MYC Phosphorylation at Novel Regulatory Regions Suppresses Transforming Activity

The c-Myc (MYC) oncoprotein is a prominent contributor to tumorigenesis. Understanding pathways that both regulate and cooperate with MYC has the potential to identify novel therapeutic strategies to inhibit this oncogene (1). The powerful antitumor effect of both direct (e.g., Omomyc) and indirect (e.g., JQ1) MYC inhibition has been experimentally reinforced by important recent publications, underscoring the need to better understand the mechanisms both regulating and contributing to MYC-induced transformation (2–5). Phosphorylation has been shown to be an important mechanism, regulating MYC activity. For example, two phosphorylation sites within MYC homology box I (MBI), threonine 58 (T58), and serine 62 (S62) undergo hierarchical phosphorylation. S62 can be phosphorylated by a number of kinases, including extracellular signal–regulated kinase (ERK), c-jun–NH₂–kinase (JNK), CDK1, and DYRK2 (6, 7). This subsequently promotes T58 phosphorylation by GSK3, leading to ubiquitylation by SCF-FBXW7 and proteasomal degradation of MYC (6, 8). The significance of these phosphorylation events has been characterized using alanine point mutations. Specifically, the T58A point mutant is more transforming than wild-type (WT) MYC in both cell line and in vivo models, which has been attributed to both an increase in protein stability and decreased apoptosis (8–13). Furthermore, a recent report using the MMTV-MYC transgenic model found fewer activating mutations in RAS when T58A was expressed compared with the WT allele (14), highlighting how gain-of-function phosphorylation mutants of MYC can be used to understand the cooperating pathways contributing to transformation. Clearly, signaling through T58 is an important mechanism to negatively regulate MYC-induced transformation and the T58A gain-of-function mutant has been an important experimental tool to understand signaling regulating MYC activity.

Despite the impact of the T58A mutant, additional gain-of-function MYC phosphorylation mutants have not been identified. This may, in part, be due to the general absence of somatic MYC mutations in human cancers. Although direct deregulation of MYC through point mutations is uncommon, understanding MYC regulation through the analysis of signaling mutants is a valuable experimental approach as many signaling pathways upstream of MYC are altered in cancers. A number of MYC phosphorylation sites have been identified through two-dimensional (2D) phosphopeptide mapping, yet the majority has not been functionally characterized (15–17). There are four described phosphorylation sites in the N-terminus of MYC including T58, S62, S71, and S81. Two additional clusters of residues have also been identified: T247/T248/S249/S250/S252 and T343/S344/S347/S348. There is also a single site at S293. S71 can be phosphorylated by JNK to promote MYC-induced apoptosis (9, 18). Both clusters of phosphorylation sites between amino acids 247–252 and 343–348 have been shown to be targets of casein kinase 2 (CK2; ref. 17), but to the best of our knowledge there have been no reports establishing a biologic role for signaling through these residues.

Herein, we assign biologic activity to these sites by conducting a thorough evaluation of these phosphorylation mutants on MYC-induced transformation. We have identified two mutants that show a gain of transformation compared with WT MYC: the double S71A/S81A mutant and the phosphorylation cluster mutant T343A/S344A/S347A/S348A (MYC-4A). Furthermore, investigation into regulation of these phosphorylation sites may provide new opportunities for therapeutic targeting of cancers with deregulated MYC.

MCF10A and SH-EP cells were cultured as described previously (13, 19).

Transformation was evaluated by anchorage-independent colony growth in soft agar using standard protocols and as previously described (13). Colony formation was imaged at, ×1.6 magnifications on a Leica Stereomicroscope (Leica MZ FLIII). Colony number and average colony size were quantified with ImageJ software (NIH, Bethesda, MD). For the combined statistical analysis of both colony number and size, as presented in Fig. 1C and D, Euclidean distance from c(0,0) to each data point was calculated and a subsequent paired Student t test analysis between the Euclidean distances for each mutant against the Euclidean distances for MYC was conducted.

MYC chromatin immunoprecipitation (ChIP) experiments were carried out as previously described (20). Briefly, 10⁷ cells were used per ChIP reaction with each 1.5 μg normal rabbit immunoglobulin G (IgG; Santa Cruz Biotechnology; sc-2027), or home-made purified and validated rabbit polyclonal N262. Whole genome-amplified DNA (4 μg; Genomeplex Whole Genome Amplification Kit, Sigma; WGA2) was hybridized to Agilent 2 × 244 promoter arrays at the University Health Network Microarray Center (Toronto, ON, Canada). ChIP-on-chip data was analyzed as described previously (8). Raw array data are publicly available through Gene Expression Omnibus (GSE14264).

MCF10A acini formation assays were completed as previously described (2). Morphogenesis was monitored and acini formation was imaged on an AxioObserver microscope at, ×5 magnifications.

Total RNA was isolated from day 4 acini using TRIzol (Invitrogen) as per the manufacturer's protocols. Expression was evaluated on GeneChip Human Genome U133 Plus 2.0 Arrays. Complete protocol is available through Affymetrix (www.affymetrix.com). Complete analysis methodology is provided in the Supplementary Materials and Methods. Raw array data are publicly available through Gene Expression Omnibus (GSE51123).

Assays were conducted using a Seahorse Bioscience XF96 instrument using standard protocols. For each biologic replicate the basal oxygen consumption rate (OCR) was measured at three time points for triplicate wells, averaged, and normalized to DNA content (CyQuantNF; Invitrogen) from a parallel plate. Please see supporting information in the Supplementary Material and Methods.

We first constructed a panel of MYC phosphorylation mutants by substituting serine and threonine residues for alanines. Given the requirement of phosphorylation at S62 for the subsequent phosphorylation of T58 and the knowledge that certain kinases can phosphorylate more than one residue within the N-terminus of MYC, we generated a panel of T58, S62, S71, and S81 mutants that contained every possible combination of these four N-terminal phosphorylation sites mutated to alanine (15, 18). As clusters of phosphorylation sites have been shown to be phosphorylated by the same kinase (21, 22) and/or regulate protein activity as a consequence of net charge rather than by the traditional site-specific manner (23), we created single mutants in which all five phosphorylation sites around amino acid 240 (MYC-5A) and all four phosphorylation sites near amino acid 340 (MYC-4A) were mutated to alanines (Fig. 1A). We also generated the single S293A mutant. In total, our panel contained 18 phosphorylation site mutants.

We evaluated our mutant panel by stable overexpression in nontransformed MCF10A mammary epithelial cells, as we have shown that expression of WT MYC alone is sufficient to promote anchorage-independent colony growth, which is further potentiated by the T58A point mutant (13). The cell panel was seeded in soft agar and evaluated for colony formation compared with WT MYC. This initial screen identified two mutants (S71A/S81A and MYC-4A) that formed larger colonies and no phosphorylation mutants that had reduced activity compared with WT MYC (data not shown). On a smaller panel of cells, comparable protein expression was confirmed (Supplementary Fig. S1A) and no change in 2D morphology was observed (Supplementary Fig. S1B). Soft agar colony formation experiments were quantified for both colony number and average colony size (Fig. 1B and C and Supplementary Fig. S1C and S1D). As we have reported previously, the T58A point mutation significantly increased colony number 1.8-fold compared with WT MYC (13). The S71A/S81A double mutant had a statistically significant 1.8-fold increase in average colony number and 1.5-fold increase in average colony size. Notably, the individual S71A and S81A mutants alone were insufficient to promote increased transformation, as measured by colony growth in soft agar. The MYC-4A exhibited a trend toward increased transformation compared with WT MYC (Supplementary Fig. S1C and S1D). In addition, with the MYC-4A mutant, there were also a number of macroscopic colonies consistently formed on each plate, a phenomenon rarely observed with WT MYC (Fig. 1B, upper panel). When both colony number and colony size were considered and compared with WT MYC, the T58A, S71A/S81A, and MYC-4A mutant were all significantly more transforming (P < 0.05; Fig. 1C). Combined, these results suggest that signaling through these residues negatively regulates MYC-induced transformation of MCF10A cells.

We next evaluated whether these results could be generalized beyond a single cell line model. We therefore evaluated these gain-of-function mutants in SH-EP neuroblastoma cells. Like the MCF10A cells, ectopic expression of MYC is sufficient to promote anchorage-independent growth of SH-EP cells (13). SH-EP cells stably expressing control, WT MYC and the MYC phosphorylation mutants (T58A, S71A/S81A, and MYC-4A) were generated with comparable protein expression across the panel (Supplementary Fig. S3A). Soft agar colony formation results were similar to MCF10A cells, in which T58A served as the positive control, the S71A/S81A mutant significantly increased average colony size 1.5-fold over WT MYC and the MYC-4A mutant showed a trend toward increased colony size (Fig. 1B, lower panel, and D and Supplementary Fig. S2B and S2C).

To confirm phosphorylation of these sites in vivo, we identified Myc phosphopeptides using mass spectrometry. Briefly, 293Tv cells were transiently transfected with a MYC expression plasmid, and MYC protein was immunoprecipitated and subjected to trypsin digestion. Phosphopeptides were enriched using titanium dioxide (TiO₂) or iron-based immobilized metal affinity chromatography material and were subjected to mass spectrometry analysis. Several putative serine and threonine phosphorylation sites were identified in tryptic peptides containing amino acids 343–348, including peptides phosphorylated at a single site or at multiple residues within this region (Supplementary Table S1). We further identified a peptide in multiple runs that contained a phosphorylation site on one of S67 or S71. We were, however, unable to assign the specific residue from the spectra. Phosphorylation of S71 has also been independently identified in two other studies (24, 25). Our analysis also identified peptides phosphorylated at S293. Together, these findings independently validate the previous MYC 2D phosphopeptide mapping studies (15–17).

To better understand the mechanisms of how our two phosphorylation mutants could be contributing to enhanced oncogenic activity, we evaluated MYC protein stability. Cycloheximide experiments revealed that the half-life of both the S71A/S81A and MYC-4A mutants was comparable with both endogenous and ectopic WT MYC, at approximately 30 minutes (Supplementary Fig. S3). Consistent with previous reports in other cell systems, the T58A point mutation increased MYC protein half-life from 30 to 60 minutes (26–28). These data suggest that the mechanism(s) responsible for the gain-of-function activity of S71A/S81A and MYC-4A mutants are independent of changes to MYC protein stability and signaling through T58.

To begin to understand the downstream mechanisms contributing to MYC-induced transformation of MCF10A cells, we conducted genome-wide MYC ChIP-on-chip experiments in asynchronously growing MCF10A cells. Our binding data revealed that both endogenous and ectopic WT MYC binds a large fraction of the genome, consistent with previous reports in other cell systems (Supplementary Table S2). Volcano plots representing the by-probe fold change in binding from MYC ChIP compared with IgG control are presented in Fig. 2A. Remarkably, genome-wide binding sites were very consistent between endogenous and ectopic Myc (Fig. 2B), suggesting that MYC-induced transformation is not mediated by an altered promoter-binding profile in these cells.

Next, we tested the relative ability of ectopic WT and mutant MYC to interact with MAX and bind candidate promoters. Specifically, the proximity of the MYC-4A residues to the B-HLH domain of MYC suggests that these interactions might be especially susceptible to changes in posttranslational modifications. Indeed, a previous study has shown that PAK2 phosphorylation of residues within the B-HLH domains regulate both MAX and DNA binding (29). To address this question, it was important to evaluate these parameters with minimal contributions from endogenous MYC. We therefore, serum-starved MCF10A cells for 2 hours before harvest, which depletes endogenous MYC expression and allows us to more effectively compare mutant versus ectopic WT MYC (see GFP input lane 1; Fig. 2C; Supplementary Fig. S4). MYC and all MYC mutants robustly and comparably bind MAX by coimmunoprecipitation (Fig. 2C). In addition, WT and all phosphorylation mutants bound a subset of candidate MYC target genes, both activated (CAD and APEX) and repressed (CDKN1B and SERPINE1), to similar extents as assayed by ChIP-quantitative PCR (qPCR; Fig. 2D). The negative controls for this experiment were an exonic E-box within the HNT gene on chromosome 11 (HNT) and nongenic, non-E-box on chromosome 6 (Ch6). Combined these data suggest that the mechanisms of increased transformation of these mutants are not mediated through an increased ability to bind MAX or DNA.

Growing MCF10As on a bed of extracellular matrix (Matrigel) allows cells to form polarized acinar structures that have many of the same characteristics of normal breast tissues in vivo (19). Overexpression of oncogenes, such as ERBB2, has been shown to disrupt the normal morphogenesis of MCF10A cells in three-dimensional (3D) culture (19), and this model has advanced our ability to study early events in breast tumorigenesis. We cultured MCF10A cells with stable overexpression of the phosphorylation mutant panel on Matrigel and evaluated morphogenesis. Cells with constitutive overexpression of MYC formed phenotypically normal acini, indistinguishable from cells expressing empty vector control (Fig. 3A and B). This contrast with a previous report suggesting that overexpression of MYC alone is sufficient to transform MCF10A acini (30), which may be a reflection of relative levels of ectopic MYC expression. In our system, the level of ectopic MYC expression is similar to that of endogenous MYC in asynchronously growing GFP control cells (Supplementary Fig. S1A; ref. 13). We further characterized MYC levels at different time points after seeding on Matrigel. To control for degradation of the highly labile MYC protein, we conducted denatured immunoprecipitation to enrich for MYC, then visualized expression by immunoblotting analysis and compared levels of expression relative with DNA content. These experiments show that both endogenous and ectopic MYC protein expression is high on day 4 in culture, but is similarly decreased on day 8 (Supplementary Fig. S4). These data show that both endogenous and ectopic MYC proteins are expressed at comparable levels and are similarly regulated through morphogenesis of MCF10A cells, as previously described (Supplementary Fig. S5; ref. 30). In cells overexpressing the T58A point mutant, which increases MYC protein stability and further deregulates MYC, there was robust and significant transformation of acini, including the formation of large multiacinar structures (Fig. 3A and B). The S71A/S81A and MYC-4A mutants both significantly and effectively promoted transformation of MCF10A acini, with the S71A/S81A mutant showing slightly more transformation than T58A itself (Fig. 3A and B).

With three MYC gain-of-function alleles, we were in a unique position to interrogate the molecular events associated with MYC-induced transformation of MCF10A acini. RNA was isolated from three independent biologic replicates on day 4 of 3D culture, a time point with comparable protein expression and before robust morphologic differences (Supplementary Fig. S6). RNA was evaluated on Affymetrix U133 Plus 2.0 mRNA microarrays. After standard data preprocessing, we first identified statistically significant mRNA abundance changes between cells expressing ectopic WT or mutant MYC and cells expressing GFP empty vector control. Consistent with the absence of morphologic changes, the WT MYC–overexpressing cells had the fewest significant gene expression changes (136 genes at q < 0.025; Table 1; Fig. 3C; Supplementary Table S3). All three gain-of-function mutants had a greater number of significant gene expression changes compared with control cells: T58A with 922 genes, S71A/S81A with 882, and MYC-4A with 402. More than 85% of the gene expression changes with WT MYC were also evident with each of the individual phosphorylation mutants (Fig. 3C), further concluding that these mutants represent a gain-of-function, as they alter the expression of the same genes as WT MYC and regulate many additional targets.

To understand the mechanisms contributing to MYC-induced transformation by the phosphorylation mutants, we conducted Gene Ontology (GO) term enrichment analysis on the significant gene expression changes (Table 2 and Supplementary Table S4). The most significant GO terms enriched with the S71A/S81A point mutant include those related to DNA synthesis and cell-cycle progression (Table 2 and Supplementary Table S4), suggesting that these residues might be important for negatively regulating cell-cycle control. Two of the top four most significantly enriched terms for the MYC-4A mutant were noncoding RNA processing and metabolism (Table 2 and Supplementary Table S4). These results suggest interesting avenues for further research about transforming mechanisms of MYC.

Comparing the gene expression changes among all MYC alleles was also informative. Remarkably, 82% (112 of 136) of the significant gene expression changes in WT MYC–expressing cells were also significantly altered by all of the phosphorylation mutants (Fig. 4A and Supplementary Table S5 and S6). These results suggest that these MYC-specific gene expression changes contribute to, but are insufficient for the transformation of MCF10A acini. Notably, there is a unique subset of 158 genes commonly regulated in response to the three gain-of-function mutants (Fig. 4A and Supplementary Table S6 and S7). Taken together, these 270 genes may represent a common program essential for the MYC-induced transformation of MCF10A cells. When these data were merged with our ChIP-on-chip data for WT MYC, over 70% of significant expression changes occur on genes that can be bound by MYC in MCF10A cells, suggesting that the majority of these changes are a result of direct transcriptional regulation (expression and binding data are compiled in Supplementary Table S8).

GO term enrichment analysis of both the common (112 genes) and phosphorylation mutant-specific (158 genes), as well as combined (270 genes) subsets revealed significant enrichment in a number of terms, including those involved in glutamine metabolism and purine and pyrimidine synthesis and metabolism, which are consistent with known functions of MYC (Fig. 4B; Table 3; Supplementary Table S9). There were also terms associated with cell–matrix attachment, epithelial morphogenesis, and epithelial-to-mesenchymal transition, processes that are fundamental to this culture system.

We next used these data to confirm whether the increased transformation exhibited with the phosphorylation mutants was a result of the increased number of gene expression changes, or whether it might also be mediated by a greater change in expression of the common or core subset of 112 genes. Focusing specifically on genes that contributed to the most significantly enriched GO terms, gene expression changes relative to GFP control–expressing cells are very similar for WT MYC and the MYC phosphorylation mutants (Fig. 4C), suggesting that it is the increased number of transcriptional targets regulated by the gain-of-function mutants that is driving the increased cellular transformation.

Reviewing our gene expression results, there was a common and frequent trend in enriched metabolic pathways. To evaluate cellular metabolism, we measured the OCR of MCF10A 3D cultures using a Seahorse Bioanalyzer. We conducted this experiment on both day 4 and 8 of culture. Although there were no robust differences on day 4 (data not shown), WT MYC and the gain-of-function mutants all increased the OCR on day 8, relative to GFP cells (Fig. 4D). These results further support a role for MYC deregulation in altering cell metabolism as a mechanism contributing to cellular transformation.

MYC has long been appreciated as a nuclear phosphoprotein, however, the functional significance of MYC phosphorylation at a number of residues remains to be established. Moreover, there have been a surprisingly limited number of studies investigating MYC signaling mutants. It is tempting to speculate that this mirrors the general lack of somatic MYC mutations found in human cancers, an observation that has been confirmed through the abundant tumor sequencing studies now being regularly published (31–34). Of these studies, the tumor type with the most frequent mutations is uterine corpus endometriod carcinomas, with a frequency of only 3.3% (8 of 240), and all mutations occurring at different residues (34). Important exceptions to this, however, are Burkitt- and AIDS-associated lymphomas, which have been long known to have frequent mutations in MYC exon 2, specifically with a hotspot at T58 (35–37). Intriguingly, while T58 to alanine substitutions are observed in Burkitt lymphomas, the more common mutation is T58I (summarized in ref. 37), the biologic significance of which remains elusive. Through studies of the T58A point mutant, it is clear that studying MYC signaling mutants can be used to uncover and better understand the upstream mechanisms regulating MYC. Our data also suggest that the field may have been challenged by inherent redundancy or the need for cooperating modifications, as the two novel gain-of-function phosphorylation mutants we have characterized, are composed of more than a single amino acid substitution. Although a caveat to our approach is that altering multiple amino acids may alter the overall structure of MYC, especially for the MYC-4A mutant, our mass spectrometry results clearly show that multiple residues in this cluster simultaneously undergo phosphorylation and structurally altered mutants generally lose, rather than gain, activity. Future studies to understand the interplay or cooperation between these residues as well as other known sites of MYC posttranslational regulation will be important (6, 38–40).

Through this work, we have independently validated previously reported phosphorylation sites and identified functionally important sites of MYC posttranslational regulation. The majority of studies investigating the downstream mechanisms of MYC-induced transformation model deregulation through robust overexpression of the WT protein. Although this effectively models MYC amplification in human tumors, it is not the only means of MYC deregulation. Alterations in upstream signaling pathways may not necessarily impact the abundance of MYC protein, but rather its posttranslational modification landscape thereby impacting regulation. Using the MCF10A model system, we have effectively modeled MYC deregulation independent of pathologic overexpression. With the established T58A gain-of-function allele as well as our two novel gain-of-function mutants, we have conducted thorough expression array analysis in a 3D cellular context, which effectively shows MYC-induced transformation. Our mRNA expression analysis further suggested that these mutants indeed represented gain-of-function mutants as they altered the expression of a core subset of MYC genes, but also many additional targets. Using gain-of-function mutants in this manner may help identify a signature for MYC deregulation in specific cell/tissue types. The observation that there are many significant gene expression changes unique to each gain-of-function mutants (Fig. 4A), also suggests that posttranslational modifications may play an important role in the regulation of specific subsets of Myc target genes and biologic functions downstream of MYC deregulation.

Our MYC:MAX CoIPs and MYC ChIP-qPCR show that the effect of the MYC phosphorylation mutants is independent of enhanced or altered binding to either MAX or DNA. The altered transcriptional programs downstream of the phosphorylation mutants therefore suggests that posttranslational regulation influences interactions between MYC and other transcriptional cofactors on target genes, and thereby contributes to transcriptional regulation of specific subsets of targets genes (Fig. 4E). Future work investigating the protein complexes formed both by phosphorylated and unphosphorylated MYC will be instructive. These results can then be extended to genome-wide binding and expression studies for both MYC and the necessary cofactors to advance our understanding of the mechanism of MYC-induced cellular transformation. Furthermore, as MYC activity is context-dependent, future work to optimize cell isolation from 3D culture for ChIP assays could make this a powerful system to investigate MYC activity alone and in cooperation with other interacting and regulatory proteins.

In conclusion, we have functionally characterized two novel MYC gain-of-function phosphorylation mutants, providing new evidence that the regulation of MYC occurs not only at the level of expression, but can also be directly controlled through signaling events. A better understanding of mechanisms contributing to MYC deregulation and MYC-induced transformation has the potential to identify novel approaches for the therapeutic targeting of MYC activity, either through the inhibition of upstream or downstream pathways.

No potential conflicts of interest were disclosed.

The views expressed in the article do not necessarily reflect those of the Ontario Ministry of Health and Long Term Care (OMOHLTC).

Conception and design: A.R Wasylishen, L. Huang, L.Z. Penn

Development of methodology: A.R Wasylishen, P.-K. Chan, P.J Mullen, L. Huang, N. Meyer, B. Raught, P.C. Boutros, L.Z. Penn

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.R Wasylishen, C. Bros, D. Dingar, W.B. Tu, M. Kalkat, P.-K. Chan, P.J Mullen, B. Raught, P.C. Boutros, L.Z. Penn

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.R Wasylishen, M. Chan-Seng-Yue, C. Bros, P.-K. Chan, P.J Mullen, B. Raught, P.C Boutros, L.Z. Penn

Writing, review, and/or revision of the manuscript: A.R Wasylishen, W.B. Tu, D. Dingar, B. Raught, P.C Boutros, L.Z. Penn

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Bros, P.-K. Chan, N. Meyer, B. Raught, P.C Boutros, L.Z. Penn

Study supervision: P.C Boutros

The authors thank Drs. Senthil Muthuswamy and Manfred Schwab for providing MCF10A and SH-EP Tet21/N-MYC cells, respectively. The authors also thank Dr. Senthil Muthuswamy and the members of the Penn Lab for helpful discussions and critical review of this article.

This research was funded by a grant from the Canadian Cancer Society Research Institute (018298, 020276; L.Z. Penn) and Canadian Institutes of Health Research (MOP-275788; L.Z. Penn and B. Raught), the support of the Ontario Institute for Cancer Research through funding provided by the Government of Ontario (P.C. Boutros), a Canadian Breast Cancer Foundation Ontario Region Doctoral Fellowship (A.R. Wasylishen), and an Ontario Graduate Scholarship (M. Kalkat). Additional support was provided by the OMOHLTC.