

MYC is an attractive target in cancer therapy

The abnormal activation of the c‐Myc (Myc) oncogene, either due to transcriptional overexpression (gene amplification, translocation, alterations in upstream signaling pathways) and/or protein stabilization, is one of the most common features of cancer cells. Indeed, high MYC protein levels are not only able to drive tumor initiation and progression, but are also essential for tumor maintenance: sustained MYC overexpression is required by cancer cells, and growth arrest, apoptosis and differentiation occur upon reduction in MYC levels. This has not only been described for MYC‐driven mouse tumor models, but also in tumors driven by other oncogenes (reviewed in 1), making c‐MYC a highly attractive target for anti‐cancer therapy. Unfortunately, MYC itself is not an easily “druggable” protein, due to lack of enzymatic activity or any deep pocket, which could be traditionally targeted by small molecule inhibitors. For this reason, one of the priorities in the field is the inhibition of MYC co‐factors and/or downstream effectors that might play critical roles in tumor development or maintenance.

MYC functions as a transcription factor

MYC is a transcription factor of the basic helix loop helix leucine zipper (bHLH‐LZ) family. Together with its partner protein, MAX 2, 3, MYC can bind to target DNA sequences, the E‐boxes (including the canonical CACGTG and other non‐canonical sites), when they are embedded in a euchromatic context 4.

Once bound to chromatin, MYC is able to transcriptionally regulate protein coding and non‐coding RNAs 5 that are produced by RNA Pol I, RNA Pol II, and RNA Pol III 6. Notwithstanding the vast amount of gene expression analysis performed in the last decades with the aim of finding a core of MYC regulated genes, identifying the relationship between MYC binding to chromatin and the consequent transcriptional output is still an open debate. With the advent of genome‐wide profiles, and the realization that MYC not only binds every active promoter, but also active enhancers, this long‐standing issue has become even more complex.

In the last few years, two opposite views have emerged. One suggests that MYC amplifies the transcriptional output of all active promoters 7, 8; the other that MYC regulates, either positively or negatively, discrete sets of genes mainly involved in proliferation, cell growth, metabolic reprogramming, and RNA biogenesis 9. We will not go into the details of this debate here, because it has been extensively covered elsewhere very recently 5. Instead, we will focus our attention on the plethora of equally important post‐transcriptional mechanisms that are regulated, both directly and indirectly, by MYC (Fig. 1). Because these contribute to shaping the transcriptomic and the proteomic landscape of MYC overexpressing cells, we will then propose how this knowledge can be used to target MYC's oncogenic function.

MYC promotes mRNA capping

RNA Pol II transcripts are subject to a specific post‐transcriptional modification that consists of the addition of a 7‐methylguanosine to the first transcribed nucleotide (5′ cap). This cap structure not only guarantees RNA stability, but also proper pre‐mRNA downstream processing. For example, CBC (cap binding complex) and eIF4E (eukaryotic initiation factor 4E) are recruited to the mRNA by interaction with the cap structure, and subsequently mediate processing and translation initiation, respectively 10, 11.

MYC directly promotes cap addition at its target genes by recruiting TFIIH, which then phosphorylates the RNA Pol II carboxy‐terminal domain (CTD), a step that is required for the recruitment of the capping machinery 12. It can also act indirectly on the same process, as demonstrated by the ability of the MYC transactivation domain alone (without DNA binding capacity) to increase capping of CDK transcripts 13. Moreover, MYC transcriptionally up‐regulates S‐Adenyl‐l‐homocysteine hydrolase (SAHH), which is required to metabolize SAH, an inhibitory byproduct of the capping reaction, thus antagonizing a potential negative feedback loop 14.

Aside from cap‐dependent translation, selected mRNAs can also be translated in a cap‐independent manner from internal ribosome entry sites (IRES). The overexpression of the translation initiation factors, 4EBP1 and eIF4G, has been observed in various cancer types, and under hypoxic conditions, they promote the IRES‐mediated translation of mRNAs that confer survival advantages 15. Interestingly, mTOR inhibitors that block the phosphorylation of 4EBP1 have been shown to be synthetic lethal in MYC‐driven cancers (discussed later) 16.

MYC regulates the abundance of splicing factors

MYC has been shown to directly modulate the transcription of various splicing factors, such as hnRNPA1 17, hnRNPA2, and PTB (polypyrimidine tract binding protein) 18. These proteins regulate, among several other events, the alternative splicing of pyruvate kinase (PKM), specifically, generating more of the embryonic/tumor isoform, PKM2, which promotes aerobic glycolysis, and less of the adult isoform, PKM1, which favors oxidative phosphorylation 18, 19 Additionally, these splicing factors have been implicated in the generation of constitutively active androgen receptor splice variants, which may contribute to the development of castration‐resistant prostate cancer 20.

Similarly, MYC regulates hnRNPH, which is necessary for the correct splicing of oncogenic a‐raf pre‐mRNA 21. In normal cells, low MYC and hnRNPH expression allows the production of a short isoform of A‐raf, which encodes a truncated A‐Raf protein that suppress Ras activation and transformation. Conversely, in cancer cells, full‐length A‐Raf can be produced, due to the high levels of MYC and hnRNPH, and in this form, it can effectively inhibit the activity of the MST2 pro‐apoptotic kinase.

The promoter region of arginine/serine‐rich splicing factor Srsf1 has been shown to contain two non‐canonical E‐boxes, through which MYC can directly activate its transcription 22. Srsf1 and MYC have also been shown to cooperate in the transformation of mammary epithelial cells, possibly by synergistically activating eIF4E phosphorylation 23. Srsf1 is not only involved in splicing, but also in several other aspect of RNA metabolism, such as nuclear export, nonsense‐mediated RNA decay (NMD), and translation (reviewed in 24).

During somatic reprogramming, MYC directly upregulates the expression of Spt‐Ada‐Gcn5‐acetyltransferase (SAGA) components, which are then recruited on several MYC target genes, activating their transcription 25. Many of these genes encode proteins that are involved in RNA splicing, and are essential both during the early stages of reprogramming, as well as for the maintenance of the established pluripotent stem cell state. Consequently, MYC and SAGA indirectly modulate alternative splicing, primarily promoting exon inclusion, during cell reprogramming. Interestingly, many of the alternatively spliced transcripts are involved in cell migration, transcriptional regulation, or RNA processing 25. This supports the possibility that in other contexts, such as during the initiation of cancer metastasis, the regulation of various splicing factors by MYC may contribute to EMT/MET. This notion is supported by the well‐established link between splicing 26, as well as MYC and EMT 27, 28.

We recently found that MYC directly binds to the promoters and upregulates the expression of the core snRNP assembly genes 29. In particular, we showed that depletion of PRMT5 results in apoptosis and cell cycle arrest, and that it was accompanied by the aberrant splicing of pre mRNAs with weak 5′ splice sites. The phenotypic and molecular changes following PRMT5 depletion were significantly more pronounced in B cells from Eµ‐myc mice than those from wild‐type mice, suggesting that MYC overexpressing cells have an increased dependence on a functional core splicing machinery. We hypothesize that given that MYC‐overexpressing cells are actively transcribing mRNAs, this places an increased demand on the splicing machinery to maintain the splicing fidelity necessary to generate properly spliced transcripts, which can subsequently be translated into functional proteins. Along similar lines, in MYC‐hyperactive cells, the inhibition of another core spliceosome protein, BUD31, resulted in global intron retention and aberrant pre mRNA processing 30. Indeed, as we will discuss later, MYC overexpressing cells are more sensitive to splicing inhibition 30, 31.

Table 1 summarizes MYC‐regulated and MYC‐synthetic lethal splicing factors involved in alternative and constitutive splicing.

MYC indirectly regulates several pathways of RNA degradation

1. The AU‐binding proteins (AUBPs) regulate RNA stability by binding to AU‐rich elements (AREs), which are found in up to 16% of human protein‐coding transcripts 32. Among the AUBPs, MYC represses Tristetraprolin (TTP/Tis11/Zfp36) and its family members, Tis11b (Brf1/ Zfp36l1) or Tis11d (Brf2/Zfp36l2), which are involved in RNA degradation. In contrast, MYC positively regulates HuR, Auf1, Auf2, and Nucleolin (Ncl), which stabilize ARE‐containing transcripts. In this way, MYC may promote a general increase in the stability of short‐lived RNAs 33. The targets of TTP‐induced degradation are mainly transcripts encoding genes involved in cancer and inflammation. Since many other RBPs are also subject to TTP‐induced degradation 34, the effects of MYC on the RNA processing machinery are thus amplified. Interestingly, the down‐regulation of TTP by MYC has been shown to be a critical step in MYC‐induced lymphomagenesis. Since the forced overexpression of TTP in Eµ‐myc lymphomas delays tumor onset, TTP has been ascribed as a tumor suppressor 33. Similarly, an increase in Nucleolin levels can ensure the stabilization of the antiapoptotic, pro‐oncogenic BCL‐XL factor 35, which is essential for MYC‐driven cancer progression 36.

2. Nonsense mediated RNA decay (NMD) is a safeguard mechanism that mediates the degradation of transcripts with premature termination codons (PTCs), which are often produced by incorrectly spliced RNA. This prevents their translation into truncated proteins. Of note, many genes encoding regulatory and basal splicing factors themselves can undergo alternative splicing events that introduce PTCs, which subsequently signal for NMD 37, 38. NMD also post‐transcriptionally regulates several normal transcripts, in particular, genes encoding proteins involved in the unfolded protein response (UPR) and stress‐related genes 39, 40. In contrast, endoplasmic reticulum (ER)‐stress, along with other cellular stresses (such as reactive oxygen species, hypoxia, nutrient deprivation, etc.), inhibits the NMD response 41 in a complex regulatory circuit. MYC overexpression has been shown to inhibit NMD in B lymphocytes via the activation of the UPR, and in particular, the induction of PERK‐mediated phosphorylation of eIF2A 42. MYC is also required for the inhibition of NMD by 5‐azacytidine, in an eIF2A‐independent manner 43. One possibility would be that this occurs via miRNAs, since they are widely regulated by MYC 44 and have been shown to act on NMD 45.

3. A different RNA quality surveillance pathway is represented by the exosome, a complex containing nuclease activity that is involved in the degradation of aberrant RNA, the maturation of ribosomal RNA and sn/snoRNAs and the turnover of the products of RNA maturation 46. This complex is also recruited by AUBPs to degrade ARE containing RNAs 47. The expression of several subunits of this complex is positively regulated in a MYC‐dependent manner in fibroblasts 48 and B cells 9. Whether the catalytic activity of the complex is also increased by MYC is still an open question.

Thus, on one hand, MYC augments the RNA processing capacity of the cells (by increasing the level of capping and splicing factors), while on the other hand, it selectively modulates the stability of certain oncogenic mRNAs (by upregulating/downregulating AUBPs and inhibiting the NMD pathway), while ensuring overall fidelity via the upregulation of exosome components. These concerted MYC‐driven actions lead to the (often) observed increase in total mRNA levels and consequently, proteins and cell size 49.

MYC promotes translation/ribosome biogenesis

MYC regulates ribosomal biogenesis and mRNA translation in a highly orchestrated manner. First, it stimulates the transcription of ribosomal RNAs by RNA polymerase I 50, 51, 52 by binding to E‐boxes in the promoters of the rDNA clusters and recruiting TRRAP, further remodeling the chromatin structure to a more permissive state. Additionally, MYC positively regulates the expression and/or recruitment of RNA polymerase I co‐factors. These include UBF 53, which enhances promoter escape, and SL1 51, which is part of the basal RNA polymerase I transcriptional complex.

MYC also stimulates the RNA polymerase III‐driven transcription of tRNA and 5S rRNA 6, 54. This occurs via increased recruitment of TFIIIB, RNA polymerase III, the acetyltransferases, GCN5 and TRRAP, and the selective acetylation of histone H3 at the corresponding promoters 55.

Additionally, in order to obtain mature rRNAs, the rRNA must be processed and post‐transcriptionally modified. The pre‐rRNA is processed into the 18S, 5.8S, and 28S rRNA by several MYC‐regulated proteins, such as Nucleolin, nucleophosmin, fibrillarin, Nop56, and Bop1. snoRNPs play an essential role in the rRNA modification process, by providing target specificity to methylation and pseudouridylation reactions 56. Both the RNA (snoRNAs) and protein components of these complexes are also regulated by MYC 9, 57.

Finally, MYC directly activates the expression of ribosomal components such as RPL and RPS family members 58, as well as eIF family members that are involved in translational initiation 59, 60.

Thus, MYC plays an essential role in ensuring not only transcription, but also the proper maturation of the translation machinery.

MYC transcriptionally modulates microRNAs and long non‐coding RNAs

Many microRNAs 61, which are small (∼22 nucleotide) non‐coding RNAs, and long non‐coding RNAs 62, which are defined as being longer than 200 nucleotides, have been identified as direct MYC targets. MYC‐regulated miRNAs may be present in the intronic sequences of protein‐coding genes or within non‐protein‐coding loci, and are subject to both positive and negative regulation by MYC.

MYC has been reported to directly induce various oncogenic miRNAs (e.g. the mir‐17‐92 cluster), as well as repress tumor‐suppressive miRNAs (e.g. miR‐26a and miR‐34a), thereby favoring cancer cell proliferation (reviewed in 63). MYC also indirectly modulates miRNA expression, for example, via the induction of the long‐noncoding RNA, HOTAIR, which in turn, represses the tumor‐suppressive miR‐130a 64.

Furthermore, MYC controls the expression of proteins involved in miRNA biogenesis: among them, lin28B is an RNA‐binding protein that binds to the stem‐loops of precursor miRNAs, like let7 and miR‐150, and inhibits their Dicer‐ and Drosha‐mediated processing, accelerating their decay 65. On the other hand, MYC has also been reported to promote pri‐miRNA processing by transcriptionally regulating Drosha 66.

By regulating miRNA biogenesis and processing, that in turn regulate the expression of several mRNAs, MYC is thus able to greatly expand the number of its indirect targets.

Can we directly target MYC transcription?

Transcription of the MYC gene is controlled in a complex manner, by the action of numerous transcriptional regulators 67, as well as enhancers 68. BET bromodomain proteins, which bind acetylated lysine residues, in conjunction with the mediator co‐activator complex, can also regulate MYC transcription. Blocking their function with small molecule inhibitors, such as JQ1, i‐BET, and MMS417, have been shown to downregulate MYC transcription and consequently, the expression of MYC target genes 69, 70, 71. However, a caveat regarding the efficacy of decreasing the transcription of MYC via Brd4 inhibition is that this strategy is limited to cases in which Brd4 is the predominant regulator of MYC transcription 72. The strategy may not be as effective, for example, if gene amplification or protein stabilization is the major underlying cause of its overexpression. An additional consideration is that BET proteins modulate the expression of numerous genes in addition to MYC 73 and accordingly, bromodomain inhibition is by no means MYC‐specific.

Altering the topology of the DNA upstream of the MYC gene, such as by the stabilization of the MYC G‐quadruplex with small molecules 74, 75, has also been shown to be effective in reducing MYC transcription.

Can we target MYC translation and protein turnover/stability?

Once transcribed, the MYC mRNA can be bound by RNA binding proteins that regulate its stability 76 or translation 77, 78, for example, CELF1 and HuR. Following the depletion of polyamines by inhibition of ornithine decarboxylase, the increase in CELF1 reduces MYC levels. Interestingly, ornithine decarboxylase heterozygosity has also been shown to delay MYC‐driven lymphomagenesis 79, and the inhibition of Srm, which is also involved in polyamine biosynthesis, is chemopreventative in B‐cell lymphomas 80.

Myc mRNA can be translated both by 5′ cap‐ and internal ribosome entry site (IRES)‐ dependent mechanisms 81. In multiple myeloma cells, increased Myc IRES activity, which is dependent on hnRPA1 and Rps25, has been observed following ER stress 78. In these cells, a compound that blocked the binding of hnRNPA1 to the Myc IRES was only toxic in the presence of ER stress. Furthermore, a small molecule inhibitor of the translation initiation factor eIf4a, silvestrol, was recently shown to reduce tumor growth in vivo in a mouse model of colorectal carcinogenesis by suppressing both cap‐ and IRES‐dependent translation of Myc 82.

The stability and turnover of the MYC protein is regulated by a series of post‐translational modifications that are controlled by a myriad of proteins 83. As the expression of many of these proteins is also often deregulated in cancer, their inhibition with small molecules could reduce the stability of MYC, resulting in its degradation, and could, therefore, have therapeutic potential.

Can we design Antisense/ASO approaches to reduce Myc mRNA or protein abundance?

Antisense phosphorothioate oligonucleotides targeting Myc mRNA have demonstrated efficacy in multiple cancer types. These highly stable oligonucleotides bind to mRNA and trigger RNAse H activity, resulting in the hydrolysis of the DNA/RNA duplex 84. Alternative strategies have used morpholino oligonucleotides either designed to block Myc translation 85, 86 or alter its correct splicing, thereby reducing functional MYC protein levels 87.

Can we target Myc interactions?

As an alternative to the depletion of MYC levels, the interaction between MYC and its oligodimerization partner, MAX, can also be targeted in order to inhibit MYC function 88. Omomyc is a version of MYC mutated in the residues of the leucine zipper that are critical for its dimerization specificity. As a result, whereas MYC does not homodimerize, Omomyc can form both dimers with MYC, as well as with MAX. Importantly, it does not bind MAD or other HLH proteins. The MYC/Omomyc dimers bind DNA with low affinity, and thus Omomyc acts in a dominant negative fashion by sequestering MYC away from MAX and DNA. As a consequence, MYC‐mediated transcriptional activation is affected 88. However, MIZ‐1‐mediated MYC binding to promoters and trans‐repression are not affected by Omomyc. Consequently, Omomyc maintains the repression of genes that are usually negatively regulated by MYC, while enabling the dampening of MYC‐dependent transcriptional activation 89. Not surprisingly, MYC inhibition with Omomyc results in reduced proliferation and increased apoptosis. Interestingly, the apoptotic phenotype was more pronounced in cells expressing particularly high levels of MYC 90. Notably, the systemic inhibition of MYC in vivo with Omomyc was well tolerated by normal regenerating tissues, and the effects of MYC inhibition could be reversed completely and quickly 91.

Several small molecule inhibitors of MYC/MAX dimerization have also been developed. These compounds bind to and stabilize the monomeric form of MAX and prevent the association of MYC and MAX, thereby reducing MYC function in cells (reviewed in 92). More recently, alpha‐helix mimetics that prevent the binding of MYC/MAX heterodimers to DNA, but do not cause the dissociation of MYC and MAX, have also been developed 93.

Other transcription factors have been targeted by similar strategies based on directly or indirectly preventing protein:protein or protein:DNA interactions. For example, S3I‐201 prevents the dimerization of Stat3 and the binding of Stat3 to DNA 94, and Co^III Schiff base complexes can alter the structure of zinc fingers, which are a common feature of transcription factors, and selectively disrupt their DNA‐binding ability 95.

Synthetic lethal screens can be performed to identify the Achilles heel of MYC‐overexpressing cells

In addition to targeting the “oncogene addiction” to MYC observed in many cancer types by the approaches described above, strategies aimed at exploiting “non‐oncogene addiction” in cancer cells with elevated MYC levels have also proven useful. These studies are based on the concept of synthetic lethality, whereby the perturbation of two or more genes in combination, but not of either gene singly, results in a significant deleterious phenotype, such as decreased proliferation or increased cell death 96. In this respect, synthetic lethal screens have been useful in identifying critical MYC effectors that may be potential therapeutic targets 97. These screens have employed functional genomics approaches with siRNA or shRNA‐based knockdown, and compared the phenotypic and molecular responses of cells with “normal” MYC levels with cells with elevated MYC levels (and therefore hyperactive MYC‐induced transcriptional programs), in order to identify vulnerabilities in particular pathways that are hyperactive or compromised, specifically in MYC overexpressing cells.

Toyoshima et al. compared the effects of knocking down specific genes in isogenic control and MYC‐overexpressing fibroblasts 98. Using a high‐throughput approach, their screen included siRNA designed to target ∼3,300 druggable genes and 200 miRNAs. Importantly, siRNAs that had significant growth inhibitory effects on the control cells were excluded. Not surprisingly, genes that showed synthetic lethality with MYC‐overexpression belonged to key cellular pathways known to be regulated by MYC. These included DNA damage repair, transcription and transcriptional elongation, senescence, ribosome biogenesis, chromatin modification, metabolism, apoptosis, and mitotic control. Further analysis of one of the MYC synthetic lethal genes that was identified, CSNK1e, verified that its knockdown slowed the growth of MYCN‐amplified glioblastomas and that its expression correlated with MYC in other human cancer types.

More recently, Huang and co‐workers also screened a shRNA library targeting 442 genes encoding proteins for which small molecule inhibitors exist in a murine hepatocellular carcinoma model driven by MYC overexpression and p53 loss 99. They found that the kinase component of the p‐TEFb complex, CDK9, is essential to sustain the proliferation of MYC overexpressing HCC cells.

In another synthetic lethal screen, SAE2, a SUMO‐activating enzyme, was also identified as showing potent synthetic lethality in cells with hyperactive MYC 100. SAE2 knockdown resulted in spindle defects and consequently, aneuploidy and apoptosis, but only in MYC‐overexpressing cells. Mechanistically, the loss of SAE2 switched the expression pattern of several MYC dependent genes in a SUMOylation‐dependent manner, causing several MYC‐induced genes to become repressed. Several of these genes are known to be involved in the assembly and maintenance of the integrity of the mitotic spindle.

In RNAi‐based screens focused on kinases, PRKDC 101, AMPK‐related kinase 5 (see later) 102 and GSK3beta 103 were identified to exhibit synthetic lethality with MYC overexpression. Interestingly, the downregulation of the Gsk3b/Fbw7 signaling pathway potentiated TNF‐related apoptosis by stabilizing MYC and increasing the expression of a death ligand receptor, which promoted cell death 103.

How to target selected MYC downstream effectors

While screens provide an unbiased method to identify genes that are synthetic lethal with MYC, many directed gain or loss‐of‐function studies on MYC target genes have also demonstrated similar observations of synthetic lethality.

Replicative stress stimulated by hyperactive MYC is associated with the activation of the DNA‐damage response mediated by Atr and Chk1. In MYC‐driven tumors with high levels of replicative stress, the reduction of Atr levels prevented disease development. Additionally, Chk1 inhibitors caused the apoptosis of MYC‐driven lymphomas, but not K‐RAS(G12V)‐driven pancreatic adenomas 104. Chk2 deficiency and the inhibition of Chk1 and Chk2 with small molecules also promoted apoptosis and delayed lymphoma progression in vivo, and the combination of Chk2 and Parp inhibition resulted in a synergistic effect in MYC‐overexpressing cells 105.

Aurora kinase A and B have been shown to be upregulated by MYC in human and murine B cell lymphomas, and their pharmacological inhibition resulted in mitotic arrest, polyploidy, and apoptosis 106. In a diverse panel of cancer cell lines, the therapeutic efficacy of PF‐03814735, an Aurora kinase A/B inhibitor, was shown to be significantly associated with MYC overexpression or amplification, and this was confirmed in vivo using small cell lung cancer models 107. The efficacy of Aurora kinase B inhibition has been further validated in a MYC‐overexpressing medulloblastoma model using the small molecule inhibitor, AZD1152‐HQPA 108.

MYC‐overexpressing tumors have also been shown to display increased sensitivity to inhibition of cyclin dependent kinases, including Cdk1 109, 110 and Cdk2 111, the activation of the death receptor pathway 112 and Pim1 inhibition 113, 114.