Society: AGA

The speakers in this session will provide insights into diverse molecular mechanisms of metaplasia, dysplasia and neoplasia across the esophagus, stomach, and colon.

Obesity epidemic is associated with increased colon cancer (CC) risk and progression. One of the mechanisms of obesity mediated CC progression may include Lipid Droplets (LDs). They are dynamic organelles storing intracellular triacylglycerols (TG) and providing fuel for diverse cell functions. LDs initiation and growth is mediated by LDs- coat acyltransferases, DGAT1/2, that catalyze the final step in TG synthesis. We/others demonstrated elevated LDs in colonic tumors (vs normal). Further, we identified a self-reinforcing negative regulatory loop involving LDs and FOXO3 that drives metabolic and tumorigenic remodeling in CC. Here, we aim to elucidate the mechanisms of DGATs mediated LDs dynamics that facilitate CC in obesity. Methods. Human CC: Tissue array, Local CC patients; TCGA and NCBI’s GEO databases, Human colon cells (CCC): HT29, HCT116, SW620, SW480; NCM460 (non-transformed), Colonospheres; Mouse: APC^min/+ colon cancer model, AOM/DSS, High Fat Diet (HFD)- obese; Oleic acid (OA, 300mM); Enzymatic inhibitors: DGAT1- A922500, DGAT2- PF-06424439; Differentially expressed genes (DEGs), RNA/cDNA, CUT&RUN, RNAseq (Illumina HiSeq 2500), RSEM/EBseq DEGs analysis, Ingenuity Pathway Analysis (IPA), Kaplan-Meier survival estimates; IHS; IFS; IB; qPCR; ChIP assay; Colony formation assay; FACS; Luciferase-Reporter Assay. Results. We found significantly higher DGATs levels in human CC tissue (vs normal) that is augmented by both the severity of the disease and obesity. Similarly, their levels are increased in CCC (vs non-transformed cells) and augmented by obesity mediator (OA). This increase in DGATs expression is mediated by MYC activation and binding to DGAT2 promoter (Fig 1A-I). Moreover, blockade of DGATs shut down the self-reinforcing regulatory loop between LDs and FOXO3. Specifically, it protected FOXO3 by attenuating PI3K and lowered LDs accumulation by reducing MYC dependent DGATs expression. Further, blockade of DGATs lowered OA facilitated growth of CCC and colonosphere (cell cycle G0/G1 arrest via FOXO3/p27kip1 protection). Moreover, treatment with DGATs inhibitors reduced colonic tumor burden in APC^min/+ mouse, on regular diet and HFD (obese). Further, this inhibition attenuated DEGs linked to metabolic and tumorigenic remodeling in CCC, colonospheres and colon of HFD-obese APC^min/+ mouse (IPA, FDR<0.05). It additionally targeted pathways that are activated in human CC and colon cancer crypts (IPA, FDR<0.05) (Fig 2A-R). Selected DEGs were validated in CCC, colonospheres, and APC^min/+ mouse colon (qPCR, p<0.05); their similarly altered transcriptional levels were found in human CC. Conclusions. We identified a novel mechanism of DGATs dependent metabolic and tumorigenic remodeling mediated by LDs and FOXO3 regulatory loop in obesity facilitated CC. This establishes a platform for new diagnostics and for development of effective treatments.

Figure 1A-I

Figure 2A-R

Background: Colorectal cancer (CRC) is highly heterogeneous at the genetic and molecular level, where immunotherapy with checkpoint inhibition has shown potent antitumor activity in microsatellite instability (MSI) metastatic cancer, while microsatellite stable (MSS) CRC has long been considered resistant to immunotherapy. Given that metastatic sites may predict the response or resistance to checkpoint blockade, additional efforts are required to study the mechanism of MSS CRC metastasis and to develop novel therapeutic strategies to overcome its immune resistance. Here we report a novel role of HNF4A in MSS CRC progression.
Methods: HNF4A/CXCL12/TGFBI expression was examined by qRT-PCR, Western blot and IHC staining in MSS CRC tissues and cells. The clinical significance of HNF4A/CXCL12/TGFBI in two independent cohorts of CRC was assessed by KM analysis and the Multivariate Cox hazards model. The biological roles of HNF4A in tumor growth and metastasis were detected both in vivo and in vitro. The Inflammatory Cytokines & Receptors Profiler was used to explore the targets of HNF4A in MSS CRC. Serial deletion, site-directed mutagenesis luciferase report assays and chromatin immunoprecipitation were used to determine transcriptional regulation of CXCL12 promoters by HNF4A. MDSCs were isolated by flow cytometry and the Gr-1 monoclonal antibody was utilized to block the recruitment of MDSCs into tumors.
Results: HNF4A expression was upregulated in MSS CRC and negatively correlated with PDL1 expression. Upregulation of HNF4A promoted the proliferation and metastasis of MSS CRC cells and indicated poor prognosis. Meanwhile, HNF4A overexpression facilitates immunosuppression, with accumulated M0 and Tregs while reduced CD8+T, aNK, aDendritic, Monocytes and Neutrophils. The levels of HNF4A are positively correlated with the numbers of M0, Tregs and memory B cells, whereas negatively correlated with CD8+T, aNK, aDendritic, Neutrophils, Eosinophils and M1. Furthermore, CXCL12 was found to be the most upregulated cytokine upon HNF4A overexpression in MSS CRC cells. The proportion of MDSCs was remarkably enhanced in HNF4A-high CRC tissues of mice. While the use of Gr-1 antibody or CXCL12 knockdown rescued HNF4A-mediated CRC metastases. In addition, TGFBI/PI3K-AKT signaling was hyperactive in MSS CRC. TGFBI-induced HNF4A expression was blocked by PI3K inhibitor. The combination of the Gr-1 antibody and PI3K inhibitor attenuated HNF4A-mediated MSS CRC metastasis.
Conclusions: TGFBI-induced HNF4A overexpression promotes MSS CRC metastasis by transactivating the cytokine CXCL12 and thereby recruiting MDSC cells to suppress anti-tumor immunity in MSS CRC. The combination of Gr-1 antibody and PI3K inhibitor attenuated HNF4A-mediated MSS CRC metastasis, suggesting that HNF4A and CXCL12 as promising prognostic biomarkers and novel therapeutic targets for MSS CRC.

Figure 1. Upregulated HNF4A expression promoted the proliferation and metastasis of MSS CRC cells, predicting poor prognosis in CRC patients. (A-B)Differentially expressed genes between MSI and MSS CRC. (C-D)HNF4A expression was upregulated in MSS CRC and negatively correlated with PDL1 expression. (E-F)qRT-PCR and Immunohistochemistry staining analysis of HNF4A expression in adjacent non-tumorous and CRC tissues. (G-I)CRC patients with high HNF4A levels had an increased recurrence rate and decreased survival rate. HNF4A overexpression was positively correlated with lymphatic and distant metastasis of CRC. HNF4A upregulation was proved as an independent prognostic factor for CRC. (J-K)CCK-8 and transwell analysis of CRC cell proliferation and migration. (L-M) <i>In vivo</i> tumorigenesis and metastasis assay. Representative bioluminescent imaging, the overall survival, representative H&E staining of the liver and lung tissues, and the number of metastatic foci in different groups are shown.

Figure 1. Upregulated HNF4A expression promoted the proliferation and metastasis of MSS CRC cells, predicting poor prognosis in CRC patients. (A-B)Differentially expressed genes between MSI and MSS CRC. (C-D)HNF4A expression was upregulated in MSS CRC and negatively correlated with PDL1 expression. (E-F)qRT-PCR and Immunohistochemistry staining analysis of HNF4A expression in adjacent non-tumorous and CRC tissues. (G-I)CRC patients with high HNF4A levels had an increased recurrence rate and decreased survival rate. HNF4A overexpression was positively correlated with lymphatic and distant metastasis of CRC. HNF4A upregulation was proved as an independent prognostic factor for CRC. (J-K)CCK-8 and transwell analysis of CRC cell proliferation and migration. (L-M) In vivo tumorigenesis and metastasis assay. Representative bioluminescent imaging, the overall survival, representative H&E staining of the liver and lung tissues, and the number of metastatic foci in different groups are shown.

Figure 2. TFGBI-induced HNF4A promotes metastasis of CRC by recruiting MDSC via CXCL12. (A-B)Upregulation of HNF4A facilitates immunosuppression, with increased M0 and Tregs while decreased CD8+T, aNK, aDendritic, Monocytes and Neutrophils in HNF4-high CRC. The levels of HNF4A are positively correlated with the numbers of M0, Tregs and mB cells, while negatively correlated with CD8+T, aNK, aDendritic, Neutrophils, Eosinophils and M1. (C)Cytokine microarray showed that CXCL12 was significantly upregulated upon HNF4A overexpression. The proportion of MDSC was remarkably enhanced in HNF4A-high CRC tissues of mice. (D-G)Representative BLI, tumor growth, overall survival, the number of metastatic foci and H&E staining in different groups are shown. (H)TGFBI was significantly upregulated in MSS CRC cells. (I)The PI3K-AKT pathway was hyperactive in MSS CRC. (J)TGFBI-induced HNF4A expression in a dose-dependent manner. (I)Blocking the PI3K-AKT pathway inhibits TGFBI-induced HNF4A expression.

Figure 2. TFGBI-induced HNF4A promotes metastasis of CRC by recruiting MDSC via CXCL12. (A-B)Upregulation of HNF4A facilitates immunosuppression, with increased M0 and Tregs while decreased CD8+T, aNK, aDendritic, Monocytes and Neutrophils in HNF4-high CRC. The levels of HNF4A are positively correlated with the numbers of M0, Tregs and mB cells, while negatively correlated with CD8+T, aNK, aDendritic, Neutrophils, Eosinophils and M1. (C)Cytokine microarray showed that CXCL12 was significantly upregulated upon HNF4A overexpression. The proportion of MDSC was remarkably enhanced in HNF4A-high CRC tissues of mice. (D-G)Representative BLI, tumor growth, overall survival, the number of metastatic foci and H&E staining in different groups are shown. (H)TGFBI was significantly upregulated in MSS CRC cells. (I)The PI3K-AKT pathway was hyperactive in MSS CRC. (J)TGFBI-induced HNF4A expression in a dose-dependent manner. (I)Blocking the PI3K-AKT pathway inhibits TGFBI-induced HNF4A expression.

Introduction
Tuft cells represent a rare population of epithelial cells in the gastrointestinal tract to sense diverse inflammatory conditions to trigger immune responses. Tuft cells have also been shown to expand in response to oncogenic Kras activation in the pancreas and influence tumor progression. However, few studies have examined tuft cells in the gastric pathologies. We have previously generated Lrig1-Kras and Mist1-Kras mouse models to recapitulate gastric metaplasia to dysplasia progression. Here, we used these models to investigate changes in tuft cell population upon Kras activation in the stomach. In addition, we examined tuft cells in various human gastric lesions including gastritis, intestinal metaplasia (IM) and gastric tumors.
Method
Lrig1-CreERT2;LSL-Kras(G12D) (Lrig1-Kras) mice were treated with 2 mg of tamoxifen and sacrificed 2 months after tamoxifen injection. Mist1-CreERT2;LSL-Kras(G12D) (Mist1-Kras) mice were treated with 5 mg of tamoxifen and sacrificed at 1, 3, and 4 months after tamoxifen injection. Gastric organoids were generated from Mist1-Kras mice at 3 and 4 months after tamoxifen treatment. Tumor formation assay was performed with subcutaneous flank injection in the nude mice. Immunostaining using antibodies against Dclk1, POU2F3, ChAT was performed to examine changes in tuft cell populations.
Result
Two months after induction of active Kras in Lrig1+ cells, foveolar hyperplasia developed and tuft cells increased in number compared to control mice, whereas neuroendocrine cells decreased (Fig 1). Mist1-Kras mice displayed progression from SPEM to IM and dysplasia and tuft cells substantially increased in the metaplastic mucosa. Organoids derived from dysplastic glands were implanted into nude mice and engraftments at 7 and 13 weeks after injection were harvested. Interestingly, tuft cells were mostly observed in the cystic or well-differentiated tubular glands with mucin expression and low Ki-67, while rarely seen in the poorly-differentiated glands. This pattern was consistently observed in tumors formed by re-implanting tumor cells in the nude mice. With POU2F3 and ChAT staining, we found that tuft cells are normally absent in human antrum and corpus (Fig 2). Tuft cells began to appear in the mucosa with inflammation and IM. Compared to tuft cells in the normal intestine and colon, which are usually POU2F3 and ChAT double-positive, tuft cells in the gastric mucosa with gastritis or IM were often POU2F3 single-positive. Tuft cells were frequently observed in gastric adenomas, however they become very rare in gastric cancers.
Conclusion
Our study demonstrates that tuft cells expand upon Kras activation during gastric metaplasia and dysplasia in mice. In the human stomach, tuft cells also increased in association with inflammation, metaplasia and benign tumors, but are then markedly decreased in gastric cancers.

<b>Fig. 1 Expansion of Dclk1-positive cells in Lrig1-Kras mice.</b> (<b>A)</b> Representative images of H&E and co-immunostaining for Dclk1, UEA1, P120 or Chromogranin A (CgA) in the corpus from control (Ctl) and Lrig1-Kras mice. (<b>B)</b> Quantitation of the number of Dclk1-positive and CgA-positive cells in control (n = 3) and Lrig1-Kras (n = 3) mice. (<b>C</b>) Co-immunostaining for Dclk1, Ki-67, and UEA1 in control and Lrig1-Kras mice.

Fig. 1 Expansion of Dclk1-positive cells in Lrig1-Kras mice. (A) Representative images of H&E and co-immunostaining for Dclk1, UEA1, P120 or Chromogranin A (CgA) in the corpus from control (Ctl) and Lrig1-Kras mice. (B) Quantitation of the number of Dclk1-positive and CgA-positive cells in control (n = 3) and Lrig1-Kras (n = 3) mice. (C) Co-immunostaining for Dclk1, Ki-67, and UEA1 in control and Lrig1-Kras mice.

<b>Fig. 2 Newly emerged tuft cells in human gastric lesions. (A) </b>Co-immunostaining for POU2F3, Choline acetyltransferase (ChAT), and VILLIN (VIL) in normal small intestine and colon. (<b>B</b>) Co-immunostaining for POU2F3 and ChAT in the inflamed gastric mucosa with prominent foveolar hyperplasia. (<b>C</b>) Co-immunostaining for POU2F3, ChAT, and VIL in the gastric mucosa with intestinal metaplasia.

Fig. 2 Newly emerged tuft cells in human gastric lesions. (A) Co-immunostaining for POU2F3, Choline acetyltransferase (ChAT), and VILLIN (VIL) in normal small intestine and colon. (B) Co-immunostaining for POU2F3 and ChAT in the inflamed gastric mucosa with prominent foveolar hyperplasia. (C) Co-immunostaining for POU2F3, ChAT, and VIL in the gastric mucosa with intestinal metaplasia.

Background: Gastroesophageal Reflux Disease (GERD) is a common digestive disorder that affects approximately 20% of the adult population in North America and is considered to be one of the strongest risk factors for esophageal adenocarcinoma. The exposure of esophageal cells to chronic gastroesophageal reflux induces extensive DNA damage through the formation of reactive oxygen species (ROS) causing genomic instability and cancer. A surveillance mechanism known as DNA damage response (DDR) ensures proper repair of the DNA lesions. In our previous study, we found a unique way of protein adduction by isolevuglandins (isoLGs) in precancerous conditions of the esophagus. One of the most adducted proteins is p53 that controls a critical branch of the DDR. In this study, we investigated how the protein adduction affects DNA damage response in conditions of esophageal reflux.
Methods and Results: For the first time, we found the adduction of multiple proteins by isoLGs in the esophagus of GERD patients and esophagojejunostomy mice that recapitulates human GERD conditions using immunohistochemistry. IsoLGs are formed through lipid peroxidation caused by ROS and enzymatic cyclooxygenation of polyunsaturated fatty acids. IsoLGs are highly reactive with free amines on lysine residues forming LG-lysine lactam protein adducts. Despite significant DNA damage caused by treatment with acidic bile salts (ABS), the p53-associated DDR was not activated in normal and precancerous esophageal cells. In contrast to p53, DNA damage induced by ABS led to a strong upregulation of p73 protein that shares some similar p53 targets, in the same cells, confirming a selective inhibitory effect of ABS on p53. Investigating the consequence of p53 protein adduction using focus array (84 key genes of human p53 signaling pathway), qPCR, cell cycle analysis and Chromatin immunoprecipitation sequencing (ChIP-seq) showed that the modification of p53 protein with isoLGs diminishes the genome-wide activation of p53 target genes that are critical for cell cycle arrest, apoptosis and DNA repair. Using dot blot, double-immunofluorescence, native gel, proteostat aggregation assay and amyloid staining, we found the accumulation of adducted p53 protein in intracellular amyloid-like aggregates both in vitro and in vivo. We also tested the isoLG scavengers, such as 2-hydroxybenzylamine (2-HOBA), and found that these compounds are able to restore p53-associated DDR in esophageal cells.
Conclusions: Combined, our studies revealed, for the first time how isoLGs regulate DDR in esophageal cells exposed to reflux by inactivating p53 that may play a previously unrecognized role in human tumorigenesis. Our work also helps to explain how chronic reflux induce tumorigenic process in the esophagus and opens new opportunities for the development of novel cancer chemopreventive agents.

Background: C-to-U RNA editing is mediated by APOBEC1 and two RNA binding proteins (A1CF and RBM47). To date only a single target (ApoB) has been functionally characterized in mouse and human tissues. We previously demonstrated reduced polyp burden in Apc^min/+ mice crossed into Apobec1^–/– mice, but the mechanisms and targets are unknown. Approach: We crossed Apc^min/+ mice with conditional Rbm47 intestinal knockout mice (Apc-RIKO) and with intestinal Apobec1 transgenic (Apc-A1tg) to examine gain- and loss-of-function (GOF/LOF) phenotypes (1a). RNA from Apc^min/+ polyp and uninvolved mucosa (UM) were pooled from >5 mice and deep sequenced. Genomic DNA was examined in parallel and variants were Sanger-sequenced to verify C-to-U RNA editing. We also examined tissue from patients with familial adenomatous polyposis (FAP) and sporadic colorectal cancer (CRC). Findings: We observed 29 newly identified RNA editing targets in Apc^min/+ mice, of which 2 were edited only in polyp, 11 edited only in UM and 16 altered in both UM and polyp (1b). All 29 RNA targets were APOBEC1-dependent, but only 16 were edited in Apc-A1tg mice (5 at higher frequency, 4 at lower frequency and 7 at similar frequency than Apc^min/+ mice) suggesting a key stoichiometric requirement for APOBEC1. 23/29 RNA targets were RBM47-dependent as demonstrated in Apc-RIKO mice (1b). All C-to-U RNA editing events were localized in the 3’ untranslated regions of target RNAs with no coding variants or truncations. GOF Apc-A1tg mice demonstrated increased polyp burden while LOF Apc-RIKO mice demonstrated larger polyps but similar overall burden (1a). We found no correlation between editing efficiency and steady state mRNA expression by QPCR. However, we observed reduced mRNA and protein abundance of a novel RBM47- and APOBEC1-dependent target Flnb (95% edited in Apc^min/+ mice) in Apc-RIKO mice (2a). We generated a composite RNA sequence/folding prediction model and examined candidate RNAs from UM and polyp from FAP patients. We found 2 RNA targets with 60% sequence identity between human and mouse orthologs over a 120 nt-sequence surrounding the editing site containing a requisite cis-acting motif but neither candidate target (FLNB and BCLAF1) demonstrated C-to-U RNA editing, suggesting findings in Apc^min/+ mice are likely species-specific. However, we observed that FNLB mRNA abundance in UM from FAP patients correlated with APOBEC1 mRNA abundance (2b). We further observed decreased FNLB mRNA abundance in CRC samples and a correlation in UM with RBM47 RNA expression (2c). Conclusions: We identified 29 novel C-to-U RNA editing targets in Apc^min/+ mice that exhibit potential roles in intestinal polyposis with selective cofactor (APOBEC1/RBM47) dependence. These APOBEC1/RBM47-dependent RNA targets likely function through altered mRNA processing/stability rather than C-to-U RNA editing.

Figure 1. <b>a</b>.Gross mophorlogy of polyps in small intestine jejunum of <i>Apc <sup>min/+</sup></i>, <i>Apc.<sup>min/+</sup></i> <i>Apobec1 <sup>–/–</sup>,</i> <i>Apc <sup>min/+</sup> Apobec1<sup>+/Tg</sup></i> and <i>Apc <sup>min/+</sup> RIKO</i> mice. Polyp number (#) and size in each genotype relative to <i>Apc <sup>min/+</sup></i>. <b>b</b> Venn diagram showing that out of 29 APOBEC1-dependent RNA targets identified in <i>Apc <sup>min/+</sup></i>, 23 are also RBM47-dependent. Distribution of the 23 APOBEC1/RBM47-dependent RNAs targets is as follows: 11 edited only in uninvolved tissue (yellow), 2 edited only in polyps (green) and 10 edited in both, uninvolved and polyp tissue (orange).

Figure 1. a.Gross mophorlogy of polyps in small intestine jejunum of Apc ^min/+, Apc.^min/+ Apobec1 ^–/–, Apc ^min/+ Apobec1^+/Tg and Apc ^min/+ RIKO mice. Polyp number (#) and size in each genotype relative to Apc ^min/+. b Venn diagram showing that out of 29 APOBEC1-dependent RNA targets identified in Apc ^min/+, 23 are also RBM47-dependent. Distribution of the 23 APOBEC1/RBM47-dependent RNAs targets is as follows: 11 edited only in uninvolved tissue (yellow), 2 edited only in polyps (green) and 10 edited in both, uninvolved and polyp tissue (orange).

Figure 2. <b>a</b>. Top <i>Flnb </i>RNA is significantly down regulated in <i>Apc</i> <i><sup>min/+</sup></i> <i>RIKO</i> uninvolved and tumor tissue. (* p <0.05, *** p < 0.001). Lower panel FLNB expression is reduced in <i>Apc <sup>min/+</sup> RIKO </i>compared to <i>Apc <sup>min/+</sup></i>. ACTIN is used as loading control (n=3 per <i>Apc</i> genotype). <b>b</b>. Left Quantitative analysis of <i>FLNB</i> RNA expression in FAP patients. Right, positive correlation between <i>FLNB</i> and <i>APOBEC1</i> RNA expression in FAP uninvolved tissue by Q-PCR analysis of 8 uninvolved-tumor pairs. <b>c</b>. Left, <i>FLNB</i> RNA is significantly downregulated in CRC patients (* p < 0.05) (n=7 uninvolved-tumor pairs). Right, Positive correlation between <i>FLNB</i> and <i>RBM47 </i>RNA in CRC uninvolved tissue by Q-PCR analysis.

Figure 2. a. Top Flnb RNA is significantly down regulated in Apc ^min/+ RIKO uninvolved and tumor tissue. (* p <0.05, *** p < 0.001). Lower panel FLNB expression is reduced in Apc ^min/+ RIKO compared to Apc ^min/+. ACTIN is used as loading control (n=3 per Apc genotype). b. Left Quantitative analysis of FLNB RNA expression in FAP patients. Right, positive correlation between FLNB and APOBEC1 RNA expression in FAP uninvolved tissue by Q-PCR analysis of 8 uninvolved-tumor pairs. c. Left, FLNB RNA is significantly downregulated in CRC patients (* p < 0.05) (n=7 uninvolved-tumor pairs). Right, Positive correlation between FLNB and RBM47 RNA in CRC uninvolved tissue by Q-PCR analysis.

Nicholas Davidson

Appearances