S-adenosyl-L-methionine supplementation alleviates damaged intestinal epithelium and inflammatory infiltration caused by Mat2a deficiency

The intestinal epithelium consists of proliferative crypts and differentiated epithelia (finger-shaped villi in the small intestine and flat surface in the large intestine). Intestinal stem cells self-renew and produce transit-amplifying (TA) cells in the crypts, then TA cells migrate upward along the crypt-villus axis and differentiate into various epithelial cells, which have a turnover of 3-5 days throughout adult life, maintaining the integrity of intestinal epithelium (Allaire et al., 2018; Barker, 2014). The intestine functions include digestion and absorption of nutrients, secretion of mucins and immunoglobulins, and selective barrier protection against harmful antigens and pathogens (Peterson and Artis, 2014). Damaged intestinal epithelium and barrier dysfunction are closely related to the occurrence of intestinal inflammation (Iwamoto et al., 1996; Turner, 2009; Yukawa et al., 2002). Abnormal alteration of intestinal homeostasis leads to increasing risk of intestinal diseases, such as inflammatory bowel diseases (Maloy and Powrie, 2011).

The gut is in direct contact with the nutritional environment, and intestinal development and homeostasis are tightly controlled by metabolites, such as amino acids, cholesterol, ketone body, lipids, etc. (Beyaz et al., 2016; Cheng et al., 2019; Wang et al., 2018, 2009). Methionine is an amino acid that is essential for normal growth and development. A study shows that only 20% of the dietary methionine intake is metabolized by gastrointestinal tissues in piglets (Riedijk et al., 2007). The effects of methionine on the intestinal tracts have been reported in many studies. Small intestinal villi height and width are increased after administration of methionine to rats (Seyyedin and Nazem, 2017). Sulfur amino acid deficiency suppresses epithelial growth in neonatal pigs (Bauchart-Thevret et al., 2009). Furthermore, stem cell proliferation and markers are decreased when mouse intestinal organoids are cultured in the methionine-depleted medium (Saito et al., 2017). Methionine metabolism also plays important roles in intestinal immune and anti-oxidative responses (Grimble, 2006; Luo and Levine, 2009; Tsiagbe et al., 1987). Methionine in the diets of nursery pigs enhanced duodenal morphology by improving glutathione production and reducing oxidative stress in mucosa cells (Shen et al., 2014).

Methionine adenosyltransferase (MAT) is a key enzyme in the methionine cycle. Among the three different isoforms of mammalian MAT, MAT I and MAT III are expressed in liver, whereas MAT II is widely expressed in most tissues (Hiroki et al., 1997). MAT II consists of the catalytic subunit MAT2A and the regulatory subunit MAT2B, and they are separately encoded by MAT2A and MAT2B (Kotb and Geller, 1993). MAT2A catalyzes the biosynthesis of S-adenosyl-L-methionine (SAM) from methionine and ATP. SAM is an important intermediate metabolite and the universal methyl donor required for various methyltransferases (Hirata and Axelrod, 1980; Lipson and Clarke, 2007; Svedruzic, 2008). Many effects of methionine metabolism are mediated by SAM. Short-term dietary methionine supplementation affects one-carbon metabolism and DNA methylation in mouse gut and alters normal gut physiology (Miousse et al., 2017). Dietary methionine modulates histone methylation in the liver in response to SAM and S-adenosyl-L-homocysteine (SAH) levels (Mentch et al., 2015). Previous studies suggest that genetic or pharmacological manipulation of SAM catabolism is sufficient to extend Drosophila lifespan (Obata and Miura, 2015; Parkhitko et al., 2016). Furthermore, some studies have provided evidence that maintenance and differentiation of embryonic and induced pluripotent stem cells are regulated by SAM (Shiraki et al., 2014; Shyh-Chang et al., 2013; Sperber et al., 2015). A recent study shows that dietary methionine regulates gut stem cell division by SAM in Drosophila (Obata et al., 2018). However, the underlying molecular mechanisms of how methionine metabolism affects mammalian intestinal development and homeostasis remain unknown.

To address this question, we generated transgenic mice that lack MAT2A expression in the intestinal epithelium. Overall, we found that reduced SAM synthesis mediates the effects of Mat2a deletion on shortened gut, atrophic intestinal epithelium, decreased cell proliferation, apoptotic intestinal epithelial cells (IECs) and intestinal inflammatory infiltration. We demonstrate that SAM is essential for intestinal homeostasis.

Our study provides new insights into the functional importance of SAM in intestinal development and homeostasis, and has significance for methionine metabolism-related neonatal intestinal health and intestinal pathology.

To characterize the biological function of Mat2a in the intestine, we generated intestine-specific Mat2a-knockout mice by intercrossing Vil-Cre mice and Mat2a^fl/fl transgenic mice (Fig. 1A). Unexpectedly, no Vil-Cre; Mat2a ^fl/fl (KO) mice were identified at weaning, indicating that Mat2a deletion in the intestine affected development. By counting genotypes of embryos and neonates from Vil-Cre; Mat2a ^fl/+ (Het) and Mat2a ^fl/fl (WT) intercrosses, we found that the number of KO mice was lower than the expected Mendelian ratio of 25% (16.9%, 19.1%, 13.3%, respectively) in three age groups from embryonic day (E) 14.5 to postnatal day (P) 0, indicating the occurrence of embryonic lethality (Table 1). For live KO mice at P0, their body length and body weight were significantly reduced compared with that of WT and Het littermates (Fig. 1B-D); these animals died within 24 h. Furthermore, the intestinal tracts of KO mice were shorter than those of WT and Het littermates (Fig. 1E,F) and small intestinal villi had been shed almost completely in the KO mice, as indicated by Hematoxylin and Eosin (H&E) staining (Fig. 1G,H). Next, we analyzed changes in intestinal epithelium at different embryonic stages by histological staining and found that the small intestinal villi became sparser and shorter, gradually worsening from E15.5 onwards, in the KO mice compared with corresponding WT mice (Fig. 1I,J). We also observed a significant reduction of proliferation in the intervillous regions, as shown by 5-ethynyl-2′-deoxyuridine (EdU) or Ki67 (Mki67) staining at stages of E16.5 and E17.5, respectively (Fig. S1A-D). More importantly, changes in intestinal epithelial cell type in small intestinal tissues from E18.5 fetuses were examined by quantitative PCR (qPCR). mRNA levels of the stem cell markers Lgr5 and Olfm4, the enterocyte marker Fabp2 and the goblet cell marker Tff3 were all significantly decreased in the KO fetuses compared with WT and Het littermates (Fig. S1E), but the enteroendocrine cell marker Chga was unchanged.

Based on the major effects of Mat2a on intestinal development and fetal survival, we set out to explore the function of Mat2a on intestinal homeostasis using Villin-CreERT2; Mat2a ^fl/fl transgenic mice, which could be treated with tamoxifen to knock out Mat2a in the IECs (Fig. 2A). The body weight of tamoxifen-induced Villin-CreERT2; Mat2a ^fl/fl mice (Mat2a ^iIEC-KO) was gradually reduced over time along with Mat2a knockout (Fig. 2B). By anatomical observation, we found significant changes in the intestines of Mat2a ^iIEC-KO mice at 4 days post-induction (dpi), including shorter, distended intestine and intestine filled with watery feces (Fig. 2C,D, Fig. S2A). Most KO mice died at 4 dpi accompanied by diarrhea and a few of them died at 5 dpi.

Next, we examined changes in intestinal epithelial structure and intestinal cell proliferation by histological analyses. MAT2A immunohistochemistry (IHC) results showed an absence of MAT2A expression in the small intestinal epithelial cells at 3 and 4 dpi. H&E staining of Mat2a ^iIEC-KO mice showed severely altered crypt-villus architecture (including shorter villi and progressively reduced numbers of crypts), broken epithelium and formation of subepithelial spaces (caused by retraction of basement membrane from the basal pole of the epithelial cells) in the small intestine (Fig. 2E-G, Fig. S2B). Phenotypes including broken epithelium and progressively reduced amounts of crypts were also observed in Mat2a-deficient colon tissues (Fig. S2C,D). Ki67 IHC staining showed that crypts of Mat2a ^iIEC-KO mice contained a significant decrease of Ki67-positive cells at 3 dpi and had nearly no Ki67-positive cells at 4 dpi (Fig. 2E,H, Fig. S2C,E). We also employed the small intestinal organoids model, which provides a measure of the ability of proliferation and differentiation of intestinal stem cells by enlargement and budding of crypts (Li et al., 2020) to detect the effect of Mat2a on intestinal proliferation. We constructed small intestinal organoids by culturing the jejunoileal crypts from villin-CreERT2; Mat2a ^fl/fl or C57BL/6J mice in medium supplemented with epidermal growth factor (EGF), R-spondin 1 and noggin. Then, 4-hydroxytamoxifen (4-OHT) or FIDAS-5 were used to delete Mat2a or inhibit the enzyme activity of MAT2A in the organoids, respectively. Notably, spheroids treated with 4-OHT grew slowly and had fewer buds compared with the control group, which displayed extensive budding starting from 24 h after treatment (Fig. 2I,K). Spheroids treated with FIDAS-5 had nearly no bud and maintained the morphology of cystic enteroids compared with the control group, which started budding at 48 h after treatment (Fig. 2J,L).

Given that the loss of Mat2a led to destroyed epithelial villi and reduced numbers of crypts, we hypothesized that Mat2a deletion likely caused the death of IECs. We performed colocalization experiments using terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) and immunofluorescence staining of epithelial cell markers in ileum tissues of Mat2a ^iIEC-KO mice and found that apoptosis signals were mainly located in the lower epithelium, close to the crypt zone. As for the cell populations with apoptosis, more apoptosis signals were located in proliferative cells (Ki67-positive cells) and stem cells (Olfm4-positive cells) at 2 and 3 dpi, as indicated by the white arrows (Fig. 3A). Incidentally, we also found that there were no Olfm4-positive cells and few Ki67-positive cells at 3 dpi. Little apoptosis signal was present in the enterocytes (Fabp1-positive cells) at 3 dpi and no apoptosis signal was detected in the Paneth cells (lysozyme staining-positive cells) at any time point (Fig. 3A,B). Based on the function of caspase 8 (Casp8) and caspase 3 (Casp3) as initiator and executioner, respectively, during apoptosis (Fraser and Evan, 1996), we next detected whether Casp8 and Casp3 were activated in the small intestinal crypts by western blot (WB). We found that protein levels of Pro-Casp8 and cleaved caspase 8 (CC8) were significantly elevated at 3 dpi, and reached higher levels at 3.75 dpi. Correspondingly, protein levels of Pro-Casp3 and CC3 were also significantly elevated at 3 and 3.75 dpi, but activation of Casp3 at 3.75 dpi was less pronounced than at 3 dpi (Fig. 3C). Furthermore, IHC staining of CC8 and CC3 on serial paraffin sections showed that CC8⁺ and CC3⁺ cells were mainly located in the crypt zone at 3 dpi, and there were fewer positive cells at 3.75 dpi (Fig. 3D-F). Based on studies that methionine regulates the mTORC1 signaling by SAM sensing (Gu et al., 2017) and gut epithelial TSC1/mTOR controls RIPK3-dependent necroptosis in intestinal inflammation and cancer (Xie et al., 2020), we detected changes of the mTOR signaling pathway and necroptosis signals in the Mat2a-deficient intestinal crypts by WB. The results of necroptosis markers p-RIPK3 and p-MLKL showed that there were only minimal p-RIPK3 signals at 3 and 3.75 dpi, and no detectable signal of p-MLKL (Fig. S3A). The results of p-4E-BP1 and p-S6K, hallmarks of mTOR activation, indicated their high expression at 3 and 3.75 dpi, and p-S6K signal was detected at 2 dpi (Fig. S3B). These results suggested that RIPK3-dependent necroptosis regulated by mTOR may happen in the late stage of Mat2a deficiency. Collectively, we conclude that Mat2a deletion mainly induces apoptosis of IECs.

To explore the mechanism of elevated Casp8/3 protein levels, we first investigated changes in DNA methylation levels based on previous studies showing that DNA methylation can regulate the expression of Casp8 and Casp3 (Barzi et al., 2020; Capper et al., 2009; Wu et al., 2010). However, methylation-specific PCR results of the Casp8/3 promoter showed that DNA methylation levels of Casp8 and Casp3 relatively unaffected after Mat2a deletion (loss of the band representing the methylated Casp8 promoter was only found in one Mat2a ^iIEC-KO mice) (Fig. S4A). We then examined changes in different methylated histone markers and found that levels of H3K4/9/27/79 trimethylation were all dramatically decreased in Mat2a-deficient intestines (Fig. S4B). Therefore, the increase in Casp3 and Casp8 protein levels was most likely epigenetically regulated by histone methylation.

Based on the fact that we observed both cell apoptosis and reduced cell proliferation, we explored the epistatic relationship between these two phenotypes using the organoid model. To test whether decreased proliferation could be rescued by apoptosis blockage under Mat2a deficiency conditions, Mat2a ^iIEC-KO organoids (induced by 4-OHT) were cultured in medium with or without Z-DEVD-FMK (a Casp3 inhibitor), then we observed the effect of apoptosis inhibition on proliferation by counting the number of buds in the organoids. We found that Mat2a ^iIEC-KO organoids treated with Z-DEVD-FMK kept budding at 72 h, whereas Mat2a ^iIEC-KO organoids without Z-DEVD-FMK began the process of apoptosis at the same time point (Fig. S5A,B). This result suggests that apoptosis blockage maintains the proliferation of Mat2a-deficient intestines.

A variety of differentiated cell types exist within the intestinal epithelium. The small intestinal epithelium consists of enterocytes, tuft cells, M cells and secretory cells, such as goblet cells, enteroendocrine cells and Paneth cells. The colonic epithelium contains colonocytes, goblet cells, enteroendocrine cells and tuft cells. Based on the results that intestinal epithelium was disturbed markedly after Mat2a deficiency, it is possible that Mat2a deficiency resulted in a change of intestinal cell-type composition. Thus, we examined changes in cell types by histological staining. There were abnormal cells with characteristics of both Paneth cells and goblet cells in Mat2a-deficient crypts, as shown by lysozyme IHC staining and Alcian Blue labeling on the same paraffin sections (Fig. 3G). The numbers of goblet cells and Paneth cells were all significantly decreased at 4 dpi (Fig. 3H,I). In the colon, the numbers of goblet cells were steadily decreased along with time after tamoxifen administration (Fig. 3J,K). Moreover, qPCR analysis revealed that the stem cell markers Lgr5 and Olfm4 were all significantly decreased at 3 and 4 dpi (Fig. 3L), consistent with the results of Olfm4 IF staining in 3 dpi ileum tissues (Fig. 3A). mRNA levels of the goblet cell marker Tff3 and the enteroendocrine cell marker Chga were unchanged at 3 dpi, but decreased at 4 dpi. mRNA levels of the tuft cell marker Dclk1 was unchanged both at 3 and 4 dpi (Fig. 3L, Fig. S6A).

As a crucial enzyme in the methionine metabolic pathway, loss of MAT2A is bound to cause changes of downstream metabolites in the methionine cycle. Therefore, we measured contents of SAM, SAH and homocysteine (Hcy) in ileum and colon tissues by liquid chromatography-mass spectrometry (LC-MS). There was a significant decrease of SAM content in both ileum and colon tissues, whereas a decrease of SAH content was only found in colon tissues of the Mat2a^iIEC-KO mice, as shown by relative quantification results (Fig. 4A, Fig. S6B).

Next, we wondered whether phenotypes caused by Mat2a deficiency could be rescued by SAM supplementation. As we expected, SAM content was elevated in ileum and colon tissues with SAM supplementation, with a simultaneous increase of SAH and Hcy content compared with Mat2a^iIEC-KO +Vehicle mice (Fig. 4B, Fig. S6C). Moreover, the trend of body weight loss was slowed down (Fig. 4C) and the length of intestinal tract was rescued by SAM supplementation in Mat2a^iIEC-KO mice (Fig. 4D,E). By contrast, SAM supplementation did not affect body weight or length of the intestinal tract of Mat2a ^fl/fl mice.

Given the results that Mat2a deficiency led to apoptosis signals mainly appearing at 3 dpi and death of most cells by 3.75 dpi, we chose the time point of 3 dpi to compare and clarify the recovery effects of SAM on apoptosis and other abnormal intestinal alterations. We first evaluated the effects of SAM on intestinal epithelium by histological analyses of 3 dpi intestines. Although expression of MAT2A was lost in all small intestinal and colonic epithelial cells, the length of small intestinal villi was restored, and there were partly recovered numbers of crypts and reappearing proliferative cells in both the small intestine and colon tissues (Fig. 4F-I, Fig. S6D-F). Moreover, qPCR analysis of 3 dpi small intestinal tissues showed that mRNA levels of Lgr5 and Olfm4 in Mat2a^iIEC-KO +SAM mice were elevated compared with Mat2a^iIEC-KO +Vehicle mice (Fig. S6A).

We then determined the effect of SAM on apoptosis of crypt cells. WB results showed a remarkable decrease of protein levels of Pro-Casp8 and Pro-Casp3 and their obvious inactivation in small intestinal crypts samples from Mat2a^iIEC-KO +SAM mice compared with Mat2a^iIEC-KO +Vehicle mice (Fig. 4J). CC8 and CC3 IHC staining on ileum tissues confirmed the inactivation of Casp8 and Casp3 in Mat2a^iIEC-KO +SAM mice compared with Mat2a^iIEC-KO +Vehicle mice (Fig. 4K-M), which is consistent with the WB results. Also, a suppressive effect of SAM on crypt cell apoptosis in colon tissues was observed by CC8 and CC3 IHC staining: the proportion of both CC8- and CC3-positive crypts were significantly decreased in Mat2a ^iIEC-KO +SAM mice compared with Mat2a ^iIEC-KO +Vehicle mice (Fig. S6G-I).

The reduced body weight and diarrhea in the Mat2a^iIEC-KO mice at 4 dpi (Fig. 2B, Fig. S2A) suggested the presence of intestinal inflammation. First, we detected the infiltration of leukocytes by IHC staining of CD45 (PTPRC; leukocyte marker), F4/80 (ADGRE1; macrophage marker) and CD3 (T cell marker). Indeed, we observed an expanding infiltration of leukocytes, including macrophages and T cells, in the intestinal epithelium and lamina propria of Mat2a^iIEC-KO mice, suggesting the occurrence of strong bowel inflammation after Mat2a deficiency (Fig. 5A, Fig. S7). Next, we measured changes in inflammatory cytokines and chemokines by qPCR. The results showed upregulation of the genes encoding TNF, pro-inflammatory cytokines and chemokines, such as Il6, Il1b, Cxcl1 and Ccl2, in Mat2a-deficient intestines (Fig. 5B).

Consistent with the restoration effect of SAM on intestinal epithelial integrity, relief of diarrhea and inhibition of immune cell invasion to some extent in the Mat2a^iIEC-KO +SAM mice were documented (Fig. 5C, Fig. S8A,B). Also, the increased gene expression levels of inflammatory cytokines and chemokines were downregulated in Mat2a^iIEC-KO +SAM mice compared with Mat2a^iIEC-KO +Vehicle mice (Fig. 5D), suggesting that SAM reduced inflammatory infiltration while restoring intestinal epithelial integrity. Additionally, SAM supplementation effectively elongated the lifespan of dying Mat2a ^iIEC-KO mice by one or two extra days (Fig. 5E, Movie 1).

To explore whether SAM has a general effect on intestinal inflammation, we established the dextran sulfate sodium (DSS)-induced colitis model and measured changes in SAM content and Mat2a expression in the colitis tissues following the scheme shown in Fig. S9A. SAM content was significantly reduced in the colitis tissues at day 9 compared with the control group, and its downstream metabolites, including SAH and Hcy, were also slightly reduced (Fig. 5F). The protein level of MAT2A was decreased in these colitis tissues (Fig. 5G). SAM supplementation was then performed for 1 week before DSS addition and was continued through the course of colitis induction (Fig. S9B). The results revealed that the reduction ratio of body weight in the mice that had been supplemented with SAM was slightly less than that of the vehicle mice in the late stage of DSS treatment (Fig. S9C) and the corresponding colon length was longer than that of the vehicle mice (Fig. 5H, Fig. S9D). Furthermore, we performed histological analyses of colitis tissues collected from day 9 diseased mice. Compared with vehicle treatment, SAM addition led to less crypt damage, slight decrease of ulceration (Fig. 5I,J, Fig. S9E) and a subtle reduction of inflammatory infiltration (Fig. 5K, Fig. S9E), as shown by H&E or IHC staining of inflammatory cell markers, respectively. Moreover, qPCR analysis of inflammatory cytokines and chemokines showed downregulation of the genes encoding TNF, pro-inflammatory cytokines and chemokines, such as Il6, Il1b and Cxcl1, in the mice supplemented with SAM compared with the vehicle mice (Fig. 5L). These results demonstrate that SAM has a beneficial effect in attenuation of DSS-induced colitis. Collectively, reduced SAM content caused by Mat2a deficiency induces the IEC apoptosis, destroys the integrity of intestinal epithelium and triggers bowel inflammation sequentially (Fig. 6), and SAM supplementation may be useful in the treatment of intestinal inflammation.

In this study, we show for the first time that SAM synthesis catalyzed by MAT2A is required for mammalian intestinal development and homeostasis in vivo. The effects of Mat2a deficiency on intestinal epithelium are region specific. Differentiated epithelium is disrupted grossly (broken intestinal epithelium and almost completely detached small intestinal villi) after Mat2a deficiency. Mat2a deletion also causes reduced proliferation, elevated IEC apoptosis (mainly proliferative cells and stem cells), and depletion of stem cells confirmed by Lgr5 and Olfm4 qPCR and Olfm4 IHC staining. Continuous renewal of the intestinal epithelium is indispensable for the intestine to be maintained as an intact and effective barrier and to protect the gut from commensal microbiota and digested food (van der Flier and Clevers, 2009). Thus, the extensive death of crypt cells leads to a deficient supply of IECs, breaks the balance of proliferation and differentiation, and further compromises the intestinal barrier integrity, resulting in inflammatory infiltration and diarrhea in mice.

We observed that the apoptosis signals caused by Mat2a deficiency are mainly in the crypt zone. This effect, combined with reduced cell proliferation, supports the notion that Mat2a is very important for the function of intestinal crypt cells. Also, a study by Jani et al. suggests that MAT2A plays an important role in activated proliferating T leukemic cells (Jani et al., 2009). These results show that Mat2a is very important for the survival of proliferating cells.

Previous studies have shown the importance of Mat2a for survival: a report of the International Mouse Phenotyping Consortium (https://www.mousephenotype.org/data/genes/MGI:2443731) shows that a homozygous Mat2a knockout mutation results in embryonic lethality, and a study by Mendel et al. reports that METTL16 has an essential role in mouse early embryonic development via regulation of Mat2a mRNA (Mendel et al., 2018). Consistent with this, the IEC-specific deficiency of Mat2a caused a certain percentage of embryonic death, postnatal lethality and adult rapid death.

Of note, this study expands the role of SAM as an important regulator of intestinal homeostasis and inflammation, especially under conditions of MAT2A dysfunction. In such a situation, dietary supplementation of SAM is beneficial to intestinal homeostasis and inflammation by alleviating defects in the crypt zone. In addition, attenuation of DSS-induced colitis by SAM strengthens the potential application of SAM supplementation to treatment of the intestinal inflammation. More broadly, people who are undergoing chemoradiotherapy may benefit from SAM supplementation, because their intestines can be severely damaged (Ch'ang et al., 2005; Potten, 2004).

It has also been reported that Mat2a expression is regulated by the microbiota (Jabs et al., 2020). Our study showed that the protein level of MAT2A was decreased in the DSS-induced colitis tissues, in which reduced microbial diversity and dysbiosis of gut microbiota had been reported by Munyaka et al. (2016); these results suggest a strong link between Mat2a and gut microbes. In our case, mice with IEC-specific deficiency of Mat2a all died once they displayed the symptom of diarrhea despite simultaneous application of SAM. This phenomenon indicates that the benefits of exogenous SAM administration for repair of intestinal epithelial integrity and alleviation of intestinal inflammation are insufficient to completely rescue intestinal disorders in late stages when more comprehensive pathological changes may happen, such as intestinal dysbiosis. We thus infer that the combination of antibiotic treatment and SAM supplementation probably keeps mice alive longer.

Finally, MAT2A is widely reported to support protein synthesis and tumor growth (Kalev et al., 2021; Villa et al., 2021; Wang et al., 2019; Yang et al., 2015), and previous studies have shown that MAT2A is highly expressed in colon cancer and provides a growth advantage for colon cancer cells (Chen et al., 2007; Wang et al., 2016). Based on our study, we suggest that SAM synthesis mediated by elevated MAT2A might help tumor cells escape apoptosis.

In summary, our study provides insights into the importance of Mat2a in intestinal development and homeostasis at the genetic level, elucidating some of the mechanisms underlying intestinal diseases and suggesting potential therapies.

Mat2a ^flox/flox mice were generated by CRISPR-Cas9 technology with loxP sites flanking exon 4 and exon 6, and were backcrossed to the C57BL/6J strain for five generations and then crossed with villin-Cre (Vil-Cre) mice to ablate Mat2a in the IECs. villin-CreERT2 (Vil-CreERT2) mice were kindly provided by Dr Jun Qin from Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences. They were crossed with Mat2a ^flox/flox mice to yield tamoxifen-induced Vil-CreERT2; Mat2a ^flox/flox mice. For induction of Vil-CreERT2-mediated recombination, 8-week-old mice were given one dose of tamoxifen (dissolved in corn oil, 50 mg/ml) at 400 mg per kg body weight (mg/kg) by intragastric administration. Mouse genotypes were identified by PCR with the primer sequences listed in Table S1. Animals were randomly allocated to the control or experimental groups. All experimental mice were matched by age and sex; sex preference did not exist in the mouse experiments presented here except the DSS-induced colitis model. All mice were in the C57BL/6J background and maintained in pathogen-free facilities in a conventional environment under a 12-h light/dark cycle, with free access to food and water. All mouse experiments were performed following the national guidelines for housing and care of laboratory animals (Ministry of Health, China) and protocols were in compliance with institutional regulations after review and approval by the Institutional Animal Care and Use Committee at Fudan University, Shanghai.

Ileum or distal colon tissues were selected as representative of small intestine or colon, respectively.

The initial induction time point of tamoxifen was designated as 0 dpi; the time point at which mice were dying was 4 dpi. Therefore, 0, 3 and 4 dpi were used as representative time points to display morphological changes in the tissues. Based on the result that the intestinal epithelium was severely damaged at 4 dpi, 3.75 dpi was selected as the final time point to display the changes of apoptosis signals and inflammatory signals. Based on the result that apoptosis signals mainly appeared at 3 dpi and most cells died at 3.75 dpi, the recovery effect of SAM on apoptosis and inflammatory infiltration was assessed at 3 dpi.

Antibodies in the IHC and immunofluorescence experiments were: rabbit anti-MAT2A (Atlas Antibodies, HPA043028), rabbit anti-Ki67 (Abcam, ab15580), rabbit anti-Olfm4 (Cell Signaling Technology, 39141), rabbit anti-lysozyme (Dako, A0099), rabbit anti-Fabp1 (Cell Signaling Technology, 13368), rabbit anti-cleaved caspase 8 (Cell Signaling Technology, 8592), rabbit anti-cleaved caspase 3 (Cell Signaling Technology, 9664), rat anti-CD45 (Thermo Fisher Scientific, MA5-17687), rabbit anti-F4/80 (Cell Signaling Technology, 70076), rat anti-CD3 (Thermo Fisher Scientific, 14-0032-82), goat anti-rabbit IgG conjugated with HRP (Servicebio, G1213), goat anti-rat IgG conjugated with HRP (Servicebio, GB23302), goat polyclonal antibody to rabbit IgG (Alexa Fluor 594) (Abcam, ab150080).

Antibodies in the WB experiments were: mouse anti-MAT2A (GeneTex, GTX50027), mouse anti-β-actin (Proteintech, 66009-1-Ig), rabbit anti caspase 8 (ABclonal, A0215), rabbit anti-caspase 3 (Cell Signaling Technology, 9662), rabbit anti-RIPK3 (ABclonal, A5431), rabbit anti-p-RIPK3 (Abcam, ab195117), rabbit anti-MLKL (ABclonal, A19685), rabbit anti-p-MLKL (Abcam, ab196436), rabbit anti-H3 (Cell Signaling Technology, 4499), rabbit anti-H3K4me3 (Cell Signaling Technology, 9751), rabbit anti-H3K9me3 (Cell Signaling Technology, 13969), rabbit anti-H3K27me3 (Cell Signaling Technology, 9733), rabbit anti-H3K79me3 (Cell Signaling Technology, 4260), rabbit anti-S6K (Cell Signaling Technology, 9202), mouse anti-p-S6K (Cell Signaling Technology, 9206), rabbit anti-4E-BP1 (Cell Signaling Technology, 9644) and rabbit anti-p-4E-BP1 (Cell Signaling Technology, 2855).

Reagents used for small intestinal organoids experiments were: EGF mouse recombinant (Sigma-Aldrich, SRP3196), recombinant murine noggin (PeproTech, AF-250-38), human R-spondin 1, recombinant (PeproTech, 120-38), Y-27632 (Sigma-Aldrich, Y0503), N2 supplement (Thermo Fisher Scientific, 17502048), B-27 supplement (Thermo Fisher Scientific, 17504044), GlutaMax (100×) (Thermo Fisher Scientific, 35050061), penicillin-streptomycin solution (HyClone, SV30010), Advanced DMEM/F12 Medium (Thermo Fisher Scientific, 12634010), Matrigel (Corning, 356231). Reagents used to knockout Mat2a were: tamoxifen (Sigma-Aldrich, T5648), 4-hydroxytamoxifen (Sigma-Aldrich, H6278). Inhibitor of MAT2A was FIDAS-5 (MedChemExpress, HY-136144).

For experiments at embryonic stages, the day when a vaginal plug was observed was considered as E0.5. For EdU in vivo labeling of fetal intestine, a dose of 50 mg/kg EdU solution (dissolved in PBS, 10 mg/ml) was injected between the inner thigh and genitals of pregnant mice to ensure no organs were punctured. The fetuses were collected 2 h after the injection, and intestinal sections were generated in paraffin. Three nonadjacent sections (technical replicates) for each of the three littermates (biological replicates) were stained with the Click-iT^® EdU Alexa Fluor^® 647 Imaging Kit (Invitrogen, C10340).

For SAM supplementation experiments, commercial ademetionine 1,4-butanedisulfonate enteric coated tablets (XiMeiXin, SFDA approval number: H20133197) were used. Mice involved in the experiments were pretreated with tablets (500 mg/kg) by intragastric administration every day for 1 week before tamoxifen treatment. SAM tablet supplementation was continued daily after induction with tamoxifen until the time point of observation and sampling.

Intestinal crypts were isolated from 6-week-old Vil-CreERT2; Mat2a ^flox/flox mice or C57BL/6J mice, following the published protocol from Mahe et al. (2013). In brief, 10 cm pieces of small intestines adjacent to the stomach were dissected and rinsed immediately with PBS. These were then cut into two pieces and were cut open longitudinally in PBS. Then, they were incubated in cold PBS containing 2 mM EDTA for 10 min on ice with slight oscillation after scraping off the villi using a coverslip. This EDTA digestion step was repeated once. Intestinal segments were then shaken vigorously in PBS solution containing 1% fetal bovine serum 50 times to strip crypts and then filtered through a 70-μm strainer. This step was repeated again and the supernatant collected from two filtrations was centrifuged at 900 rpm (163 g) for 5 min to gather crypts. Then, single cells and broken fragments were expelled by centrifuging in sorbitol/sucrose solution at 900 rpm (163 g) for 5 min to obtain purified crypts. For intestinal organoid culture, the purified crypts were counted, pelleted and resuspended at the desired density and plated in a 24-well plate in Matrigel matrix. The plate was incubated at 37°C for 15 min until the Matrigel was solidified. Then, organoid growth medium containing growth factor (advanced DMEM/F12 supplemented with B27, N2, EGF, noggin and R-spondin 1) was added to the plate and it was then incubated under at 37°C and 5% CO₂.

For 4-OHT and FIDAS-5 treatment experiments, primary crypts were allowed to grow for 12 h before treatments were carried out. Once separated crypts from Vil-CreERT2; Mat2a ^flox/flox mice closed up, they were treated with 4-OHT (5 μM) to delete Mat2a or treated with DMSO as control. For FIDAS-5 inhibition experiments, once separated crypts from C57BL/6J mice closed up, they were treated with FIDAS-5 (6 μM) to inhibit the enzyme activity of MAT2A or treated with DMSO as control.

For Z-DEVD-FMK treatment experiments, organoids treated with 4-OHT (5 μM) were cultured in the medium with or without Z-DEVD-FMK (40 μM) separately; normal organoids cultured in medium with DMSO or Z-DEVD-FMK separately were used as control groups.

Intestinal tissues were homogenized using TRIzol. Total RNA was extracted with chloroform followed by isopropanol precipitation and reverse-transcribed to cDNA following the manufacturer's instructions (Takara Bio). qPCR was performed using SYBR Premix ExTaq (Takara Bio, RR420A). Relative gene expression level was calculated by the comparative CT method using Hprt as an endogenous control. The sequences of primers for the qPCR are listed in Table S2.

Total protein of separated small intestinal crypts (separation method of crypts as for organoid experiments) was extracted with RIPA lysis buffer supplemented with protease inhibitors. Protein samples were mixed with SDS loading buffer and subjected to SDS-PAGE and immunoblotting. Antibodies against MAT2A (1:2000), caspase 8 (1:1000), caspase 3 (1:1000), RIPK3 (1:1000), p-RIPK3 (1:1000), MLKL (1:1000), p-MLKL (1:1000), H3 (1:3000), H3K4/9/27/79me3 (all 1:2000), S6K (1:1000), p-S6K (1:1000), 4E-BP1 (1:1000), p-4E-BP1 (1:1000) and β-actin (1:8000) were used at 4°C overnight. After that, secondary antibody was incubated for 1 h at room temperature.

Intestinal tissues were flushed with cold PBS and fixed in 4% paraformaldehyde solution for 2 days, embedded in paraffin and sectioned at 3 μm thickness. For H&E staining, sections were deparaffinized, rehydrated and then stained with H&E following a standard protocol. For IHC staining, sections were processed by antigen retrieval with citrate buffer solution and quenching endogenous peroxidases with 3% H₂O₂ solution (in methanol) after deparaffinization and rehydration. Sections were then incubated with goat serum for 30 min, followed by incubation with primary antibody overnight at 4°C. Then, sections were washed with PBS and incubated with goat anti-rabbit IgG conjugated with HRP (1:200) for 1 h, followed by detection using a DAB (3,3′-diaminobenzidine) kit (Servicebio, G1212). Primary antibodies used for staining were: rabbit anti-MAT2A (1:200), rabbit anti-Ki67 (1:2000), rabbit anti-lysozyme (1:2000), rabbit anti-cleaved caspase 3 (1:200), rabbit anti-cleaved caspase 8 (1:100), rat anti-CD45 (1:1000), rabbit anti-F4/80 (1:250) and rat anti-CD3 (1:700). For Alcian Blue/Nuclear Red staining, sections were incubated with Alcian Blue for 20 min and then stained with Nuclear Red for 3 min after deparaffinization and rehydration. All sections above were finally dehydrated in gradient ethanol, immersed in xylene for 5 min and mounted with neutral balsam for microscopy. For co-staining of TUNEL and immunofluorescence, sections were processed by antigen retrieval with citrate buffer solution after deparaffinization and rehydration. Then sections were permeated with 0.5% Triton X-100 for 10 min and then labeled using a TUNEL kit (C1086, Beyotime) for 1 h following the manufacturer's instructions. Next, sections were incubated with goat serum for 30 min, followed by incubation with primary antibody overnight at 4°C. Then sections were washed with PBS and incubated with goat polyclonal antibody to rabbit IgG (Alexa Fluor 594) (1:1000) for 90 min. Then, sections were mounted with ProLong Gold antifade reagent with DAPI (Invitrogen, P36935). Primary antibodies used for immunofluorescence staining were: rabbit anti-Olfm4 (1:150), rabbit anti-Ki67 (1:2000), rabbit anti-lysozyme (1:3000) and rabbit anti-Fabp1 (1:50).

First, genomic DNA was extracted from small intestinal crypts samples, then bisulfite conversion of 1 μg DNA was performed with EpiArt DNA Methylation Bisulfite Kit (Vazyme, EM101) following the manufacturer's instructions. Then, 20 ng purified bisulfate-modified products were used as templates to perform the MSP reactions using 2×EpiArt HS Taq Master Mix (Vazyme, EM202). After that, 10 μl PCR products were loaded to detect the changes of methylated and non-methylated bands by 1.5% agarose gel electrophoresis. PCR primers used in the MSP reactions were designed from the gene sequences containing the first exon and the 1000 bases upstream of the promoter region using the online tool MethPrimer (http://www.urogene.org/cgi-bin/methprimer/methprimer.cgi). The primer sequences are listed in Table S3. The PCR conditions were set according to the instructions of the 2×EpiArt HS Taq Master Mix, but the annealing temperature was changed to 53°C. The methylated or non-methylated primers of Casp8 amplified a 151 bp fragment or 155 bp fragment, respectively. The methylated or non-methylated primers of Casp3 amplified a 180 bp fragment or 183 bp fragment, respectively. Genomic DNA not converted with bisulfite was not specifically amplified by these primers (data not shown).

The detection of metabolite intensity was carried out as previously described (Ma et al., 2022). Briefly, 10 mg of intestinal tissues were homogenized in 1 ml ice-cold 80% methanol solution (HPLC-grade methanol diluted with HPLC-grade water), then processed by freezing-thawing with liquid nitrogen; the cycle was repeated twice. Lysates were then centrifuged at 13,000 rpm (16,200 g) for 15 min to pellet cell debris, and supernatants were collected and evaporated. Then, 50 μl of extraction solution was added to resuspend the powder, followed by vertexing and centrifugating at 16,000 g for 10 min at 4°C. Then, 3 μl solution was loaded to detect the intensity of metabolites by LC-MS. Changes of metabolite intensity in Mat2a ^iIEC-KO mice or colitis mice were normalized to control mice.

Briefly, 8-week-old male mice on a C57BL/6J background were given 2.5% (w/v) DSS (molecular mass 36-40 kDa, MP Biomedicals) in drinking water for 9 days. Time-matched mice given only normal drinking water were used as the control group.

Mice were randomly divided into four groups, with eight mice per group. For SAM-supplemented colitis mice, mice were pretreated with SAM tablets (500 mg/kg) by intragastric administration every day for 1 week before DSS induction. SAM tablets continued to be given every day after DSS induction until day 9. Time-matched mice given DSS-containing drinking water were used as the colitis group. Time-matched mice given only normal drinking water or only SAM tablet treatment were used as control groups.

The method of histopathological scoring of colitis was modified from the literature (Xu et al., 2016). Briefly, the degrees of crypt damage, inflammatory infiltration and ulceration were assessed in three individuals based on the following scoring system. The inflammatory infiltration score (0-5) was defined based on IHC staining results of distal colon sections as follows: 0=no infiltrate, 1=occasional inflammatory cells in the lamina propria, 2=many inflammatory cells in the lamina propria, 3=inflammatory cells existed in the both lamina propria and submucosa, 4=large areas of mucosa including surrounding blood vessels were infiltrated by inflammatory cells and 5=transmural inflammation. The crypt damage score (0-4) and the ulceration score (0-3) were defined based on the H&E staining results of distal colon sections as follows: for scoring of crypt damage, 0=none, 1=partial crypt damage, with gaps between crypts, 2=loss of goblet cells, some shortened crypt and larger gaps between crypts, 3=widely absent of crypt and 4=no crypts; for scoring of ulceration, 0=none, 1=small focal ulcers, 2=frequent small ulcers and 3=largely absent of colonic epithelium. For statistical analysis of each indicator, three mice per group were assessed by three individuals who were unaware of the treatment groups. Briefly, five 200× micrographs per mouse were selected for scoring, then the average score rated by three people was calculated.

Histological staining images were captured using an Olympus BX43 upright microscope or a Nikon microscope connected to a CCD (charge-coupled device) camera. Macro morphologies of mice and intestines were captured using a Nikon single-lens reflex camera. Images of organoids were taken using an Olympus inverted microscope. Fluorescence signals were captured using a Leica SP8 confocal laser microscope.

For measurement of various parameters, villus length was measured from the top of the crypt to the tip of the villus using the measurement tool in ImageJ, body length was measured from the tip of the nose to the base of the tail manually, and the length of the intestinal tract was measured from the start point of the duodenum to the end of the colon manually. For fields used to count crypts amounts and crypts containing Paneth cells, 200× micrographs were used, and crypts were counted within the field of view. Crypt-villus axes and crypt units were used to count various staining parameters, including Ki67⁺ cells, TUNEL signals, inflammatory cells, goblet cells, etc., were from 200× micrographs and positive signals were counted manually. For calculating the percentage of ileal crypts with CC8/CC3⁺ cells, crypts containing positive signals indicated by the arrows were counted manually (e.g. Fig. 4K). Briefly, 225 crypts per mouse from 200× photographs were identified, then the numbers of crypts containing positive signals were counted; the proportion of positive crypts in 225 crypts was calculated to obtain the percentage of crypts with CC8/3⁺ cells. Three or four mice were used for the statistics. Data shown in graphs are mean±s.e.m. Data from animal studies were collected by individuals unaware of the treatment groups. No data were excluded from the data analysis. For analysis of the statistical significance of differences between two groups, two-sided, unpaired Student's t-test was used. For analysis of three or more groups, one-way ANOVA with Tukey's test was used. The exact unit of statistics for each parameter is given in the figure legends.

We thank members of the Lei Laboratory for discussion throughout this study and the Biomedical Core Facility Leica SP8 LSCM, Ultimate 3000 chromatograph and 6500qtrap mass spectrometer of Fudan University Institutes of Biomedical Sciences for technical support. We thank Dr Jun Qin for providing Vil-CreERT2 mice. We thank Drs Yi-Fan Zhang and Qing Wu for guidance in some experiments and data analysis.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201135.reviewer-comments.pdf