Fungal lysozyme leverages the gut microbiota to curb DSS-induced colitis

ABSTRACT Colitis is characterized by colonic inflammation and impaired gut health. Both features aggravate obesity and insulin resistance. Host defense peptides (HDPs) are key regulators of gut homeostasis and generally malfunctioning in above-mentioned conditions. We aimed here to improve bowel function in diet-induced obesity and chemically induced colitis through daily oral administration of lysozyme, a well-characterized HDP, derived from Acremonium alcalophilum. C57BL6/J mice were fed either low-fat reference diet or HFD ± daily gavage of lysozyme for 12 weeks, followed by metabolic assessment and evaluation of colonic microbiota encroachment. To further evaluate the efficacy of intestinal inflammation, we next supplemented chow-fed BALB/c mice with lysozyme during Dextran Sulfate Sodium (DSS)-induced colitis in either conventional or microbiota-depleted mice. We assessed longitudinal microbiome alterations by 16S amplicon sequencing in both models. Lysozyme dose-dependently alleviated intestinal inflammation in DSS-challenged mice and further protected against HFD-induced microbiota encroachment and fasting hyperinsulinemia. Observed improvements of intestinal health relied on a complex gut flora, with the observation that microbiota depletion abrogated lysozyme’s capacity to mitigate DSS-induced colitis. Akkermansia muciniphila associated with impaired gut health in both models, a trajectory that was mitigated by lysozyme administration. In agreement with this notion, PICRUSt2 analysis revealed specific pathways consistently affected by lysozyme administration, independent of vivarium, disease model and mouse strain. Taking together, lysozyme leveraged the gut microbiota to curb DSS-induced inflammation, alleviated HFD-induced gastrointestinal disturbances and lowered fasting insulin levels in obese mice. Collectively, these data present A. alcalophilum-derived lysozyme as a promising candidate to enhance gut health.


Introduction
Host-microbe mutualism has emerged as a potent regulator of both gastro-intestinal (GI) and extraintestinal diseases, including inflammatory bowel disease (IBD), obesity, and insulin resistance (IR). Overnutrition and high fat diet (HFD) consumption alter gut microbiota composition and reduce intestinal barrier function, thereby promoting lowgrade inflammation. [1][2][3][4][5][6] Gut bacteria are speciesdependently affecting IR 7-9 thus advocating a causal link between aberrant microbiota and altered metabolic health. Bacterial remnants are, as a result of intestinal barrier dysfunction, further capable of translocating to circulation, where lipopolysaccharides (LPS) from Gram-negative bacteria exacerbate IR. 10 Host defense peptides (HDPs) patrol the intestinal barrier to keep intestinal microbes in check, hence mitigating bacterial translocation. 11 These peptides are generally downregulated in human obesity and Type 2 Diabetes (T2D), 12,13 potentially enabling microbiota encroachment and translocation to extra-intestinal tissues. 14 Microbiota encroachment, i.e. a reduced distance from bacteria to the intestinal epithelial cells, correlates with impaired glycemic control. 15 Recently, we and others reported bacterial translocation to key metabolic tissues in human obesity, where the site-specific community structures discriminated between weight-matched individuals with and without T2D. 16,17 These findings add to the accumulating evidence of bacterial involvement in the onset and prevention of glucoregulatory impairments.
Compared to obesity, gut health is further disrupted in IBD with clear genetic links to diminished HDP production. 18,19 Bacterial translocation is well established in patients with IBD 20,21 associating with reduced gut barrier function 22 and a change in gut microbiota composition. 23,24 Thus, obesity and IBD exhibit similar GI complications, although the links between the two phenotypes remain inadequately described. 25 When the gut microbiota was evaluated by 16S rRNA gene amplicons, no similarities in microbiota composition changes within patients with obesity and IBD were reported. 26 However, with more advanced network analysis, common regulation of specific enzymes involved in the phosphotransferase system or the nitrate reductase pathway has been identified between these patient groups 27 supported by a change in bacterial coabundances. 28 Similarly, frameshift mutations in the pattern recognition receptor, nucleotidebinding oligomerization domain-containing protein 2 (NOD2), predispose for human IBD, 29 while genetic ablation of the same protein promotes IR in HFD-fed mice. 30 Patients with IBD exhibit increased prevalence of IR and nonalcoholic fatty liver disease (NAFLD), thus supporting a close connection between the mentioned diseases. 31,32 Still, IR and IBD are rarely studied together, although both pathologies are increasing worldwide. 33,34 We have recently demonstrated beneficial traits of the two HDPs, human α defensin 5 (HD-5) and -β defensin 2 (hBD-2) on HFDinduced IR and experimental colitis, respectively. 35,36 Lysozyme, also known as muramidase, was the first identified HDP. 37 Lysozyme degrades peptidoglycans (PGN), thereby targeting both Gram negative andpositive bacteria, ultimately protecting the host from bacterial invasion. 11 Enhanced bacterial translocation in human obesity 16,17 may partly relate to decreased HDP secretion, 14 wherein particular fecal lysozyme abundance inversely correlates with hyperglycemia. 12 In humans, lysozyme is most abundantly produced by small intestinal Paneth cells. Paneth cell hyperplasia, where small intestinal Paneth cells 'leaks' to descending colon, is a well-described phenomenon following gut inflammation in IBD 38 enhancing colonic lysozyme activity. 39 Endogenous lysozyme therefore seems discordantly regulated and spatially separated in obesity 12 and IBD. 39 Curiously, Yu et al. recently reported that lysozyme-mediated degradation of hallmark IBDassociated microbes, such as Rumunicoccus gnavus, facilitated gut inflammation rather than protecting against disease burden, 40 hence presenting lysozyme as a potential doubleedged sword. 41 Conversely, hen egg white (HEW)-derived lysozyme has previously been reported to reduce chemically induced colitis in animal models. 42 Lysozyme is also produced by fungi, including Acremonium alcalophilum, with no observable bacteriolytic activity 43 and thus considered safe to use in, e.g., animal feed. 44,45 This fungal lysozyme belongs to the glycoside hydrolase family (GH) 25 and is thereby distinct from human and HEW lysozymes included in the GH22 family. 46 Recently, a non-purified feed additive of A. alcalophilum-derived lysozyme was reported to degrade bacterial PGN 46 to NOD2-activating muramyl dipeptides (MDP) reducing duodenal inflammation 43 and reduce the abundance of CD45-positive cells in the ileum of broiler chickens, 47 thus pointing toward enhanced barrier integrity.
Considering the above-described GI implications of A. alcalophilum-derived lysozyme, we here investigated the disease-preventing potential of a highly purified version of this lysozyme in both HFD-induced obesity-linked IR and DSS-provoked colitis. We report that purified A. alcalophilum-derived lysozyme protects against HFD-induced microbiota encroachment, mitigates fasting hyperinsulinemia, and further leverages the gut microbiota to alleviate DSSinduced gut inflammation. Pathway analysis of the gut microbiota revealed similar traits followed by lysozyme supplementation in both in vivo models, despite a vastly different microbiota community structure between the two mouse strains. These findings suggest that A. alcalophilum lysozyme may target shared GI manifestations in obesity and colitis.

Lysozyme alleviates diet-induced fasting hyperinsulinemia, increases mucus production and reduces microbiota encroachment
We first investigated the potential effect of lysozyme administration in diet-induced obese mice (Fig S1a). Mice fed HFD for 12 weeks gained more body weight compared to LFD-fed mice, with no measurable impact from daily lysozyme supplementation on body composition or energy intake ( Figure 1a, Fig  S2a-c). Lysozyme-receiving mice did, however, exhibit lower fasting insulin levels compared to vehicletreated counterparts (Figure 1b). Lysozyme further tended to reduce blood glucose post, but not prior, glucose challenge (two-way ANOVA RM = 0.10) compared to vehicle-treated HFD counterparts (Figure 1c-d). Still, glucose stimulated insulin concentration was unaffected by lysozyme supplementation (Fig S2d). Akin to diminished fasting insulin, known to modulate gut leakiness, 48 we observed a numerical protection from HFD-induced gut permeability in lysozyme-gavaged animals ( Figure 1e). Despite these in vivo indications of increased gut permeability in HFD-fed mice, circulating LPS, intestinal inflammation, and expression of selected tight junction genes in colon and ileum remained similar between groups (Fig S2e-i). Still, lysozyme normalized the HFD-induced reduction in Tlr2 expression in colon (Figure 1f). TLR2 activation has been linked to mucus regulation by colonic goblet cells, 49 prompting us to investigate mucus production. Lysozyme supplementation prevented HFD-induced colonic Muc2 downregulation ( Figure 1g). Moreover, such lysozyme-mediated preservation of mucus integrity was associated with a more than sixfold increase in the average distance from intestinal epithelial cells to the nearest bacteria, thus demonstrating that lysozyme supplementation sufficiently prevents microbiota encroachment induced by HFD consumption (Figure 1h-i).

Lysozyme progressively modifies gut microbiota composition and its predicted functions
As diminished encroachment might reflect bactericidal activity, we next assessed longitudinal changes in the fecal microbiota of experimental mice. Microbiota composition analysis revealed similar signatures at the beginning of the study, where all mice were fed a low-fat reference diet (Fig S3a-b). HFD-feeding induced a vastly different microbiota composition in both small intestine and fecal samples (Figure 2a-d).
Longitudinal sampling revealed instant changes to the fecal microbiota by dietary intervention, while the impact of lysozyme administration occurred gradually. Mice receiving daily doses of orally administered lysozyme thus exhibited a unique and progressive trajectory in their microbiota community structure compared to any other group (Figure 2c). Despite this progressive trajectory, several of the HFD-induced changes to the fecal, but not small intestinal, microbiota were countered by lysozyme supplementation (Figure 2a-f). The alpha diversity of the microbiota was not affected by lysozyme administration, pointing toward alterations in relative microbial abundances rather than an antimicrobial effect of this HDP (Fig S3c-d). In agreement with this notion, the relative fecal abundances of numerous bacterial ASVs were modified in lysozyme receiving mice compared to their vehicle-receiving counterparts. A majority of the regulated ASVs in fecal samples belonged to the Firmicutes phylum. These included a reduction of Akkermansia muciniphila and Oscillospira, while Ruminococcus gnavus, an unclassified species of the Ruminococcus genus, and Bifidobacterium pseudolongum and Allobaculum were increased in relative abundance in lysozyme receiving mice (Figure 2e). Four bacterial genera with modified ASV abundances showed a change in the entire genus followed by lysozyme supplementation (Figure 2f), while other ASV modifications appeared species dependent (Fig S3e). We next mapped the predicted functions of the bacterial community, revealing a significant shift in fecal but not small intestinal samples of mice receiving lysozyme (Figure 2g, Figure3f).

Lysozyme dose-dependently prevents DSS-induced colitis
To further investigate the beneficial outcome of lysozyme supplementation on intestinal health, we assessed colon damage in chemically induced mild colitis. Mice received one of the two doses of lysozyme or vehicle from 2 days before and during the 5 days of DSS-challenge (Fig S1b). Mice receiving 1.0, but not 0.04, mg lysozyme were protected against DSS-induced  6 h fasting insulin levels in week 10 of the study period. c) 6 h fasting blood glucose levels in week 10 of the study. d) Blood glucose levels during oral glucose tolerance test (oGTT) in week 10 with 2 µg glucose per g lean mass measured by MR scans. e) Fluorescence measured in plasma sampled at indicated minutes post oral sulfonic acid administration. f) Relative gene expression of Toll-like receptor (Tlr) 2 in ileum and colon tissue assessed by RT-qPCR. g) Relative Mucin (Muc) 2 gene expression in ileum and colon tissue assessed by RT-qPCR. h) Distances of the closest bacteria to intestinal epithelial cells (IEC) in proximal colon per condition over three high-powered fields per mouse, with each dot representing the average distance to bacteria per field of 2-3 representative mice per group. i) Representative slides of confocal microscopy analysis of microbiota localization; Muc2 (green), actin (purple), bacteria (red) and DNA (blue). Each slide measures 70 × 70 µm. a, d, e) Graphs depict group mean ± SEM. Repeated-measures two-way ANOVA with Geisser-Greenhouse correction and Dunnet's multiple comparisons test to HFD+Vehicle. b, c, f, g, h) Graphs depict group mean ± SEM and individual data points. Oneway ANOVA with Dunnet's multiple comparisons test comparing to HFD+Vehicle. a-h) * = p < .05, ** p < .01, *** p < .001. Grey asterisk indicates significant difference between LFD+Vehicle vs HFD+Vehicle and black indicates HFD+Vehicle vs HFD+Lyso. reduction in body weight (Figure 3a-b). Despite a lack of protection against diminished bodyweight upon the DSS-challenge, 0.04 mg lysozyme effectively curbed intestinal inflammation (Figure 3c-h). We observed a highly consistent trend toward dose-dependency across all measured readouts (Figure 3a-h). A trait that was replicated in an independent study-cohort using an additional third, intermediary dose (Fig S4ah). Importantly, 1.0 mg of lysozyme administered to non-DSS-challenged mice did not induce measurable changes in colonic damage or cytokine levels compared to vehicle-treated control mice, thus indicating that lysozyme alone did not affect immune homeostasis in healthy mice (Figure 3c-h, Fig S4c-h). Missing entries indicate unclassified family and/or genus. c) PCoA using weighted UniFrac distances of overall bacterial presence in fecal samples week 4, 8, and 12 after initiation of LFD/HFD feeding and Vehicle/Lysozyme administration, sampled approximately 24 hours since the latest lysozyme/vehicle administration. Centroids indicate group mean at the indicated week after study start. PERMANOVA test between LFD+Vehicle vs. HFD+Vehicle p < .01 at all sampled times. PERMANOVA test between HFD+Vehicle vs. HFD+Lyso p = .078 at week 4, p < .001 at week 8 and 12. d) Mean relative abundance in % of 17 most abundant aggregated bacterial genera in fecal samples in week 12 of the study. Missing entries indicate unclassified family and/or genus. e) ASVs with differential abundance (FDR adjusted p < .05) by DEseq2 analysis comparing fecal microbiota of HFD+Lysozyme to HFD+Vehicle at week 12 of the study. ASVs are categorized with their classified genus and colored by their classified phylum. Missing entries indicate unclassified genus. f) Relative abundance in % of genera found differentially abundant in e from baseline (0) over the study period (4)(5)(6)(7)(8)(9)(10)(11)(12) with similar baseline abundance between the groups and a tendency across the entire aggregated genus abundance. Graphs depict group mean and 75% CI. g) PCoA of predicted KEGG Orthologs in fecal samples from week 12 of the study using Bray-Curtis distance sampled approximately 24 hours since the latest lysozyme/vehicle administration. Centroids indicate group mean. PERMANOVA test between LFD+Vehicle vs HFD+Vehicle p = .67, HFD +Vehicle vs HFD+Lyso p = .006.

Lysozyme partially protects against DSS-induced microbiota changes and modifies its predicted functions
Healthy chow-fed mice receiving 1.0 mg of lysozyme for 2 days had similar changes in fecal microbiota composition from their baseline samples as vehicle-treated reference mice ( Figure 4a). We next assessed gut microbial composition in ileum, cecum and colon at the end of the study. Consistent with previous reports, we observed pronounced interlocation variations in community structures ( Figure 4b). 50 In agreement with the colonic target of DSS-induced colitis, DSS-challenge induced consistent changes in the distal intestines but only modest changes in the ileum (Figure 4b-c, Fig S4i-j).
Lysozyme partially protected against DSSinduced gut microbiota disturbances, further corroborated by a clear separation from vehicle receiving counterparts in the most distal intestinal locations, cecum and colon (Figure 4b-c, Fig  S4j). Analogous to our HFD-fed obesity model, we observed a pronounced increase in numerous ASVs from the mucus degrading genus Akkermansia in the colon of DSS-challenged control mice (Fig S4i). Lysozyme protected against this DSS-related trajectory with a general decrease in several ASVs classified as A. muciniphila and within the Prevotella and Oscillospira genera, while we observed an increase in R. gnavus, hence aligning the lysozyme mediated improvements in DSS challenged mice to the traits observed in HFD-fed mice (Figure 4c-d). The predicted functions of the gut microbiota were, similar to the microbiota composition, mostly affected by lysozyme in colon samples (Figure 4e, Figure4k). We then compared the lysozyme-modified pathways based on the KEGG orthologs across the HFDand DSS-study. To this end, we identified 21 pathways regulated in both studies of which 13 were consistently regulated by lysozyme across studies. These consistently affected pathways related to amino acid and nucleotide/nucleoside synthesis, alterations in carbohydrate metabolism, as well as three distinct pathways involving peptidoglycan synthesis and maturation (Table 1).

Lysozyme-mediated mitigation of DSS-induced colitis relies on a functional gut microbiota
We next sought to investigate the role played by the intestinal microbiota in the promotion of lysozyme-mediated protection against DSS-induced colitis. To this end, we subjected mice to a similar challenge and treatment regime as described above and included an additional study arm of microbiota depleted animals (Fig S1c), known to precipitate similar DSS-induced intestinal injury as conventional mice. 51 Moreover, the experiment was performed in a new vivarium and with mice from a different vendor in order to ensure reproducibility of lysozyme-mediated protection against DSSinduced colitis (DSS-induced colitis is notoriously difficult to phenocopy across cohorts 52 ). Albeit the DSS-challenged vehicle-treated mice in this study did not lose weight, the colitis phenotype mimicked our previous observations (Figure 5a-c). Moreover, daily lysozyme gavage conferred strong protection against DSS-induced colon shortening and spleen weight, as well as both macro-and microscopic colitis scores (Figure 5a-g). Importantly, while antibiotics themselves are not sufficient to inhibit lysozyme activity in vitro (Fig S4l), lysozyme-mediated protection was abolished in microbiota depleted mice (Figure 5h-n). These data importantly suggest that lysozyme impact on colitis is dependent on the gut microbiota.

Discussion
In the present study, we first investigated the preventive effects of lysozyme administration on HFDinduced intestinal changes. Here, lysozyme reduced fasting insulin levels, commonly used for clinical determination of IR, 53 despite not improving HFDinduced obesity. Diminished fasting insulin in lysozyme-gavaged mice is associated with improved barrier function and markedly diminished microbiota encroachment. These beneficial traits were accompanied by a progressive change in fecal and small intestinal microbiota composition where lysozyme administration specifically modified the predicted functions of the fecal microbiota. Encouraged by the observed benefits of lysozyme administration in the obesity setting primarily targeting the large intestine, we next evaluated the efficacy of a short-term DSS-challenged mouse model of mild colitis. Here, lysozyme is reproducibly and dose-dependently protected against DSSinduced colon damage and mitigated diseaseassociated microbiota changes in cecum and colon. The lysozyme-induced alterations in the predicted functions of the gut microbiota shared numerous traits between the models, despite differences in sex, mouse strain, vendor and experimental facility. We further reproduced the DSS-induced colitis model in a new vivarium, where groups of antibiotic-treated mice were also included. Depletion of the commensal microbiota by broadspectrum antibiotics abrogated lysozyme efficacy, pointing toward microbiota-dependency in the lysozyme mediated protection against DSSinduced colitis. Collectively, these observations corroborate consistent and biologically relevant adaptations in the fecal microbiota of lysozyme-gavaged mice.
Although rarely studied in parallel, HFDinduced intestinal deterioration and DSSinduced colitis share common features in etiology, including changes in gut microbiota composition and microbiota encroachment 15,54 ( Figure 6). In both disease models, the relative A. muciniphila abundances were negatively associated with gut health markers. A. muciniphila is a mucus-degrading species and generally considered a beneficial microbe able to protect against both HFD-induced hyperinsulinemia 55 and DSSinduced colitis. 56 Despite the widely accepted benefits of A. muciniphila, numerous studies also challenge this binary view reporting significant colitis-associated blooms in Akkermansia abundances. 52,57-60 The species has further been shown to exaggerate proton pump inhibitor (PPI)-induced small intestinal injury by diminishing the jejunal mucus layer, hence enhancing inflammatory burden 61 and promoting IL-1a secretion from colonic tissue. 62,63 Such reports are in agreement with observations from graft versus host diseases, similarly characterized by substantial gut disturbances. 64 In our study, we observed a robust induction of A. muciniphila in vehicle treated HFD-fed and DSS-challenged control mice, an increase that was fully prevented by lysozyme in both models. Corroborating a causal link between lysozyme administration and A. muciniphila regulation, Lyz1 knockout mice, incapable of mounting an endogenous lysozyme response, were recently reported to exhibit increased Akkermansia abundances. 40 Although bacterial (i.e. Akkermansia) induced mucus degradation might stimulate renewal,  group. a, c-e, h, j-l) Graphs depict mean ± SEM with individual data points. One-way ANOVA with Dunnet's multiple comparisons test to DSS+Vehicle group. a-f, h-m) * = p < .05, ** p < .01, *** p < .001. Grey asterisk indicates significant difference between Vehicle+Vehicle vs DSS+Vehicle and black indicates comparisons between DSS+Vehicle vs Lyso-supplemented groups in both conventional mice and mice receiving antibiotics.
hence improving barrier function, in mild low-grade inflammatory diseases, 65 the same traits might facilitate detrimental outcomes during more severe gut inflammation as observed in experimental colitis and immunocompromised individuals. 66 Still, it remains elusive if lysozyme administration inhibits further growth or actively degrades mucolytic bacteria, such as Akkermansia. Future studies are therefore warranted to elucidate if such degradation takes place and to what extent it would liberate bioactive compounds from Akkermansia to mitigate gut inflammation. 67 Intriguingly, a recent study elegantly demonstrated how endogenous lysozyme, when abundantly expressed in the colon of IBD patients, degraded colonic R. gnavus, thereby liberating its proinflammatory polysaccharide 68 exaggerating intestinal inflammation. 40 This novel finding presents endogenous lysozyme as a double-edged sword 41 that may be tweaked by inflammatory status and microbiota composition. To this end, lysozyme-mediated lysis of Lactococcus lactis facilitates delivery of colitis-attenuating superoxide dismutase to the inflamed colon. 69 Such descriptions add complexity to the otherwise well-described HDPmediated improvements in barrier function to protect against bacterial translocation. 11 Peculiarly, we observed increased R. gnavus abundances in both disease models following lysozyme administration. Our data on dose-dependent downregulation of intestinal inflammation are thus in sharp contrast to the recent report by Yu et al. 40 Future studies are thus warranted to elucidate if the mentioned discrepancy relates to differences in sources (endogenous versus A. alcalophilum-derived) and/or family (GH22 vs GH25 lysozymes), microbial community structure or inflammatory tone upon lysozyme administration.
In conclusion, A. alcalophilum-derived lysozyme administration reshaped the gut microbiota composition along with its predicted functions, diminished HFD-induced intestinal permeability and bacterial encroachment, as well as microbiota-dependently diminishing DSS-induced colitis. The beneficial impact across two different mouse strains, in both sexes, combined with the immune regulating properties reported in broiler chickens, 47,70 merits further investigation to assess its prospects of clinical translatability.

Animals
Male C57BL6/J mice (Jackson) or female BALB/c mice (Charles River) of 6 weeks of age were acclimatized for 2 weeks on either compositionaldefined low-fat diet (D12450H, Research Diets) or standard, fiber-rich, chow diet (RM1P diet, Special Diet Services (First DSS studies), or LabDiet 5021 (DSS-study including antibiotics)), respectively. All mice were co-housed with free access to water and 12 hours light-dark cycle at room temperature.
C57BL6/J mice were then switched to a compositional defined HFD (D12451, Research Diets) or kept on the matched LFD for the duration of 12 weeks while receiving daily oral gavage 4 h ± 30 min into the light cycle of 100 µL PBS (Gibco) with or without 0.5 mg lysozyme from A. alcalophilum grown in the production organism Trichoderma reesei, as previously described 44 and further purified 46 to a purity of >95%; endotoxins <0.004 EU/mL. Body weights were monitored weekly and feed intake measured and exchanged thrice a week during the study period, while body composition by magnetic resonance (MR) scans (Minispec LF90, Bruker) and fecal sampling was assessed every 4 weeks. After 12 weeks, the mice were anesthetized with isoflurane (Fresenius Kabi) after 6 h fasting from 1 h into the light cycle with their usual daily gavage 4 h into the light cycle and euthanized by cardiac puncture followed by cervical dislocation. See study outline Fig S1a. The study in C57BL6/J mice was conducted in accordance with the Canadian Council on Animal Care guidelines and regulations approved by Laval University, Canada (license 2017-086-1).
After acclimatization, BALB/c mice received a daily oral gavage 3.5 h into the light cycle of 200 µL PBS (Gibco) with or without 0.04 mg or 1 mg A. alcalophilum-derived lysozyme for 7 days to the day before euthanasia while continuously fed chow diet. After 2 days of daily lysozyme or vehicle administration colitis was induced in selected groups by addition of DSS in the drinking water (3%w/v, Product code 42867, Sigma Aldrich) for the following 5 days until euthanasia. Body weights were monitored daily. At day 5 of DSS-challenge mice were anesthetized with isoflurane (UK National Veterinary Service Ltd) and euthanized by cardiac puncture followed by pentobarbitone (UK National Veterinary Service Ltd) overdose and cervical dislocation. See study outline Fig S1b. The primary outcomes of the BALB/c study were reproduced in an identical study setup adding an intermediary dose of 0.2 mg lysozyme. The studies in BALB/c mice were conducted with relevant guidelines and regulations approved by the UK Home Office Scientific Procedures Act under project license 80/2613. The BALB/c studies were further reproduced in a different study facility including the 1.0 mg lysozyme dose where half of the mice were administered antibiotics in the form of ampicillin (1.0 g/L) and neomycin (0.5 g/L) in the drinking water. Antibiotics were administered from 5 days prior to the first administration of lysozyme and/or 7 days prior to the addition of DSS, as outlined in Fig S1c, and were otherwise identical to the study outlined in Figure1b. The antibiotic study followed the relevant guidelines and regulations approved by the French Ministry of Research and Innovation under license APAFIS#247 88-2019102806256593 v8.

Oral glucose tolerance test (oGTT), glucose-stimulated insulin concentration (GSIC), and intestinal permeability assay
An oGTT and an sulfonic acid assay for intestinal permeability were carried out in week 10 of the experimental protocol in C57BL6/J mice. Mice were fasted 6 h from 2 h into the light cycle and gavaged with their usual daily supplementation at 4 h into the light cycle. Fasting blood glucose was measured and plasma insulin sampled from the tail vein prior to oral gavage with 2 µg/g lean mass dextrose (Hospira) and 1.5 mg fluorescein-5 (6)sulfonic acid (Invitrogen) dissolved in 150 μL suspension of 0.5% Carboxymethylcellulose Sodium Salt (CMC) (Sigma) in distilled water. Blood glucose was measured at 0, 15, 30, 60, 90, and 120 min after dextrose challenge, and blood samples for quantification of plasma insulin, and sulfonic acid levels were sampled in EDTA-prepared capillary tubes (Sarstedt) at 0, 15, 30, 60, and 120 min postgavage. Mice were subcutaneously injected with 0.5 mL saline (Hospira) after the procedure for rehydration. Blood samples were centrifuged for 10 min at 1000 rcf at 4°C. Plasma insulin was quantified by Mouse Ultrasensitive Insulin ELISA (Alpco) following the manufacturer's protocol. Gut permeability was assessed by quantification of fluorescence from the sulfonic acid fluorescein in 5 µL plasma transferred to black 96-well optical-bottom plates (Nunc) and kept on wet ice protected from light until the addition of 150 μL of 0.5% CMC in distilled water and read at excitation/emission 485/ 528 nm wavelength.

Tissue cytokine quantification
Snap-frozen liver, ileum, and colon tissue from C57BL6/J mice were cryo-grinded in liquid nitrogen and washed by adding 200 µL cooled PBS added protease inhibitors (Sigma-Aldrich), centrifuged at 16,000 rcf for 20 sec at 4°C and the liquid aspirated. For liver tissue, the wash-step was repeated 10 times, while intestinal tissues were washed 2 times. Afterward, 400 µL T-PER Tissue Protein Extraction Buffer (ThermoFisher) and beads were added and vortexed on TissueLyzer (Qiagen) for 2 × 1 min at 50 os/sec. Samples were centrifuged at 4,000 rcf for 1 min at 4°C and the supernatant transferred to new tubes that were then centrifuged again 13,000 rcf for 10 sec at 4°C. Protein extracted from intestinal tissue was filtered through 0.22 µm centrifugal filters (Merck-Millipore) and centrifuged 12,000 rcf for 4 min. Protein concentration was measured in triplicates using Pierce BCA Protein Assay kit (ThermoFisher) using the manufacturer's instructions. Cytokine levels in 50 µg extracted protein were measured using multiplex assays (Merck Millipore for liver and colon, BioRad for ileum) using a Bio-plex Multiplex System (BioRad). Colon tissue from BALB/c mice was placed in a lysing tube containing lysis solution added protease inhibitor and tissue protein extraction reagent at a ratio of 1 g of tissue to 5 mL lysis solution. The tissue was homogenized 3 times at 4247 rcf for 30 sec after which the samples were centrifuged at 92 rcf for 5 min at 4°C to extract the protein. The supernatants were evaluated using a multiplex assay (Merck Millipore) for a range of Th1 and Th17 characteristic cytokines using a Magpix system (Luminex).

Lipopolysaccharide (LPS)
Blood from euthanasia of C57BL6/J mice was sampled with EDTA-coated syringes and immediately kept on ice and centrifuged for 10 min at 1000 rcf at 4°C and the plasma stored at −80°C until further processing. LPS was quantified by ELISA (MyBioSource) following the manufacturer's instructions.

Quantitative reverse transcriptase PCR (qPCR) of intestinal tissues
RNA from snap-frozen ileum and colon tissues from C57BL/6 J mice were extracted using the Directzol RNA Miniprep kit (Zymo Research). cDNA synthesis was made from 2 µg ileum or 1.5 µg colon RNA using High Capacity cDNA Reverse Transcriptase kit (Applied Bioscience) following the manufacturer's protocol. qPCR was carried out using 4 µl of cDNA, 5 µl of Advanced qPCR MasterMix (Wisent Bioproducts) and 0.5 µl of each primer (diluted at a concentration of 10 µM) in a total reaction volume of 10 µL with the following cycle setting: 95°C for 2 minutes (95°C for 20 sec, 61.5-62°C for 20 sec, 72°C for 20 sec) x 40 ending with a melting Curve: 65°C to 95°C. Each target in each tissue was evaluated and accepted in the case of a unified peak from melting curves and an efficiency of 100%±15 and R 2 > 0.95 from a standard curve. Relative expression was calculated by 2 ΔCq of target Cq to 18S Cq of the sample accepting replicates with coefficient of variation <0.05. Target primer sequences and annealing temperatures are found in Supplementary  Table 1.

Immuno-fluorescent in situ hybridization (FISH) of mucins and localization of bacteria
Mucus immunostaining was paired with FISH to analyze the localization of bacteria at the surface of the intestinal mucosa, as previously described. 71 Briefly, proximal colon tissue was places in Carnoy's solution (60% methanol, 30% chloroform, 10% glacial acid) and washed in methanol 2 × 30 min, ethanol 2 × 15 min, ethanol/xylene (1:1) 15 min, and xylene 2 × 15 min, followed by embedding in paraffin. Sections of 5 µm were made and dewaxed by preheating for 10 min at 60°C, xylene for 10 min at 60°C, and 99.5% ethanol for 10 min. Hybridization was carried out overnight at 50°C with EUB338 probe (5ʹ-GCTGCCTCCCGTAGGAGT-3ʹ, with a 5ʹ labeling using Alexa 647) diluted to a final concentration of 10 μg/mL in hybridization buffer (20 mM Tris-HCl, pH 7.4, 0.9 M NaCl, 0.1% SDS, 20% formamide). After 10 min in wash buffer (20 mM Tris-HCl, pH 7.4, 0.9 M NaCl) and 3 × 10 min in PBS, we used PAP pen (Sigma-Aldrich) to mark the section and added block solution of 5% fetal bovine serum in PBS for 30 min at 4°C. The primary antibody for Mucin-2 (MUC2) (rabbit H-300, Santa Cruz Biotechnology) was diluted 1:1500 in block solution and added overnight at 4°C. Washing was done by 3 × 10 min in PBS, after which block solution containing anti-rabbit Alexa 488 secondary antibody diluted 1:1500, Phalloidin-Tetramethylrhodamine B isothiocyanate (Sigma) at 1 μg/mL and Hoechst 33258 (Sigma) at 10 μg/ mL was applied to the section for 2 h. Washing 3 × 10 min in PBS was then followed by mounting using Prolong anti-fade mounting media (Life Technologies). Observations were performed with a Zeiss LSM 700 confocal microscope with software Zen 2011 version 7.1. This software was used to determine the distance between bacteria of the microbiota and epithelial cell monolayer. 72

Microbiota profiling using 16S rRNA gene amplicon sequencing
Fecal samples from C57BL6/J mice were collected before the study started and at 4, 8, and 12 weeks of the study and small intestine content from the necropsy. Fecal samples from BALB/c mice were collected prior the study start (day −3) and 2 days after vehicle or lysozyme administration started prior to the DSS-challenge (day 0). The contents of the ileum, cecum, and colon were collected at necropsy (day 5). DNA was extracted using the Nucleospin Soil Extraction Kit (Machery-Nagel) on heat-inactivated samples by 15 min of heating to 95°C. 16S rRNA gene amplification targeting the V3-V4 region using forward primer S-D-Bact-0341-b-S-17 and reverse primer S-D-Bact-0785a-A-21 with Illumina adapters 73 74 Primer binding regions were removed with fastx_truncate and reads were filtered to contain less than one error per read. The quality filtered reads were denoised with unoise3. ASV abundance was calculated by mapping with usearch global using a 97% identity threshold. The phylogenetic tree was made by aligning the 16S sequences with mafft, and the tree was inferred by FastTree. Taxonomical classification was done with the qiime classifier (qiime2-2019.4) trained on the Silva database (Silva_132). Analysis of microbiota composition was carried out in R version 4.0.2 and R Studio version 1.3.1056 using the phyloseq, vegan, and DESeq2 packages 75-78 on nonrarefied data, including samples with a minimum of 8,000 reads. The dataset had a median of 19,035 and a mean of 20,968 reads per sample with a standard deviation of 8,414. Alpha diversity was calculated based on rarefied data to 8,000 reads per sample (Fig S1cd). Predicted functionality of the gut microbiota was assessed using PICRUSt2 generated Kyoto Encyclopedia of Genes and Genomes (KEGG) orthologs and pathways. 79

Histological scoring of colon
Inflammation grading of the colon of BALB/c mice was performed macroscopically using a light microscope and was conducted by two blinded observers based on the Wallace scoring method. Criteria for the scoring of colonic damage were: 0) no damage; 1) Hyperemia. Normal bowel wall. No ulcers; 2) Hyperemia and thickening of bowel wall. No ulcers; 3) One ulcer without thickening of the bowel wall; 4) Two or more sites of ulceration or inflammation; 5) Two or more major sites of ulceration and inflammation or one site of ulceration/inflammation extending >1 cm along the length of the colon; 6-10) Covering damage >2 cm along the length of the colon. The score was increased by 1 for each additional centimeter of colon damage.

H&E Staining of colonic tissue and histopathologic analysis
Following euthanasia, 1 cm of the proximal mouse colons were fixed in 10% buffered formalin for 24 hours at room temperature and then embedded in paraffin. Tissues were sectioned at 5-mm thickness and stained with hematoxylin and eosin (H&E) using standard protocols. H&E stained slides were next scored. Each colon was assigned four scores based on the degree of epithelial damage and inflammatory infiltrate in the mucosa, submucosa and muscularis/serosa, as previously described. 80 A slight modification was made to this scoring system as follows: each of the four scores was multiplied by 1 if the change was focal, 2 if it was patchy and 3 if it was diffuse. The 4 individual scores per colon were added, resulting in a total scoring range of 0-36 per mouse. 81

Lysozyme activity assay in the presence and absence of antimicrobials
A. alcalophilum-derived lysozyme was added to an assay plate containing fluorescent-labeled PGN from Micrococcus lysodeikticus (EnzChek Lysozyme Assay Kit, Invitrogen) with or without antibiotics (10 mg/mL neomycin and 20 mg/mL ampicillin) and incubated at 37 degrees for 2 hours. The fluorescence signal (excitation/emission of 485/520 nm) was continuously measured and is proportional to the enzymatic activity. Assays were conducted with lysozyme concentration ranging from 0.0001 to 2.5 mg/mL. The data is represented as the normalized linear curve where each data point represents the mean from three independent measurements.

Statistical analysis
GraphPad Prism 9 was used to calculate the statistical analysis except for the analysis of the microbiota composition described in the designated section. Each dataset was tested for normality by D'Agostino & Pearson test. Datasets not passing normality (p < .05) were log transformed prior testing where specific tests are described in figure legends. All repeated-measure ANOVAs include the Geisser-Greenhouse correction for adjusting for lack of sphericity. Statistical significance was considered at p < .05. In the study of C57BL6/J mice statistically significant differences between HFD+Vehicle and LFD+Vehicle are indicated by gray asterisks and difference between HFD+Vehicle and HFD+Lyso by black asterisks. This study includes n = 10 mice in LFD+Vehicle group, n = 11 in HFD+Vehicle and HFD+Lyso groups. The two first studies of BALB/c mice include n = 12 mice per group. The study featuring antibiotic-treated mice includes n = 8 mice per group.

Data availability
All data and R scripts generated and analyzed in this manuscript are available from the corresponding author upon reasonable request.

Disclosure statement
NNK, EGWS, PBO and LBSH are employed at Novozymes A/ S (NZAS). NZAS provided the test compound and filed a patent based on the enclosed findings (PCT/EP2018/ 050189). The views presented in this manuscript are those of the authors and not necessarily those of NZAS.

Funding
This work was primarily funded by Novozymes A/S, Denmark. BAHJ was supported by Lundbeck Foundation (grant number: R232-2016-2425) and Novo Nordisk