DOT1L affects colorectal carcinogenesis via altering T cell subsets and oncogenic pathway

ABSTRACT

Chronic inflammation and oncogenic pathway activation are key-contributing factors in colorectal cancer pathogenesis. However, colorectal intrinsic mechanisms linking these two factors in cancer development are poorly defined. Here, we show that intestinal epithelial cell (IEC)-specific deletion of Dot1l histone methyltransferase (Dot1l^ΔIEC) reduced H3K79 dimethylation (H3K79me2) in IECs and inhibited intestinal tumor formation in Apc^Min- and AOM-DSS-induced colorectal cancer models. IEC-Dot1l abrogation was accompanied by alleviative colorectal inflammation and reduced Wnt/β-catenin signaling activation. Mechanistically, Dot1l deficiency resulted in an increase in Foxp3⁺RORϒ⁺ regulatory T (Treg) cells and a decrease in inflammatory Th17 and Th22 cells, thereby reducing local inflammation in the intestinal tumor microenvironment. Furthermore, Dot1l deficiency caused a reduction of H3K79me2 occupancies in the promoters of the Wnt/β-catenin signaling genes, thereby diminishing Wnt/β-catenin oncogenic signaling pathway activation in colorectal cancer cells. Clinically, high levels of tumor H3K79me2 were detected in patients with colorectal carcinomas as compared to adenomas, and negatively correlated with RORϒ⁺FOXP3⁺ Treg cells. Altogether, we conclude that DOT1L is an intrinsic molecular node connecting chronic immune activation and oncogenic signaling pathways in colorectal cancer. Our work suggests that targeting the DOT1L pathway may control colorectal carcinogenesis. Significance: IEC-intrinsic DOT1L controls T cell subset balance and key oncogenic pathway activation, impacting colorectal carcinogenesis.

Introduction

Colorectal cancer is one of the most common causes of cancer mortality in the United States and worldwide. Immune checkpoint therapy (ICB) has been approved to treat patients with unresectable or metastatic microsatellite instability-high (MSI-H) or mismatch repair deficient (dMMR) colorectal cancer.^1,2 Unfortunately, because few patients have these particular genetic alterations, a vast majority of patients with colorectal cancer are not responsive to ICB therapy, highlighting the critical need to unveil previously unknown cellular and molecular determinants of colorectal cancer etiology. Colorectal cancer development is mediated in part by accumulated somatic mutations in selected oncogenes and tumor suppressor genes.^3–5 The Wnt/β-catenin pathway regulates colorectal cancer cell proliferation, apoptosis, and contributes to colorectal carcinogenesis.^6,7 Histone acetyltransferase (HAT) proteins recruited by β-catenin/T cell factor (TCF) transcription factor complexes to selected target genes have been implicated in the activation of β-catenin/TCF target genes.^8,9 These findings highlight the likely important roles that chromatin modifications have in modulating the Wnt/β-catenin signaling pathway to regulate gene expression in colorectal carcinogenesis.

The disruptor of telomeric silencing 1-like (DOT1L) protein mediates histone 3 lysine 79 (H3K79) methylation.¹⁰ DOT1L plays a role in mixed-lineage leukemia (MLL),^11,12 and T cell polyfunctionality and survival.¹³ It has been elucidated that DOT1L is activated by Th22 cells in human colorectal cancer cells.¹⁴ Furthermore, a DOT1L-containing complex has been linked to the Wnt signaling pathway in Drosophila¹⁵ and zebrafish.¹⁶ Thus, prior works suggest a potential role of DOT1L in colorectal carcinogenesis. To address this possibility, we pursued studies of intestinal epithelial cell (IEC)-intrinsic abrogation of Dot1l function in the Wnt/β-catenin pathway activation in mouse colorectal tumorigenesis.

On the other hand, chronic immune activation-associated inflammation has been implicated in colorectal cancer development.^17,18 However, whether there exists an intestinal intrinsic mechanism controlling the balance among gut T cell subsets, particularly inflammatory Th17 and Th22 cells, and RORϒ⁺Foxp3⁺ regulatory T cells, and affects colorectal cancer pathogenesis remains unknown. To answer this question, using mouse colorectal cancer models with IEC-specific deletion of Dot1l histone methyltransferase (Dot1l^ΔIEC), we examined a potential impact of intestinal epithelial cell DOT1L on immune cell subsets in the colorectal cancer microenvironment, and extended our studies to patients with colorectal adenoma and carcinoma.

In the work presented here, we demonstrate that intestinal DOT1L-mediated H3K79 methylation, as an intrinsic mechanism, controls both the Wnt/β-catenin activation and immune cell subset balance, thereby contributing to colorectal cancer development.

Materials and methods

Human subjects

Treatment-naïve patients with colorectal adenoma and early colorectal carcinoma were recruited for this study.¹⁹ Human subjects were from Renji Hospital affiliated to Shanghai Jiaotong University School of Medicine between 2012 and 2017. Patients were pathologically and clinically diagnosed with colorectal adenoma or early colorectal cancer. Informed consent was obtained from the patients before sample collection in accordance with institutional guidelines. The Ethics Committees in the Renji Hospital approved the study protocol. Patient characteristics were detailed in the Supplementary Table S1.

Mouse strains

Dot1l^FF mice¹² were bred to C57BL/6 mice expressing Cre-recombinase under the control of the Villin promoter²⁰ to generate intestinal epithelial cell (IEC) specific Dot1l knockout (Dot1l^ΔIEC) mice. Genotypes were determined by PCR. Age- and gender-matched mice were used for all experiments. Apc^Min mice (C57BL/6 J-Apc^Min/J; stock 002020, the Jackson Laboratory) were bred with Dot1l^FF; Vil-Cre mice to generate Apc^Min; Vil-Cre; Dot1l^FF (Dot1l^ΔIECApc^Min) mice, and Apc^Min; Dot1l^FF (Dot1l^FFApc^Min) control littermates. Mouse studies were approved by the Institutional Animal Care & Use Committee of the University of Michigan.

Murine colorectal inflammation and tumor models

Dot1l^FF and Dot1l^ΔIEC littermate mice were injected intraperitoneally with 10 mg of azoxymethane (AOM) (Sigma) per kilogram body weight. Five days later, 1.8% Dextran sulfate sodium salt (DSS) (colitis grade (36,000–50,000) (MP Biomedicals) was given in drinking water for 5 days followed by regular drinking water for 16 days. This cycle was repeated twice, and mice were sacrificed on day 70 or at the indicated time. In colorectal inflammation model, mice were fed with 1.8% DSS water for 5 days and followed with regular water for 4 days. Dot1l^FFApc^Min and Dot1l^ΔIECApc^Min littermates were maintained for 6–7 months to assess spontaneous colorectal cancer development or were given drinking water with 1.8% DSS for 5 days, followed by regular water for 30 days to access colorectal cancer development.

Colorectal tumor organoid culture

Colorectal tumor organoid cultures were performed as described previously²¹ with minor modifications. Briefly, tumor tissues were mechanically separated from surrounding normal tissues and washed in PBS containing 100 U/ml penicillin and 100 µg/ml streptomycin. These antibiotics were added to all solutions used in the following procedure. After washing with PBS, tumor tissues were cut into small pieces and digested with 200 U/ml type IV collagenase for 2 hours. The digested tumor tissue was passed through a 70-µm cell strainer. Then, 10000 tumor cells were mixed with Matrigel (Fisher Scientific) and seeded in a 24-well plate. 15 minutes after the Matrigel polymerization, DMEM/F12 medium (ThermoFisher Scientific) contains 1× Glutamax (ThermoFisher Scientific), 1 M HEPES (Life Technologies), 1× N2 supplement (ThermoFisher Scientific), 1× B27 supplement (ThermoFisher Scientific), and 50 ng/µl epidermal growth factor (EGF; Sigma) was added to the plate.

IEC harvest, real-time PCR, and Western blotting

IECs were isolated from murine samples by shaking intestinal tissue in 1 mM EDTA/1 mM DTT and 5% FCS at 37°C for 10 min, resulting in 80–90% EpCAM⁺ IEC purity. RNA was isolated from cells using TRIzol (ThermoFisher Scientific) reagent then subjected to reverse transcription with AMV reverse transcriptase (Promega). Real-time PCR was performed using SYBR green chemistry (Applied Biosystems). Reactions were run on a real-time PCR system (StepOne Plus Real-Time PCR System; Applied Biosystems) Specific primers are listed in Supplementary Table S2. For Western blotting, IECs were lysed in a modified RIPA buffer (ThermoFisher Scientific) and lysates were subjected to immunoblot analysis. Blots were probed with rabbit anti-H3K79me2, H3K79me3, H3K4me2, H3K4me3, H3K27me3 (Abcam), total H3 (Cell Signaling), and GAPDH (Invitrogen).

Human colorectal cancer cell culture and lentiviral transduction

Human colorectal cancer cell lines DLD-1 and HT29 were obtained from ATCC. Lentiviral vectors were used to transduce colorectal cancer cells and establish stable cell lines. The lentiviral transduction efficiency was confirmed by GFP which was co-expressed by the lentiviral vector. The knockdown efficiency was assessed by immunoblotting. The vectors included pGIPZ lentiviral vector encoding gene-specific shRNAs for DOT1L or scrambled shRNA (Puromycin resistant).¹⁴ Specific primers are listed in Supplementary Table S2.

RNA extraction and quantitative PCR

Total RNA was isolated from cells by column purification (Direct-zol RNA Miniprep Kit, Zymo Research) with DNase treatment. cDNA was generated using High-Capacity cDNA Reverse Transcription Kit (ThermoFisher Scientific). Quantitative PCR (qPCR) was performed on cDNA using Fast SYBR Green Master Mix (Thermo Fisher Scientific) on a StepOnePlus Real-Time PCR System (Thermo Fisher Scientific). Gene expression was quantified using the primers listed in Supplementary Table S2.

Chromatin immunoprecipitation (ChIP) assay

ChIP assay was performed according to the protocol (Upstate, Millipore) as previously described.^14,22 Crosslinking was performed with 1% formaldehyde or 1% paraformaldehyde for 10 minutes. To enhance cell lysis, we ran the lysate through a 27 g needle three times and flash froze it in −80°C. Sonication was then performed with the Misonix 4000 water bath sonication unit at 15% amplitude for 10 minutes. Protein/DNA complex was precipitated by specific antibodies against H3K79me2 (Abcam) and IgG control (Millipore). Then, DNA was purified using DNA Purification Kit (Qiagen). ChIP-enriched chromatin was used for Real-Time PCR, relative expression level is normalized to Input. Specific primers are listed in Supplementary Table S3.

Flow cytometry

To isolate lamina propria immune cells, IEC and intraepithelial lymphocyte layers were first stripped by shaking sections of large intestine in 5 mM EDTA/1 mM DTT. Remaining tissue was digested with collagenase (0.5 mg/ml) to obtain single cell suspensions. For flow cytometry, cells were stained with a combination of the following fluorescence-conjugated monoclonal antibodies, anti-CD3 (clone: 500A2), anti-CD4 (clone: RM4-5), anti-Foxp3 (clone: FJK-16s), anti-CD45 (clone: 30-F11), anti- RORγt (clone: Q21-559), anti-IL-17A (clone: TC11-18H10), and anti-IL-22 (clone: 1H8PWSR) (BD-Biosciences or Therma-Fisher). Samples were acquired on an LSR II (BD Biosciences) and were analyzed with FACSDiva software (BD Biosciences).

Histology, immunohistochemistry, and immunofluorescence

For bromodeoxyuridine (BrdU) staining, 200 µL of BrdU Labeling Reagent (ThermoFisher Scientific) was injected into mouse peritoneal cavity. Mice were sacrificed 2 hours after BrdU injection. The same segment of distal colorectal was fixed in 10% neutral formalin and paraffin embedded. Proliferating cells were detected with BrdU detection kit (BD Bioscience). Tissues were counterstained with hematoxylin. The number of BrdU-positive cells was quantified by number of cells in intact well-orientated crypts.

Immunofluorescence staining in mouse tissues. Frozen mouse colorectal tissue sections were incubated with anti-mouse β-catenin (1:200, BD Biosciences) and anti-mouse Axin2 (1:200, Abcam), followed by Alexa Fluor 488- or Alexa Fluor 594-conjugated secondary antibodies (Life Technologies) and DAPI (1 µg/ mL, Sigma-Aldrich) for 1 h at room temperature and embedded using FluorSave (Merck Millipore). Images were taken using a Leica confocal microscope.

Conventional immunohistochemistry staining in human tissues. Immunohistochemical staining on human colorectal colorectal adenoma and colorectal cancer tissue sections was carried out using an MaxVisionTM kit (MXB Biotech, Fujian, China) according to the manufacturer’s recommendation. Serial slides of de-paraffinized tissue sections were labeled with rabbit polyclonal antibodies against human H3K79me2 (Abcam, ab177184). Nuclear H3K79me2 was scored using the immunoreactive score (IRS) as a method of assessing the extent of nuclear immunoreactivity. The IRS takes into account the percentage of positive cells (1–4 score for 0–25%, 25–50%, 50–75%, and 75–100%, respectively) in each intensity category (0–3) and computes a final score, on a continuous scale between 0 and 12.

Immunofluorescence staining in human tissues.^23,24 Human colorectal adenoma and colorectal cancer tissues were de-paraffinized. The staining was performed with PerkinElmer’s Opal™ multiplexed reagents according to the manufacturer’s instructions: Opal 520, Opal 570, and Opal 650.^23,24 Multiple antigen retrieval steps using either AR6 (CD3) or AR9 (Foxp3, RORϒt, and DAPI) antigen retrieval buffer was used for the removal of primary and secondary antibodies. The following primary and secondary antibodies were used: rabbit anti-human/mouse FOXP3 (Clone: D6O8R, Cell Signaling), rabbit polyclonal antibody against CD3 (DAKO), anti-RORϒt antibody (Clone: EPR20006, Abcam), immPRESS™ HRP Goat anti-rabbit (Vector), and immPRESS™ HRP Horse anti-mouse IgG (Vector). Tissue slides were scanned on the Mantra Automated Quantitative Pathology Imaging System and analyzed using Inform software (Perkin-Elmer). At least 3 fields were acquired across the entire tissue sections and analyzed for FOXP3, RORϒt, and CD3 expression. Any discrepancies were resolved by subsequent consultation with a diagnostic pathologist.

Statistical analysis

Statistical analysis was performed in GraphPad Prism statistical software (version 6). P values were calculated using T test, Mann-Whitney U-test, Spearman’s correlation test, χ² test, or one-way ANOVA as indicated. P value < .05 was considered statistically significant.

Results

IEC-Dot1l supports colorectal tumorigenesis

DOT1L is involved in the regulation of cancer stemness in the in vitro cultured human colorectal cancer cell lines.¹⁴ We wondered if DOT1L directly affects intestinal tumorigenesis. To address this in an in vivo model, we generated intestinal epithelial cell (IEC)-specific Dot1l-deficient (Dot1l^ΔIEC) mice by crossing mice carrying a floxed allele of Dot1l (Dot1l^FF)¹² to Villin-cre mice.²⁰ Dot1l^ΔIEC mice were born at normal Mendelian frequencies (Supplementary Figure 1a). We isolated and sorted epithelial cell adhesion molecule (EpCAM)⁺ IECs (Supplementary Figure 1b) from colon of these mice. Real-time PCR (Supplementary Figure 1c) and Western blotting (Supplementary Figure 1d) confirmed IEC-specific Dot1l deletion. We observed no apparent alterations in intestinal morphology in Dot1l^ΔIEC mice (Supplementary Figure 1e).

To investigate the potential role of epithelium-intrinsic Dot1l function in intestinal tumorigenesis, we established two different mouse intestinal tumor models. We generated Dot1l^ΔIECApc^Min (Apc^MinΔIEC) mice by breeding Dot1l^ΔIEC mice with Apc^Min mice and challenged these mice with DSS to promote colorectal cancer development. We observed low tumor numbers (Figure 1a–b) and small tumor load (Figure 1c) in Dot1l^ΔIECApc^Min mice as compared to Dot1l^fl/flApc^Min (Apc^MinFF) mice. The data suggests that epithelium intrinsic Dot1l plays a role in colorectal carcinogenesis. In addition, we established an AOM/DSS-induced colorectal cancer model in Dot1l^FF and Dot1l^ΔIEC mice. Again, we found that tumor numbers (Figure 1d –e), tumor load (Figure 1f), and larger tumors (>4 mm) (Figure 1g) were reduced in Dot1l^ΔIEC mice, as compared to control littermates. Furthermore, to explore a role of Dot1l in spontaneous intestinal tumor development, we maintained Dot1l^FFApc^Min and Dot1l^ΔIECApc^Min mice for 6–7 months in the absence of DSS and AOM administration. We found less tumors in small and large intestines in Dot1l^ΔIECApc^Min mice than Dot1l^FFApc^Min mice (Figure 1h, i and j). Together, the results reveal a pro-tumor role of intrinsic intestinal epithelium Dot1l in colorectal tumorigenesis.

Figure 1.

IEC-Dot1l supports colorectal tumorigenesis. (a) Macroscopic view of the representative colons from Apc^MinDot1l^FF or Apc^MinDot1l^ΔIEC mice on day 30 in the DSS induced tumor model. (b and c) Colorectal tumor numbers (b) and loads (c) are quantified in Apc^MinDot1l^FF (n = 9) or Apc^MinDot1l^ΔIEC (n = 13) mice on day 30 in the DSS induced tumor model. ****P < .0001, Mann-Whitney U-test for tumor number, ****P < .0001, Student’s t-test for tumor load. (d) Macroscopic view of the representative mouse colon from Dot1l^FF or Dot1l^ΔIEC mice on day 70 in the cancer model. (e and f) Colorectal cancer numbers (e) and tumor loads (f) are quantified in Dot1l^FF (n = 7) and Dot1l^ΔIEC (n = 9) mice on day 70 of the cancer model. ***P = .001, Mann-Whitney U-test for tumor number, ***P = .0002, Student’s t-test for tumor load. (g) Tumor size distribution is shown in Dot1l^FF and Dot1l^ΔIEC mice. P = .0408 (tumor < 2 mm), *P < .05, Chi-squared test. (h) Macroscopic view of the representative mouse distal small (left) and large (right) intestines in Apc^MinDot1l^FF or Apc^MinDot1l^ΔIEC mice on day 210 in the spontaneous tumor model. (i and j) Tumor numbers were counted in small (i) and large intestines in Apc^MinDot1l^FF and Apc^MinDot1l^ΔIEC mice on day 210 of spontaneous tumor model. *P = .0109 for small intestine, ***P = .0095 for large intestine, Mann-Whitney U-test.

Intestinal Dot1l affects IEC proliferation

Cancer cell proliferation contributes to cancer phenotype and progression. We wondered if intrinsic intestinal epithelium Dot1l affects colorectal cancer cell proliferation. We performed hematoxylin and eosin (H & E) staining in intestinal tissues and observed no histological difference between Dot1l^FF and Dot1l^ΔIEC mice (Supplementary Figure 2a). Mice were treated with BrdU. We detected similar levels of BrdU⁺ intestinal epithelial cells between Dot1l^FF and Dot1l^ΔIEC mice (Supplementary Figure 2b and c). The data suggest that genetic deficiency of Dot1l does not affect intestinal epithelial cells in homeostasis. Interestingly, we observed tubular colorectal adenomas with high-grade dysplasia in Dot1l^FF mice and low-grade dysplasia in Dot1l^ΔIEC mice in AOM-DSS induced intestinal tumor model (Figure 2a). Furthermore, we detected less BrdU⁺ proliferative cells in Dot1l^ΔIEC mice than Dot1l^FF mice (Figure 2b and Supplementary Figure 2d). In line with this, we observed high-grade dysplasia (Figure 2c) and high levels of BrdU⁺ proliferative cells (Figure 2d and Supplementary Figure 2e) in Dot1l^FFApc^Min mice as compared with Dot1l^ΔIECApc^Min mice. We conducted intestinal tumor organoid culture assay. We found that the growth and the size of tumor organoids were reduced in Dot1l^ΔIEC mice compared to Dot1l^FF mice (Figure 2e –f). In addition, treatment of Dot1l^FFApc^Min tumor-derived organoids with a DOT1L methyltransferase specific inhibitor, EPZ004777²⁵ can also reduce organoids growth (Supplementary Figure 2 f and g). Thus, epithelium intrinsic Dot1l supports tumor cell proliferation.

Figure 2.

Intestinal Dot1l affects IEC proliferation. (a and b) Representative Hematoxylin and eosin (H & E) histology (a) and BrdU histology (b) of colorectal from Dot1l^FF and Dot1l^ΔIEC mice on day 70 of the colorectal cancer model. One of 5 is shown. (c and d) Representative H&E histology (c) and BrdU histology (d) of mouse colorectals from Apc^MinDot1l^FF and Apc^MinDot1l^ΔIEC mice on day 30 of the DSS induced colorectal cancer model. One of 5 is shown. (e) Representative images of tumor organoids from Dot1l^FF and Dot1l^ΔIEC mice on day 70 of the colorectal cancer model. Scale bar, 75 µm. Data represents one of three independent experiments. (f) Tumor organoid numbers and sizes are shown in Dot1l^FF and Dot1l^ΔIEC mice on day 70 of the colorectal cancer model. n = 3, ***P < .001 for all groups in different sizes of tumor organoids, Student’s t-test.

Intestinal Dot1l affects the Wnt/β-catenin pathway

We next explored molecular targets of DOT1L in promoting tumorigenesis ang tumor cell proliferation. The Wnt/β-catenin signaling pathway is involved in colorectal carcinogenesis.^6,7,26 Mutations in exon 3 of the β-catenin gene was also found in AOM/DSS induced mouse colorectal tumors.²⁷ We examined whether Dot1l mediated methylation of H3K79me2 affected canonical Wnt/β-catenin signaling pathway in vivo. Expression levels of Sox4, a typical Wnt/β-catenin target gene, were comparable in naïve Dot1l^FF and Dot1l^ΔIEC colon epithelial cells (Supplementary Figure 3a). Interestingly, real-time PCR (Figure 3a) and Western blotting (Figure 3b) demonstrated higher expression levels of canonical Wnt target genes, such as Axin2, Sox4, Ephb3, and Lgr5 in Dot1l^FF tumors than in Dot1l^ΔIEC tumor in the AOM/DSS induced tumor. In line with this, immunofluorescence staining showed higher levels of β-catenin and Axin2 in early-stage colorectal cancer epithelial cells in Dot1l^FF mice than Dot1l^ΔIEC mice in the AOM/DSS model (Figure 3c). Axin2 is a canonical Wnt target gene.^6,7 The result indicates a higher Wnt/β-catenin signaling in Dot1l^FF mice than Dot1l^ΔIEC mice.

Figure 3.

Intestinal Dot1l affects the Wnt/β-catenin pathway. (a) Wnt/β-catenin downstream genes mRNA expression from Dot1l^FF and Dot1l^ΔIEC mice day 70 of the colorectal cancer model, n = 7–8, **P = .0069 for Axin2, ****P < .0001 for Sox4, **P = .0038 for Lgr5, ***P = .0006 for Ephb3, Student’s t-test. (b) Western blot of Axin2 expression from Dot1l^FF and Dot1l^ΔIEC mice day 70 of the colorectal cancer model. One of three is shown. (c) Immunofluorescence staining of β-catenin and Axin2 in Dot1l^FF and Dot1l^ΔIEC colorectal on day 14 of AOM-DSS model. Green: β-catenin, Red: Axin2, Blue: DAPI. (d and e) Effect of EPZ004777 on Wnt/β-catenin downstream gene mRNA expression in DLD-1 (d) and HT29 (e) cells. (d) **P = .0023 for AXIN2, **P = .0026 for SOX4, *P = .0475 for EPHB3, **P = .0026 for LGR5. (e) **P = .0081 for AXIN2, **P = .0046 for SOX4, **P = .007 for EPHB3. Student’s t-test. (f and g) Effect of DOT1L knockdown on Wnt/β-catenin downstream gene mRNA expression in DLD-1 (f) and HT29 (g) cells. (f) *P < .05. (g) **P = .0081 for AXIN2, **P = .0046 for SOX4, **P = .007 for EPHB3, Student’s t-test. (h and i) H3K79me2 ChIP on the promoter (p) and different gene body regions (B1, B2, and B3) of SOX4 (h) and AXIN2 (i) in DLD-1 cells. NS, P > .05, *P ≤ .05, **P ≤ .01, ***P ≤ .001, ****P ≤ .0001, one-way ANOVA.

Human colorectal cancer cell lines DLD-1 and HT29 have increased β-catenin-regulated gene transcription.²⁸ To determine whether DOT1L regulated the Wnt target genes in human colorectal cancer cells, we treated human colorectal cancer cells with EPZ004777. EPZ004777 treatment inhibited H3K79 methylation in DLD-1 and HT29 cells (Supplementary Figure 3b). Accordingly, EPZ004777 suppressed the expression of AXIN2, SOX4, EPHB3, and LGR5 in DLD-1 (Figure 3d) and HT29 (Figure 3e) cells. Genetic knock down of DOT1L with specific shRNAs caused reduced expression of AXIN2, and SOX4 in DLD-1 (Figure 3f) and HT29 cells (Figure 3g) and reduced EPHB3 in HT29 cells (Figure 3g). DOT1L can bound to both promoter²⁹ and gene body³⁰ of actively transcribed genes. Chromatin immunoprecipitation (ChIP)-seq matrix from ENCODE showed H3K79me2 occupation in the promoter and gene body regions of SOX4 and AXIN2 (Supplementary Figure 3c and d). We used ChIP assay to evaluate the occupancy of H3K79me2 at the promoter and gene body regions of SOX4 and AXIN2 in DLD-1 cells (Supplementary Figure 3e and f). We found that H3K79me2 was enriched in both SOX4 and AXIN2 loci. Knockdown DOT1L resulted in reduced H3K79me2 level in the gene body region of SOX4 loci (Figure 3h). While knockdown DOT1L caused the reduction of H3K79me2 level at both promoter region and early gene body regions of AXIN2 loci (Figure 3i). These results suggest that DOT1L targets the Wnt/β-catenin signaling gene transcription via H3K79me2.

Intestinal Dot1l alters immune cell subsets

In addition to activation of the Wnt/β-catenin pathway, chronic inflammation is a risk factor in colorectal carcinogenesis.¹⁷ Activation of WNT/β-catenin signaling also has been linked with immune escape and T cell exclusion.³¹ In support of this notion, in the AOM-DSS mode, Dot1l^ΔIEC mice exhibited higher body weight (Figure 4a), longer colon length (Supplementary Figure 4a), and reduced disease score (Supplementary Figure 4b) as compared to their Dot1l^F littermates. This suggests that DOT1L may facilitate intestinal inflammation, thereby affecting tumorigenesis. As local immune cell activation contributes to inflammation, we compared immune cell subsets in lamina propria in the intestinal microenvironment in Dot1l^FF and Dot1l^ΔIEC mice. We detected a decrease in IL-17⁺CD4⁺ T (Th17) and IL-22⁺CD4⁺ T (Th22) cells in colorectal tissues in Dot1l^ΔIEC mice as compared to Dot1l^FF littermates (Figure 4b –c). This was not observed in mesenteric lymph nodes (mLNs) (Supplementary Figure 4c). Accordingly, the levels of several pro-inflammatory gene transcripts, including Cox2, Il1b, and Il23a, were lower in Dot1l^ΔIEC tumor tissues than those in Dot1l^FF colorectal tissues (Supplementary Figure 4d). These pro-inflammatory molecules might be from myeloid cells³² and can induce Th17 and Th22 cells.^33–35 Hence, Dot1l supports inflammatory immune cells, including Th17 and Th22 cells, in colorectal cancer. It has been reported that there is a balance between Th17 cells and Treg cells in gut.^33,36,37 Gut-specific RORγt⁺Foxp3⁺ Treg cells are highly suppressive and protect chronic intestinal inflammation.^38–40 The percentage of lamina propria Foxp3⁺CD4⁺ Tregs and RORϒt⁺Foxp3⁺ Tregs were comparable in Dot1l^FF and Dot1l^ΔIEC mice in homeostasis condition (Supplementary Figure 4e and f). However, we found that the percentage of tumor infiltrating Foxp3⁺CD4⁺ Treg cells, particularly RORϒt⁺Foxp3⁺ Treg cells were increased in Dot1l^ΔIEC mice as compared to Dot1l^FF mice Dot1l^FF in the AOM-DSS model (Figure 4d –e) and Apc^Min model (Figure 4f –g). The difference of RORϒt⁺Foxp3⁺ Tregs was not observed in mesenteric lymph nodes (Figure 4f –g). Altogether, the data suggest Dot1l in IECs affects intestinal T cell subsets, thereby altering colorectal inflammation.

Figure 4.

Intestinal Dot1l alters T cell subsets. (a) Body weight changes at the indicated time points in the AOM-DSS induced cancer model. DSS water treated windows are highlighted. Dot1l^FF (n = 7), Dot1l^ΔIEC (n = 8), *P = .0253 (Day 35), *P = .0266 (Day 55), 2way ANOVA. (b, c) Flow cytometry analysis shows IL-22 and IL-17 expressing CD4⁺ T cells in lamina propria in colorectal cancer model. n = 6/group, *P = .04 for IL-17; *P = .05 for IL-22, Student’s t-test. (d, e) Dot plots show RORϒt⁺ Foxp3⁺ CD4⁺ T cell subsets in lamina propria in Dot1l^FF and Dot1l^ΔIEC mice on day 70 of the colorectal cancer model. Mean± SEM, n = 7. *P = .05, Student’s t-test. (f, g) Dot plots show RORϒt⁺ Foxp3⁺CD4⁺ T cell subsets in mesenteric lymph nodes (mLNs) and colon tumor tissues in Apc^MinDot1l^FF and Apc^MinDot1l^ΔIEC mice on day 210 of spontaneous tumor model. n = 5–6. **P = .009, Student’s t-test.

H3K79 methylation status correlates with RORγ⁺ Tregs in human colorectal cancer

Given the interplay between intestinal Dot1l and lamina propria immune cell subsets, and its potential link to colorectal carcinogenesis in mouse models, we extended our studies to patients with colorectal adenoma and early colorectal cancer. To this end, we detected H3K79 di-methylation with immunohistochemistry in human colorectal adenoma and early colorectal cancer tissues (Figure 5a). The levels of H3K79me2 expression were lower in colorectal adenoma than colorectal cancer (Figure 5a –b). Based on mouse data (Figure 4f –g), we assumed that there might exist a correlation between H3K79 methylation level and RORγ⁺FOXP3⁺ Treg cells in patients with colorectal adenoma and early carcinoma. To this end, we conducted multiple color immunofluorescence tissue staining in human colorectal adenoma and early carcinoma tissues to identify and quantify RORγ⁺FOXP3⁺ Treg cells (Figure 5c). We detected high RORγ⁺FOXP3⁺ T cells in human colorectal colorectal adenoma and colorectal cancer tissues. There were more RORγ⁺FOXP3⁺ T cells in colorectal adenoma than early colorectal cancer tissues (Figure 5d). Thus, there is a link between H3K79 methylation level and RORγ⁺FOXP3⁺ Tregs in the gut during colorectal cancer development.

Figure 5.

H3K79 methylation status correlates with RORγ⁺ Tregs in human colorectal cancer. (a and b) H3K79me2 immunochemistry staining in human colorectal adenoma and early colorectal cancer tissues. (a) 4 representative images showed different levels of H3K79me2 expression. 40X magnifications. (b) Quantification of H3K79me2 levels from 5a in colorectal adenoma (green) and early carcinoma (blue) patients. Colorectal adenoma, n = 87; early colorectal cancer, n = 82, ***P = .00018, Mann-Whitney U-test. (c and d) RORϒ⁺FOXP3⁺ T cell staining in human colorectal adenoma and early colorectal cancer tissues. (c) Human colorectal adenoma tissues (left) and early colorectal tissues (right) were stained for CD3 (blue), FOXP3 (green), RORϒt (red), and DAPI (white). One representative image is shown. 40X magnifications. (d) RORϒ⁺FOXP3⁺ T cells were quantified and compared in colorectal adenoma (green) and early carcinoma (blue) patients. Colorectal adenoma, n = 82; early colorectal cancer, n = 72, **P = .009, Mann-Whitney U-test.

Discussion

Chronic immune stimulation and the Wnt/β-catenin pathway activation are two critical contributors to colorectal carcinogenesis. Our current work demonstrates that intestinal epithelial cell methyltransferase DOT1L is an intrinsic mechanism connecting these 2 factors in colorectal cancer.

DOT1L mediates H3K79 methylation.¹⁰ We have previously studied how Th22 cells affect cancer stemness via the DOT1L and H3K79me2 axis in the human colorectal cancer microenvironment.¹⁴ Using intestinal epithelium-specific Dot1l knockout mice, we have demonstrated a pro-tumor role of Dot1l in the Apc^Min and AOM/DSS-induced colorectal tumor models. Furthermore, we have observed an increase in H3K79me2 levels from colorectal adenoma to carcinoma in humans. The data reveal that intestinal epithelial intrinsic DOT1L directly affects colorectal carcinogenesis. Given that colorectal cancer development is associated with chronic inflammation and oncogenic pathway activation, we hypothesized that DOTL1 contributes to colorectal carcinogenesis via altering both chronic immune activation and oncogenic gene pathway activation in the intestinal microenvironment. In support of this, we observed a decrease in Th17 and Th22 cells, and an increase in RORγ⁺Foxp3⁺ regulatory T cells in lamina propria in Dot1l^ΔIEC mice. In line with mouse data, high levels of Th17 and Th22 cells are detected in patients with colorectal cancer and chronic bowel diseases,³³ and RORγ⁺Foxp3⁺ T cells negatively correlated with colorectal tumor stage and H3K79me2 level in patients with colorectal cancer. Thus, it seems that elevated DOT1L may tilt the balance between inflammatory and regulatory T cell subsets in the gut, thereby supporting and maintaining a chronic immune active microenvironment. In line with this possibility, loss of RORγ⁺Foxp3⁺ T cells can lead to an increase in Th17 cells and exacerbated 2,4,6-trinitrobenzene sulfonic acid (TNBS)-induced colitis in mice.³⁸ Thus, DOT1L mediated-H3K79 methylation correlates with chronic immune activation and colorectal cancer progression.

In addition to chronic immune activation, we have found a close relationship between DOT1L and the Wnt/β-catenin signaling pathway. We have detected reduced levels of several downstream genes in the Wnt/β-catenin signaling pathway in Dot1l^ΔIEC mice. Accordingly, Dot1l^ΔIEC intestinal epithelium cells form less organoids ex vivo compared to wild type cells. These results are supported and validated in human colorectal cancer cells with genetic and biochemical manipulation of DOT1L. Furthermore, ChIP-seq and ChIP-PCR assay reveal high levels of H3K79me2 signals in the promoters of the Wnt/β-catenin target genes in colorectal cancer cells. Thus, DOT1L mediated-H3K79 methylation can regulate the Wnt/β-catenin signaling pathway, thereby impacting colorectal cancer progression.

In summary, DOT1L, as an intrinsic molecular node, orchestrates two key elements, chronic inflammation and oncogenic pathway activation, to promote colorectal cancer development. Apart from patients with mixed-lineage leukemia,^12,41 the current study provides a rationale for initiating small-molecule screening to target DOT1L activity in patients at high risk of and/or with colorectal cancer.

Funding Statement

This work was supported by the University of Michigan endowment.

Author contributions

Conceptualization, D.S; E.R.F.; J.Y.F; W.Z.; Methodology, D.S., W.W.; F.G.; M.P.; W.D.; S.W.; S.G.; L.V.; Y.C.; Investigation, D.S., W.W.; F.G.; W.D.; S.W.; L.V.; Y.C. I.K.; Writing-Original Draft, D.S.; W.W.; W.Z.; Writing-Reviewing & editing, D.S., W.W.; F.G.; E.R.F; I.K.; J.Y.F.; W.Z; Funding Acquisition, W.Z.; Supervision, I.K.; J.Y.F; W.Z.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website