Society: AGA

Background: Immune checkpoint blockade has revolutionized cancer treatment by harnessing the patient’s native anti-tumor T cells, but most colorectal cancers do not respond, for unclear reasons. A major challenge in the field is understanding how patient charactistics, such as age, modify cancer cell-mediated antigen presentation and T cell activation. Patient age across cancer types is associated with increased tumor mutational burden, increased expression of immune checkpoint genes, and increased pro-inflammatory signaling. Here, we investigated the effects of aging on immune pathways in an inducible, genetically engineered mouse model of colorectal cancer.

Materials and Methods: We induced Apc-null tumors in the distal colons of young (age 2-3 months) and old (age 18-22 months) Apc^fl/fl;Villin^CreERT2 mice (C57BL/6J background) via colonoscopy-guided injection of 4-hydroxytamoxifen. Tumor size was measured by ex vivo weight. We used flow cytometry to measure the proportions of T cell subtypes, monocytes, macrophages and dendritic cells. Epithelial (Epcam+) cells were isolated from young and old tumors with flow cytometry, and then analyzed with RNA-Seq and ATAC-Seq.
Results: Tumor weight was significantly lower in old mice compared to tumors from young mice (N=6 in each group) (0.36 gr vs 0.46 grams, P=0.04). Compared to tumors from young mice (N=3), tumors from old mice (N=3) demonstrated a significantly higher percentage of dendritic cells ( 36.5 vs. 23.5, P=0.03), CD3+ T cells ( 27.6 vs. 15.4, P=0.005) (Figure 1), and IFN-gamma producing Th17 cells (22.1 vs. 14.9, P=0.02), and lower proportion of Tregs ( 25.4 vs. 45.6, P=0.02). CD4+, CD8+, and exhausted T cell population frequencies were not significantly different between old and young cohorts. Gene ontology (GO) analysis from RNA-Seq data (Figure 2) showed that differentially expressed genes in old vs. young colon tumor epithelial cells were significantly enriched in multiple immune functions, including inflammatory pathways, T-cell function, and antigen presentation (MHC class II). We found increased expression and chromatin accessibility of Ciita (a master regulator of MHC class II) , and increased expression of H2-Ab1 (which encodes the MHC class II protein complex), in cancer epithelial cells from old vs. young mice. Higher expression of MHC-II-related genes in tumor epithelial cells from old vs. young mice was confirmed by flow cytometry and in situ hybridization.

Conclusion: Results from this study demonstrate that aging promotes an anti-tumor immune microenvironment in an inducible, genetically engineered mouse model of colorectal cancer, possibly by enhancing cancer cell-mediated antigen presentation by MHC class II machinery.

<b>Figure 1. Aging is associated with expansion of the intratumoral T cell population.</b> Representative quantification of the percentage of CD45+, CD3+ T cells in colorectal tumors from old versus young mice (N=3); P<0.01.

Figure 1. Aging is associated with expansion of the intratumoral T cell population. Representative quantification of the percentage of CD45+, CD3+ T cells in colorectal tumors from old versus young mice (N=3); P<0.01.

<b>Figure 2. Aging is associated with upregulation of immune signaling and antigen processing pathways in colorectal tumor epithelial cells</b>. Gene Ontology (GO) pathways (Biological Process category) enriched for up-regulated genes in sorted epithelial cells from old mice compared to sorted epithelial cells from young mice. The top GO term items with the most significant enrichment (FDR < 0.05) were selected.

Figure 2. Aging is associated with upregulation of immune signaling and antigen processing pathways in colorectal tumor epithelial cells. Gene Ontology (GO) pathways (Biological Process category) enriched for up-regulated genes in sorted epithelial cells from old mice compared to sorted epithelial cells from young mice. The top GO term items with the most significant enrichment (FDR < 0.05) were selected.

Gastric cancer (GC) is one of the most common cancers in humans. Despite advances in GC therapy, it is still among the world's highest-mortality tumours due to metastases and recurrences, which usually occur within 2 years. Recently, a high ratio of GC, metastasis and cancer recurrence has been associated with asymptomatic Helicobacter pylori (Hp) infections. Hp penetrates into the deeper layers of the mucosa, including the gastric glands, where it interacts with mucous, parietal or progenitor cells and with non-glandular stroma cells, including fibroblasts. Gaseous mediator hydrogen sulfide (H₂S) has been shown to have anti-inflammatory and antioxidant properties and to inhibit NFκB signaling in conjunction with the reduction of pro-inflammatory cytokines such as IL-6, IL-1β and TNFα. We have undertaken study to determine whether the H₂S-released from its fast donor, NaHS, is able to inhibit the activity of Hp-infected cancer-associated fibroblasts (CAF). Such activated fibroblasts (Hp-AGF) have recently been shown to induce phenotypic reprogramming of normal rat gastric epithelial cells towards a propluripotent and highly invasive phenotype. Herein, we attempted to assess the effect of Hp infection on activation of human gastric fibroblasts and to define the possible mechanisms of this activation. Human fibroblasts isolated from biopsies of patients without systemic inflammatory, autoimmune and Hp infection, who underwent the laparoscopic sleeve gastrectomy, were infected with 1x10⁹ live Hp (cagA+;vacA+) per plate and incubated for 96 hrs. To determine inhibition of TGFβ signaling, a non-toxic dose of SB-431542 (10µM, ALK5/TGF-β type I receptor inhibitor) alone or in combination with NaHS (non-toxic dose 50µM) was used. Fibroblast activation markers and their corresponding signaling pathways were determined by RT-PCR, Western Blot and immunofluorescence. Hp-infection upregulated a subset of genes characterizing phenotype of CAFs. The α-SMA was upregulated and incorporated into stress fibres, which was mimicked by co-incubation with TGFβ1 (p<0.05). Hp-infection triggered inflammatory pathways upregulating TLR2, TLR4, STAT3 and NFκB (relA) signaling (p<0.05), resulting in a Snail⁺Twist⁺ phenotype with subsequent incorporation of Twist and Snail into the nucleus. Expression of TLR2, relA and STAT3 genes was dependent on autocrine TGFβ1 signaling. Co-incubation with NaHS evoked a transcriptomic downregulation of components of proinflammatory pathway including TLR2 and TLR4, STAT3 and NFκB (p65) with subsequent downregulation of Twist (p<0.05). We conclude that H₂S donors deserve a great attention as promising anti-inflammatory drugs capable of reducing fibroblasts activation during Hp-infection, thus limiting the risk of Hp-induced GC development and possibly constituting novel candidates acting as adjuvants in future eradication pharmacotherapy.

Background: RNA N⁶-Methyladenosine (m⁶A) modification is closely associated with tumor progression and therapy resistance. In this study, we aim to investigate the function and mechanism of m⁶A reader YTHDF1 in CRC stemness and chemoresistance.
Methods: The protein expression of YTHDF1 and cancer stem cells (CSCs) markers (CD133 and LGR5) was evaluated in human CRC tissue microarray (n=174) by immunohistochemistry staining. The effects of YTHDF1 on CRC stemness and chemoresistance were investigated in colon-specific YTHDF1 knockin mice (Ythdf1^KILgr5-Cre^ER) and in human colorectal CSC spheroids. Molecular targets and pathways targeted by YTHDF1 were identified by integrated m⁶A-seq, Ribo-seq, and RNA-seq, and further verified by RNA immunoprecipitation, qPCR and Western blot.
Results: YTHDF1 protein expression was positively correlated with colorectal CSCs markers (CD133 and LGR5) in human CRC tissues (P<0.001 for both markers). Functionally, YTHDF1 overexpression increased self-renewal capacity and sphere formation of human colorectal CSC spheroids in vitro, whereas YTHDF1 knockdown exerted opposite effects. In line with in vitro findings, YTHDF1-overexpression increased the tumor-initiating ability of CRC, while YTHDF1-knockdown suppressed tumor initiation ability, as evidenced by limiting dilution assay in mice. Moreover, colon-stem cell specific conditional Ythdf1 knockin mice accelerated azoxymethane/dextran sodium sulfate (AOM/DSS)-induced colorectal tumorigenesis with increased tumor multiplicity (P<0.05), implying Ythdf1 initiated and promoted CRC pathogenesis in a stem cell-specific manner. Mechanistically, integrated analysis of RNA-seq, m⁶A-seq, and Ribo-seq on YTHDF1-overexpressing colorectal CSCs revealed that YTHDF1 upregulated NOTCH signaling pathway. YTHDF1 could recognize and directly bind to m⁶A-modified NOTCH1 mRNA, leading to the up-regulation of NOTCH1 in CSCs. Furthermore, knockdown of NOTCH1 or NOTCH1 inhibitor DAPT abolished promoting effect of YTHDF1 on the self-renewal capacity of colorectal CSC spheroids, indicating a m⁶A-YTHDF1-NOTCH1 axis underlies CRC stemness. Consistently, YTHDF1 protein expression was positively correlated with NOTCH1 protein expression in human CRC tissues (n=174, P<0.01). In keeping with the increased stemness, YTHDF1 overexpressing colorectal CSCs showed significant resistance to the chemotherapies of oxaliplatin and fluorouracil. This was further confirmed in colon-stem cell specific Ythdf1 knockin mice, that showed resistance to oxaliplatin treatment as compared to their wildtype counterparts.
Conclusion: YTHDF1 promotes CRC stemness and chemoresistance via binding to m⁶A-modified NOTCH1 mRNA and inducing m6A-YTHDF1-NOTCH1 axis. YTHDF1 is a potential novel therapeutic target for CRC patients with chemoresistance.

Introduction: Therapeutic implications of genomic alterations such as immunotherapy in MSI status (MSI-H) and/or cetuximab in K-ras mutations are increasingly recognized in cancers including colorectal cancer (CRC). Our group has previously shown that the downregulation of the cohesin stromal antigen-1 (SA-1) is an early event in CRC and could be a therapeutic vulnerability. Cohesins are responsible for chromatin looping and increase susceptibility to DNA damage and may regulate the transcription of DNA repair proteins. Single-strand DNA break is repaired by Poly ADP-ribose polymerase (PARP) which relies at least partially on cohesin function. In breast cancer, DNA repair deficient tumors (BRCA mutations) can be exploited by using PARP inhibitors such as Olaparib. Therefore, we wanted to assess whether SA-1 loss in CRC would increase susceptibility to a PARP inhibitor.
Material and Methods: We prepared SA-1 knockdown (SA1_KD) cells using siRNA transfection. The control vs SA-1_KD cells (n=3) were analyzed using microarray. We also prepared stable SA-1_KD cells using shRNA constructs (n=3). Dysregulated genes were validated in SA-1_KD cells using qPCR assay. Olaparib sensitivity was analyzed using WST assay, and apoptosis induction was analyzed using Annexin V/PI and ApoTox-Glo Assay. PARP activity and DNA damage repair pathways were analyzed using Western blot analysis. The statistical difference between the two groups was analyzed using a t-test and multiple group comparison was done using the ANOVA in GraphPad Prism v.9.
Results: Microarray analysis indicated a dysregulation of genes involved in base excision repair viz. POLB (FC= -1.2), PARP4 (FC=-1.1), etc. Quantitative PCR (qPCR) analysis of stable SA-1_KD cells (~37-52% KD at protein level) supported that POLB (-29.98, p=0.003) and PARP4 (-36.33, p= 0.029) get downregulated (Figure 1). Dose-curve analysis using Olaparib indicated a higher sensitivity of SA-1_KD cells compared to the control (figure 2). Annexin V/PI (p=0.0148, p>0.05 and p=0.0108 respectively for shRNA1, 2, and 3) and Apotox assays (p<0.0001) (cleaved caspase 3/7 luminescence assay) post Olaparib treatment (50µM) also showed higher apoptosis and cytotoxicity in SA1_KD cells. In-silico data analysis of TCGA COAD cohort also revealed a significant correlation of SA-1 with PARP4 (r=0.45, p<0.0001) and LIG3 (r=0.39; p<0.0001).
Conclusion: We demonstrate, for the first time, that SA-1 loss in CRC leads to susceptibility to Olaparib. This may result through the regulation of genes involved in DNA repair including POLB and PARP4 in conjunction with vulnerability to chromatin looping challenges. Since the epigenetic loss of SA-1 is common in CRC, this synthetic lethality may have real potential for clinical impact. This study shows the potential of novel therapeutic vistas in CRCs with cohesin loss.

Nearly 50% of all colorectal cancer patients progress to develop metastatic lesions (mCRC) and despite on-going efforts the survival rate for mCRC patients remains significantly low at 14%. This, to a great extent, can be attributed towards a substantial lack of understanding of the genomic and epigenomic alterations in mCRC tumours, that would ultimately allow for identification of novel diagnostic and/or therapeutic targets⁽²⁾.
With this in mind, we applied methylcapture sequencing to DNA derived from 57 mCRC patients and identified a tumour specific 376-loci methylation signature specific to mCRC patients, predominantly comprised of hypomethylated enhancers. Importantly, subsequent validation of the signature using publicly available CRC methylation datasets (n=860 in total) revealed that that the signature consistently differentiates early stage adenoma from carcinoma, indicating its potential to predict CRC progression.
To ascertain the therapeutic value of the 376-signature loci, and other enhancers identified in CRC literature, we next performed a CRISPR inhibition (CRISPRi) loss-of-function screen to sequentially silence 2,274 loci (using 15,114 single gRNAs) in HCT116 mCRC cells. In total, 31 loci essential (adj.P < 0.05) for HCT116 proliferation were identified. Among them, silencing of a hypomethylated cis-regulatory element (CRE) of interest from the 376-methylation signature caused one of the most significant decreases in proliferation, which was confirmed in subsequent validation of the screen using three independent gRNAs. Interestingly, in silico analysis revealed this CRE featured marks of active enhancers (H3K27ac and H3K4me1) and enrichment for transcription factor motifs such as YY1 and OSR2. We next investigated the genome-wide interactions of this enhancer using chromosome conformation capture-on-chip (4C-seq) and also showed that hypomethylation of this enhancer significantly correlated with increased expression of proximal oncogenic lncRNAs CCAT1 and CASC19 in the mCRC discovery cohort. Finally, in vitro phenotypic assays (apoptosis/invasion) carried out following perturbation of this enhancer strongly demonstrated its potential oncogenic function, which may ultimately drive a mCRC-associated phenotype.
Collectively, we present a novel DNA methylation-based signature with potential diagnostic utility in mCRC patients and identify a novel hypomethylated enhancer which drives CRC cell proliferation, thus rationalizing it as a therapeutic target for mCRC.

Background: Colorectal cancer (CRC) is the second most common cause of cancer-related deaths worldwide. While screening reduces CRC mortality, its effectiveness is limited by poor uptake. Current stool screening tests are cumbersome and have poor compliance, while colonoscopy is resource-intensive and invasive. A blood-based screening test would address an unmet need for an accurate, affordable and non-invasive screening test.
Objective: To develop a blood-based miRNA panel for the early detection of CRC.
Methods: The study population enrolled 4 groups of patients, CRC, advanced adenomas (AA), adenomas and controls. CRC cases were newly-diagnosed and had blood collected prior to any treatment. The other 3 groups had blood samples collected prior to colonoscopies. AA were adenomas >10mm in diameter, or with villous histological features or high-grade dysplasia. Controls were age, gender and race-matched patients who underwent colonoscopies and had no polyps or other colonic pathology. Plasma from the blood samples were profiled for the expressions of 723 candidate miRNAs using high-throughput qPCR (ID3EAL Premium Profiling, MiRXES, Singapore) and readouts normalized to spike-in control RNAs. CRC-, AA- and adenoma- specific miRNA biomarkers were identified using Ridge regression model.
Results: A total of 841 subjects were enrolled, comprising 217 CRCs, 190 AAs, 217 adenomas and 217 controls. Among CRC subjects, the mean age was 64.5 years (SD ±10.2), 65.0% male, 81.6% were Chinese, 33.2% were current or previous smokers and 13.4% had a positive first degree of family history of CRC. The proportion of stage I, II, III and IV were 19%, 32%, 33% and 15% respectively. More than half of the miRNAs (n=374, 52%) were prevalent in >95% of the population, of which 38 miRNAs were differentially expressed between cancer cases and controls (FDR < 0.001). Our CRC-specific miRNA panel included miR-21, a previously identified biomarker. Of note, 83% of upregulated and 25% of downregulated miRNAs in CRC tissue samples, as reported by The Cancer Genome Atlas, were also found to have congruent directionality in our plasma samples. Using recursive feature selection analysis, a panel of 125 miRNAs best identified CRC cases with an AUC of 88.7%. This miRNA panel can further discriminate early (stage 1 and 2) and late stage CRC (stage 3 and 4) with an AUC of 87% and 90% respectively (Figure 1).
Conclusions: These preliminary results demonstrate plasma miRNA biomarkers combined with clinical factors were able to accurately predict CRC (AUC 90%), suggesting that a targeted miRNA panel is a potential novel, non-invasive blood test for early detection of CRC. Further optimization to develop a parsimonious miRNA panel is ongoing and external validation in an independent cohort will be performed.

<b>Figure 1.</b> (A) Heat-map showing expression levels of plasma miRNAs that were differentially regulated between colorectal cancer and controls (FDR<0.001). Absolute miRNA expression levels (copy/ml) of miRNAs were presented in log2 scale and standardized to zero mean. Hierarchical clustering was carried out for both samples and miRNAs based on Euclidean distance. (B) Diagnostic performance of multi-feature biomarker panels with 2-200 features (198 miRNAs, drinking and smoking status) as determined by mean area under ROC curve (AUC). Ten iterations of five-cold cross-validation were performed and the highest mean cross-validated AUC was observed at 113 features. (C) ROC curves of the 113-feature panel for differentiating early-stage/late-stage CRC from control subjects. AUC was reported as the mean AUC from 10 iterations of 5-fold cross-validation using Ridge model.

Figure 1. (A) Heat-map showing expression levels of plasma miRNAs that were differentially regulated between colorectal cancer and controls (FDR<0.001). Absolute miRNA expression levels (copy/ml) of miRNAs were presented in log2 scale and standardized to zero mean. Hierarchical clustering was carried out for both samples and miRNAs based on Euclidean distance. (B) Diagnostic performance of multi-feature biomarker panels with 2-200 features (198 miRNAs, drinking and smoking status) as determined by mean area under ROC curve (AUC). Ten iterations of five-cold cross-validation were performed and the highest mean cross-validated AUC was observed at 113 features. (C) ROC curves of the 113-feature panel for differentiating early-stage/late-stage CRC from control subjects. AUC was reported as the mean AUC from 10 iterations of 5-fold cross-validation using Ridge model.

Khay-Guan Yeoh

Appearances