Our functional genomics studies demonstrate that the coordinate functions of Myc and Miz-1 go well beyond the regulation of the limited set of genes previously identified by a candidate target gene approach. Myc/Miz-1 function involves a much broader set of targets involved in the maintenance of ES cell biology and more specifically in ES cell differentiation. Our findings also implicate Hox
genes as key Myc target genes. Hox
genes are repressed by Polycomb group proteins, which catalyze methylation at histone H3 lysine 27 (H3K27me3) of Hox
genes to promote their repression in human ES cells [36
]. Differentiation of ES cells leads to erasure of H3K27me3 marks and upregulation of Hox
gene expression [37
]. In our study we found that Hox
gene expression is also repressed by the Myc/Miz-1 complex, which represents a novel mechanism of Hox
gene repression in stem cells. The Hox
genes targeted by Myc and Miz-1 possessed enrichment for H3K27me3 marks as well, suggesting a potential link between Myc and Polycomb as has been observed in Drosophila
Recent studies in mouse ES cells defined three functionally separate ES modules: Core (74 genes), Polycomb (449 genes), and Myc (355 genes) [39
] suggesting that Myc largely functions independently of Polycomb. We found that out of 4,398 Myc target genes identified in our study in human ES cells, 862 genes (20%) are also identified as Myc target genes in the study by Kim et al.
in mouse ES cells. We have identified 119 Myc target genes that fell into the Polycomb module, 84 target genes that fell into Myc module and 13 target genes belonging to the Core module. The enrichment of the genes identified as belonging to the Polycomb module in identified Myc target genes in our study further suggest the potential link between Myc and Polycomb in human ES cells. An additional study [40
] found that Myc exhibits substantial overlap in genomic binding in mouse ES cells with Polycomb binding, but Myc activated those Polycomb-targeted genes and had no effect on H3K27me3 levels. These findings and our data here leave open the precise role, if any, of Myc together with Polycomb in ES cells, but at least suggest that Myc and Polycomb are cooperative in ES cells, most likely not through regulating H3K27me3.
We found evidence of a complex relationship between Myc and cell survival in ES cells. The link between Myc and apoptosis is well documented [41
]. While Myc overexpression is linked to cell death in some contexts, so is the loss of Myc in other contexts [42
]. Our ChIP-chip study demonstrates that Myc directly controls the expression of a large number of cell death related genes. The absence of Myc, however, also leads to the activation of proapoptotic program, which can be attributed to the global dysregulation of a variety of cellular pathways that in turn elicits secondary response involving activation of a cell death program.
Our findings on localization of Miz-1 genomic binding almost exclusively away from TSS were particularly unexpected and of interest because the predominant model in the field is that Miz-1 exclusively targets INR sequences in the immediate proximity of TSS [1
]. However, these previous studies only examined core promoter regions of a few known candidate Miz-1 target genes, likely not fully representing the global distribution of Miz-1 genomic binding and function.
What might be the mechanism underlying the Miz-1 INR-independent function? One possibility is that Miz-1 is recruited to the upstream promoter region through interaction with other DNA-binding proteins. Alternatively, Miz-1 may directly bind as yet unknown motifs that are distinct from INRs in terms of both location and sequence. We were not able to identify any novel predicted consensus motifs for Miz-1 binding within Myc and Miz-1 cobound regions by MEME algorithms, but this may reflect the expected large size and complexity of Miz-1 DNA binding sites as it has 13 zinc finger DNA binding domains. Since we did not find Myc E-boxes significantly enriched in Myc/Miz-1 cobound sequences, it remains unclear if Myc binds DNA directly in such complexes.
More generally, we found two main types of Miz-1 genomic targets in human ES cells: (1) targets of putative Miz-1 activation that are associated strongly with euchromatic, transcriptionally active histone marks, and (2) targets of possible Miz-1 repression that lack or have very low relative levels of euchromatic marks. The pool of former, active genomic domains bound by Miz-1 comprises genes involved in regulation of metabolism, cell cycle control, chromatin and protein ubiquitination, indicating potential mechanisms by which Miz-1 regulates ES cell biology in a Myc-independent manner. A relatively small group of Miz-1 target genes (36 genes out of 734 genes) displayed simultaneous enrichment in activating and silencing histone modifications, including several developmentally regulated genes such as MEIS2, PAX6, GSC
. These genes might be poised, having 'bivalent domains', which harbor both activation and silencing marks [44
]. Myc binding to a subset of genes within 'bivalent domains' was previously reported [45
], but here we implicate Miz-1 in this process.
Another model is that Myc and Miz-1 are strictly antagonistic. Consistent with this model, Myc loss-of-function in human ES cells leads to increased expression of many Myc/Miz-1 targets. Interestingly, Myc and Miz-1 may also be antagonistic in another, novel way in the context of Myc activation. We found a statistically very significant group of Myc targets of activation that were not only downregulated by loss of Myc, but also upregulated by loss of Miz-1. If Miz-1 indeed functions to antagonize transactivation by Myc of a specific subset of target genes, it must be through an as yet new mechanism of action by Miz-1 as it has not been previously reported to repress transcription or recruit corepressors. The lack of a statistically significant population of genes that are either downregulated in both Myc and Miz-1 KD, or upregulated in both KDs, clearly demonstrates that Myc and Miz-1 do not act synergistically to either activate or repress the expression of their target genes. This further supports the model of Myc and Miz-1 generally being antagonists.