m6A modification and SUMOylation of PHGDH regulates ulcerative colitis through autophagy Authors: Hui Xue, Dandan Wang, Chengjian Pang, Yue Shen, Jingjing Xu, Lize Zhang Published: 2025 ABSTRACT AbstractBackgroundUlcerative colitis (UC) is a chronic inflammatory bowel disorder with increasing incidence globally. However, the understanding of UC pathogenesis remains limited.MethodsDSS-induced UC mouse model and lipopolysaccharide (LPS)-treated colonic epithelial cell (CEC) model were established. The pathological changes in colons were evaluated by hematoxylin and eosin. Colon injury was graded by pathological scoring, evaluation of colon length and disease activity index. qRT-PCR, Western blot, immunofluorescent staining, and ELISA assay were used to detect the expression and secretion of key molecules. CCK-8 was employed to monitor cell viability. The m6A modification of phosphoglycerate dehydrogenase (PHGDH) was assessed by MeRIP assay, and the SUMOylation of PHGDH, as well as the association between PHGDH and ubiquitin conjugating enzyme 9 (UBC9), were detected by co-immunoprecipitation.ResultsPHGDH knockdown alleviated DSS-induced colitis in mice. Inhibiting PHGDH promoted autophagy and inhibited inflammation in DSS-induced colitis and LPS-treated CECs. Methyltransferase-like 3 (METTL3) mediated m6A modification of PHGDH in LPS-treated CECs. PHGDH overexpression reversed METTL3 knockdown-facilitated autophagy in the in vitro UC model. Additionally, UBC9-mediated SUMOylation regulated PHGDH protein stability in vitro. Enhanced SUMOylation of PHGDH rescued LPS-impaired cell viability via promoting autophagy. METTL3 inhibited autophagy in LPS-treated CECs by regulating UBC9/PHGDH axis. METTL3 knockdown and UBC9 overexpression synergistically alleviated DSS-induced colon injury. Furthermore, curcumin promoted autophagy via modulating METTL3/UBC9/PHGDH axis, thereby alleviating colonic damage in ulcerative colitis.Conclusionm6A modification and SUMOylation of PHGDH regulated UC through Beclin-1-dependent autophagy.Supplementary informationThe online version contains supplementary material available at 10.1007/s00018-025-05970-9.Keywords: PHGDH, METTL3, UBC9, Ulcerative colitis Introduction Ulcerative colitis (UC) is a chronic inflammatory bowel disease (IBD) with global increasing incidence [1]. Patients with UC suffer from bloody diarrhea and abdominal pain. Current treatments include anti-inflammatory agents, immunosuppressants, biologics and surgery [2]. Despite advances in the UC therapy, UC is still associated with reduced life expectancy and a high risk of colorectal cancer (CRC) [3]. Additionally, the pathogenesis of UC remains ambiguous. It is well-established that UC predominantly damages the colonic mucosa and submucosa where colonic epithelial cells (CECs) serve as both a critical physical barrier and immunological defense [4–7]. Given the earliest compromised CEC homeostasis and their critical pathophysiological role, unraveling the molecular mechanisms regulating CECs will provide growing silver linings for UC treatment. Autophagy is a highly conserved cellular process which degrades intracellular materials [5, 8]. Emerging evidence indicates that autophagy plays a key role in UC pathogenesis and works in concert with apoptosis, endoplasmic reticulum stress, NOD-, LRR-, and pyrin domain-containing protein 3 (NLRP3) inflammation and gut microbiota to regulate the progression of UC [9]. Impaired autophagy has been observed in UC and is considered as a therapeutic target [9–11]. For instance, a number of natural products, such as curcumin and resveratrol, alleviate UC through modulating autophagy [12, 13]. These findings indicate that targeting autophagy is a promising therapeutic approach for UC. Phosphoglycerate dehydrogenase (PHGDH) catalyzes serine synthesis and regulates immune cell function [14, 15]. Myeloid-specific knockout of PHGDH ameliorates lipopolysaccharide (LPS)-induced inflammation in mice; PHGDH activates NLRP3 inflammasome to induce interleukin-1β (IL-1β) maturation [15]. In various cancers, high PHGDH expression is associated with increased proliferation an unfavorable prognosis. For instance, PHGDH is highly expressed in CRC, and its depletion enhances the sensitivity to stemregenin 1 in CRC cells by modulating autophagy [16]. What’s more, PHGDH inhibition induces autophagy in a phosphorylated m-TOR-independent but Beclin-1-dependent manner, and facilitates multilineage differentiation in embryonal carcinoma stem-like cells [17]. Our preliminary bioinformatics analysis predicted elevated PHGDH expression in the colons of UC patients and DSS-induced mouse colitis, raising the possibility that PHGDH might contribute to UC pathogenesis through regulating Beclin-1-dependent autophagy. Gene expression, protein stability, and functionality are governed by a multistep regulatory cascade encompassing transcriptional control, translational regulation, and post-translational modifications (PTMs) [18, 19]. N6- methyladenosine (m6A) is the most prevalent and abundant transcriptional modification in eukaryotes [18, 20]. Recently, m6A modification has emerged as a key player in UC pathogenesis, and several putative regulatory axes has been identified [21–26]. m6A is regulated by methyltransferase (such as methyltransferase-like 3 (METTL3) and methyltransferase-like 14 (METTL14)), demethylases (such as WTAP) and m6A reader [27]. A study found that METTL3 is upregulated in colonic tissues of mice with ulcerative colitis (UC) induced by dextran sulfate sodium (DSS) and 2,4,6-trinitrobenzenesulfonic acid (TNBS) [24]. Our previous bioinformatics prediction showed that PHGDH mRNA contains multiple m6A modification sites. However, whether PHGDH expression is regulated by m6A modification in UC remains unclear. It is worth noting that multiple studies have found that m6A modification and SUMOylation can interact with each other [28, 29]. SUMOylation is a PTM in which small ubiquitin-like modifier (SUMO) is covalently and reversibly attached onto target proteins [30]. Ubiquitin conjugating enzyme 9 (UBC9) serves as the only E2 conjugating enzyme [31]. SUMOylation has gained increasing attention due to its critical role in various biological processes, such as carcinogenesis, immune responses, cell apoptosis and DNA damage repair [32, 33]. SUMOylation plays a crucial role in UC and other IBDs [34, 35]. Bioinformatic analysis also predicted the presence of SUMOylation sites in the PHGDH protein. However, whether and how m6A modification and SUMOylation of PHGDH regulate its expression and biological function remains undefined. In this study, we demonstrated that METTL3 enhances the expression of PHGDH by promoting the m6A modification of PHGDH and inhibiting UBC9-mediated SUMOylation of PHGDH protein, thereby inhibiting autophagy of CECs and aggravating colon injury. Targeting the epigenetic regulation of PHGDH may be an effective therapy for UC. Materials and methods Bioinformatics analysis The dysregulated genes in UC patients and DSS-induced colitis mouse model were predicted by using GEO database (GSE243625 and GSE13705). Screening criteria: | foldchange | >1.5, P < 0.05. Bioinformatics analysis based on SRAMP (http://www.cuilab.cn/sramp) database predicted the m6A modification sites in PHGDH. Phosphosite database (https://www.phosphosite.org/homeAction) was employed to predict the SUMOylation sites of PHGDH. Clinical specimens A cohort of 22 UC patients and 30 healthy volunteers were recruited from the Affiliated Hospital of Qingdao University. This study was approved by Medical Ethics Committee of the Affiliated Hospital of Qingdao University, and written consents were obtained from all participants. Biopsies were taken from the inflamed mucosa in UC patients and the healthy mucosa in healthy controls by colonoscopy, respectively. Animal study All animal studies were approved by Medical Ethics Committee of the Affiliated Hospital of Qingdao University (QYFY WZLL 28915). Male C57BL/6 mice (6 ~ 8-week-old, n = 6 per group) were bought from Qingdao Tianxing Biotechnology Co., LTD. To establish UC mouse model, mice were fed with 2.5% (w/v) DSS (42867, Sigma-Aldrich, St. Louis, MO, USA) in drinking water for 7 days, followed by recovery for 3 days [36]. Mice were sacrificed on day 10, and the colons were dissected for subsequent analysis. The disease activity index (DAI) was scored as described: a, body weight loss: 0, no loss; 1, 1–5% loss; 2, 5–10% loss; 3, 10–20% loss. b, diarrhea: 0, normal; 2, loss stools; 4, watery diarrhea; c, hematochezia:0, no bleeding; 2, slight bleeding; 4, gross bleeding [36]. To conduct in vivo knockdown of PHGDH, mice were injected with adeno-associated virus (AAV)-negative control shRNA (sh-NC), AAV-sh-PHGDH, AAV-sh-METTL3, AAV-overexpression negative control (oe-NC), AAV-oe-UBC9 and AAV-oe-PHGDH (4 × 1010 vg in 100 µL PBS, OBio Technology, Shanghai, China) intracolonically for 2 weeks prior to DSS treatment [37]. For curcumin treatment, mice were administered curcumin via oral gavage at doses of 25, 50 and 100 mg/kg/day for 7 consecutive days as described [38]. Cell culture and treatment Human colonic epithelial cells (CECs, including Caco-2, HT-29 and FHC) and HEK293T cells were from ATCC (Manassas, VA, USA). Caco-2, HT-29 and HEK23T cells were cultured in DMEM containing 10% FBS (Gibco, Grand Island, NY, USA) at 37℃/5% CO2. FHC cells were grown in RPMI1640 supplemented with 10% FBS at 37℃/5% CO2. To establish the in vitro UC model, CECs were treated with 10 µg/mL LPS for 24 h, and these cells were treated with 1, 5, 10 µM curcumin (820354, Sigma-Aldrich) for 48 h. For protein stability assay, CECs were treated with 5 µg/mL cycloheximide (CHX, 239763, Sigma-Aldrich) for 0, 3, 6, 9 and 12 h. To inhibit autophagy, CECs were treated with autophagy inhibitor 10 mM 3-Mdthyladenine (3-MA, M9281, Sigma-Aldrich) for 3 h [39]. PHGDH activity assay The activity of PHGDH was measured as described [40]. Briefly, Flag-tagged PHGDH-WT and Flag-tagged PHGDH-E108A/I121A proteins were purified using Pierce anti-Flag magnetic agarose (A36797, Thermo Fisher Scientific) and added into the 96-well plate and incubated with the assay buffer. Phosphoserine Aminotransferase 1 protein was included to prevent the product inhibition of PHGDH. The enzyme activity was calculated by detecting the change in Nicotinamide Adenine Dinucleotide absorbance at Ex/Em 340/460 nm using a microplate reader (Thermo Fisher Scientific). Plasmid construction and cell transfection The sh-NC, sh-PHGDH, sh-METTL3 and sh-UBC9 were from GenePharma (Shanghai, China). For overexpression, the full-length of human PHGDH, METTL3 or UBC9 was cloned into pcDNA3.1 vector. Flag-tagged wild-type (WT) PHGDH, Flag-tagged PHGDH K21R and Flag-tagged PHGDH E108A/I121A mutants were cloned into p3xFLAG-CMV10 vector (Sigma-Aldrich) and HA-tagged UBC9 were constructed into pcDNA3.1-HA (#128034, Addgene, Cambridge, MA, USA). To study the SUMOylation of PHGDH, the PHGDH-SUMO1 and K21R-SUMO1 fusion proteins were generated. The PHGDH point mutant K21R and E108A/I121A were generated using QuickChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA). PHGDH-WT, PHGDH-K21R, PHGDH-E108A/I121A and SUMO1 were amplified by PCR. WT-PHGDH or K21R PHGDH (amino acids 1–533) was cloned downstream of SUMO1 (amino acids 1–96) as described [41]. CECs were transfected with shRNA or/and overexpression plasmid using Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA) for 48 h. qRT-PCR and RNA stability assay Total RNA was extracted from CECs using Trizol reagent (Invitrogen), and reverse transcribed using iScript Reverse Transcription Supermix (Bio-Rad, Hercules, CA, USA). qRT-PCR was conducted using SYBR Green Supermix (Bio-Rad) and quantified using 2−ΔΔCT method. For RNA stability assay, CECs were treated with Actinomycin D (ActD, 5 µg/mL, 114666-5MGCN, Sigma-Aldrich) for 0, 3, 6, 9 and 12 h. The stability of PHGDH mRNA was analyzed by qRT-PCR. Primers used in qRT-PCR were listed in Table 1. Table 1Primer information for qRT-PCR testPrimerSequence 5’−3’PHGDH human ForwardCTGCGGAAAGTGCTCATCAGTPHGDH human ReverseTGGCAGAGCGAACAATAAGGCPHGDH mouse ForwardATGGCCTTCGCAAATCTGCPHGDH mouse ReverseAGTTCAGCTATCAGCTCCTCCMETTL3 human ForwardTTGTCTCCAACCTTCCGTAGTMETTL3 human ReverseCCAGATCAGAGAGGTGGTGTAGUBC9 human ForwardCCCAGGAGAGGAAAGCATGGUBC9 human ReverseTTCCCACGGAGTCCCTTTCTGAPDH human ForwardAGGTCGGAGTCAACGGATTTGAPDH human ReverseTGACGGTGCCATGGAATTTGGAPDH mouse ForwardAGCCCAAGATGCCCTTCAGTGAPDH mouse ReverseCCGTGTTCCTACCCCCAATG Primer information for qRT-PCR test Western blot Total protein lysates were extracted from colons and CECs using RIPA lysis buffer (Pierce, Rockford, IL, USA), and protein estimation was conducted using BCA Protein Assay Kit (Pierce). After gel electrophoresis, proteins were transferred onto PVDF membrane (Bio-Rad) and blocked with 5% non-fat milk. The membranes were incubated with primary antibody at 4℃ overnight. The membrane was then incubated with secondary antibody, and the signal was visualized using SuperSignal Pico PLUS ECL (Pierce). Primary antibodies used in western blot: anti-PHGDH (1:2000, ab240744, Abcam, Cambridge, UK), anti-Beclin-1 (1:2000, ab207612, Abcam), anti-ATG5 (1:500, PA1-46178, Invitrogen), anti-ATG7 (1:5000, ab133528, Abcam), anti-LC3II/I (1:1000, PA1-46286, Invitrogen), anti-SQSTM1 (1:5000, ab109012, Abcam), anti-METTL3 (1:1000, ab195352, Abcam), anti-METTL14 (1:1000, ab309096, Abcam), anti-WTAP (1:1000, PA5-144681, Invitrogen), anti-UBC9 (1:1000, ab75854, Abcam), anti-Flag tag (1:2000, ab236777, Abcam), anti-HA tag (1:2000, ab9110, Abcam), anti-SUMO1 (1:1000, ab32058, Abcam), and anti-GAPDH (1:2000, ab8245, Abcam) antibodies. Hematoxylin and eosin (H&E) and Immunofluorescence (IF) staining The distal colons were fixed with 4% paraformaldehyde (PFA) and embedded in paraffin. For H&E solution, the sections were then stained with H&E solution (Beyotime, Shanghai, China). The histological damages were graded as described: a, inflammation: 0, no damage; 1, mild; 2, moderate; 3, severe. b, the lesion depth: 0, none; 1, mucosal layer; 2, submucosal layer; 3, muscle layer; 4, transmural. c, crypt damage: 0, none; 1, basal 1/3 damaged; 2, basal 2/3 damaged; 3, only surface epithelium intact; 4, entire crypt and epithelium lost. d, lesion range: 1, 1–25%; 2, 26–50%; 3, 51–75%; 4, 76–100% [42]. For IF staining, the frozen section of distal colon was fixed with 4% PFA and permeabilized with 0.3% Triton X-100. After blocking with 1% BSA, the slides were stained with anti-PHGDH (1:500, ab236763, Abcam), anti-EpCAM (1:100, ab213500, Abcam), anti-F4/80 (1:50, ab16911, Abcam), anti-α-SMA (1:100, ab7817, Abcam) anti-Beclin-1 (1:400, PA1-16857, Invitrogen) or anti-LC3B (1:400, PA1-46286, Invitrogen) antibody at 4℃ overnight, followed by secondary antibody incubation. Images were acquired using a confocal microscope (Nikon, Tokyo, Japan). Pharmacokinetics Male C57BL/6 mice were administered 100 mg/kg curcumin by oral gavage. Blood samples were collected at 0, 10 min, 30 min, 1, 2, 3, 4, 6 and 8 h after oral administration. The plasma were subjected to HPLC analysis as described [43]. The non-compartmental analysis of the mean concentration values was conducted using WinNonlin 5.2 program (Pharsight, Mountain View, CA, USA). ELISA assay The levels of TNF-α, IL-6 and IL-1β in mouse serum and cell culture supernatant were measured using commercial ELISA kits (mouse TNF-α, BMS607-3; human TNF-α, BMS2034; mouse IL-6, BMS603-2; human IL-6, BMS213-2; mouse IL-1β, BMS6002-2; human IL-1β, BMS224INST, Invitrogen) following the manufacturer’s protocols. A450 was measured using a microplate reader (Bio-Rad). CCK-8 assay CECs were seeded in 96-well plates 24 h prior to LPS stimulation. At 0, 24, 48 and 72 h, cells were incubated with 10 µL/well CCK-8 solution (Beyotime) at 37 °C for 1 h and A450 was measured. Methylated RNA immunoprecipitation (MeRIP)-qPCR PolyATract mRNA Isolation System (Promega, Madison, WI, USA) was employed to obtain mRNA. MeRIP was conducted using Magna MeRIP m6A Kit (Millipore, Billerica, MA, USA). Briefly, mRNA was subjected to RNA fragmentation, followed by the incubation with 5 µg anti-m6A antibody. The m6A-containing RNAs were immunoprecipitated by Protein A/G beads. qRT-PCR was employed to detect m6A modification of PHGDH and UBC9 in the elutes. Co-immunoprecipitation (Co-IP) The SUMOylation and ubiquitination of PHGDH, as well as the interaction between PHGDH and UBC9, were assessed by Co-IP. Briefly, Caco-2, HT-29 or HEK293T cells were transfected with the corresponding plasmids (Flag-PHGDH, Flag-PHGDH K21R, HA-UBC9, GFP-SUMO1) or treated with LPS according to the design of the experiments for 48 h. Cells were lysed using IP buffer. The immunocomplexes were then enriched using Protein A/G agarose (Pierce), and the elutes were analyzed by western blot with specific antibody overnight. Statistical analysis Data were presented as mean ± SD. Statistical differences between two-groups or multi-groups were analyzed using two-tailed Student’s t test and one-way ANOVA by SPSS22.0 software, respectively. P < 0.05 was considered to be statistically significant. Results Silencing of PHGDH alleviated DSS-induced colitis in mice Bioinformatics analyses based on GEO database (GSE243625 and GSE13705) predicted the upregulated genes in colonic mucosa of UC patients and colonic tissues of DSS-induced colitis mouse model. It was found that PHGDH was elevated in both UC patients and mouse model of DSS-induced colitis (Fig. 1A-B). In addition, clinical samples from UC patients similarly showed upregulated PHGDH mRNA and protein levels in the colons (Fig. 1C-D). Consistently, qRT-PCR and western blot showed the remarkable elevation of PHGDH in the colons of DSS-treated mice (Fig. 1E-F). Then we examined the co-localization of PHGDH and epithelial cell marker EpCAM. IF result found that PHGDH was upregulated in CECs of DSS-treated mice (Fig. S1A). What’s more, the co-localization between PHGDH and macrophage marker F4/80, as well as between PHGDH and fibroblast marker α-SMA, were also examined in mouse colons. No significant change in PHGDH expression was observed in macrophages or fibroblasts from the DSS-induced colitis model compared to those from normal mice (Fig. S1B-C), suggesting that UC-associated alteration of PHGDH primarily occurs in CECs. To further delineate the role of PHGDH in UC, in vivo knockdown of PHGDH was conducted prior to DSS administration. DSS-upregulated PHGDH was attenuated in DSS + sh-PHGDH group (Fig. 1G-H), indicating the successful knockdown of PHGDH in the colons. Moreover, silencing of PHGDH ameliorated DSS-induced body weight loss (Fig. 1I), DAI increasing (Fig. 1J) and colonic shortening (Fig. 1K) in vivo. Furthermore, H&E staining revealed that PHGDH knockdown protected against DSS-induced infiltration of inflammatory cells, mucosal damage and crypt loss (Fig. 1L). These findings indicate that PHGDH acts as a key player in DSS-induced colitis. Silencing of PHGDH alleviated DSS-induced colitis in mice. (A-B) The upregulated genes in UC were predicted based on GEO database (GSE243625 and GSE13705). (C) The mRNA level of PHGDH in healthy control (n=30) and UC colons (n=22) was detected by qRT-PCR. (D) The protein level of PHGDH in healthy control (n=6) and UC colons (n=6) was detected by western blot. Mice were randomly divided into two groups (n=6 per group): Normal and DSS groups. (E) The mRNA level of PHGDH in mouse colons was detected by qRT-PCR. (F) The protein level of PHGDH in mouse colons was detected by western blot. Mice were randomly divided into four groups (n=6 per group): Normal, DSS, DSS+sh-NC and DSS+sh-PHGDH groups. (G) The mRNA level of PHGDH in mouse colons was detected by qRT-PCR. (H) The protein level of PHGDH in mouse colons was detected by western blot. The body weight (I), DAI (J), the representative image of colons and length of colon (K) were examined. (L) The histological changes of mouse colons were assessed by H&E staining. Scale bar, 50 μm. *P<0.05, **P<0.01, ***P<0.001. Silencing of PHGDH facilitated autophagy and inhibited inflammation in DSS-induced colitis mouse model As shown in Fig. 2A, IF staining found that DSS dramatically promoted the infiltration of F4/80+ macrophages in the colons, while PHGDH knockdown decreased the number of F4/80+ macrophages. In line with this finding, ELISA assay revealed that DSS-upregulated serum levels of pro-inflammatory cytokines TNF-α, IL-6 and IL-1β were reduced in DSS + sh-PHGDH group (Fig. 2B). In addition, IF staining also showed that silencing of PHGDH reversed DSS-mediated reduction of autophagy-related proteins Beclin-1 and LC3B in the colons (Fig. 2C). Western blot revealed that DSS decreased LC3 II/LC3 I, Beclin-1, ATG5 and ATG7 expression, along with the upregulation of SQSTM1. By contrast, knockdown of PHGDH counteracted DSS-induced changes of these autophagy-related proteins in the colons (Fig. 2D). Collectively, these data suggest that PHGDH knockdown alleviates DSS-induced colitis via modulating inflammation and autophagy in vivo. Silencing of PHGDH facilitated autophagy and inhibited inflammation in DSS-induced colitis mouse model. Mice were randomly divided into four groups (n=6 per group): Normal, DSS, DSS+sh-NC and DSS+sh-PHGDH groups. (A) The infiltration of F4/80-positive macrophages in mouse colons was detected by IF staining. Scale bar, 50 μm. (B) The serum levels of TNF-α, IL-6 and IL-1β were assessed by ELISA assay. (C) The immunoreactivities of Beclin-1 and LC3B in mouse colons were detected by IF staining. Scale bar, 50 μm. (D) The protein levels of LC3 II/LC3 I, Beclin-1, ATG5, ATG7 and SQSTM1 were detected by western blot. ***P<0.001 PHGDH knockdown promoted autophagy and inhibited inflammation in LPS-treated CECs We next validated these the role of PHGDH in LPS-treated CECs (Caco-2, HT-29 and FHC cells) in vitro. Consistent with the in vivo data, the mRNA and protein levels of PHGDH in Caco-2, HT-29 and FHC cells were greatly upregulated by LPS (Fig. 3A-B, Fig. S2A-B). Knockdown experiments were next carried out. As expected, transfection of sh-PHGDH successfully downregulated PHGDH expression in CECs (Fig. 3C-D, Fig. S2C-D). LPS-increased PHGDH was also reversed by sh-PHGDH (Fig. 3E-F, Fig. S2E-F). Additionally, CCK-8 assay revealed that silencing of PHGDH rescued LPS-impaired cell viabilities of CECs (Fig. 3G, Fig. S2G). Furthermore, LPS downregulated the expression of LC3B in Caco-2, HT-29 and FHC cells, whereas PHGDH knockdown led to a rebound of LC3B in these cells as detected by IF staining (Fig. 3H, Fig. S2H). Similar with the in vivo findings, the protein levels of Belin 1, ATG5, ATG7 and LC3II/I in CECs were reduced by LPS, along with the upregulation of SQSTM1. PHGDH silencing counteracted LPS-mediated changes of these autophagy-related proteins (Fig. 3I, Fig. S2I). What’s more, knockdown of PHGDH reversed the effect of LPS on the release of inflammatory cytokines (TNF-α, IL-6 and IL-1β). Intriguingly, the protective effects of PHGDH knockdown on inflammation were abrogated by the autophagy inhibitor 3-MA (Fig. 3J, Fig. S2J). These findings suggest that PHGDH promotes LPS-induced inflammatory responses by suppressing autophagy in CECs. PHGDH knockdown promoted autophagy and inhibited inflammation in LPS-treated CECs. Caco-2 and HT-29 cells were treated with LPS. (A-B) The mRNA and protein levels of PHGDH in Caco-2 and HT-29 cells were detected by qRT-PCR and western blot. (C-D) The mRNA and protein levels of PHGDH in Caco-2 and HT-29 cells were detected by qRT-PCR and western blot after transfected with sh-NC/sh-PHGDH. Caco-2 and HT-29 cells were transfected with sh-NC/sh-PHGDH, followed by LPS treatment. (E-F) The mRNA and protein levels of PHGDH were detected by qRT-PCR and western blot. (G) Cell viability was monitored by CCK-8 assay. (H) The immunoreactivity of LC3B in Caco-2 and HT-29 cells were detected by IF staining. Scale bar, 10 μm. (I) The protein levels of LC3 II/LC3 I, Beclin-1, ATG5, ATG7, and SQSTM1 were detected by western blot. Caco-2 and HT-29 cells were transfected with sh-NC/sh-PHGDH, followed by LPS or/and 3-MA treatment. (J) The secretions of TNF-α, IL-6 and IL-1β were assessed by ELISA assay. *P<0.05,**P<0.01, ***P<0.001 Reducing PHGDH activity facilitated autophagy in CECs and alleviated inflammation It is known that PHGDH is a key enzyme in de novo serine biosynthesis. Previous study has reported that mutations of Glutamic acid (E) 108 and Isoleucine (I) 121 impair its enzyme activity [44]. To investigate the regulatory role of PHGDH enzyme activity in autophagy and inflammation, wild-type PHGDH (PHGDH-WT) and PHGDH-E108A/I121A mutant were generated (Fig. S3A). As anticipated, PHGDH activity assay showed that the mutation of E108/I121 markedly impaired the enzymatic activity of PHGDH (Fig. S3B). Silencing of PHGDH rescued LPS-impaired viability of FHC cells, while PHGDH-WT abrogated this protective effect and PHGDH-E108A/I121A failed to decrease cell viability (Fig. S3C). Consistently, sh-PHGDH-restored LC3B expression in LPS-treated FHC cells was reduced by PHGDH-WT, whereas PHGDH-E108A/I121A lacked this effect (Fig. S3D). PHGDH knockdown rescued LPS-induced changes of Belin 1, LC3II/I, and SQSTM1, which were further counteracted by PHGDH-WT but not by PHGDH-E108A/I121A (Fig. S3E). ELISA assay further revealed that inhibition of PHGDH reversed LPS-induced secretions of pro-inflammatory cytokines, whereas PHGDH-WT led to rebounds of these cytokines. By contrast, PHGDH-E108A/I121A had no effect on cytokine secretion (Fig. S3F). These findings suggest that suppressing PHGDH activity facilitates autophagy and alleviates inflammation in CECs. METTL3 mediated m6 A modification of PHGDH in LPS-treated CECs We next studied the mechanism underlying the upregulation of PHGDH in UC. Bioinformatics analysis based on SRAMP database predicted the m6A modification sites in PHGDH (Fig. 4A). MeRIP showed that LPS enhanced m6A modification of PHGDH (Fig. 4B). We next sought to identify the methyltransferase responsible for the m6A modification of PHGDH. Western blot revealed the upregulation of METTL3 and METTL14 in LPS-treated Caco-2 and HT-29 cells, while no significant change of WTAP was observed in CECs upon LPS treatment (Fig. 4C). Since LPS-mediated upregulation of METTL3 was more prominent, it was selected for the subsequent studies. Intriguingly, the elevation of METTL3 was also observed in the colon tissue of UC patients, compared with that of healthy control (Fig. S4A-B). In addition, the m6A modification of PHGDH was also increased in the colons of UC patients as detected by MeRIP (Fig. S4C). qRT-PCR and western blot confirmed the downregulation of METTL3 in METTL3-knockdown Caco-2 and HT-29 cells (Fig. 4D-E). Knockdown of METTL3 accelerated the decay of PHGDH mRNA in the presence of ActD in both CECs as detected by RNA stability assay (Fig. 4F). As anticipated, MeRIP revealed that loss of METTL3 reduced m6A modification of PHGDH (Fig. 4G). In LPS-treated CECs, METTL3 and PHGDH were upregulated, while silencing of METTL3 downregulated METTL3 and PHGDH expression (Fig. 4H-J). Moreover, LPS-enhanced PHGDH mRNA stability was attenuated in METTL3-knockdown CECs (Fig. 4K). These data suggest that METTL3 is responsible for the m6A modification of PHGDH in LPS-treated CECs, and METTL3 is required for the maintenance of PHGDH mRNA stability. METTL3 mediated m6A modification of PHGDH in LPS-treated CECs. (A) The m6A modification sites in PHGDH were predicted by SRAMP database. Caco-2 and HT-29 cells were treated with LPS. (B) The m6A modification of PHGDH was detected by MeRIP. (C) The protein levels of METTL3, METTL14 and WTAP were detected by western blot. Caco-2 and HT-29 cells were transfected with sh-NC or sh-METTL3. (D) The mRNA level of METTL3 was detected by qRT-PCR. (E) The protein level of METTL3 was detected by western blot. (F) The mRNA stability of METTL3 was assessed by RNA stability assay. (G) The m6A modification of PHGDH was detected by MeRIP. Caco-2 and HT-29 cells were transfected with sh-NC/sh-METTL3, followed by LPS treatment. The mRNA levels of METTL3 (H) and PHGDH (I) were detected by qRT-PCR. (J) The protein levels of METTL3 and PHGDH were detected by western blot. (K) The mRNA stability of METTL3 was assessed by RNA stability assay. *P<0.05,**P<0.01, ***P<0.001 Overexpression of PHGDH reversed METTL3 knockdown-facilitated autophagy in LPS-treated CECs To further investigate the function of METTL3/PHGDH axis in the in vitro UC model, gain- and loss-of-function experiments were carried out. Firstly, CECs cells were transfected with oe-PHGDH to upregulate PHGDH expression. Figure 5A-B found that oe-PHGDH dramatically upregulated PHGDH mRNA and protein levels in CECs. In addition, silencing of METTL3 abrogated LPS-induced PHGDH upregulation, whereas the effect of sh-METTL3 was compensated by PHGDH overexpression (Fig. 5C-D). CCK-8 assay showed that loss of METTL3 protected against LPS-impaired cell viability in CECs, while the protective effect of METTL3 knockdown was abolished in LPS + sh-METTL3 + OE-PHGDH group (Fig. 5E). Moreover, IF staining revealed that LPS led to a reduction of LC3B, while knockdown of METTL3 rescued LPS-suppressed LC3B which was further counteracted in PHGDH-overexpressing CECs (Fig. 5F). Consistently, silencing of METTL3 reversed LPS-mediated changes of autophagy-related proteins, whereas overexpression of PHGDH reversed these effects of sh-METTL3 (Fig. 5G). These findings suggest that METTL3 suppresses autophagy in LPS-treated CECs through increasing PHGDH expression. Overexpression of PHGDH reversed METTL3 knockdown-facilitated autophagy in LPS-treated CECs. aco-2 and HT-29 cells were transfected with vector alone (OE-NC) or PHGDH overexpression construct (OE-PHGDH). (A-B) The mRNA and protein levels of PHGDH was detected by qRT-PCR and western blot. Caco-2 and HT-29 cells were transfected with sh-NC/sh-METTL3 and OE-NC/OE-PHGDH, followed by LPS treatment. (C-D) The mRNA and protein levels of PHGDH was detected by qRT-PCR and western blot. (E) Cell viability was monitored by CCK-8 assay. (F) The immunoreactivity of LC3B in Caco-2 and HT-29 cells were detected by IF staining. Scale bar, 10 μm. (G) The protein levels of LC3 II/LC3 I, Beclin-1, ATG5, ATG7 and SQSTM1 were detected by western blot. *P<0.05,**P<0.01, ***P<0.001 UBC9-mediated SUMOylation regulated PHGDH protein stability in LPS-treated CECs We further explored the possible mechanisms underlying the upregulation of PHGDH protein levels under UC condition. Interestingly, through database analysis, we found that PHGDH may be regulated by SUMOylation at the post-translational level. SUMOylation is a PTM in which a series of enzymes catalyze SUMO attach onto target proteins, and UBC9 is the only E2 enzyme [31]. More importantly, the decreased expression of UBC9 were found in the colon tissues of UC patients (Fig. S4D-E). Then the Flag-tagged PHGDH, HA-tagged UBC9 and GFP-tagged SUMO1 were transfected into HEK293T cells to construct an exogenous SUMOylation system, and co-IP result revealed that regulated by SUMOylation (Fig. 6A). Similarly, endogenous SUMOylation of PHGDH was also observed in CECs (Fig. 6B). As shown in Fig. 6C, PHGDH SUMOylation at the lysine (K) 21 site was highly conserved among human, mouse and rat. Co-IP further revealed that the mutation of PHGDH SUMOylation site (K21R) attenuated the SUMOylation of PHGDH in HEK293T cells, as well as in CECs (Fig. 6D), indicating the significant role of K21 SUMOylation site on PHGDH. More importantly, LPS suppressed the SUMOylation of PHGDH in CECs (Fig. 6E), and it also decreased the association between PHGDH and UBC9 in these cells (Fig. 6F). In Caco-2 and HT-29 cells, transfection of HA-UBC9 vectors also promoted the endogenous SUMOylation of PHGDH (Fig. 6G). Moreover, protein stability assay showed that UBC9 overexpression greatly accelerated the degradation of PHGDH proteins time-dependently, while K21R mutation of PHGDH abolished this effect in the presence of CHX (Fig. 6H). Furthermore, Co-IP assay also revealed the overexpression of UBC9 increased ubiquitination of PHGDH in CECs (Fig. 6I). Consistently, LPS upregulated PHGDH expression in both CECs, whereas UBC9 overexpression counteracted LPS-induced PHGDH in CECs (Fig. 6J). These data indicate that the protein stability of PHGDH in UC model intestinal epithelial cells is regulated by UBC9-mediated SUMOylation. UBC9-mediated SUMOylation regulated PHGDH protein stability in LPS-treated CECs. (A) The SUMOylation of PHGDH in HEK293T cells was detected by Co-IP. (B) The SUMOylation of PHGDH in Caco-2 and HT-29 cells was examined by Co-IP. (C) Conservation of PHGDH SUMOylation sites across different species. (D) The SUMOylation of PHGDH after transfected with WT or PHGDH K21R (the SUMOylation site was mutated) vectors in HEK293T cells or CECs was detected by Co-IP. (E) The endogenous SUMOylation of PHGDH in LPS-treated Caco-2 and HT-29 cells was monitored by Co-IP. (F) The association between PHGDH and UBC9 in LPS-treated Caco-2 and HT-29 cells was assessed by Co-IP. (G) The SUMOylation of PHGDH in HA-UBC9-overexpressing Caco-2 and HT-29 cells was examined by Co-IP. (H) The protein stability of PHGDH in HEK293T cells was monitored by western blot in the presence of CHX when transfected with WT or PHGDH K21R vectors. (I) The ubiquitination of PHGDH in Caco-2 and HT-29 cells was monitored by Co-IP. (J) The protein levels of PHGDH in Caco-2 and HT-29 cells were detected by western blot when cells transfected with HA-tagged UBC9 and treated with LPS.**P<0.01 Enhanced SUMOylation of PHGDH rescued LPS-impaired cell viability via promoting autophagy Considering that PHGDH becomes unstable when UBC9 and SUMO1 are co-expressed. To exclude the interference of SUMO-modifying enzyme, we generated the Flag-tagged wild-type (WT) and K21R PHGDH-SUMO fusion protein (K21R-SUMO). In K21R-SUMO group, SUMO1 was directly fused to the PHGDH mutant (PHGDH-K21R) vectors, faithfully mimicking the endogenous SUMOylated PHGDH, to directly study the direct effect of SUMOylation of PHGDH on cellular functions. Western blot showed the successful overexpression of Flag-tagged proteins in CECs. The molecular weight of K21R -SUMO was slightly higher than WT control (Fig. 7A). For functional testing, cells were transfected with K21R-SUMO or K21R and treated with LPS. LPS-impaired cell viability was rescued by SUMOylated PHGDH (K21R-SUMO), while K21R mutant did not show this improvement in CECs (Fig. 7B). In addition, IF staining and western blot unequivocally showed LPS-mediated changes of autophagy-related proteins were reversed by K21R-SUMO. By contrast, whereas transfection with K21R failed to show similar changes (Fig. 7C-D). These findings suggest that enhanced SUMOylation of PHGDH protects UC model cells against LPS-induced impairments through facilitating autophagy. Enhanced SUMOylation of PHGDH rescued LPS-impaired cell viability via promoting autophagy. (A) The Flag-tagged WT PHGDH or K21R-SUMO (K21R PHGDH-SUMO fusion protein) were transfected into Caco-2 and HT-29 cells and tested by western blot. Caco-2 and HT-29 cells were transfected with K21R or K21R-SUMO vectors followed by treatment of LPS. (B) Cell viability in LPS-treated Caco-2 and HT-29 cells was monitored by CCK-8 assay. (C) The immunoreactivity of LC3B in LPS-treated Caco-2 and HT-29 cells was detected by IF staining. Scale bar, 10 μm. (D) The protein levels of LC3 II/LC3 I, Beclin-1, ATG5, ATG7 and SQSTM1 in LPS-treated Caco-2 and HT-29 cells were detected by western blot. *P<0.05, **P<0.01, ***P<0.001 METTL3 suppressed autophagy in LPS-treated CECs by regulating UBC9/PHGDH axis The above results revealed that both METTL3 and UBC9 can regulate the expression of PHGDH. Therefore, we hypothesized whether there is crosstalk or compensation between the m6A modification mediated by METTL3 and the SUMOylation mediated by UBC9. SRAMP database predicted the putative m6A modification sites of UBC9 (Fig. S5A). Moreover, silencing of METTL3 reduced the m6A modification of UBC9 mRNA (Fig. S5B), concomitant with increased UBC9 mRNA and protein levels (Fig. S5C-D). Moreover, we next sought to study the function of UBC9/PHGDH axis by loss-of-function experiments. As presented in Fig. S5E-F, qRT-PCR and western blot confirmed the successful knockdown of UBC9 in CECs. As expected, LPS-downregulated UBC9 was restored by sh-METTL3, while UBC9 knockdown reduced its expression. Conversely, knockdown of METTL3 attenuated LPS-mediated upregulation of PHGDH, whereas silencing of UBC9 led to a rebound of PHGDH (Fig. S5G). CCK-8 assay further showed that METTL3 knockdown rescued LPS-impaired viability of both Caco-2 and HT-29 cells, while UBC9 depletion abolished the protective effect of METTL3 knockdown (Fig. S6A). As shown in Fig. S6B, LPS deceased the immunoreactivity of LC3B in CECs, an effect reversed by METTL3 knockdown which was further abrogated by UBC9 depletion. Western blot revealed that silencing of METTL3 counteracted LPS-induced changes of LC3II/I, Belin 1 and SQSTM1, whereas the effects of sh-METTL3 were attenuated by UBC9 knockdown (Fig. S6C). Collectively, these data suggest that METTL3 can also reduce UBC9 in an m6A-dependent manner, thereby indirectly promoting the level of PHGDH protein and influencing autophagy in CECs. Mettl3 knockdown and UBC9 overexpression synergistically alleviated DSS-induced colon injury in mice To investigate whether Mettl3 and UBC9 affect the severity of colitis by regulating PHGDH expression, gain- and loss-of-function studies were conducted in DSS-induced UC mice. Western blot showed that Mettl3 knockdown or UBC9 overexpression reduced PHGDH, but induced UBC9 expression in the colons of DSS mice. Simultaneous Mettl3 silencing and UBC9 overexpression caused more pronounced changes in PHGDH and UBC9 expression (Fig. S7A). In DSS mice, colon length was remarkably reduced. Either Mettl3 depletion or UBC9 overexpression alleviated colon shortening, with the combined intervention demonstrating superior efficacy (Fig. S7B). Additionally, either Mettl3 knockdown or UBC9 overexpression alone ameliorated DSS-induced body weight loss, DAI increasing and histological injury, and the combined intervention showed greater protective effects (Fig. S7C-E). Furthermore, DSS-mediated changes of autophagy markers were reversed by either Mettl3 deletion or UBC9 overexpression, with synergistic restorative effects observed when both interventions were combined (Fig. S7F). These findings indicate that Mettl3 knockdown and UBC9 overexpression synergistically alleviate DSS-induced colon injury in vivo. Curcumin promoted autophagy via modulating METTL3, UBC9 and PHGDH in the in vitro UC model Previous study has illustrated that curcumin protects against DSS-induced UC via triggering autophagy [13]. Our findings suggest that METTL3, UBC9 and PHGDH served as crucial players in the regulation of CEC autophagy, therefore, we next sought to study if these key molecules were implicated in curcumin-mediated protection in UC. Western blot showed that LPS upregulated METTL3 and PHGDH protein levels, but downregulated UBC9 expression in CECs. Curcumin reversed LPS-mediated changes of METTL3, PHGDH and UBC9 expression dose-dependently (Fig. 8A). Additionally, curcumin also ameliorated LPS-impaired the viabilities of CECs (Fig. 8B). This was accompanied with enhanced autophagy in which LPS-mediated changes of autophagy-related proteins were counteracted by curcumin (Fig. 8C). To further investigate if curcumin’s regulatory effect on PHGDH was modulated by METTL3/UBC9, METTL3 overexpression or UBC9 knockdown were conducted. As presented in Fig. S8A-B, transfection of METTL3 overexpression construct successfully induced METTL3 expression in CECs. In LPS and curcumin co-treatment group of CECs, overexpression of METTL3 moderately increased PHGDH expression, and UBC9 knockdown synergistically potentiated this upregulation (Fig. S8C). These data indicate that curcumin protects CECs against LPS-induced impairment through regulating METTL3/UBC9/PHGDH. Curcumin promoted autophagy via modulating METTL3, UBC9 and PHGDH in the in vitro UC model. Caco-2 and HT-29 cells were treated with LPS (10 μg/mL) or/and curcumin (1, 5, and 10 μM). (A) The protein levels of METTL3, UBC9 and PHGDH were detected by western blot. Caco-2 and HT-29 cells were treated with LPS (10 μg/mL) or/and curcumin (10 μM). (B) Cell viability was monitored by CCK-8 assay. (C) The protein levels of LC3 II/LC3 I, Beclin-1, ATG5, ATG7 and SQSTM1 in Caco-2 and HT-29 cells were detected by western blot. *P<0.05,**P<0.01, ***P<0.001. Overexpression of PHGDH attenuated the effects of curcumin on autophagy and inflammation in the in vitro UC model We next investigated whether curcumin regulated autophagy in CEC via modulating PHGDH expression. As shown in Fig. 9A-B, oe-PHGDH resulted in the upregulation of both PHGDH mRNA and protein levels compared to the corresponding controls. Moreover, the protective effect of curcumin on cell viability was abrogated by PHGDH overexpression in LPS-treated CECs (Fig. 9C). Furthermore, ELISA assay revealed that LPS-stimulated secretions of TNF-α, IL-6 and IL-1β in CECs were dramatically rescued by curcumin, while the levels of these pro-inflammatory cytokines were increased in PHGDH-overexpressing CECs (Fig. 9D). Western blot also showed that curcumin-upregulated LC3 II/LC3 I, Beclin-1, ATG5 and ATG7, as well as curcumin-downregulated SQSTM1, were reversed by PHGDH overexpression in CECs upon LPS treatment (Fig. 9E). These data suggest the critical role of PHGDH in curcumin-mediated protection in the in vitro UC model. Overexpression of PHGDH attenuated the protective effects of curcumin on autophagy and inflammation in the in vitro UC model. Caco-2 and HT-29 cells were transfected with OE-NC/OE-PHGDH, followed by the treatment of LPS or/and curcumin. (A) The mRNA level of PHGDH was detected by qRT-PCR. (B) The protein level of PHGDH was detected by western blot. (C) Cell viability was monitored by CCK-8 assay. (D) The secretions of TNF-α, IL-6 and IL-1β were assessed by ELISA assay. (E) The protein levels of LC3 II/LC3 I, Beclin-1, ATG5, ATG7 and SQSTM1 in Caco-2 and HT-29 cells were detected by western blot. *P<0.05, **P<0.01,***P<0.001 Overexpression of PHGDH counteracted curcumin-mediated amelioration of colonic damage in DSS-induced UC mice The in vivo findings were next validated in DSS mice. In mouse colons, DSS upregulated PHGDH expression, while curcumin dose-dependently reduced PHGDH mRNA and protein levels (Fig. S9A-B). H&E staining revealed that DSS induced colonic mucosal damage, the infiltration of inflammatory cells and crypt loss in mice. Curcumin rescued these injuries in a dose-dependent manner, whereas PHGDH overexpression counteracted the protective effects of curcumin (Fig. S9C). In addition, curcumin reversed DSS-mediated changes of autophagy-related proteins, while PHGDH overexpression antagonized curcumin (Fig. S9D). These findings indicate that PHGDH overexpression abrogated the protective effects of curcumin on colonic damage in DSS mice. Furthermore, the pharmacokinetics parameters of curcumin were also examined. After oral administration of curcumin (100 mg/kg) to mice, the highest concentration in plasma was observed at 30 min with a peak concentration (Cmax) of 17.97 ng/mL (Fig. S9E). The Area Under the Curve (AUC) is 1948. The Half-life (t1/2) is 103.37 min. Understanding the pharmacokinetics of curcumin is essential to guide the rational design of drug formulations and dosing regimens, and to maximize the modulation of intracellular PHGDH levels. Discussion UC, a subtype of IBD, involves chronic inflammation and ulceration of the colon. Due to its unclear pathogenesis, there is no cure for UC currently. Therefore, further investigation is urgently needed to unravel the precise etiology of UC. In this study, we demonstrated that METTL3-mediated m6A modification and UBC9-mediated SUMOylation of PHGDH regulated autophagy and inflammation in UC, and aforementioned molecules contributed to curcumin-mediated protection in UC. Our findings identified PHGDH as a promising therapeutic target for UC, and elucidated the mechanism underlying curcumin’s therapeutic benefits in UC. CECs are key players in UC pathogenesis by secreting cytokines to recruit immune cells and regulate immune responses [45]. A compromised intestinal epithelial barrier which is formed by CECs also lead to mucosal inflammation [46]. Besides immune dysregulation and inflammation, previous study has also reported that autophagy in CECs protects against UC by the secretion of mucus and maintaining normal gut microflora [47]. More importantly, dysregulated autophagy in CECs resulted in excessive inflammation and aberrant immune responses, leading to the development of UC [48, 49]. Consistently, this study illustrated that DSS-induced impairments were associated with decreased autophagy, enhanced macrophage infiltration and elevated inflammation. Conversely, autophagy activation ameliorates UC via pathogen clearance, bacterial toxin resistance and immune modulation [5, 49]. Our findings found that PHGDH silencing or enzymatic inhibition facilitated autophagy and inhibited inflammation in CECs, thus alleviating UC in vivo and in vitro. PHGDH was identified as a novel therapeutic target for UC, and its precise mechanism in autophagy and inflammation merit in-depth investigation in the future study. Given the critical role of PHGDH in UC pathogenesis, this study further investigated its regulatory mechanism and discovered that m6A modification acts as an important upstream modulator of PHGDH in UC. m6A modification regulates RNA stability, splicing, translation and translocation through m6A writers (e.g., METTL3, METTL14 and WTAP), erasers and readers [20]. A recent study has illustrated that serum METTL3 and WTAP are correlated with the disease extent/type of UC, and WTAP is correlated with IL-6 level in UC patients [50]. Moreover, METTL3 is elevated in IBD samples and DSS-induced UC model in mice [51]. Dysregulated METTL14 expression in UC is associated with CEC apoptosis, ferroptosis, autophagy and inflammation [52–54]. The above suggests that abnormal m6A modification is closely related to UC progression. In line with this finding, this study reported that METTL3 and METTL14 were upregulated in LPS-treated CECs. However, since the change of METTL14 was less pronounced than METTL3 alteration, we did not further investigate if METTL14 participates in the regulation of PHGDH expression during UC pathogenesis. Our results revealed that METTL3 maintains the stability of PHGDH mRNA in an m6A-dependent manner. METTL3 overexpression exacerbates LPS-induced inflammatory responses in CECs and DSS-induced colitis in mice [51]. Li et al. have reported that METTL3 exacerbates LPS-induced UC [55]. Consistently, our findings showed that knockdown of METTL3 protected against LPS-impaired cell viability of CECs through facilitating autophagy, while PHGDH counteracted this rescue effect, indicating the significance of METTL3-mediated m6A modification of PHGDH in UC. Notably, some literature also reported conflicting findings. A study has illustrated that METTL3 deficiency impairs the self-renewal and differentiation capacities of intestinal stem cells [56]. Furthermore, a more recent study has demonstrated that METTL3 depletion aggravates DSS-induced colitis by impairing Lgr5+ stem cell maintenance [57]. These findings imply that both excessive and insufficient METTL3 expression might dysregulate colonic function. These discrepancies may be attributed to methodological differences. In this study, we modulated METTL3 expression specifically in CECs, while Ding et al. [57] employed complete METTL3 knockout affecting multiple types of cells and more complex regulatory mechanisms. These observations indicate that METTL3 may play cell-type-specific roles in the colon, warranting cell-specific METTL3 manipulation approaches to clarify its precise function. This study also explored SUMOylation as another crucial regulatory mechanism for PHGDH. SUMOylation is a dynamic PTM involving E1 enzyme SAE1/SAE2, E2 enzyme UBC9 and E3 SUMO ligases that catalyzes maturation, activation, ligation and deconjugation [31, 33]. It is generally accepted that SUMOylation regulates protein stability, activity and localization [31]. In various physiological and pathological processes, SUMOylation cooperates with other PTMs, such as phosphorylation, ubiquitination and acetylation [19, 58]. For instance, arsenic-mediated PML or PML-RARα SUMOylation triggers polyubiquitination and proteasomal degradation [59], suggesting the crosstalk between SUMOylation and ubiquitination in maintaining protein stability. A recent study has illustrated that SUMOylation-related signature genes are implicated in immune infiltration in UC, and NF-κB signaling regulates SUMOylation in UC [35], further supporting the role of SUMOylation in UC. Ritesh Kumar et al. have reported the decreased expression of UBC9 in UC colonic tissues [60]. Additionally, UBC9 is decreased in the colons of DSS mice which is accompanied with increased epithelial inflammation and reduction of Akt1 and SUMOylated Akt1 in IBD [34]. The present study found that the protein stability of PHGDH was regulated by UBC9-mediated SUMOylation, and SUMOylated PHGDH promoted autophagy in CECs upon LPS stimulation. Interestingly, we also found that METTL3-mediated m6A modification not only directly affects PHGDH but also indirectly intervenes in PHGDH by regulating the m6A modification and expression of UBC9. Moreover, simultaneous knockdown of METTL3 and overexpression of UBC9 can synergistically alleviate UC. Curcumin, an active polyphenol in the turmeric root, exhibits potent anti-tumor, anti-inflammatory and anti-oxidative properties. Clinical trials have illustrated its safety and efficacy in UC treatment [61], and animal studies have demonstrated that curcumin inhibits the release of pro-inflammatory cytokines and suppresses NF-κB signaling in UC [62]. Moreover, a recent report has demonstrated that curcumin protects against DSS-induced injury through modulating Treg/Th17 balance and cytokine release [63]. Besides the immunomodulatory effects, previous studies have demonstrated that curcumin exerts the protective effects in UC via regulating autophagy [13, 64]. Building upon these findings, the present study identified that METTL3/UBC9/PHGDH axis as a key molecular mechanism of curcumin-mediated autophagy in UC, and PHGDH was identified as a crucial downstream effector of curcumin in UC. In conclusion, this article found that PHGDH plays a key role in the UC process. Mechanically, METTL3-mediated m6A modification and UBC9-dependent SUMOylation coordinately regulated PHGDH expression and autophagy in CECs. Further investigation into the role of PHGDH in other cell types (e.g., dendritic cells and endothelial cells) and validation of these regulatory axes in clinical specimens are warranted in the future. Such studies will not only expand our understanding of PHGDH in UC but also strengthen its potential as a therapeutic target. Curcumin, as a drug to interfere with this pathway, combined with PHGDH inhibitors is expected to become a promising strategy for the treatment of UC. Supplementary information Below is the link to the electronic supplementary material. (PNG 3.46 MB) The localization of PHGDH in mouse colons. (A) The immunoreactivities of PHGDH and EpCAM in mouse colons were detected by IF staining. Green, EpCAM (epithelial cell marker); Red, PHGDH; Blue, DAPI. Scale bar, 20 μm. (B) The co-localization between PHGDH and F4/80 in mouse colons was detected by IF staining. Green, F4/80 (macrophage marker); Red, PHGDH; Blue, DAPI. Scale bar, 20 μm. (C) The co-localization between PHGDH and α-SMA in mouse colons was detected by IF staining. Green, α-SMA (fibroblast marker); Red, PHGDH; Blue, DAPI. Scale bar, 20 μm, ***P<0.001 High Resolution Image (TIF 10.3 MB) (PNG 620 KB) PHGDH knockdown promoted autophagy and inhibited inflammation in LPS-treated FHC cells. FHC cells were treated with LPS. (A-B) The mRNA and protein levels of PHGDH in FHC cells were detected by qRT-PCR and western blot. (C-D) The mRNA and protein levels of PHGDH in FHC cells was detected by qRT-PCR and western blot after transfected with sh-NC/sh-PHGDH. FHC cells were transfected with sh-NC/sh-PHGDH, followed by LPS treatment. (E-F) The mRNA and protein levels of PHGDH were detected by qRT-PCR and western blot. (G) Cell viability was monitored by CCK-8 assay. (H) The immunoreactivity of LC3B in FHC cells were detected by IF staining. Scale bar, 10 μm. (I) The protein levels of LC3 II/LC3 I, Beclin-1 and SQSTM1 were detected by western blot. FHC cells were transfected with sh-NC/sh-PHGDH, followed by LPS or/and 3-MA treatment. (J) The secretions of TNF-α, IL-6 and IL-1β were assessed by ELISA assay. *P<0.05, **P<0.01,***P<0.001. High Resolution Image (TIF 2.25 MB) (PNG 789 KB) Reducing PHGDH activity facilitated autophagy and alleviated inflammation in CECs. (A) The mutations of E108A/I121A (mutated glutamic acid (E) at site 108 to alanine (A), or mutated isoleucine (I) at site 121 to A). FHC cells were transfected with PHGDH-WT or PHGDH-E108A/I121A. (B) The enzymatic activity of PHGDH was accessed by PHGDH activity assay. FHC cells were transfected with sh-NC/sh-PHGDH or/and PHGDH-WT/PHGDH-E108A/I121A, followed by LPS treatment. (C) Cell viability was monitored by CCK-8 assay. (D) The immunoreactivity of LC3B in FHC cells were detected by IF staining. Scale bar, 10 μm.(E) The protein levels of LC3 II/LC3 I, Beclin-1 and SQSTM1 were detected by western blot. (F) The secretions of TNF-α, IL-6 and IL-1β were assessed by ELISA assay. *P<0.05,**P<0.01, ***P<0.001. High Resolution Image (TIF 2.67 MB) (PNG 291 KB) The levels of METTL3, PHGDH m6A and UBC9 in the colon tissues of UC patients. (A) The mRNA level of METTL3 in healthy control (n=30) and UC colons (n=22) was detected by qRT-PCR. (B) The protein level of METTL3 in healthy control (n=6) and UC colons (n=6) was detected by western blot. (C) The m6A modification of PHGDH in healthy control (n=6) and UC colons (n=6) was detected by MeRIP. (D) The mRNA level of UBC9 in healthy control (n=30) and UC colons (n=22) was detected by qRT-PCR. (E) The protein level of UBC9 in healthy control (n=6) and UC colons (n=6) was detected by western blot. ***P<0.001. High Resolution Image (TIF 1.16 MB) (PNG 416 KB) METTL3 upregulated PHGDH expression by promoting UBC9 m6A modification. (A)The putative m6A modification sites of UBC9 were predicted by SRAMP database. Caco-2 and HT-29 cells were transfected with sh-NC or sh-METTL3. (B) The m6A modification of UBC9 in CECs was detected by MeRIP. (C-D) The mRNA and protein levels of UBC9 in CECs was detected by qRT-PCR and western blot. Caco-2 and HT-29 cells were transfected with sh-NC or sh-UBC9. (E-F) The mRNA and protein levels of UBC9 in CECs was detected by qRT-PCR and western blot. Caco-2 and HT-29 cells were transfected with sh-NC/sh-METTL3 or/and sh-UBC9, followed by LPS treatment. (G) The protein levels of PHGDH and UBC9 were detected by western blot. *P<0.05, **P<0.01, ***P<0.001. High Resolution Image (TIF 2.04 MB) (PNG 927 KB) METTL3 suppressed autophagy in LPS-treated CECs by downregulating UBC9. Caco-2 and HT-29 cells were transfected with sh-NC/sh-METTL3 or/and sh-UBC9, followed by LPS treatment. (A) Cell viability was monitored by CCK-8 assay. (B) The immunoreactivity of LC3B in CECs were detected by IF staining. Scale bar, 10 μm.(C) The protein levels of LC3 II/LC3 I, Beclin-1 and SQSTM1 were detected by western blot. *P<0.05, **P<0.01, ***P<0.001. High Resolution Image (TIF 2.81 MB) (PNG 1.58 MB) Mettl3 knockdown and Ubc9 overexpression synergistically alleviated DSS-induced colon injury. Mice were randomly divided into six groups (n=6 per group): Normal, DSS, DSS+sh-NC+oe-NC and DSS+sh-Mettl3+oe-NC, DSS+sh-NC+oe-Ubc9 and DSS+sh-Mettl3+oe-Ubc9 groups. (A) The protein levels of PHGDH and Ubc9 in mouse colons were detected by western blot. The length of colon (B), body weight (C) and DAI (D) were examined. (E) The histological changes of mouse colons were assessed by H&E staining. Scale bar, 50 μm. (F) The protein levels of LC3 II/LC3 I, Beclin-1 and SQSTM1 in mouse colons were detected by western blot. *P<0.05,**P<0.01, ***P<0.001. High Resolution Image (TIF 5.07 MB) (PNG 354 KB) Curcumin downregulated PHGDH expression in CECs by modulating METTL3 and UBC9. Caco-2 and HT-29 cells were transfected with oe-NC or oe-METTL3. (A-B) The mRNA and protein levels of METTL3 in CECs were detected by qRT-PCR and western blot. Caco-2 and HT-29 cells were transfected with oe-METTL3 or/and sh-UBC9, followed by LPS and curcumin (10 μM, 48 h) treatment. (C) The protein level of PHGDH in CECs was detected by western blot. *P<0.05, **P<0.01,***P<0.001. High Resolution Image (TIF 8.46 MB) (PNG 1.17 MB) Overexpression of PHGDH counteracted curcumin-mediated amelioration of colonic damage in DSS-induced UC mice. Mice were randomly divided into five groups (n=6 per group): Normal, DSS, DSS+curcumin (25 mg/kg) and DSS+curcumin (50 mg/kg) and DSS+curcumin (100 mg/kg) groups. (A-B) The mRNA and protein levels of PHGDH in mouse colons were detected by qRT-PCR and western blot. Mice were randomly divided into seven groups (n=6 per group): Normal, DSS, DSS+curcumin (25 mg/kg) and DSS+curcumin (50 mg/kg), DSS+curcumin (100 mg/kg), DSS+curcumin (100 mg/kg)+oe-NC and DSS+curcumin (100 mg/kg)+oe- PHGDH groups. (C) The histological changes of mouse colons were assessed by H&E staining. Scale bar, 50 μm. Mice were randomly divided into five groups (n=6 per group): Normal, DSS, DSS+curcumin (100 mg/kg), DSS+curcumin (100 mg/kg)+oe-NC and DSS+curcumin (100 mg/kg)+oe-PHGDH groups. (D) The protein levels of LC3 II/LC3 I, Beclin-1 and SQSTM1 in mouse colons were detected by western blot. (E) Mean concentration–time profiles of curcumin at a dose of 100 mg/kg after oral gavage to mice (n=6). *P<0.05, **P<0.01, ***P<0.001. High Resolution Image (TIF 10.7 MB) Abbreviations adeno-associated virus Actinomycin D Area Under the Curve colonic epithelial cells cycloheximide peak concentration co-immunoprecipitation colorectal cancer disease activity index dextran sulfate sodium Gene Expression Omnibus hematoxylin and eosin inflammatory bowel diseases immunofluorescence interleukin-1β lipopolysaccharide N6-methyladenosine methylated RNA immunoprecipitation methyltransferase-like 3 NOD-,LRR-,and pyrin domain-containing protein 3 phosphoglycerate dehydrogenase post-translational modification small ubiquitin-like modifier 2,4,6-trinitrobenzenesulfonic acid Half-life ubiquitin conjugating enzyme 9 ulcerative colitis Wilms-tumor associated protein Authors’ contributions Hui Xue: Conceptualization; Writing - Original Draft; Visualization; Dandan Wang: Conceptualization; Methodology; Supervision; Chengjian Pang: Validation; Investigation; Resources; Yue Shen: Formal analysis; Data Curation; Jingjing Xu: Visualization; Writing - Review & Editing; Lize Zhang: Writing - Review & Editing; Project administration; Funding acquisition. Funding This work was supported by Shandong Province medicine Health Science and Technology Development Project (202203030855); Shandong Traditional Chinese Medicine Science and Technology Development Project (M-2023027); Qilu Health Leading Talents Training Project Special Fund; Shandong Provincial Natural Science Foundation(ZR2025MS1522). Data availability Data sharing not applicable to this article as no datasets were generated or analysed during the current study. Declarations Ethics approval and consent to participate All animal studies were approved by Medical Ethics Committee of the Affiliated Hospital of Qingdao University (QYFY WZLL 28915). Consent for publication Not applicable. Competing interests The authors declare that there is no conflict of interest. Footnotes Hui Xue and Dandan Wang are the co-first authors. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials (PNG 3.46 MB) The localization of PHGDH in mouse colons. (A) The immunoreactivities of PHGDH and EpCAM in mouse colons were detected by IF staining. Green, EpCAM (epithelial cell marker); Red, PHGDH; Blue, DAPI. Scale bar, 20 μm. (B) The co-localization between PHGDH and F4/80 in mouse colons was detected by IF staining. Green, F4/80 (macrophage marker); Red, PHGDH; Blue, DAPI. Scale bar, 20 μm. (C) The co-localization between PHGDH and α-SMA in mouse colons was detected by IF staining. Green, α-SMA (fibroblast marker); Red, PHGDH; Blue, DAPI. Scale bar, 20 μm, ***P<0.001 High Resolution Image (TIF 10.3 MB) (PNG 620 KB) PHGDH knockdown promoted autophagy and inhibited inflammation in LPS-treated FHC cells. FHC cells were treated with LPS. (A-B) The mRNA and protein levels of PHGDH in FHC cells were detected by qRT-PCR and western blot. (C-D) The mRNA and protein levels of PHGDH in FHC cells was detected by qRT-PCR and western blot after transfected with sh-NC/sh-PHGDH. FHC cells were transfected with sh-NC/sh-PHGDH, followed by LPS treatment. (E-F) The mRNA and protein levels of PHGDH were detected by qRT-PCR and western blot. (G) Cell viability was monitored by CCK-8 assay. (H) The immunoreactivity of LC3B in FHC cells were detected by IF staining. Scale bar, 10 μm. (I) The protein levels of LC3 II/LC3 I, Beclin-1 and SQSTM1 were detected by western blot. FHC cells were transfected with sh-NC/sh-PHGDH, followed by LPS or/and 3-MA treatment. (J) The secretions of TNF-α, IL-6 and IL-1β were assessed by ELISA assay. *P<0.05, **P<0.01,***P<0.001. High Resolution Image (TIF 2.25 MB) (PNG 789 KB) Reducing PHGDH activity facilitated autophagy and alleviated inflammation in CECs. (A) The mutations of E108A/I121A (mutated glutamic acid (E) at site 108 to alanine (A), or mutated isoleucine (I) at site 121 to A). FHC cells were transfected with PHGDH-WT or PHGDH-E108A/I121A. (B) The enzymatic activity of PHGDH was accessed by PHGDH activity assay. FHC cells were transfected with sh-NC/sh-PHGDH or/and PHGDH-WT/PHGDH-E108A/I121A, followed by LPS treatment. (C) Cell viability was monitored by CCK-8 assay. (D) The immunoreactivity of LC3B in FHC cells were detected by IF staining. Scale bar, 10 μm.(E) The protein levels of LC3 II/LC3 I, Beclin-1 and SQSTM1 were detected by western blot. (F) The secretions of TNF-α, IL-6 and IL-1β were assessed by ELISA assay. *P<0.05,**P<0.01, ***P<0.001. High Resolution Image (TIF 2.67 MB) (PNG 291 KB) The levels of METTL3, PHGDH m6A and UBC9 in the colon tissues of UC patients. (A) The mRNA level of METTL3 in healthy control (n=30) and UC colons (n=22) was detected by qRT-PCR. (B) The protein level of METTL3 in healthy control (n=6) and UC colons (n=6) was detected by western blot. (C) The m6A modification of PHGDH in healthy control (n=6) and UC colons (n=6) was detected by MeRIP. (D) The mRNA level of UBC9 in healthy control (n=30) and UC colons (n=22) was detected by qRT-PCR. (E) The protein level of UBC9 in healthy control (n=6) and UC colons (n=6) was detected by western blot. ***P<0.001. High Resolution Image (TIF 1.16 MB) (PNG 416 KB) METTL3 upregulated PHGDH expression by promoting UBC9 m6A modification. (A)The putative m6A modification sites of UBC9 were predicted by SRAMP database. Caco-2 and HT-29 cells were transfected with sh-NC or sh-METTL3. (B) The m6A modification of UBC9 in CECs was detected by MeRIP. (C-D) The mRNA and protein levels of UBC9 in CECs was detected by qRT-PCR and western blot. Caco-2 and HT-29 cells were transfected with sh-NC or sh-UBC9. (E-F) The mRNA and protein levels of UBC9 in CECs was detected by qRT-PCR and western blot. Caco-2 and HT-29 cells were transfected with sh-NC/sh-METTL3 or/and sh-UBC9, followed by LPS treatment. (G) The protein levels of PHGDH and UBC9 were detected by western blot. *P<0.05, **P<0.01, ***P<0.001. High Resolution Image (TIF 2.04 MB) (PNG 927 KB) METTL3 suppressed autophagy in LPS-treated CECs by downregulating UBC9. Caco-2 and HT-29 cells were transfected with sh-NC/sh-METTL3 or/and sh-UBC9, followed by LPS treatment. (A) Cell viability was monitored by CCK-8 assay. (B) The immunoreactivity of LC3B in CECs were detected by IF staining. Scale bar, 10 μm.(C) The protein levels of LC3 II/LC3 I, Beclin-1 and SQSTM1 were detected by western blot. *P<0.05, **P<0.01, ***P<0.001. High Resolution Image (TIF 2.81 MB) (PNG 1.58 MB) Mettl3 knockdown and Ubc9 overexpression synergistically alleviated DSS-induced colon injury. Mice were randomly divided into six groups (n=6 per group): Normal, DSS, DSS+sh-NC+oe-NC and DSS+sh-Mettl3+oe-NC, DSS+sh-NC+oe-Ubc9 and DSS+sh-Mettl3+oe-Ubc9 groups. (A) The protein levels of PHGDH and Ubc9 in mouse colons were detected by western blot. The length of colon (B), body weight (C) and DAI (D) were examined. (E) The histological changes of mouse colons were assessed by H&E staining. Scale bar, 50 μm. (F) The protein levels of LC3 II/LC3 I, Beclin-1 and SQSTM1 in mouse colons were detected by western blot. *P<0.05,**P<0.01, ***P<0.001. High Resolution Image (TIF 5.07 MB) (PNG 354 KB) Curcumin downregulated PHGDH expression in CECs by modulating METTL3 and UBC9. Caco-2 and HT-29 cells were transfected with oe-NC or oe-METTL3. (A-B) The mRNA and protein levels of METTL3 in CECs were detected by qRT-PCR and western blot. Caco-2 and HT-29 cells were transfected with oe-METTL3 or/and sh-UBC9, followed by LPS and curcumin (10 μM, 48 h) treatment. (C) The protein level of PHGDH in CECs was detected by western blot. *P<0.05, **P<0.01,***P<0.001. High Resolution Image (TIF 8.46 MB) (PNG 1.17 MB) Overexpression of PHGDH counteracted curcumin-mediated amelioration of colonic damage in DSS-induced UC mice. Mice were randomly divided into five groups (n=6 per group): Normal, DSS, DSS+curcumin (25 mg/kg) and DSS+curcumin (50 mg/kg) and DSS+curcumin (100 mg/kg) groups. (A-B) The mRNA and protein levels of PHGDH in mouse colons were detected by qRT-PCR and western blot. Mice were randomly divided into seven groups (n=6 per group): Normal, DSS, DSS+curcumin (25 mg/kg) and DSS+curcumin (50 mg/kg), DSS+curcumin (100 mg/kg), DSS+curcumin (100 mg/kg)+oe-NC and DSS+curcumin (100 mg/kg)+oe- PHGDH groups. (C) The histological changes of mouse colons were assessed by H&E staining. Scale bar, 50 μm. Mice were randomly divided into five groups (n=6 per group): Normal, DSS, DSS+curcumin (100 mg/kg), DSS+curcumin (100 mg/kg)+oe-NC and DSS+curcumin (100 mg/kg)+oe-PHGDH groups. (D) The protein levels of LC3 II/LC3 I, Beclin-1 and SQSTM1 in mouse colons were detected by western blot. (E) Mean concentration–time profiles of curcumin at a dose of 100 mg/kg after oral gavage to mice (n=6). *P<0.05, **P<0.01, ***P<0.001. High Resolution Image (TIF 10.7 MB) Data Availability Statement Data sharing not applicable to this article as no datasets were generated or analysed during the current study.