MTG16 (CBFA2T3) regulates colonic epithelial differentiation, colitis, and tumorigenesis by repressing E protein transcription factors

Aberrant epithelial differentiation and regeneration contribute to colon pathologies including inflammatory bowel disease (IBD) and colitis-associated cancer (CAC). MTG16 (CBFA2T3) is a transcriptional corepressor expressed in the colonic epithelium. MTG16 deficiency in mice exacerbates colitis and increases tumor burden in CAC, though the underlying mechanisms remain unclear. Here, we identified MTG16 as a central mediator of epithelial differentiation, promoting goblet and restraining enteroendocrine cell development in homeostasis and enabling regeneration following dextran sulfate sodium (DSS)-induced colitis. Transcriptomic analyses implicated increased E box-binding transcription factor (E protein) activity in MTG16-deficient colon crypts. Using a novel mouse model with a point mutation that disrupts MTG16:E protein complex formation (Mtg16P209T), we established that MTG16 exerts control over colonic epithelial differentiation and regeneration by repressing E protein-mediated transcription. Mimicking murine colitis, MTG16 expression was increased in biopsies from patients with active IBD compared to unaffected controls. Finally, uncoupling MTG16:E protein interactions only partially phenocopied the enhanced tumorigenicity of Mtg16-/- colon in the azoxymethane(AOM)/DSS-induced model of CAC, indicating that MTG16 protects from tumorigenesis through additional mechanisms. Collectively, our results demonstrate that MTG16, via its repression of E protein targets, is a key regulator of cell fate decisions during colon homeostasis, colitis, and cancer. Graphical Abstract


Introduction 66
The colonic epithelium is a complex, self-renewing tissue comprised of specialized cell 67 types with diverse functions (1). Stem cells at the base of the colon crypt divide and differentiate 68 into absorptive and secretory cells. Conversely, colonic epithelial regeneration in response to 69 injury occurs through de-differentiation of committed absorptive and secretory lineage cells (2-70 4). Secretory lineage dysregulation is implicated in inflammatory bowel disease (IBD), diabetes, 71 and even behavioral changes (5-9), and aberrant regenerative programs may lead to 72 uncontrolled cell proliferation and dysplasia (10). Thus, these processes must be tightly 73 controlled by coordinated networks of transcriptional activators and repressors (1). 74 induced inflammatory carcinogenesis and that these were epithelial-specific phenotypes (18). 82 However, the precise mechanisms by which MTG16 controls colonic homeostasis, 83 regeneration, and tumorigenesis remain unknown. 84 One family of transcription factors repressed by MTG16 are basic helix-loop-helix 85 (bHLH) transcription factors (14,(19)(20)(21)(22). Ubiquitously expressed (class I) and tissue-specific 86 (class II) bHLH transcription factors dimerize to control stem cell dynamics, lineage allocation, 87 and differentiation (23-25). Class I bHLH transcription factors are also known as E proteins 88 because they bind to consensus Ephrussi box (E box) DNA sequences and include E12 and 89 E47 (splice variants of E2A), HEB, and E2-2 (22, 23). However, the functions of E proteins have 90 mainly been described outside of the colon (26-32). 91 In this study, we defined the topology of Mtg16 expression in the colon and discovered 92 new functions of MTG16 in the colonic epithelium. We used a novel MTG16:E protein 93 uncoupling point mutant mouse model to demonstrate that the mechanism driving these 94 phenotypes was repression of E protein transcription factors by MTG16. We tested the 95 functional impact of this regulatory relationship in murine models of IBD and colitis-associated 96 cancer (CAC) and correlated our observations with patient data. Overall, we demonstrate novel, 97 context-specific roles for MTG16 in colonic epithelial lineage allocation and protection from 98 colitis and tumorigenesis that depend on its repression of E protein-mediated transcription. 99 Thus, we hypothesized that MTG16 drives goblet cell differentiation from secretory progenitors 175 by inhibiting E protein-mediated transcription of key enteroendocrine transcription factors. 176 We next tested this hypothesis by selectively disrupting the MTG16:E protein interface in 177 vivo. To do so, we leveraged a CRISPR-generated mouse with a point mutation resulting in a 178 proline-to-threonine substitution in MTG16 (Mtg16 P209T Fig. 3A-C). We next performed bulk RNA-seq on proximal 184 and distal colon crypt isolates. In both colon segments, goblet cell genes were downregulated 185 compared to WT, while enteroendocrine cell genes were upregulated compared to WT, in a 186 pattern resembling that of Mtg16 -/colon ( Fig. 3D-E). Notably, the distal colon exhibited 187 significant upregulation of Neurog3 (Fig. 3E). Additionally, GSEA of the overall enteroendocrine 188 signature, key enteroendocrine progenitor genes, Bmi1 + and mTert + stem cell signatures, and E 189 protein signatures all largely phenocopied the Mtg16 -/distal colon ( Fig. 3F-H including a large peak at an E box-rich site upstream of Neurog3 (Fig. 4B), suggesting that 198 these genes are repressed by MTG16. Finally, Neurog3 + enteroendocrine progenitor cells were 199 markedly increased in Mtg16 -/and Mtg16 T/T colon crypts by RNAscope (Fig. 4C-D) and healthy control patients (60). We found that MTG16 was upregulated in UC patients in both 210 datasets ( Fig. 5A-B). MTG16 expression decreased with remission, although remained 211 elevated compared to unaffected control patients (Fig. 5B). Lastly, a microarray dataset from 212 pediatric colon biopsies (61) demonstrated upregulation of MTG16 in both UC and Crohn's 213 colitis (Fig. 5C). Collectively, these data show that increased MTG16 expression is correlated 214 with active injury and regeneration in IBD. 215 MTG16 contributes to colon crypt regeneration following DSS-induced colitis. As stated 216 previously, MTG16 deficiency increases disease severity in DSS-induced colitis (17), a mouse 217 model of IBD (schematic in Fig. 5D) that recapitulates histologic features of human UC (62). We 218 first assessed Mtg16 expression in DSS-induced colitis to determine whether it would 219 recapitulate the increased MTG16 expression in human IBD. Because DSS treatment largely 220 affects the distal colon (62), we compared distal WT colon crypt epithelial isolates post-DSS-221 induced injury and regeneration to WT distal colon crypts at baseline. Indeed, Mtg16 expression 222 was significantly increased in regenerating colon crypts (Fig. 5E) differentiating population (2) yielded a peak at a second transcription start site of Mtg16,231 suggesting that ASCL2 induces Mtg16 expression during colonic epithelial regeneration (Fig.  232

5F). 233
To confirm that the ASCL2-driven regeneration program was relevant to crypt 234 regeneration following DSS-induced injury, we constructed a gene set consisting of the 238 235 genes upregulated in Ascl2 + regenerating cells (2) ( Table S3) Fig. 5J-K), leading us to hypothesize that the contribution of 243 MTG16 to colon crypt regeneration occurs via repression of E protein-mediated transcription. 244

MTG16-mediated colonic epithelial regeneration is dependent on its repression of E 245
proteins. To test whether MTG16-dependent regeneration occurs via an E protein-dependent 246 mechanism, we next analyzed colonic epithelial regeneration following DSS-induced injury (Fig.  247 6A) in Mtg16 T/T vs. WT mice. Like Mtg16 -/mice, Mtg16 T/T mice developed worse colitis and poor 248 regeneration measured by increased weight loss (Fig. 6B), decreased colon length (Fig. 6C), 249 and increased histologic injury-regeneration score (Fig. 6D, Table S2) compared to WT. 250 Mtg16 T/T mice also exhibited increased extent of injury along the length of the Mtg16 T/T colon 251 (Fig. 6E). Next, we injected mice with 5-Ethynyl-2'-deoxyuridine (EdU) 1 h prior to sacrifice 252 following DSS injury and regeneration and found that Mtg16 T/T mice had fewer S-phase cells in 253 injury-adjacent crypts (Fig. 6F), indicating a proliferation defect in the crypts that actively 254 regenerate the ulcerated colonic epithelium following DSS-induced colitis (63). Finally, like 255 induced CAC (18). We next tested whether the mechanism driving these phenotypes was E 274 protein-dependent by treating Mtg16 T/T mice with AOM/DSS (Fig. 7C). Llike Mtg16 -/mice, 275 Mtg16 T/T mice did exhibit increased histologic injury, tumor multiplicity, and tumor size compared 276 to WT (Fig. 7D-G). Interestingly, Mtg16 T/T mice had more tumors and larger tumors in the 277 middle and proximal colon than their WT counterparts ( Fig. 7L-K). This was consistent with a 278 greater extent of DSS-induced injury (Fig. 6E). These data suggest that this phenotype could be 279 driven by increased injury in the proximal colon. Mtg16 T/T tumors (Fig. S8). Overall, these data indicate that uncoupling MTG16:E protein 287 complexes recapitulates some, but not all, Mtg16 -/phenotypes, suggesting that MTG16 protects 288 from tumorigenesis through additional mechanisms. 289

Discussion 290
In this study, we investigated the mechanism underlying context-specific functions of the 291 transcriptional corepressor MTG16. We identify previously unappreciated roles for MTG16 as a 292 critical regulator of colonic secretory cell differentiation and active colon regeneration. To define 293 the mechanism underlying these phenotypes, we utilized a novel mouse model with a single 294 point mutation in MTG16, Mtg16 P209T , that disrupts MTG16 binding to the E proteins E2A and 295 HEB(20). Mtg16 T/T mice homozygous for this MTG16:E protein uncoupling mutation largely 296 recapitulated Mtg16 -/lineage allocation and regeneration phenotypes, although the effect of 297 MTG16 loss in dysplasia appears to only partially be due to increased E protein activity. Thus, 298 these data indicate a new role for MTG16:E protein complexes in the colonic epithelium. 299 First, we demonstrated a novel role for MTG16 in colonic epithelial homeostasis. MTG16 300 has been shown to repress stem, goblet, and enteroendocrine cell genes to drive enterocyte 301 differentiation in the SI (16). However, in accordance with numerous studies demonstrating 302 important differences between SI and colon biology (2, 35, 67-70), Mtg16 -/colon did not exhibit 303 increased stem cell transcriptional signatures or an overall decrease in secretory genes. 304 Instead, we observed specific upregulation of the enteroendocrine lineage, occurring at least in 305 part due to unrepressed expression of Neurog3, the class II bHLH transcription factor required 306 for enteroendocrine cell differentiation (42-45). This could be due to the fact that MTG16 307 repression targets, and its effects on them, vary in different tissues and cell types due to 308 expression gradients of its binding partners (14). Another potential explanation for the difference 309 between SI and colonic phenotypes is that Neurog3 + progenitor cells in the SI may still increased enteroendocrine cells, and increased Neurog3 tone. One potential non-E protein-322 mediated pathway that was increased (FDR q-value = 0.11) in Mtg16 -/-, but not Mtg16 T/T colon 323 crypts, is the non-canonical, β-catenin-independent WNT-planar cell polarity (PCP) pathway. 324 This could be responsible for the slight difference in magnitude between Mtg16 -/and Mtg16 T/T 325 phenotypes, as the WNT-PCP pathway may enable enteroendocrine cells to differentiate 326 directly from "unipotentially-primed" intestinal stem cells 99 . 327 An area of considerable interest in gastrointestinal biology is the relationship between 328 epithelial differentiation and regeneration (4). Although it was long thought that dedicated Bmi1 + 329 or mTert + "reserve" stem cells at the +4/+5 position replenished the stem cell compartment in 330 intestinal epithelial regeneration, recent data indicate that these cells are enteroendocrine cells 331 that can de-differentiate into stem cells following injury (46). Recently, "committed" progenitors 332 from both the secretory and absorptive lineages were demonstrated to share this ability through 333 induction of ASCL2 (2). In this study, Ascl2 + de-differentiating cells were isolated from the 334 colonic epithelium following CBC ablation (2). Interestingly, these cells were not enriched for Clu 335 or fetal epithelial genes (2), previously thought to enhance intestinal epithelial response to injury 336  In conclusion, we discovered key new functions of MTG16 in colonic secretory lineage 363 allocation, regeneration following colitis, and colitis-associated tumorigenesis, dependent on its 364 repression of E proteins. Confirming translational relevance, we determined that MTG16 is 365 upregulated in patients with active IBD, reduced with restitution, and decreased in dysplasia. 366 Thus, MTG16 may be a candidate biomarker for disease activity or a target to modulate 367 differentiation and regeneration. 368

Methods 369
Animal models. Mtg16 -/mice were previously generated and validated (78). Mtg16 P209T  performed H&E and PAS staining. Chromogenic and immunofluorescent IHC were performed 392 as previously described (77, 84). Briefly, 5-µm FFPE sections were deparaffinized, rehydrated in 393 graded ethanol concentrations, and permeabilized using TBS-T (Tris-buffered saline with 0.05% 394 Tween-20) followed by antibody-specific antigen retrieval (Table S1). Next, for chromogenic 395 IHC, endogenous peroxidases were quenched in 0.03% H2O2 with sodium azide for 5 min. 396 Slides were then incubated in primary antibody (Table S1)   pathologist blinded to genotype (MKW) (scoring system described in Table S2). Distal colon 427 crypts were collected as described above for RNA isolation and RNA-seq as described below.   Table S3). RNA-seq of human CRC and CAC 461 samples collected from 9 regional hospitals in Finland was performed and analyzed as 462