Reg4 and complement factor D prevent the overgrowth of E. coli in the mouse gut

The expansion of Enterobacteriaceae, such as E. coli is a main characteristic of gut inflammation and is related to multiple human diseases. However, how to control these E. coli overgrowth is not well understood. Here, we demonstrate that gut complement factor D (CFD) plays an important role in eliminating E. coli. Increased E. coli, which could stimulate inflammatory macrophages to induce colitis, were found in the gut of CFD deficient mice. We also showed that gut Reg4, which is expressed in gut epithelial cells, stimulated complement-mediated attack complexes to eliminate E. coli. Reg4 deficient mice also had increased E. coli. The dominant E. coli were isolated from colitis tissues of mice and found to be sensitive to both CFD- and Reg4-mediated attack complexes. Thus, gut Reg4- and CFD-mediated membrane attack complexes may maintain gut homeostasis by killing inflammatory E. coli.

T he expansion of facultative anaerobic Enterobacteriaceae (phylum Proteobacteria) with a high level of commensal Escherichia coli (E. coli) is a common characteristic during inflammation in the colon [1][2][3][4] . These increased E. coli levels are related to the occurrence and development of multiple diseases such as obesity, diabetes, cardiaovascular and liver diseases, and tumor [5][6][7][8][9] . Interactions between the microbiota and pathogenic bacteria 10 , microcin-mediated competition among Enterobacteriaceae in the inflamed gut 11 , microbiota-activated PPAR-γ signaling 12 , dietary zinc 13 and tungstate-mediated microbiota editing 14 may affect the expansion of these bacteria. However, it is incompletely clear how to control these increased E. coli.
The membrane attack complex (MAC), a hetero-oligomeric protein assembly that kills pathogens by perforating their cellular envelopes, has an essential role in human immunoprotection against gram-negative bacteria such as E. coli [15][16][17] . The formation of the MAC depends on the sequential assembly of the soluble complement proteins C5b, C6, C7, C8, and C9, and is mainly triggered by immune complexes or pattern recognition molecules (PRMs) upon recognition of non-self or altered self-cells, such as by C1q, collectins, ficolins, and properdin 18 . The MACs are also activated by many other factors such as pentraxins, c-reactive protein, serum-amyloid P component, and pentraxin 3 19 . Complement factor D (CFD), a specific serine protease that cleaves its unique substrate to generate the C3 convertases C3(H2O)Bb and C3bBb, may amplify all complement-mediated bactericidal effects, and play a critical role in mannose-binding lectin-mediated MACs [20][21][22] . However, many pathogenic organisms also find ways to escape complement attach through a range of different mechanisms such as Pic, extracellular serine protease P, the Adr2 protein and the C3b-fH complex 18,23 . Whether increased E. coli in colitis can be eliminated by MACs, and how to stimulate the formation of MACs in gut tissues remain elusive.
To prevent overgrowth of the commensal population and prohibit pathogens from colonizing and causing infection, the host must maintain tight control over gut homeostasis. Gut epithelial cells produce and release a variety of biomolecules such as defensins, lectins, mucins, and secretory immunoglobulin A into the mucosa and lumen to contribute to immunity 24 . In this study, we found that Reg4 (REG4 in humans), a member of the C-lectin family, which is expressed by gut Paneth cells or deep crypt secretory cells 25 , may stimulate the complement system to kill inflammatory E. coli. We also demonstrated that gut epithelialderived CFD plays a critical role in eliminating gut E. coli. Thus, both gut Reg4 and CFD may induce bactericidal activity against E. coli to maintain gut homeostasis.

Results
CFD fl/fl pvillin-cre T mice are sensitive to DSS-mediated colitis through the accumulation of inflammatory macrophages in gut tissues. CFD was detected in Paneth cells in the small intestine and Paneth cell-like cells in the colon of mice (Fig. 1a, b), consistent with our and other data 26,27 . We also found that the gut microbiota may regulate the expression of CFD in gut epithelial cells. The levels of CFD in the small intestine and colon in germ-free (GF) mice were markedly lower than those in wild-type (wt) mice ( Supplementary  Fig. 1a-d). The colonization of mouse feces, E. coli or lactobacilli in GF mice could promote the expression of CFD ( Supplementary  Fig. 1e, f). Since CFD may be involved in killing gram-negative bacteria such as E. coli, which are related to colitis [15][16][17] , CFD expression in gut epithelial cells may play a role in the occurrence and development of colitis. To investigate this, we generated gut CFD conditional knockout mice (CFD fl/fl pvillin-cre T mice) along with wt littermate control CFD fl/fl pvillin-cre w mice according to the described strategy. In CFD fl/fl pvillin-cre T mice, complement factor D (CFD, Gene ID: 5068) was selectively knocked out in mouse gut epithelial cells (Fig. 1a, b). We next used dextran sulfate sodium (DSS)-mediated model to determine sensitivity of CFD fl/fl pvillincre T mice to colitis. We found that CFD fl/fl pvillin-cre T mice were more sensitive to 2.5% DSS induced colitis than their control littermate wild-type CFD fl/fl pvillin-cre w mice, and exhibited lower survival rate, more weight loss, higher disease indexes, more colonic shortening and higher histological score ( Fig. 1c-g). Higher levels of the inflammatory cytokines TNFα, IL6 and ILβ were also detected in the colonic tissues of CFD fl/fl pvillin-cre T mice. Gut tissue inflammatory macrophages play a critical role in occurrence and development of colitis [28][29][30][31] . These cells may be characterized as CD45 + CD11b + CD103 − CD64 + MHCII + ly6C + cells 17,18 . Increased inflammatory macrophages were observed in both CFD fl/fl pvillincre T mice and especially DSS-treated CFD fl/fl pvillin-cre T mice ( Fig. 1i-j), suggesting that the sensitivity of CFD fl/fl pvillin-cre T mice to DSS-mediated colitis may depend on inflammatory macrophages. Taken together, CFD fl/fl pvillin-cre T mice are sensitive to DSSmediated colitis through the markedly accumulation of inflammatory macrophages in gut tissues.
E. coli levels are markedly increased in CFD fl/fl pvillin-cre T mice. Since CFD may promote bactericidal activity (Fig. 2a), we next analyzed the composition of the gut microbiota. Sequencing analyses of 16S ribosomal RNA (V3-V4 variable region) in colonic contents did not show marked changes in the composition such as the proportion of Proteobacteria under normal physiological conditions in CFD fl/fl pvillin-cre T mice and cohoused control CFD fl/fl pvillin-cre w mice (Fig. 2b, and Supplementary  Fig. 3a). However, after DSS treatment, which may cause relative increase in the luminal abundance of Enterobacteriaceae (phylum Proteobacteria) ( Supplementary Fig. 3b) 5 , there were markedly increased levels of gut Enterobacteriaceae, mainly E. coli in the colonic contents of CFD fl/fl pvillin-cre T mice as compared to those of CFD fl/fl pvillin-cre w mice (Fig. 2c, d, and Supplementary  Fig. 3c), suggesting that CFD deficiency in gut tissues causes reduced killing on E. coli (Escherichia-Shigella). The increased E. coli were confirmed by flow cytometry, qRT-PCR and fluorescence hybridization (FISH) (Fig. 2e-g). q-PCR also showed that only the proportion of E. coli but not level of total bacteria changed in DSS-treated CFD fl/fl pvillin-cre T mice (Fig. 2f). Thus, CFD deficiency not only changes the composition of the gut microbiota but also promotes the abundance of E. coli in colitis.
E. coli promote sensitivity to DSS-mediated colitis and the accumulation of inflammatory macrophages in CFD fl/fl pvillincre T mice. We next determined whether the increased E. coli promoted the sensitivity of CFD fl/fl pvillin-cre T mice to DSSmediated colitis. We identified the dominant E. coli in DSStreated mice, named E. coli 0160 32 . After equal amount of E. coli 0160 were infused into mice, CFD fl/fl pvillin-cre T mice had more severe colitis than CFD fl/fl pvillin-cre W mice after the administration of 2% DSS, including lower survival rate, more weight loss, higher disease indexes and more colonic shortening (Fig. 3a-d).
Histological H/E staining and the expression of inflammatory cytokines also verified the worsened inflammation in CFD fl/fl pvillin-cre T mice after the infusion of E. coli (Fig. 3e, f). Since E. coli potentially promoted colitis, we selected a low concentration (2% DSS). Taken together, CFD fl/fl pvillin-cre T mice are more sensitive to DSS-mediated colitis after the infusion of E. coli. Notably, more accumulated inflammatory macrophages were observed in the colon tissues of CFD fl/fl pvillin-cre T mice (Fig. 3g, h) after the infusion of E. coli. Gram-negative pathogenic bacteria may directly activate not only inflammatory macrophages but also gut epithelial cells to produce IL-18 and induce the macrophages through IFNγ producing cells 33 . ELISA and immunostaining showed markedly increased IL-18 in gut epithelial cells in DSS-treated CFD fl/fl pvillin-cre T mice (Supplementary Fig. 4a, b). Increased levels of mature IL-18 were also present in the colonic tissues of CFD fl/fl pvillin-cre T mice that were treated ex vivo in E. coli stimulation ( Supplementary Fig. 4c-e). Grambacteria such as Salmonella may activate macrophages and gut epithelial cells through NLRC4, caspase11 etc [34][35][36][37] . We found that NLRC4, caspase-1/11, caspase 8 and PKCδ in gut epithelial cells of CFD fl/fl pvillin-cre T mice were involved in the expression of IL-18, which was induced by the administration of E. coli 0160 ( Supplementary Fig. 4c-e). Since the generation of inflammatory macrophages in gut tissues is dependent on IFNγ 33 , we also analyzed the IFNγ-producing cells. A drastic expansion of interferon γ (IFNγ)-producing CD4 + T helper cells (CD4 + IFNγ + Th1 cells) was observed in E. coli infused CFD fl/fl pvillin-cre T mice (Fig. 3h). NK cells, another IFNγ producing cell type, also accumulated in gut tissues of E. coli infused CFD fl/fl pvillincre T mice (Fig. 3h). Thus, our data indicate that E. coli promote not only colitis but also the accumulation of inflammatory macrophages in the gut tissues of CFD fl/fl pvillin-cre T mice.
CFD fl/fl pvillin-cre T mice have reduced bactericidal abilities against E. coli. We next examined whether CFD fl/fl pvillincre T mice had reduced bactericidal abilities against E. coli. We first generated GFP-labeled E. coli 0160. When equal amounts of GFP-labeled E. coli were infused into CFD fl/fl pvillin-cre T and CFD fl/fl pvillin-cre W mice, the fluorescence intensities in the small intestine, colon and cecum were higher in CFD fl/fl pvillin-cre T mice than in CFD fl/fl pvillin-cre W mice (Fig. 4a). The number of E. coli in the colon tissues and feces of CFD fl/fl pvillin-cre T mice was much higher than that in CFD fl/fl pvillin-cre W mice (Fig. 4b). These results suggest that CFD fl/fl pvillin-cre T mice have reduced bactericidal abilities against E. coli. To further determine whether the increase in E. coli was from CFD deficiency in CFD fl/fl pvillin-cre T mice, we generated CFD fl/fl pvillin-cre T and CFD fl/fl pvillin-cre W germ-free (GF) mice. When equal amounts of GFP-labeled E. coli were infused into CFD fl/fl pvillin-cre T and CFD fl/fl pvillin-cre W GF mice, the fluorescence intensity and number of E. coli in CFD fl/fl pvillin-cre T GF mice were also higher than those in CFD fl/fl pvillin-cre W GF mice (Fig. 4c, d). Flow cytometry also showed fewer dead bacteria in CFD-KO mice (Fig. 4e). Since we mainly examined the killing effect of CFD on E. coli in the gut cavity, these mice were not treated with DSS.
Notably, when equal amounts of GFP-labeled E. coli were infused into mice, there was reduced C3b deposition on E. coli in CFD fl/fl pvillin-cre T mice as compared to CFD fl/fl pvillin-cre W mice (Fig. 4f). Metabolic factors were also different in colonic tissues from CFD fl/fl pvillin-cre T and CFD fl/fl pvillin-cre W mice such as that C1, which is involved in the typical pathway; complement factor B (BF), which serves as the catalytic subunit of complement C3 convertase in the alternative pathway (AP); and C3 or the C5bC9 complex, which participate in all three pathway (Fig. 4g). More FB, and less C3b and C5bC9 were found in colonic tissues of CFD fl/fl pvillin-cre T mice than in CFD fl/fl pvillin-cre W mice; whereas C1q was not significantly changed in any analyzed samples from both types of mice (Fig. 4h). Reduced 5b-9 and C3b, and increased FB were also measured in the feces of CFD fl/fl pvillin-cre T mice (Fig. 4i). Thus, these findings suggest that there is decreased atypical complement-mediated bactericidal activity in CFD fl/fl pvillin-cre T mice.
Gut Reg4 initiates the membrane attack complex against E. coli. We next investigated how the complement system killed E. coli in gut tissues. To do this, we first measured the sensitivity of E. coli to normal serum, C3-, C1-and CFD-deficient sera. We found that C3 deficiency markedly reduced E. coli killing as compared to that of C1 deficient serum (Fig. 5a). CFD-deficient serum also exhibited reduced killing of E. coli (Fig. 5b). Thus, E. coli isolated from colitis tissues may be sensitive to an unknown C3-dependent complement-mediated killing pathway. Lectin may induce the C3-dependent complement pathway through binding with the mannose in LPS such as the "O" antigen of mannose 38 . Since E. coli 0160 is "O" antigen positive 32 , we performed lipopolysaccharide (LPS) and mannan blocking experiments. We found that LPS from E. coli (0111B4) containing "mannose" 39 and mannan but not peptidoglycan could inhibit normal serum-mediated killing on E. coli 0160 (Fig. 5c). We next examined which kind of lectin was involved in the complement-mediating killing of E. coli. There are several mannose-binding lectins such as mannose-binding lectin (MBL), ficolins (ficolin-1 or M-ficolin, ficolin-2 or L-ficolin, and ficolin-3 or H-ficolin), and collectin-11, which potentially activate complement pathways 40,41 , but Reg 4 has a typical lectin fold, which is produced by gut epithelial cells, and may potentially bind with mannose in LPS in gut tissues 42 . We generated recombinant mouse Reg4 and human REG4 proteins, and examined their effects on E. coli killing ( Supplementary Fig. 5). Both Reg4 and REG4 may promote killing by normal serum but not CFDdeficient sera; Furthermore this killing was dose-dependent ( Fig. 5d-f). However, mutant hREGIV-P91S, which has comparably reduced mannan-binding abilities 42 , only slightly promoted to complement-mediated E. coli killing, suggesting that Reg4 in gut tissues participates in E. coli killing (Fig. 5d). IgA, a major of defense factor, was also involved in the lectin-mediated activation of complement, consistent with other studies [43][44][45] (Fig. 5g). To further determine the effects of Reg4 and IgA, we measured the deposition of these components on E. coli in vivo. When equal amounts of GFP-labeled E.coli were infused into 2% DSS-treated CFD fl/fl pvillin-cre T or CFD fl/fl pvillin-cre w mice, which were treated with pan-antibiotics to eliminate the effects of other bacteria on E. coli, we found that Reg4 and IgA bound to E. coli in colon tissues (Fig. 6a). These E. coli also bound to MBLassociated serine protease (MASP)1/3 (Fig. 6a), which triggers complement activation through serine proteases 46,47 . However, E. coli did not bind with WASP2 48 . Flow cytometry also showed the binding of E. coli with Reg4 and IgA, but not collectin, which is also involved in mediating complement activation 41 (Fig. 6b). a qRT-PCR (upper) of CFD in colon tissues and immunoblotting (lower) of CFD in equal amount of colon contents of CFD fl/fl pvillin-cre W (WT) and CFD fl/fl pvillin-cre T mice (CFDKO). Total protein was stained using Coomassie. Number, different individuals. b Immunostaining of CFD in ileum (upper) and colon (lower) of CFD fl/fl pvillin-cre W mice (WT) and CFD fl/fl pvillin-cre T mice (CFDKO). Red, CFD; Green, lysozyme. Scale bar, 40 μM. c, d and e The survival rate (C), body weight (D) and disease index (E) in CFD fl/fl pvillin-cre T mice (CFDKO) and CFD fl/fl pvillin-cre W (WT) mice (male, n = 15) after DSS (2.5%) treatment. f The length of colon in CFD fl/fl pvillin-cre T mice (CFD KO) and CFD fl/fl pvillin-cre W mice (WT) after DSS (2.5%) treatment (n = 6). g H&E staining and histology score of colon tissues of CFD fl/fl pvillin-cre T mice (CFDKO) and CFD fl/fl pvillin-cre W mice (WT) after DSS (2.5%) treatment. For histological score, three slides/mouse, n = 6. Scale bar, 40 μM. h QRT-PCR of inflammatory cytokines TNFα, IL-6 and IL-1β in the colonic tissues of CFD fl/fl pvillin-cre T (CFDKO) and CFD fl/fl pvillin-cre W mice (WT) after DSS (2.5%) treatment (n = 6). i Flow cytometry of CD45 + CD11b + CD64 + CD103 -MHC + Ly6C + , F4/80 + TNFα + cells in colon tissues in normal CFD fl/fl pvillin-cre T (CFD KO) and CFD fl/fl pvillin-cre w (WT) mice (n = 5). j Flow cytometry of CD45 + CD11b + CD64 + CD103 -MHC + Ly6C + , F4/80 + TNFα + cells in the colon tissues of DSS-treated CFD fl/fl pvillin-cre T (CFDKO) and CFD fl/fl pvillin-cre w (WT) mice. Kruskal-Wallis test in c; Analysis of variance test in d and e; Two side student's t test in f and h-j; Mann-Whitney U test in g; *P < 0.05; **P < 0.01; ***P < 0.001; NS no significance. Data in c-e are a representative of three independent experiments. R.E relative expression.
Finally, we used ELISA to examine the binding of Reg4 with LPS, we found that Reg4 but not control BSA (bovine serum albumin) could bind with LPS; Furthermore this binding was dosedependent ( Fig. 6c). We also confirmed the binding of LPS with IgA 44 (Fig. 6c). Reg4 or IgA caused the deposition of C3 and the combination of two factors caused more deposition of C3 than either factor alone (Fig. 6d, e). These data indicate that Reg4 initiate MAC activation to kill E. coli. Taken together, these data suggest that Reg4 may be involved in the initiation of complement-mediated MACs killing of E. coli. COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-01219-2 ARTICLE COMMUNICATIONS BIOLOGY | (2020) 3:483 | https://doi.org/10.1038/s42003-020-01219-2 | www.nature.com/commsbio Reg4 deficient mice have reduced complement bactericidal abilities against E. coli. To further determine the role of gut Reg4 in complement-mediated killing on E. coli, we generated Reg4 deficient mice according to described strategy ( Supplementary  Fig. 6a-c). Since E. coli may accumulate in DSS-mediated colitis 5 , we analyzed and compared the proportion of E. coli in Reg4 deficient and their cohoused control wt mice after 2% DSStreatment. 16S rRNA analyses demonstrated that there were increased proteobacteria in Reg4 deficient mice as compared to the cohoused control wt mice (Fig. 7a, b, and Supplementary  Fig. 6d). Equal amount of GFP-labeled E. coli 0160 were administered to pan-antibiotics-treated mice, and the fluorescence intensities in the small intestine, colon and cecum were higher in Reg4 deficient mice than control mice, suggesting reduced bactericidal abilities in these mice (Fig. 7c). There was increased E. coli in feces collected at different times (Fig. 7d). Flow cytometry also showed fewer dead bacteria in Reg4 deficient mice than control mice (Fig. 7e). Flow cytometry also exhibited less deposition of C3b on the bacteria in Reg4 deficient mice than control mice (Fig. 7f). The data also further confirmed tht binding of IgA and Reg4 but not collectin to GFP-labeled E. coli (Fig. 7f). Reduced C3b and C5bc9, and increased FB in the colon tissues and feces of Reg4 deficient mice was also observed as compared to those of the control wt mice ( Supplementary Fig. 7a, b). Finally, the activity of CFD in Reg4 deficient mice was not different from that in wt mice ( Supplementary Fig. 8). Taken together, these data suggest that Reg4 deficient mice exhibit reduced complementmediated killing of E. coli.

Discussion
Inflammatory diseases of the gastrointestinal tract are frequently associated with changes in gut microbial communities that include the expansion of facultative anaerobic bacteria (Phylum Proteobacteria, family Enterobacteriaceae, mainly E. coli),which is a common characteristic of dysbiosis [2][3][4]8,49 . These bacteria are related to gut inflammation, and are more importantly associated with other human diseases such as obesity, tumor and liver diseases. Here, we showed that complement-mediated MAC plays an important role in eliminating gut E. coli. We found that gutderived Reg4 and/or IgA may initiate the formation of MACs to kill E. coli. Thus, our data demonstrate that both gut-derived Reg 4 and CFD are involved in maintaining gut homeostasis. These findings pave the way for the development of therapeutic strategies for gut inflammation-associated diseases.
The complement system, a proteolytic cascade, can be activated via the classic pathway (CP), the lectin pathway (LP), and the alternative pathway (AP). Complement factor D (CFD) may amplify all complement-mediated bactericidal abilities, and play an important role in mannose-binding lectin-mediated MACs that target bacteria [20][21][22] . CFD may be derived from multiple kinds of cells, especially those in adipose tissues. Here, we demonstrated that gut epithelial cell-derived CFD plays a critical role in eliminating increased E. coli in colitis.
We found that Reg4 initiated complement-mediated killing of E. coli. The Lectin family is composed of carbohydrate-binding proteins that contribute to mucosal innate immunity. Several lectins such as mannose-binding lectin (MBL) and ficolin1, 2 and 3 are pattern recognition molecules that are involved in the initiation of complement activation 40 . These pathways are involved in the protecting against E. coli infection 50 . Other lectins such as REG3α in humans and its murine homolog REG3γ are also found throughout small intestinal epithelium 51 . The interactions between REG3α and bacterial targets may kill bacteria through peptidoglycan. Reg 4 may bind to polysaccharides, mannose, and heparin in the absence of calcium 42 . We demonstrate that Reg 4 may bind with E. coli LPS to initiate the formation of MACs. Additionally, Reg4 may also bind mannose at two calcium-independent sites and directly induce damage to the bacterial cell wall 42,52 .
IgA may promote Reg4-mediated complement membrane attack complexes. Upon interaction of IgA with pathogens, IgA molecules can have diverse effector functions, including directly preventing the invasion of microorganisms, the interaction with the phagocytic IgA Fc receptor, and complement activation. Secretory IgA (sIgA) is the dominant humoral immunity at the intestinal mucosa. Due to its polymeric nature and multivalency, sIgA primarily protects mucosal surfaces by noncovalently crosslinking microorganisms or macromolecules. sIgA also plays a role in shaping and maintaining the gut microbiota from birth to adulthood 53 . Here we demonstrated that IgA and Reg4 may initiate the activation of complement pathways. Polymeric IgA has recently been suggested to activate the lectin pathway and contribute to complement activation 44 . Complement activation by IgA was previously shown to involve the alternative, but not the classical complement pathway 43,44 .
Gut mucosal immunity includes a complex network of different types of cells that work together to provide physical and chemical barriers against microbial invasion, and maintain gut homeostasis. Paneth cells can produce a variety of peptides and proteins including Reg4, CFD, α-defensins, lysozyme C, phospholipases and the C-type lectin Reg3α 54 . The degranulation and expression of these proteins occurs in response to stimuli including cholinergic agonists, bacteria, and bacterial products such as lipopolysaccharide (LPS) and lipoteichoic acid 55 . Our results also demonstrate that the expression of CFD may be regulated by gut bacteria. These small molecules may also be synthesized in vitro. Since these components may be regulated and synthesized in vitro, our findings have implications for improving strategies to eliminate E. coli.

Methods
All reagents and oligoes used in this study were listed in Supplementary Table 1.
Mice. Four-to six-week-old male or female C57BL/6 mice were obtained from Nanjing Animal Center. Caspase-1/Caspase 11−/−, and NLRC4−/− in B6 background from Prof. Meng in University of Chinese Academy of Sciences, shanghai and Prof. Shao in National Institute of Biological Sciences, Beijing were bred and kept under specific pathogen-free (SPF) condition in Nankai University. The proportion of gut bacteria after 16s rRNA analyses of colon contents in normal CFD fl/fl pvillin-cre w (WT) and CFD fl/fl pvillin-cre T (CFDKO) mice (pooled samples from six mice). Also see Fig.  S3A. c The proportion of gut bacteria after 16s rRNA analyses in CFD fl/fl pvillin-cre w (WT) and CFD fl/fl pvillin-cre T (CFDKO) mice after 2% DSS. Also see Fig. S3B. d The proportion of E. coli (Escherichia-Shigella) in CFD fl/fl pvillin-cre w (WT) and CFD fl/fl pvillin-cre T (CFDKO) mice after 2% DSS (n = 5). Also see Fig. S3C. e Flow cytometry of feces bacteria after staining using anti-LPS antibodies in CFD fl/fl pvillin-cre w (WT) and CFD fl/fl pvillin-cre T (CDKO) mice after 2% DSS. f QPCR of colonic content (upper) and colonic tissues (lower) in CFD fl/fl pvillin-cre w (WT) and CFD fl/fl pvillin-cre T (CFDKO) mice after 2% DSS (n = 6). g FISH of E. coli in the gut tissues of CFD fl/fl pvillin-cre w (WT) and CFD fl/fl pvillin-cre T (CFDKO) mice after 2% DSS. One representative from six mice. Scale bar, 40 μM; Scale bar in bottom panel, 5 μM. Two side student's t test in e and f; *P < 0.05; **P < 0.01; ***P < 0.001; NS no significance. Reg4 KO mice were generated by CYAGEN, China according to described strategy in Supplementary Fig. 6a. Reg4 KO mice were identified using following All experimental litters were bred and maintained under specific pathogen-free conditions in Animal Center of Nankai University. Experiments were carried out using age-and gender-matched mice or cohoused CFD fl/fl pvillin-cre w mice. All procedures were conducted according to the Institutional Animal Care and Use Committee of the Model Animal Research Center. Animal experiments were approved by the Institute's Animal Ethics Committee of Nankai University. All experimental variables such as husbandry, parental genotypes and environmental influences were carefully controlled.
Mouse models. For dextran sodium sulfate (DSS) induced colitis, DSS induced colitis was performed according to our previously reported method 56 with modification. Briefly, mice received 2.5% DSS (40,000 kDa; MP Biomedicals) or at the indicated dose in their drinking water for 7 days, then switched to regular drinking water. The amount of DSS water drank per animal was recorded and no differences in intake between strains were observed. For survival studies, mice were followed for 12 days post start of DSS-treatment. Mice were weighed every other day for the determination of percent weight change. This was calculated as: % weight change = (weight at day X-day 0/weight at day 0) × 100. Animals were also monitored clinically for rectal bleeding, diarrhea, and general signs of morbidity, including hunched posture and failure to groom. For microbiota transplantation, germ-free (GF) mice were orally administered 200 μl 1 × 10 9 bacteria (once/week). In wt mice, mice were first treated with pan-antibiotics (ampicillin (A, 1 g/L, Sigma), vancomycine (V, 0.5 g/L, Sigma), neomycin sulfate (N, 1 g/L, Sigma), and metronidazole (M, 1 g/L, Sigma)) via the drinking water for one week (sometime longer than one week), which could eliminate gut bacteria, and then orally administered 200 μl of 1 × 10 9 bacteria (once/week). To confirm the elimination of bacteria, stools were collected from antibiotic-treated and untreated mice and cultured in anaerobic and aerobic condition. Mice were sacrificed at the indicated days. Representative colon tissues were embedded in paraffin for hematoxylin/ eosin (H&E) staining or embedded in OCT compound (Tissue-Tek, Sakura, Torrance, CA) and frozen over liquid nitrogen for immuno-staining.
Gut microbiota analyses. For gut microbiota analyses, previously reported method was used 58 . Gut microbiota was analyzed by Majorbio Biotechnology Company (Shanghai, China) using primers that target to the V3-V4 regions of 16S rRNA. Once PCR for each sample was completed, the amplicons were purified using the QIAquick PCR purification kit (Qiagen Valencia, CA), quantified, normalized, and then pooled in preparation for emulsion PCR followed by sequencing using Titanium chemistry (Roche, Basel Switzerland) according to the manufacturer's protocol. Operational Taxonomic Unit (OTU) analysis was performed as follows: sequences were processed (trimmed) using the Mothur software and subsequently clustered at 97% sequence identity using cd-hit to generate OTUs. The OTU memberships of the sequences were used to construct a sample-OTU count matrix. The samples were clustered at genus and OTU levels using the sample-genus and sample-OTU count matrices respectively. For each clustering, Morisita-Horn dissimilarity was used to compute a sample distance matrix from the initial count matrix, and the distance matrix was subsequently used to generate a hierarchical clustering using Ward's minimum variance method. The Wilcoxon Rank Sum test was used to identify OTUs that had differential abundance in the different sample groups.
For the absolute numbers of gut bacteria, 16s rRNAs were extracted using bacterial DNA extraction kit (CWBio), including that nuclear lysis buffer split cell nuclear, protein precipitation solution precipitates and removes protein, and purified genomic DNA is precipitated by isopropanol; and then amplified using phylum such as Firmicutes, Bacteroidetes, genus or strain-specific primers. All qPCR assays were performed on CFX Connect system (BIO-RAD) in 25 μl reaction volume containing 1×SYBR Green PCR Mastermix. For each primer, a concentration of 1μM was added. The following thermal program was applied: a single cycle of DNA polymerase activation for 5 min at 95°C followed by 40 amplification cycles of 15-s denaturing step (95°C), 30-s annealing (55°C) and 30-s extension step (72°C). Afterward, melting temperature analyses of the obtained amplification products was performed using standard machine settings. The concentration of each product was detected and then exchanged into copy numbers. Primers used were listed in Supplementary Table 1.
For compositional analysis of the intestinal tract and extra intestinal tissues, previously reported method was used 6 . Briefly, whole tissues were harvested and homogenized them using a Polytron PT2100 homogenizer at 17000 rpm (Kinematica) in sterile thioglycolate media. We then serially diluted the homogenate and plated on LB agar and Schaedler agar supplemented with 5% defibrinated sheep blood plates and grew at 37°C aerobically and anaerobically for 24 h. We counted bacterial colonies and then separated based on colony appearance and performed colony 16S rDNA PCR and sequencing.
For colony PCR, cecal contents from mice were diluted and plated onto LB agar. The colonies from each animal were picked, resuspended in sterile water, and heated to 100°C for 15 min to lyse bacteria. PCR was performed and PCR products were resolved on a 1% agarose gel.
Cell isolation and flow cytometry. For the staining of lamina propria (LP) lymphocytes, previously reported method 59 was used. Briefly, colon or small intestine were isolated, cleaned by shaking in ice-cold PBS four times before tissue was cut into 1 cm pieces. The epithelial cells were removed by incubating the tissue in HBSS with 2 mM EDTA for 30 min at 37°C with shaking. The LP cells were isolated by incubating the tissues in digestion buffer (DMEM, 5% fetal bovine serum, 1 mg/ ml Collagenase IV (Sigma-Aldrich) and DNase I (Sigma-Aldrich) for 40 min. The digested tissues were then filtered through a 40-mm filter. Cells were resuspended in 10 ml of the 40% fraction of a 40: 80 Percoll gradient and overlaid on 5 ml of the 80% fraction in a 15 ml Falcon tube. Percoll gradient separation was performed by centrifugation for 20 min at 1800 rpm at room temperature. LP cells were collected at the interphase of the Percoll gradient, washed and resuspended in medium, and then stained and analyzed by flow cytometry. Dead cells were eliminated through 7-AAD staining.
For analyses of C3b-, Reg4-, IgA-or collectin-coating of E. coli in colonic contents, fresh colonic contents was resuspended in sterile PBS. An aliquot estimated to contain no more that GFP-labeled 10 6 E. coli was directly stained with a monoclonal body. After washing, secondary antibodies were added. After a final washing step, samples were analyzed on the BD flow cytometry with setting adapted for optimal detection of bacterial-sized particle.
in phosphate-buffered saline (PBS). Optical density of the solutions was measured at 600 nm using a spectrophotometer (Thermo Scientific, Waltham, MA, USA) and adjusted to an optical density of 0.700. 5 × 10 5 or 1 × 10 6 CFU was added into PBS that contains 10% Normal Human Complement Standard (Quidel), C1q-depleted serum (Quidel) or C3 depleted and CFD depleted serum (Quidel). Mixtures were incubated at 37°C and sampled at 0, 1, 2, 3 and 4 h. Bacterial suspensions were diluted 1:100 in PBS and plated on LB agar plates. Plates were incubated overnight at 37°C. Resultant bacterial CFU were counted, and CFU per mL of bacterial suspension were calculated. All serum sensitivity assays were repeated at least three times. Blocking experiments. For blocking experiments, LPS (100 mM) or mannan (100 mM) were added into PBS that contains 10% Normal Human Complement Standard (Quidel); Recombinant mouse Reg4 (rReg4, 100 ng/ml), human recombinant REG4 (100 ng/ml), mutant rREG4 (100 ng/ml) and IgA (100 ng/ml) were also respectively added into PBS that containing 10% Normal Human Complement Standard (Quidel). The tubes were further incubated at 37°C in a CO 2 incubator using a rotator (multitube tube with a fixed speed; Thermo Fisher) for 1-4 h, at which equal amount of liquids were removed and plated onto chocolate agar to quantify surviving CFUs per milliliter. To ascertain input CFUs per milliliter, aliquots were also removed at time 0 from control tubes containing buffer alone or the subjects' plasma that had been heated to 56°C for 30 min to inactivate complement.
For C1q, BP, C5b-9 and C3 assays, tissues and colonic content samples were detected for C1q, BP, C5b-9 and C3 by ELISA (Quidel Corporation, San Diego, CA, USA) respectively and according to the manufacturer's instruction.
Generation of mouse Reg4, human REG4 and mutant human REG4. To generated the mouse Reg4 and human REG4, the DNA sequence encoding mouse Reg4 and human REG4 with an N-terminal His-tag was cloned into the pEASY®-Blunt E2 Expression vector (Transgen) and transformed into E.coli BL21. The site-directed mutant human REG4 Δ P 91S was prepared by use of the Fast MultiSite Mutagenesis System (Transgen), and the coding regions were confirmed by DNA sequencing. The protein samples were expressed after stimulation with 1 mM IPTG in LB medium, and then purified using ProteinIso® Ni-NTA Resin (Transgen).
H & E staining. For hematoxylin/eosin (H&E) staining, previously reported methods were used in this experiment 60 . Briefly, the entire colon was excised to measure the length of the colon and then were fixed in 4% (w/v) paraformaldehyde buffered saline and embedded in paraffin, 5 µm sections colon sections were cut and stained with H&E.
Immunostaining. Immunostaining was performed according to reported method 58 . Briefly, colon tissues were embedded in OCT compound (Tissue-Tek, Torrance, CA) and frozen over liquid nitrogen. 5-μm-thick sections were prepared from frozen tissue and fixed in acetone (−20°C) for 10 min. After rehydration in PBS for 5 min and further washing in PBS, tissue sections were blocked with 1% (w/v) BSA and 0.2% (w/v) milk powder in PBS (PBS-BB). The primary antibody was added in PBS-BB and incubated overnight at 4°C. After washing (three times, 5 min each), tissue was detected with DAB kit or fluorescence labeled second antibody. Nuclei were stained by DAPI.
Fluorescent in situ hybridization. For fluorescent in situ hybridization (FISH), mucus immune-staining was paired with fluorescent in situ hybridization (FISH) in order to analyze bacteria localization at the surface of the intestinal mucosa according to reported method 58 . In brief, the ileum and colonic tissues (proximal colon, second centimeters from the caecum) containing fecal material were placed in methanol-Carnoy's fixative solution (60% methanol, 30% chloroform, 10% glacial acetic acid) for a minimum of 3 h at room temperature. Tissue were then washed in methanol, ethanol, ethanol/xylene (1:1) and xylene, followed by embedding in paraffin with a vertical orientation. 5-μm sections were cut and dewaxed by preheating at 60°C for 10 min, followed by bathing in xylene at 60°C for 10 min, xylene at room temperature for 10 min and 99.5% ethanol for 10 min. The hybridization step was performed at 50°C overnight with an probe diluted to a final concentration of 0.01 μg/mL in hybridization buffer (20 mM Tris-HCl, pH 7.4, 0.9 M NaCl, 0.1% SDS, 20% formamide). After washing for 10 min in wash buffer (20 mM Tris-HCl, pH 7.4, 0.9 M NaCl) and 10 min in PBS and block solution (5% FBS in PBS) was added for 30 min at 50°C. Mucin 2 primary antibody (rabbit H-300, Santa) was diluted to 1: 200 in block solution and applied overnight at 4°C. After washing in PBS, block solution containing anti-rabbit secondary antibody diluted to 1:200 was applied to the section for 2 h. Nuclei were stained using Hoechst33342.
Immunoblot. Immunoprecipitation and immunoblot were performed according to previous methods 26 . The cells were lysed with cell lysis buffer (Cell Signaling Technology), which was supplemented with a protease inhibitor 'cocktail' (Calbiochem). The protein concentrations of the extracts were measured using a bicinchoninic acid assay (Pierce). For the immunoblot, hybridizations with primary antibodies were conducted for 1 h at room temperature in blocking buffer. The protein-antibody complexes were detected using peroxidase-conjugated secondary antibodies (Boehringer Mannheim) and enhanced chemiluminescence (Amersham).
RT-PCR and qRT-PCR. RT-PCR and qRT-PCR were performed according to our previous methods 58 . Total RNA was extracted from the cells, tissues and organs using TRIzol reagent (Invitrogen). First-strand cDNA was generated from total RNA using oligo-dT primers and reverse transcriptase (Invitrogen Corp). The PCR products were visualized on 1.0% (wt/vol) agarose gels. Quantitative real-time PCR (qRT-PCR) was conducted using QuantiTect SYBR Green PCR Master Mix (Qiagen) and specific primers in an ABI Prism 7000 analyzer (Applied Biosystems). GAPDH mRNA expression was detected in each experimental sample as an endogenous control. All reactions were run in triplicate. The primers used in this study were listed in Supplementary Data 1 of the Methods.
Ex vivo stimulation. For ex vivo colon stimulation, colon from healthy mice were harvested, washed and incubated with or without 1 × 10 9 E. coli in DMEM media with ATP (2 mM) for 1 h, For analyses of IL-18, the colon epithelial cells were separated from colon tissues using 0.1% EDTA, followed by three 1 min shakings by hand, a 15-min incubation at 4°C, and passage through 70-μm filters (BD Falcon) to collect the flow through. Fraction containing intact and isolated crypts were collected by centrifugation at 75 × g for 5 min. at 4°C and washed with PBS. The lamina propria was separated from crypts to enrich for mononuclear and intestinal epithelial cells, respectively. Protein extracts were analyzed by immunoblotting for mature forms of IL-18. The supernatants were collected for IL-18 ELISA.
For PKCδ inhibition experiment, PKCδ inhibitors (20 μM) were added into culture, and then colon epithelial cells were separated at the indicated time. The expression of IL-18 was analyzed using immunoblotting and ELISA.
Statistics and reproducibility. Student's t test and ONE-way ANOVA Bonferroni's Multiple Comparison Test was used to determine significance. The statistical significance of the survival curves was estimated using Kaplan and Meier method, and the curves were compared using the generalized Wilcoxon's test. Histological scores in different groups were analyzed by a Mann-Whitney U test. A 95% confidence interval was considered significant and was defined as p < 0.05. * indicates p < 0.05, **p < 0.01, ***p < 0.001. Fig. 6 Gut reg4/IgA initiate membrane attack complex on E.coli. a Immunostaining of C3b, C5-9, Reg4, IgA, MASP1/3 and MASP2 on GFP-labeled E. coli in 2% DSS-treated GFP-labeled E. coli infused CFD fl/fl pvillin-cre T (CFDKO/0160) mice (Left) and statistic analyses (right, three images/slide, three slides/ mouse, n = 6). Scale bar, 40 μM. No, number; Iso, isotypic antibod. b Flow cytometry of Reg4, IgA, MASP1/3 and MASP2 on the GFP-labeled E. coli in untreated GFP-labeled E. coli infused CFD fl/fl pvillin-cre w (WT) and CFD fl/fl pvillin-cre T (CFDKO) mice and statistic analyses (n = 3). c ELISA for detecting binding of Reg4 with LPS (left), Reg4 with IgA (middle) or IgA with LPS (right) in different concentration of Reg4 (left and middle) or IgA (right) coated plates d ELISA of C3 in different concentration of Reg4 (left) or IgA (right) with LPS coated plates. For complement resources, 5% normal human sera were added. e ELISA of C3b in the LPS, Reg4, IgA, LPS + Reg4 (LPS/Reg4), LPS + IgA (LPS/IgA), Reg4+IgA (Reg4/IgA) and LPS + Reg4+IgA (LPS/Reg4/IgA) coated plates. For complement resources, 5% normal human sera were added. Mann-Whitney U test in a; Two side student's t test in b; ANOVA plus post-Bonferroni analysis in e; Analysis of variance test in c and d. *P < 0.05; **P < 0.01; ***P < 0.001; NS no significance. In c-e, one representative of three independent experiments. The proportion of bacteria after 16s rRNA analyses of colon contents in wt and Reg4 deficient mice after 2% DSS (n = 5). Also, see Fig. S6B. c Fluorescence intensity in small intestine (SI), colon and caecum after infusing wt and Reg4 KO using GFP-labeled E. coli for 48 h (n = 6). d E.coli clones in the feces of WT and Reg4 KO after infusing GFP-labeled E. coli at the indicated time (n = 6). e Flow cytometry of live (GFP + PI−)/dead(GFP + PI+) bacteria. The equal amount of GFP-labeled E.coli (E. coli 0160, 1 × 10 7 ) were infused into different mice, including CTR2 (WT) and REG4 KO mice. After 48 h, the feces from different mice were analyzed using flow cytometry for live (GFP + PI−)/dead(GFP + PI+) bacteria. f Flow cytometry of Reg4, IgA, C3b and collectin deposited E. coli and statistic analyses (n = 3). Two side student's t test in c, e and f; Analysis of variance test in d. *P < 0.05; **P < 0.01; ***P < 0.001; NS no significance.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
16S rRNA gene sequence data include:16S rRNA gene sequence data of colon feces derived from WT and CFDKO mice. PRJNA577385; 2) 16S rRNA gene sequence data of colon feces derived from DSS-treated WT and normal WT mice. PRJNA577387; 16S rRNA gene sequence data of colon feces derived from WT and CFDKO mice after DSS treatment. PRJNA577388; 16S rRNA gene sequence data of colon feces derived from wild type and Reg4 KO mice after DSS treatment. PRJNA577541. Full blots are shown in Supplementary Information. Source data underlying plots shown in figures are provided in Supplementary Data 2. All other data (if any) are available upon reasonable request.