Early-Life Gut Microbiota Governs Susceptibility to Colitis via Microbial-Derived Ether Lipids

Localized intestine inflammation could induce short-term increases in colonic oxygenation and leads to increases in the aerobic bacteria population and reduction in the anaerobic bacteria population by changing the intestinal environment. However, the mechanisms involved and the associated functions of intestinal anaerobes in gut health still remain unclear. Here, we found that early-life depletion of gut microbiota exacerbated later colitis, while mid-life microbiota depletion showed partially reduced colitis. Notably, we observed that early-life gut microbiota depletion confers susceptibility to ferroptosis in colitis. In contrast, restitution of early-life microbiota conferred protection against colitis and inhibited ferroptosis triggered by gut microbiota dysbiosis. Similarly, colonization with anaerobic microbiota from young mice suppressed colitis. These results may attribute to high abundance of plasmalogen-positive (plasmalogen synthase [PlsA/R]-positive) anaerobes and plasmalogens (one of the common ether lipids) in young mice but reduced abundance in the development of inflammatory bowel disease. Early-life anaerobic bacteria elimination also resulted in the aggravation of colitis, while this aggravation phenotype was reverted by plasmalogen administration. Interestingly, plasmalogens inhibited ferroptosis triggered by microbiota dysbiosis. We further find that the alkenyl-ether group of plasmalogens was critical to colitis prevention and ferroptosis inhibition. These data point to one of the mechanisms by which the gut microbiota controls susceptibility to colitis and ferroptosis early in life via microbial-derived ether lipids.


Introduction
Inflammatory bowel disease (IBD) is a chronic and relapsing gastrointestinal disorder that presents intestinal inflammation, with growing incidence worldwide [1]. While the precise cause of IBDs is unknown, several factors have been implicated in the pathogenesis, including age, genetics, immune system, and environmental factors [2]. One accepted hypothesis of its pathogenesis is gut microbiota alteration [3]. Recent work high lighted that gut microbiota dysbiosis plays an essential role in the pathogenesis of IBD [4]. Besides the microbiota, previous studies demonstrated that environmental factors, including oxygen, osmolality, nitrogen, and pH in the gut, were involved in the regulation of IBD [5]. It is well known that a healthy gut is characterized by low oxygen levels and enriched communi ties of obligate anaerobes [4], while IBD is characterized by high contents of nitrogen and reactive oxygen species (ROS). Circumstantial evidence of a role for obligate anaerobes in IBD has been accumulating for years [4]. Intestinal inflammation could induce the increases in colonic oxygenation and imbalance of bacterial population structure, especially the aerobic bacteria population and anaerobic bacteria population. Mounting evi dence showed the beneficial effects of obligate anaerobes and their products on the colonic environment in IBDs, including Bifidobacterium and Clostridium butyricum [1]. Notably, micro biota dysbiosis marked by an increase in facultative anaerobes and a reduction of obligate anaerobes has been observed in patients suffering from IBDs [6].
One of the most marked differences between obligate an aerobes and other organisms is the stark difference between aerobic and anaerobic biosynthetic routes of plasmalogen [5]. Plasmalogens, the most common form of ether lipids, are widely distributed in animals and anaerobic bacteria, sensitive to ROS, and involved in membrane structure, signaling and protection against ROS [7]. Although plasmalogen synthase (pls) has been observed to be widespread in many members of the gut micro biota, including obligate anaerobes and some facultative anaer obes [7], there is still relatively little information regarding the plasmalogenproducing bacterial in gut microbiota. Besides, plasmalogens have not been found in aerobic and facultative anaerobic bacteria until now. As such, a better understanding of and indepth characterization of bacterial communities of plasmalogenpositive species (pls [PlsA/R]positive) and their role in gut inflammatory diseases is needed to define therapeu tic targets. Although studies have documented that gut micro biota plays a crucial role in intestinal health and disease, there is still relatively little information regarding agerelated changes in gut microbiome from young adulthood to mature adult stage in mice and the mechanisms underpinning the ability of the gut microbiota on intestinal health, particularly in early life. Of note, etherlinked phospholipids have been revealed to be criti cal in driving ferroptosis, which is a regulator of intestinal dis eases [8]. In this study, we sought to address the role of the microbiota harboring the anaerobic plasmalogen biosynthetic pathway in establishing intestinal homeostasis in early life. We hypothesized that plasmalogenpositive (pls [PlsA/R]positive) bacterial species and plasmalogen conferred beneficial effects on intestine homeostasis in early life.

Early-life microbiota depletion exacerbates colitis
In the present study, to confirm whether the ageassociated alteration of gut microbiota composition between young and adult mice is involved in the pathogenesis of colitis, we used a broadspectrum antibiotic cocktail to deplete the gut micro biota ( Fig. 1A and L) in 3weekold (thereafter referred to as young mice) and 20weekold mice (thereafter referred to as mature adult mice). We then checked the impact of gut micro biota depletion in different stages of life on intestinal health. Culturedependent (Fig. S1A) and cultureindependent anal ysis (Fig. S1B) confirmed that the fecal microbiota was fully depleted in antibioticinduced microbiotadepleted (AIMD) mice. In agreement with previously published reports [9], early life antibioticinduced microbial depletion leads to further exacerbated colitis in mice after dextran sulfate sodium (DSS) challenge, as indicated by increased disease activity index score (stool consistency and colorectal bleeding, Fig. 1B and C), aggravated body weight loss (Fig. 1D), increased mortality ( Fig.  1E and F), and colon shortening ( Fig. 1H and I). Moreover, colon of antibiotic cocktailtreated mice exhibited loss of crypts and extensive ulceration after DSS exposure (Fig. 1G). Consistent with those results, antibioticinduced microbiota depletion upregulated mRNA levels of inflammatory cytokines (IL-1β, Tnf-α, and IL-6) and chemokines (Cxcl1, Cxcl1, Ccl2, and Ccl3), except S100a8, as compared with DSStreated mice (Fig. 1J and K). However, inconsistent with young mice, midlife micro biota depletion (Fig. 1L) reduced colitis upon DSS challenge ( Fig. 1M to P). Therefore, these findings showed a contrary pattern of susceptibility from earlylife and midlife microbial dysbiosis on colitis, probably reflecting the potential role of gut microbiota alteration mediated by age changes in intestinal health.

Early-life microbiota confers susceptibility to ferroptosis in colitis
Ferroptosis was identified as a cause of colitisassociated cell death of intestinal epithelial cells (IECs) [10], one of the char acteristics of IBD [11]. Thus, we hypothesized that regulation of ferroptosis may be an important process in maintaining intes tinal homeostasis. In our present study, we observed that DSS induced colitis (Fig. S2A, B, and E) was accompanied by enhanced colonic iron (Fig. S2C), malondialdehyde (MDA) levels (Fig. S2D), and COX2 (cyclooxygenase2) signals, and reduced GPX4 (glutathione peroxidase 4) levels ( Fig. S2F to P) in the colon tissues. These results suggested that ferroptosis was induced in the colonic IECs of colitis, a finding consistent with the previous study. To confirm whether ferroptosis is involved in the enhanced susceptibility from earlylife microbial dysbiosis to colitis, we then measured COX2, GPX4 levels, and iron con tents in the colonic tissues of AIMD mice. Immunohistochemistry and immunofluorescent assay of ferroptosis biomarkers revealed elevated positive signals for COX2 and reduced GPX4 signals in the epithelial cells from mice with earlylife microbial dysbi osis. We also found that colonic MDA and 4hydroxynonenal (4HNE) levels were obviously increased in mice with earlylife microbial dysbiosis (Fig. S3). Of note, after 1 week of DSS expo sure, AIMD mice exhibited high ferroptosis susceptibility as evidenced by increased lipid peroxide levels, elevated COX2 positive signals, and reduced GPX4positive signals ( Fig. 2A to E) and Fig. S4A and B). Furthermore, ferroptosis inhibitor ferrostatin1 (Fer1) extended mice survival and protect IECs against colitis in AIMD mice ( Fig. 2F to I) through inhibition of lipid peroxide accumulation ( Fig. 2J to L), which suggested that blockage of ferroptosis rescued colitis aggravated by early life microbial dysbiosis. Taken together, the data presented above indicated a potential nexus between ferroptosis and microbial dysbiosis in colitis.

Intestinal microbiota in early life contributes to suppression of colitis and ferroptosis
To further explore the process in that earlylife microbiota depletion renders mice susceptible to ferroptosis, we sought to test whether microbiota from young donor mice or adult donor mice in AIMD recipient mice alters the susceptibility to colitis and ferroptosis. According to the scheme described in Fig. 3A, young normobiotic mice (3weekold) were treated with broadspectrum antibiotics, followed by administration of 3% DSS for 7 days and fecal microbiota transplantation (FMT) to ensure gut microbiota colonization from either the young (3weekold) or adult (20weekold)donor mice. Of note, restoring gut microbiota from healthy young mice reverted clinical consequences of colitis and indicators of intestinal inflammation significantly, while the overall response of FMT Adulttreated mice was comparable to AIMD mice in colitis (Fig. 3). When compared with control mice, transplantation of gut bacteria from young mice showed reduced signs of intesti nal inflammation, as indicated by milder body weight loss (Fig.  3B), extended survival (Fig. 3D), reduction of the pathological Early-life but not mid-life antibiotic-induced microbiota depletion exacerbates DSS-induced colitis. (A) Study design using male AIMD mice at the age of 3 weeks. Threeweek-old mice were subjected to a 1-week oral administration of a broad-spectrum antibiotic cocktail and then treated with 3% DSS for 7 days. (B and C) Stool consistency and colorectal bleeding were scored in antibiotic-treated or untreated mice at the age of 3 weeks. (D) Body weight of antibiotic-treated or untreated mice at the age of 3 weeks was measured daily. (E and F) Survival curve of antibiotic-treated or untreated mice at the age of 3 weeks following treatment with 3% DSS. (G) Colon tissues of antibiotic-treated or untreated mice at the age of 3 weeks were examined histologically after H&E staining. Scale bars, 100 μm. (H and I) Antibiotic-treated or untreated mice at the age of 3 weeks were euthanized on day 17, and colon lengths were measured. (J and K) Heatmap showing the relative expression levels of genes involved in the inflammatory response in antibiotic-treated or untreated mice at the age of 3 weeks. (L) Study design using male AIMD mice at the age of 20 weeks. Twenty-week-old mice were subjected to a 1-week oral administration of a broad-spectrum antibiotic cocktail and then treated with 3% DSS for 7 days. (M) Survival curve of antibiotic-treated or untreated mice at the age of 20 weeks following treatment with 3% DSS. Stool consistency and colorectal bleeding were scored. (N) Body weight was measured daily. (O and P) Antibiotic-treated or untreated mice at the age of 20 weeks were euthanized on day 17, and colon lengths were measured. Data are shown as individual points with mean ± SEM. From (B) to (K), n = 6 mice at the age of 3 weeks (Ctrl and AIMD group), n = 10 mice (DSS and AIMD-DSS group). (L to P) n = 6 mice at the age of 20 weeks (Ctrl and AIMD group), n = 10 mice (DSS and AIMD-DSS group). ns, no significant difference, *P < 0.05 and **P < 0.01.   (Fig. 3E), and counteraction of colon shortening ( Fig. 3F and J), while FMT from adult mice resulted in high suscep tibility to colitis, as indicated by significant body weight loss (Fig. 3C), increased mortality (Fig. 3G), increased clinical score (Fig. 3H), and shortening of the colon ( Fig. 3I and K). Similarly, transplantation of gut bacteria from young mice, but not adult mice, exhibited the improvement of extensive ulceration, loss of crypts, and inflammation after DSS expo sure (Fig. 3L). On the contrary, gut microbiota colonization from adult (20weekold) donor mice had no overt effect on intestinal inflammation in AIMD mice during experimental colitis, as indicated by comparable ulceration, loss of crypts, and inflammation (Fig. 3L). Thus, these results demonstrate that colonization with early life microbiota counteracts intes tinal inflammation caused by the DSS challenge. In addition, transplantation of gut bacteria from young mice significantly reduced colonic MDA levels ( Fig. 3M) and 4HNE (Fig. 3N) levels compared with normal mice. COX2 protein expression was significantly suppressed, and GPX4 expression was in creased after transplantation of gut bacteria from young mice (Fig. 3O). These results indicate that earlylife gut microbiota colonization prevents ferroptosis through the regulation of ferroptosisrelated proteins.

Plasmalogen-positive species are enriched in the early-life human gut microbiome and reduced in the development of colitis
To validate our observations above, we here determined whether gut microbiota, represented by the bacteria residing in colonic contents, differed between young and mature adult male mice. Mounting evidence suggests that colitisassociated microbiota shifts were strongly involved in the development of colitis, thereby regulating gut homeostasis [1]. We first characterized microbiota profiles of 3weekold and 20weekold mice in the colon contents. We performed 16S ribosomal RNA (rRNA) gene sequencing on the colon contents (Fig. 4). Analysis of bacterial communities showed no significant differences in alpha diver sity ( Fig. 4A and B) and betadiversity ( Fig. 4D) analysis metrics between young and adults mice, in line with previous reports [5]. However, microbiome analysis revealed distinct differences in the gut microbiome between young and mature adult mice ( Fig. 4C and D), represented by the abundance of bacteria from the phyla and principal coordinate analysis. Plasmalogens, the most common form of ether lipids in mammals and anaerobe, contain a vinyl etherlinked fatty alcohol at position sn1 and has been reported to be essential components for governing ferroptosis [8]. Interestingly, we further confirmed that fecal ether lipids levels, including plasmalogens and etherlinked phospholipids, were remarkably reduced in mice with micro biota depletion, which suggested that those ether lipids could be microbialderived ( Fig. S5A and B). Here, on the basis of the biosynthetic characteristics of plasmalogen in anaerobe, we tried to address the role of plasmalogenpositive bacterial spe cies in establishing intestinal homeostasis in early life.
It has been previously shown that the genes from micro biomes that contained the pls operon (encodes PlsA and PlsR proteins) were classified as a species as plasmalogenpositive, and the distribution of the pls operon helps us to assess the distribution of plasmalogen biosynthesis genes [7]. Notably, among the plasmalogenrelated Pfams and plasmalogenpositive species identified by Jackson et al. [7], we observed the decreased distribution of plasmalogenpositive species in mature adult mice ( Fig. 4E) compared to young mice as indicated by the vol cano plot. On the basis of a combination of prior studies [7,12], we illustrate a schematic of proposed anaerobic synthesis path way of plasmalogens, followed by conversion of plasmalogens to dimethylacetal (DMA) in the presence of acid/methanol (Fig. 4F) [5]. DMA was formed from aliphatic aldehyde groups from the sn1 position of plasmalogens, while fatty acid methyl ester (FAME) was formed from fatty acid in diacylglycerophospholipids. We analyzed the profile and abundance of DMA derivatives (per centage relative to total FAME and DMA) that represent ali phatic chains from ether lipids (Fig. 4G). DMA was identified with the domination of C18:1 DMA, C18:0 DMA, C16:0 DMA, C12:0 DMA, and C11:0 DMA in the colon contents of 3week old and 20weekold mice (Table S1). As detection of DMA abundance (% in DMA and FAME) in the colon contents of young mice was significantly higher (Fig. 4H) than in the adult mice (P < 0.01), this suggested the number of plasmalogens from the gut flora was significantly decreased in adult mice, compared with young mice. Of note, as detection of DMA levels in the colon contents of AIMD mice and fecal etherlinked phos pholipids levels were significantly lower than in the specific pathogenfree (SPF) control mice ( Fig. S6A and B), which under scored that these ether lipids were microbiotadependent. On the basis of the reduction of plasmalogenproducing ana erobic bacteria and its plasmalogen levels in the colon contents of adult mice, these results indicate that plasmalogenpositive microbiota and its crucial metabolite (plasmalogens) decreased significantly with age, but whether this affects intestinal health was not known.
Our observation that colonic earlylife microbiota, which is identified to be enriched with plasmalogenpositive species, contribute to the prevention of colitis development and driv ing ferroptosis susceptibility prompted us to test whether plasmalogen positive microbiota plays a role in the develop ment of colitis. Herein, we sought to investigate whether the reduced abundance of plasmalogenpositive bacteria is linked to aggravated colitis. We first characterized fecal microbiota profiles of nonIBD individuals (n = 12) aged 13 to 29 years and IBD patients (n = 19) aged 13 to 30 years, which showed significant separation of microbial profiles between patients with healthy and IBD individuals ( Fig. 5A to D). Previous studies have shown that colon inflammation could induce proportional changes in the gut microbiota [13]. The above results also suggest that the microbial composition of feces from IBD patients is quite different from that of feces from the healthy population. Next, we analyzed the differences in plas malogenpositive species between IBD patients and the healthy population. Jackson et al. have classified plasmalogenpositive species on the basis of the Pfam domains related to pls (PlsA/R) [7]. Our study assessed the OTU (operational taxonomic unit) levels of plasmalogenpositive bacteria containing plasmalo gen biosynthesis genes and found that the abundance of plas malogenpositive bacteria in IBD patients was significantly lower than in healthy people (Fig. 5E)). We next characterized microbiota profiles of DSSinduced colitis in the colon con tents of young mice (3 weeks old) via 16S rRNA gene sequenc ing. Microbiome analysis revealed distinct differences in the gut microbiome between control mice and DSSinduced colitis mice, represented by the abundance of bacteria from the phyla (Fig. 5F)) and principal coordinate analysis (Fig. 5G). As sess ment of bacterial communities showed a significant decrease  in alpha diversity ( Fig. 5H and I) analysis metrics between young and adult mice. Of note, we also observed the decreased dis tribution of plasmalogenpositive bacteria species in mice with colitis ( Fig. 5J) when compared to SPF control mice. We then analyzed the distribution of DMA derivatives that represent aliphatic chains from plasmalogens (Fig. 5K). We found that DMA levels in the colon contents of DSStreated mice were sig nificantly lower than SPF control mice, which suggested that plasmalogens and plasmalogenproducing bacteria are nega tively correlated with colitis.

Intestinal anaerobic bacteria in early life protects against colitis
Accumulating evidence suggests that plasmalogens could be produced by anaerobic bacteria [7]. Therefore, to determine whether plasmalogens from the colon contents were secreted by anaerobic bacteria, we treated the mice at 3 weeks of age with metronidazole (Metro, one of the antibiotics with an aerobic bacteriaspecific inhibition) for 1 week. The culture dependent analysis confirmed that the anaerobic bacteria were depleted in the colon contents of metronidazoletreated mice (Fig. S1A). As detection of DMA levels in the colon con tents of metronidazole induced anaerobic bacteriaeliminated mice was significantly lower than the SPF control mice (Fig.  S6), which underscored that anaerobic bacterium are the mainly microbial origin for the colonic plasmalogens in mice. We next treated metronidazole induced anaerobic bacteria eliminated mice with 3% DSS (Fig. S7A). Of note, we found that depletion of anaerobic bacteria in early life by metroni dazole also leads to further exacerbated DSSinduced colitis in mice, as indicated by aggravated body weight loss (Fig.  S7D), increased mortality ( Fig. S7E and F), destruction of the epithelial architecture (Fig. S7G), and upregulated mRNA levels of inflammatory cytokines (Tnf-α, IL-6, and IL-1β) and chemokines (Cxcl1, Ccl2, and Ccl3) (Fig. S7J and K). Besides, depletion of anaerobic bacteria did not cause further aggra vated clinical scores ( Fig. S7B and C) and colon shortening ( Fig. S7H and I).
Given that neither earlylife microbiota depletion nor an aerobic bacteria elimination suppressed the development of acute colitis caused by DSS, we suggested that reduction of anaerobic bacteria in the early life of mice leads to a shift in their metabolites profile and ultimately does not show the protection against colitis. Thus, we hypothesized that the predominance of earlylife anaerobic bacteria is required for the protective effects of colitis. To further support this hypoth esis, we cultured the microbiota (Fig. S8A) under aerobic, micro aerobic, or anaerobic conditions (referred to as Aero, Mi cro Aero, and Anaero, respectively). AIMD mice were then colonized with the respective microbiota (Aero, MicroAero, and Anaero), and we treated the mice with 3% DSS for 7 days. The mice colonized with either aerobic or microaerobic microbiota developed severe and comparable intestinal in flammation in response to DSS exposure. In contrast, when compared with AIMD mice, mice colonized with anaerobic microbiota were protected against DSSinduced colitis, as in dicated by reduced body weight loss (Fig. S8B), decreased clinical score (Fig. S8C and D), and counteraction of colon shortening ( Fig. S8E and F). Moreover, histological examina tion of colon morphology revealed a marked decrease in the extent of ulceration and crypt damage in response to anaer obic microbiota colonization (Fig. S8G).

Microbiota-derived plasmalogens suppress intestinal inflammation and ferroptosis
The above findings led us to hypothesize that enriched plasma logens in the colonic microbiota environment are associated with decreased colitis susceptibility post earlylife FMT. We next sought to address whether plasmalogen conferred beneficial effects on intestine homeostasis in early life. Thus, we treated a cohort of wildtype mice at 3 weeks of age with plasmenyleth anolamine (PlsEtn, one of the ether lipids) through the oral route. PlsEtntreated or vehicletreated mice were fed with 3% DSS for 7 days (Fig. S9A). Interestingly, after DSS removal, PlsEtn treatment opposed body weight loss induced by DSS (Fig. S9B). Besides, PlsEtn treatment delayed colitis progression relative to vehicletreated mice, as indicated by counteraction of colon shortening (Fig. S9C and D) and substantial decreases in the clinical score ( Fig. S9E and F). Further, in line with these data, histological examination of colon morphology revealed reductions in the extent of ulceration, crypt damage, and sub mucosal infiltration of inflammatory cells in response to PlsEtn administration (Fig. S9G), indicating that the presence of PlsEtn in early life help to suppress colitis progression in mice. Notably, PlsEtn treatment not only suppressed colitis progression in young mice at 3 weeks of age but also in adult mice at 20 weeks of age (Fig. S10A to D).
Next, we evaluated the effect of PlsEtn on colitis and ferrop tosis in metronidazoleinduced anaerobic bacteriaeliminated mice at 3 weeks of age. Mice were firstly treated with metro nidazole for 1 week, followed by administration of PlsEtn and DSS for 7 days through the oral route (Fig. S11A). Earlylife anaerobic microbiota dysbiosis resulted in the aggravation of experimental colitis. However, PlsEtn supplementation ren dered significant resistance against body weight loss (Fig. S9B), which suggested us that PlsEtn may participate in the regula tion of anaerobic bacteria on colitis. Compared to the DSS and metronidazoletreated mice, it extended the survival of mice (Fig. S11C) and opposed the shortening of colon length ( Fig.  S11F and G) and delayed colitis progression induced by DSS treatment, as indicated by the attenuation of the overall patho logical score ( Fig. S11D and E). In addition, PlsEtn treatment significantly decreases the mRNA expression of inflammatory cytokines (Tnf-α, IL-6, and IL1β) and chemokines (Cxcl1, Cxcl1, Ccl2, and Ccl3), except S100a8, as compared with PlsEtn untreated mice (Fig. S11H). Hematoxylin and eosin (H&E) staining of colon morphology revealed a marked decrease in the extent of epithelial disruption and crypt damage (Fig. S11I) in the colon of PlsEtntreated mice. Supporting the above data, we observed that PlsEtn treatment promoted M2 macrophage polarization in the colon (Fig. S11J and K), as indicated by staining for the M1 macrophage marker, iNos, and M2 macro phages marker, CD163. The above results demonstrated that PlsEtn treatment attenuated intestinal inflammation and pro moted M1/M2 macrophages rebalance.
Notably, FAR1TMEM189 mediated endogenous poly unsaturated fatty acid (PUFA)plasmalogens synthesized in peroxisomes and endoplasmic reticulum could act as substrates for lipid peroxidation and then promote ferroptosis [14]. However, the effect of exogenous plasmalogens on ferroptosis has not been investigated. Of note, in colitis aggravated by early life microbial dysbiosis, we found that the protein levels of colonic GPX4 and COX2, wellaccepted markers of ferroptosis, show significant differences after DSS treatment (Fig. 6A to D). Meanwhile, PlsEtn treatment can significantly increase GPX4 and decrease COX2 (Fig. 6A to D). Plasmalogens synthesized in humans often contain PUFA at the sn2 position of the glyc erol moiety. In particular, iron oxidation of PUFAs and pro duction of lipid hydroperoxides initiates ferroptosis. To test the role of the alkenylether group at position sn1 in ferroptosis, lysoplasmenylethanolamine (lysoPlsEtn) with a vinyl ether link age was prepared from PlsEtn via methanolysis (Fig. 6). Indeed, lysoPlsEtn significantly alleviated arachidonic acidinduced HT29 cell damage (Fig. S12). These decreases of cell viability and cytotoxicity were partly abolished in cells exposed to the GPX4 inhibitor, ML210. Collectively, these results demonstrate that GPX4 inhibition is necessary for plasmalogens to protect against cell damage. We then performed a followup experi ment in AIMD mice switching the PlsEtn treatment to lysoPlsEtn treatment, which revealed that lysoPlsEtn treatment resulted in reduced body weight loss (Fig. 6F), extended survival (Fig.  6G), opposed shortening of colon length (Fig. 6H and I), and colonic inflammation (Fig. 6J). Consistent with the above, lysoPlsEtn treatment suppressed ferroptosis, as indicated by increased GPX4 and decreased COX2 levels. Together, these results suggest that the presence of the alkenylether group is critical to suppressing colitis and ferroptosis.

Discussion
Mounting evidence suggests that colitisassociated microbiota shifts were strongly involved in the development of colitis, thereby regulating gut homeostasis [15][16][17]. In this study, we focused mainly on bacteriaproducing ether lipids since these bacteria show specific characteristics for their anaerobic biosynthesis pathways' possible involvement in the regulation of earlylife gut microbiota alteration on colitis. In terms of the gut microbial structure alteration, the development of colitis increased the aerobic bacteria population but reduced the anaerobic bacteria population [18]. Of note, in addition to a reduction in the anaerobic bacteria population, the development of colitis could be induced by high levels of ROS [19]. Given that ether lipids (including plasmalogens) have been proposed to act as ROS scavenger molecules and endogenous antioxidants to govern ferroptosis [8,20,21], we sought to investigate the link between alteration of the gut microbial structure, ferroptosis, and the development of colitis, especially under the different stages of life. Although there is a known association between intestinal inflammation and gut microbiota structure [22,23], the potential mediators of this relationship remain unclear. Here, we pre sented the characterization of ether lipidsproducing anaerobic bacteria in patients or mice with IBDs in early life and observed that the perturbed etherlinked lipid signaling driven by anaerobic bacterial were sufficient for the increased colitis and ferroptosis susceptibility.
Shifts in gut microbiota composition affect various systems, including the development of metabolic, neural, and immune pathways and intestinal homeostasis in early life [24]. Until now, evidence of their direct or indirect impact on intestinal functions has been investigated by other researchers, especially the shift of gut microbiota early in life [25,26]. Previous studies assessing the influence of microbiota depletion or intestinal microbiota modification medicated by antibiotics on colitis in mice yielded conflicting results [27]. Antibiotic therapy, including ciprofloxacin, neomycin, or metronidazole, have been proven to play a preventive role in colitis [28,29]. Similarly, the absence of microbiota observed in germfree mice and antibiotictreated mice at 18 weeks of age reduces colonic inflammation, although it impairs barrier function [30]. However, antibiotics exacer bated colitis in interleukin 10knockout mice at 8 weeks of age [31], and germfree mice are susceptible to colitis [30,32,33]. Shifts in gut microbiota composition induced by antibiotics also could influence the effect of diet, drugs, or probiotics for colitis prevention [34,35]. These contradictory observations suggested that the ageassociated alteration of gut microbiota composition might be involved in the pathogenesis of gut inflammation. However, concerning the association between the ageassociated alteration of gut microbiota composition and the pathogenesis of colitis, we observed that earlylife microbiota depletion via antibiotic cocktail treatment exacer bated DSSinduced colitis, while midlife microbiota depletion showed a reduced clinical symptom of colitis upon DSS chal lenge. Consistent with the observed development of colitis in mice with reduced microbiota at 3 weeks of age by us, earlylife microbiota perturbation from days 5 to 10 of life of the pups also showed an increased risk of colitis later in life, which was reported by Blaser et al. [9]. Normally, one possible explanation is that earlylife antibioticinduced perturbation of the micro biome community is considered to decrease the colonic epi thelial cell cytoprotective properties of specific bacteria and its specialized metabolites, i.e., shortchain fatty acids or second ary bile acids [36,37]. Although other factors between young and mature adults differ, the different responses to perturbation of microbiome community in young and mature adult mice when treated with DSS indicate a crucial role of the perturbed earlylife microbiota in the development of colitis. Our FMT study also showed that restitution of earlylife microbiota con fers the protection to DSSinduced colitis, although no worsened disease phenotypes were observed post midlife microbiota transplantation, compared with AIMD mice. Similarly, a recent study also described that restitution of a keystone microbial strain missing in the earlylife antibioticinduced gut dysbiosis results in reduced risk for colitis [38]. Here, we also observed that earlylife microbiota dysbiosis renders mice susceptible to ferroptosis. Ferroptosis has recently emerged as one of the causes of IEC death in ulcerative colitis [11]. Indeed, our results indicated that earlylife microbiota depletion could exacerbate ferroptosis, while colonization of earlylife microbiota confers the suppression of ferroptosis.
The specific bacteria and their metabolites may be involved in the antiinflammatory effect of earlylife microbiota [39][40][41][42]. Here, we aimed to identify critical microbes required for the earlylife gut microbiome that reduce colitis risk and drives ferroptosis susceptibility. Recently, the genes from microbiomes that contained the pls operon (encodes PlsA and PlsR proteins) were classified as a species as plasmalogenpositive [7], and distribution of the pls operon helped us to assess the distribu tion of plasmalogen biosynthesis genes. We demonstrated that the young mice exhibit a higher abundance of plasmalogen positive species in the colon contents than mature adult mice via 16S rRNA sequencing. Because of the limitation of the analysis, we did not accurately quantitate overall plasmalogen molecules in the colon contents of mice. However, by converting plasmalogens to DMA in the presence of acid/methanol, the abundance of DMA derivatives to represent plasmalogens could be analyzed. We found that a higher abundance of plasmalogen positive species and DMA derivatives in the colonic microbiota community of young mice (3 weeks old) indicated that the plasmalogenproducing microbiota was enriched in the early life of individual development. The above results highlight the relevance of the earlylife plasmalogenpositive species and their associated metabolites in intestinal inflammatory condi tions. Plasmalogenpositive species from gut microbiota were not only decreased in the midlife of mice but also reduced in the individuals with colitis (Fig. 4). In a cohort of ulcer ative colitis patients and DSSinduced colitis mice, we con firmed the reduction of colonic plasmalogenproducing bacteria abundance and the decrease of plasmalogen levels mirrored by the reduction in the levels of DMA derivatives in the colon contents of DSSinduced colitis mice. In particular, as indi cated by extremely lower fecal plasmalogen levels and ether linked phospholipids levels in microbiotadepleted mice, the presence of colonic etherlink lipids was partly microbiota dependent. Furthermore, the restitution of ether lipids (plasmalo gens) in early life via the oral route promotes suppression of colitis progression in mice at 3 weeks of age, indicating that colonic plasmalogenproducing bacteria may contribute to preventing colitis via its associated metabolite, plasmalogens. Plasmalogens, one kind of phospholipids enriched in the brain and other organs of mammals and in anaerobic bacteria, were thought to be involved in the antioxidant and antiinflammatory function of humans [43]. Besides, decreased plasmalogens seem to be one of the risk factors for inflammatory diseases. Until now, plasmalogen supplementation mainly showed promising health benefits in neurodegenerative diseases and metabolic disorders [44]. Interestingly, we also suggested that plasmalo gen treatment exerts anticolitis efficacy in the wildtype mice (Figs. S9 and S10).
Although the role of plasmalogens in inflammatory diseases was reported a long time ago, until now, the function of plas malogens on intestinal inflammation remained elusive. Recently, plasmalogens were identified from some anaerobic bacteria, including Bifidobacterium longum and Clostridium butyricum [45,46]. Bifidobacterium and Clostridium butyricum have been considered probiotics to prevent colitis [47]. However, the bio logical functions of plasmalogenpositive bacteria to human health remained unknown. In the present study, metronidazole treatment not only resulted in depletion of obligate anaerobic bacteria but also eliminated the presence of plasmalogens as indicated by the entire reduction of DMA levels, which underscored that anaerobic bacterium are the mainly microbial origin for the colonic plasmalogens in mice. Several studies have shown that antianaerobic antibiotics pretreatment, including metronidazole and clindamycin, exert intestinal antiinflammatory effects and alleviate chemically induced colitis and Citrobacter rodentium induced colitis in mice [48][49][50][51][52]. Clinical trials further supported its ability to prevent colitis with modest therapeutic effects [39,[53][54][55][56]. However, the treatment of clindamycin, one of the antibiotics exceptionally efficient against anaerobic bacteria, promoted the susceptibility to Campylobacter jejuniinduced colitis. Importantly, in agreement with other previous studies [9,37], our data supported that depletion of anaerobic bacteria in early life by metronidazole leads to further exacerbated DSS induced colitis in mice, comparable with those achieved in AIMD mice. Besides, anaerobic microbiota isolated from young mice at 3 weeks of age suppressed intestinal inflamma tion caused by DSS, further demonstrating the critical role of earlylife anaerobic microbiota in the attenuation of colitis pro gression. To test whether early life anaerobic microbiota medi ates the regulation of intestinal inflammation via plasmalogens, we evaluated the preventive effects of plasmalogens on the progression of colitis in metronidazoleinduced anaerobic bacteriaeliminated mice at 3 weeks of age. Similar to coloni zation with anaerobic microbiota isolated from young mice at 3 weeks of age, plasmalogen treatment suppressed clinical symptom of colitis caused by earlylife anaerobic bacteria elimi nation and showed the protective effects of plasmalogens on the progression of colitis. In terms of the impact of plasmalo gens on colitis progression, as expected, plasmalogens attenuate inflammatory responses and revert the enhanced colitis suscep tibility of mice lacking anaerobic bacteria.
Analysis of previous reports revealed that microbiota mediated metabolism could contribute to lipid peroxidation and ferroptosis. Ferroptosis is a novel cell death modality triggered by membrane lipid peroxidation from PUFAcontaining phospholipids [8]. PUFAcontaining phospholipids are sus ceptible to peroxidation in oxygen or ironrich cellular envi ronments [8]. Cui et al. and Zou et al. [14,57] further showed that etherlinked PUFAcontaining phospholipids induce ferroptosis and switch to a ferroptosissensitive state. Moreover, it should be noted here that plasmalogens synthesized in humans often contain PUFA at the sn2 position of the glycerol moiety. However, no PUFA was observed in identified plasmalogen positive bacterial, such as Bifidobacterium longum, Clostridium difficile, Clostridium innocuum, Bifidobacterium animalis, and Clostridium beijerinckii [7,43,[58][59][60]. The presence of ether lipids in gut microbiota has been associated with its insensitiv ity to H 2 O 2 and confers bacterial survival in intestinal aerobic environments [60,61]. Further evaluation of ether lipids in colitis and ferroptosis will help us to understand the roles of this unique lipid and its linked biological function. Given that PUFAplasmalogens synthesized in mammalian cells are char acterized by the presence of the PUFA chain at position sn2 and the alkenylether group at position sn1, although the role of etherlinked lipids in defending against ferroptosis is well established, how microbiotamediated etherlinked lipids con tribute to lipid peroxidation and ferroptosis remains to be investigated. Through data from murine colitis models, we here identified ether lipids as gut microbiotaderived lipids with the potent ability to govern ferroptosis susceptibility. Specifically, ether lipids markedly promoted ferroptosis resistance, and the presence of the alkenylether group is critical to suppressing ferroptosis in colitis. Support our observations, plasmalogens produced by TMEM189 degrade FAR1 and resulted in ferrop tosis suppression in OVCAR8 cells, and plasmalogens exhibited an antiferroptosis role in Caenorhabditis elegans and zebrafish [62,63]. Interestingly, the presence of PUFA in PUFAether lipids (C18C20:4 PlsEtn) could gain sensitivity to ferroptosis in OVCAR8 cells [14] but not duo to the presence of the alkenyl ether group, which suggested the ether linkage in PUFAether lipids does not appear to be important to their function on ferroptosis. Therefore, the role of ether lipids based on the lipid structure, different tissue, and genetic background on ferroptosis came with some discrepancy, which requires further explorations.
Although the role of plasmalogen in inflammatory diseases has long been reported, the function of plasmalogen on intes tinal inflammation has not been well defined. Consistent with our findings, dietary ethanolamine plasmalogen could alleviate DSSinduced colitis by enhancing colon mucosa integrity, anti oxidative stress, and antiinflammatory responses [64]. Besides, plasmalogen modulation has been utilized in both preclinical and clinical studies to prevent onset and/or attenuate progres sion of neurodegenerative diseases and atherosclerosis [65]. However, clinical studies of plasmalogen, especially microbiota derived ether lipids, on intestinal disease have not been reported. In the present study, we recognized that plasmalogenproducing anaerobic bacteria and their derived ether lipids were involved in the response of earlylife microbiota dysbiosis to colitis, and we suggested that ether lipids signaling derived by earlylife anaerobic bacteria was sufficient to govern susceptibility to colitis and ferroptosis. Although anaerobic bacteria are not all considered probiotics, even some are pathologic bacteria causing bacterial colitis, we suggested that anaerobic bacteriamediated lipid metabolism plays a critical in the regulation of colitis and ferroptosis. Of note, gut microbiota derived ether lipids blocked ferroptosis triggered by gut microbiota dysbiosis and suppressed colitis in mice. Overall, although the mechanisms remain to be elucidated, many lines of evidence pointed out that earlylife gut microbiota governs susceptibility to ferroptosis and colitis via ether lipids and suggested the modulation of plasmalogen positive microbiota as likely targets for intestinal health and treatment of intestinal diseases.

Ethics declarations
The animal study was approved by the Ethical Committees of Jiangnan University under the specific agreement number JN.No2020930c1151121 and JN.No20201230c0200401. Human feces were obtained from patients with IBDs or healthy indi viduals of First Affiliated Hospital (FAHNMU). Patients or the public were not involved in the design, conduct, reporting, or dissemination plans of our research. It was approved by the Research Ethics Commissions of FAHNMU (2021SRFA375) and followed the tenets of the Declaration of Helsinki. Informed consent was obtained from all study participants.

Human IBD subjects
Feces obtained from patients with IBDs (n = 19) or healthy individuals (n = 12) were used in this study. Patients with active ulcerative colitis according to accepted clinical and endoscopic criteria were recruited, and the study population included male and female randomized participants of 10 to 30 years of age. Accepted clinical and endoscopic criteria included a Mayo score of ≥5 and ≤9, an endoscopic subscore of at least 2, and a rectal bleeding subscore of at least 1.

Animals and diets
Male C57BL/6J mice aged 3 and 20 weeks (Weitonglihua, Beijing, China) were used in compliance with the ARRIVE guidelines. Mice were raised in an SPF environment with free access to water and food (AIM93G). Animals were submitted to standard 12h light/dark cycles. Plasmalogens (PlsEtn) were extracted ac cording to our previous method [44]. Lysoplasmenylethanolamine were prepared via mild alkaline methanolysis according to the previous method of Hanahan et al. [66].

DSS model of colitis and bacterial culture
Colitis was induced by DSS (Meiluobio, China). Animal allo cation to treatment groups was randomized. For induction of acute experimental colitis, mice received 7 days of 3% DSS dissolved in drinking water and followed by 3 days of regular drinking water. Prior to the acute DSSinduced colitis, in a cohort of 3weekold and 20weekold mice, a cocktail of 4 antibiotics (ampicillin [25 mg/kg], vancomycin [12.5 mg/kg], metronidazole [25 mg/kg], and neomycin (25 mg/kg]) or sterile saline was gavaged to mice at 200 μl per mouse once every day for 7 days to deplete their gut microbiota. After that, AIMD mice received 7 days of 3% DSS or regular drinking water and were gavaged by plasmenylethanolamine or lysoplasmenyleth anolamine (100 mg/kg body weight) or sterile saline.
Another cohort of 3weekold mice received metronidazole (25 mg/kg) was gavaged to mice at 200 μl per mouse once every day for 7 days to deplete their anaerobic bacteria. After that, metronidazoleinduced anaerobic bacteriadepleted mice received 7 days of 3% DSS or regular drinking water. The stool of 3weekold and 20weekold mice was collected as conven tionalized microbiota of young and adult mice, respectively. Freshly collected stools were homogenized in phosphatebuffered saline (0.05% cysteine HCl) at a ratio of 5 fecal pellets/ml. For FMT, the fecal slurry from 3weekold and 20weekold mice was delivered to the AIMD mice (YoungFMT and AdultFMT) by oral gavage (200 μl) every second day. Given the coprophagic nature of rodents, these mice were singlehoused to exclude potential confounding impacts including the cage effect. During this time, the mice received 7 days of 3% DSS or regular drinking water. Using brain hart infusion or Luria broth agar plates, freshly collected stools from 3weekold mice were cul tured under aerobic, microaerobic, or anaerobic conditions. AIMD mice were orally gavaged with the microbiota cultured above (1 × 10 8 CFU, every second day). These mice also received 3% DSS or regular drinking water. Mice were weighed daily and visually inspected for diarrhea (stool consistency) and rectal bleeding. Among them, rectal bleeding was scored as 0, 2, and 4 (normal, slight bleeding, and gross bleeding, respectively). Diarrhea was scored as 0, 2, and 4 (normal, loose stools, and watery diarrhea, respectively).

Analysis of lipids
Collected colon contents of mice were freezedried, and whole lipids were extracted as the methods of Sugawa et al. [67]. The identification of DMA and FAME was analyzed by gas chromatography-mass spectrometry analysis [7].

Bacterial community analysis
Total DNA was extracted from feces, and then bacterial com munity analysis was performed as previously done by GENEWIZ, Inc. (Suzhou, China) [68]. As for plasmalogenpositive micro biota analysis, plasmalogenpositive microbiota species were provided as the results of Jackson et al. [7].

Real-time polymerase chain reaction
Total RNA was extracted and reverse transcribed using TRIzol reagent and cDNA Synthesis SuperMix (Yeasen Biotech, China). Realtime polymerase chain reaction was carried out in Applied Biosystems QuantStudio 3 with Hieff qPCR SYBR Green master mix (Yeasen Biotech, China) and genespecific primers.

Statistical analysis
All results are expressed as means ± SEM. Computations assumed that all groups were samples from populations with the same scatter. Statistical significance was determined by Student t test or 2way analysis of variance. The probability of P value < 0.05 was indicated a significant difference. formed the experiments. C.X., J.X., and Y.L. designed experi ments, and Y.L. Y.D., and Z.Y. analyzed data. P.L. and R.D. conducted blinded histology scoring, imaging, and analysis of colonic tissue. W.Z. collected the fecal samples. C.J., T.Z., Y.L., Y.Z., and Y.X. helped with data interpretation and discussion. Y.L. wrote the paper. Competing interests: The authors declare that they have no competing interests.

Data Availability
The data are available from the author (Yanjun Liu) upon rea sonable request.

Supplementary Materials
Fig. S1. Culturedependent and cultureindependent analysis confirmed that the fecal microbiota or anaerobic bacteria were depleted in AIMD mice or metronidazoleinduced anaerobic bacteriaeliminated mice. Fig. S2. Ferroptosis was triggered in DSSinduced colitis.