Yan Zhao, Chengcheng Zhang, Leilei Yu, Fengwei Tian, Jianxin Zhao, Hao Zhang,c,Wei Chen,c,e, Qixiao Zhai,d,*

a State Key Laboratory of Food Science and Technology, Jiangnan University, Wuxi 214122, China

b School of Food Science and Technology, Jiangnan University, Wuxi 214122, China

c National Engineering Research Centre for Functional Food, Jiangnan University, Wuxi 214122, China

d International Joint Research Laboratory for Probiotics, Jiangnan University, Wuxi 214122, China

e Beijing Innovation Centre of Food Nutrition and Human Health, Beijing Technology and Business University, Beijing 100048, China

Keywords:Lactobacillus fermentum Ulcerative colitis Anti-inflammation Phylogenetic analysis Comparative genomic analysisEmerging evidence shows that some Lactobacillus fermentum(L.fermentum)strains can contribute to the prevention and treatment of ulcerative colitis(UC).In this study,105 isolates of L.fermentum strains were separated from fecal samples of populations in different regions in China and their draft genomes were sequenced.Pan-genomic and phylogenetic characterizations of these strains and four model strains(L.fermentum 3872,CECT5716,IFO3956,and VRI003)were performed.Phylogenetic analysis indicated that there was no significant adaptive evolution between the genomes of L.fermentum strains and the geographical location, sex, ethnicity, and age of the hosts. Three L. fermentum strains (FWXBH115, FGDLZR121, and FXJCJ61) from different branches of the phylogenetic tree and strain type L. fermentum CECT5716 were selected and their anti-inflammatory and immune modulatory activities in a dextran sulphate sodium(DSS)-induced colitis mouse model were further investigated.Both L.fermentum FXJCJ61 and CECT5716 significantly alleviated UC by reducing all colitis-associated histological indices,maintaining mucosal integrity, and stimulating replenishment of short-chain fatty acids (SCFAs), while the other two strains failed to offer similar protection.The anti-inflammatory mechanisms of L.fermentum FXJCJ61 and CECT5716 were related to the inhibition of nuclear factor kappa-B(NF-κB)signaling pathway activation and enhancement of interleukin 10 (IL-10) production. Comparative genomic analysis of these strains identified candidate genes that may contribute to the anti-inflammatory effects of specific L.fermentum strains.

1. Introduction

Lactobacillus fermentum (L. fermentum) is a small, slender, nonmotile, Gram-positive bacterium in the genus Lactobacillus. It is pervasive in environments such as fermented vegetable feed and food, milk, cheese, artisanal starter cultures, human saliva, the human vagina, and the intestinal tract of humans and animals[1,2]. The genome of L. fermentum (IFO3956) was first sequenced in 2008 and genomes of 65 L. fermentum strains can now be obtained from the National Coalition Building Institute (NCBI)microbial genome database?[3].The genome size and G+C content of L. fermentum were (2.1 ± 0.1) Mbp and 51% ± 1%, respectively.With the advent of genome-wide sequencing, a series of genes related to the production of functional molecules (bile salt [4],exopolysaccharides [5], short-chain fatty acid (SCFAs) [6], and so on) in the genome of L. fermentum strains had been identified.

The number of genome-sequenced L. fermentum strains is relatively small when compared with some well-documented Lactobacillus strains such as L. plantarum (467), L. rhamnosus (178),and L. paracasei (178). Phylogenetic and comparative genomic analyses indicate that extensive gene loss and lateral gene transfer has occurred during the co-evolution of Lactobacillus spp. with their habitats [7]. Certain genetic variations in Lactobacillus were predicted to provide the ability to adapt to diverse niches, and efforts were made to explore the links between genome evolution and ecological versatility[8].Genes involved in sugar metabolism,the proteolytic system, and bile-salt hydrolysis were shown to be responsible for the specific habitats to which Lactobacillus strains adapt, such as to the human gut or dairy products [9].

Until now,studies have been predominantly focused on L.plantarum [10], L. reuteri [11], L. rhamnosus [12], and L. casei [13], and current evidence suggests that these species have co-evolved with their habitats,such as host species[14].To our knowledge,however,studies focused on the evolutionary characteristics of L.fermentum strains are very limited. A multilocus sequence typing analysis of 203 isolates of L. fermentum, based on the sequences of 11 housekeeping gene fragments,indicated a similar evolutionary tendency in L.fermentum isolates from food sources from the same locations[15]. Typically, L. fermentum is found in fermentation products,but it is also a general symbiont in the human gut[16].The effects of the geographic location, sex, ethnicity, and age of the hosts on the evolution of L.fermentum strains need further study.

An increasing number of studies have indicated that L. fermentum can have a positive effect on the health of its host [17,18]. It has been shown to antagonize pathogenic bacteria[19]and alleviate alcoholic liver disease [20], cardiovascular disease [21], and colitis [22]. L. fermentum was listed as a ‘‘generally recognized as safe” (GRAS) organism by the US Food and Drug Administration(FDA) in 2013. Experimental evidence from animal models has demonstrated L.fermentum plays a crucial role in the amelioration of ulcerative colitis (UC) by affecting the production of proinflammatory cytokines and inhibiting activation of nuclear factor kappa-B (NF-κB) [23]. L. fermentum CECT5716 was shown to trigger recovery of normal concentrations of SCFAs in the intestinal contents and reverse microbiota dysbiosis in mice with dextran sulphate sodium (DSS)-induced colitis [24]. In contrast, another study reported that pathogenic L. fermentum was isolated from the cholecystostomy aspirate and anaerobic blood culture of an 81-year-old male patient with cholecystitis [25]. Anderson et al.[26] also indicated that human oral isolate L. fermentum AGR1487 can cause a pro-inflammatory response in germ-free rats by increasing inflammatory cells (macrophages, lymphocytes, and neutrophils),inducing colonic myeloperoxidase and plasma serum amyloid A and activating Toll-like receptor signaling.These studies indicated that the anti-inflammatory effects of L. fermentum were strain-specific. A Bacteroides thetaiotaomicron mutant lacking a choloylglycine hydrolase gene (BT2086) responsible for bile salt hydrolase (BSH) activity could lower liver and plasma lipid levels in mice [27] and a Clostridium sporogenes mutant lacking genes for the synthesis of branched SCFAs could up-regulate IgArelated immune cells in vivo [28]. Other research has suggested that the bsh gene (coding for bile salt hydrolase) is responsible for the cholesterol-lowering activities of L. fermentum [29]. The pdu–cbi–cob(pdu,propanediol dehydratase genes;cbi,cobinamide biosynthetic genes; cob, cobalamin biosynthetic genes) gene cluster was found to encode for reuterin, and cobalamin was shown to contribute to the antibacterial properties of L. reuteri JCM 1112[3]. As these reports confirmed the tight association between the function of Lactobacillus spp.and some of its genes, the genes crucial to the activity of L.fermentum strains against inflammatory disease need to be further studied.

In this work, 105 L. fermentum strains were isolated from fecal samples of human subjects in China, and the draft genome sequences of these strains were obtained. The aim of the study was to evaluate whether genomic distinctions occurred in L. fermentum strains in populations from different geographic regions and having different physiological characteristics, to reveal the effects of L. fermentum strains from different branches of the phylogenetic tree on DSS-induced colitis in mice, and to identify the functional genes potentially responsible for the various levels of anti-inflammatory protection offered by these strains.

2. Materials and methods

2.1. Chemicals and reagents

DSS (36–50 kDa; MP Biomedicals, USA), Fast DNA spin kit for feces(MP Biomedicals,USA),TRIzol reagent(Invitrogen,USA),QIAquick gel extraction kit (QIAGEN, Germany), BCA protein assay kit(Beyotime Biotechnology, China), RIPA lysis buffer (Beyotime Biotechnology,China),ultrapure RNA kit(CWBIO,China),protease inhibitor cocktail and phosphatase inhibitor cocktail I (MedChem-Express, USA), RevertAid first strand complementary DNA (cDNA)synthesis kit (Thermo Fisher Scientific, USA), iTaqTMuniversal SYBR?green supermix(Bio-Rad,USA),enzyme linked immunosorbent assay (ELISA) kits for interleukin (IL)-10, IL-6, IL-4, IL-1β and tumor necrosis factor (TNF)-alpha (R&D Systems, USA), anti-NFκB p65 and anti-NF-κB p65 (Abcam, UK), goat anti-rabbit immunoglobulin G (IgG) antibody (GenScript, China), goat anti-mouse IgG antibody (H&L) (GenScript) and β-actin antibody(GenScript, China), ELISA kit for lactate (MLBio, China).

2.2. L. fermentum isolates and growth conditions

L.fermentum strains were isolated by a modified LAMVAB(Lactobacillus anaerobic Man–Rogosa–Sharpe (MRS) with vancomycin and bromocresol green)medium[30]developed by our laboratory(Table S1 in Appendix A). A total of 105 strains were separated from the feces of populations from 11 provinces (Anhui, Fujian,Hunan, Gansu, Guangdong, Guangxi, Jiangsu, Jiangxi, Qinghai,Shandong, Sichuan),Xinjiang autonomous region,Ningxia autonomous region,Inner Mongolia autonomous region,Shanghai municipality and Chongqing municipality of China(Appendix A Table S2).For in vitro experiments, L. fermentum strains were incubated at 37°C on MRS agar for 18–24 h.For animal experiments, final concentrations of L. fermentum strains (5 × 109colony-forming unit per milliliter (CFU?mL-1)) suspended in saline solution were adjusted and the gavage volume dosed to each mouse was 100 μL.

2.3. Comparative genome analysis and phylogenetic analysis of L.fermentum

Genomes of 109 L. fermentum strains (four from NCBI and 105 sequenced in this work) were subject to pan-genome analysis using PGAP v1.2.1 and average nucleotide identity (ANI) analysis using Python.Protein coding sequences(CDS)were predicted using Glimmer v3.02. Orthologous genes were generated using OrthoMCL v1.4 and a maximum-likelihood tree was constructed based on core genes of all L. fermentum strains and 11 housekeeping genes[15]using MAFFT v7.313.For the identification of differences in the functional genes of strains, genomes were subject to BLAST against annotated full-length CAZyme proteins in the Carbohydrate-Active Enzyme (CAZy) database and proteins in the Clusters of Orthologous Groups (COG) protein database [31].

2.4. Animals and experiment design

All procedures involving animals were approved by the Ethics Committee of Jiangnan University, China (JN. No.20180615b0950901(164)). BALB/c mice (male, eight weeks old,Shanghai SLAC Laboratory Animal Co., Ltd., China) were used for our in vitro experiment and each group comprised 10 mice. The mice were allowed to acclimate for one week before the experiment. Normal drinking water was given to mice in the normal group and water containing 4% DSS (w/v) was given to mice for 10 consecutive days in the other groups in the experiment. In the normal group and DSS group,each mouse was orally administered 100 μL saline solution per day. In the four DSS + L. fermentum groups, mice received 100 μL L. fermentum suspension (a concentration of 5 × 109CFU?mL-1) via oral gavage per day [32].

During the treatment, body weight and bloody stools were assessed every day. At the end of the experiment, mice were intraperitoneally injected with 1% pentobarbital sodium solution and euthanized [33]. The length of colons was recorded, with part of the colon used for histological observation and the remaining part stored at -80 °C. The disease activity index (DAI) of mice was evaluated according to a previous study [34].

2.5. Histological assessment

The histological assessment of distal sections of the colon was performed based on hematoxylin–eosin (H&E) staining, as previously reported [35]. The histological scores were assessed based on inflammatory infiltrates,loss of goblet cells and mucosal hyperplasia by two researchers who were blinded to the details of each section.

2.6. ELISA and immunoblotting

The concentrations of IL-4, IL-6, IL-1β, TNF-α, and IL-10 in the colons of mice were measured using ELISA kits (R&D Systems,USA). The expression of total p65 and phosphorylated p65 (pp65) in the colon of mice was determined by western blot assay,as previously described[36].The protein expression was visualized using AlphaView software v3.4.0.0.

2.7. Gene expression of tight junction protein in the colon of mice

Total RNA in the colon of mice was extracted using an Ultrapure RNA Kit(CWBIO,China)and complementary DNA was synthesized using a RevertAid first strand cDNA synthesis kit (Thermo Fisher Scientific, USA). Gene expression concentrations of occludin,

claudin-1, zonula occludens (ZO)-1 and ZO-2 were evaluated by real-time quantitative polymerase chain reaction (qPCR) [37].Information for primer sequences is shown in Table 1.

2.8. SCFAs analyses

Concentrations of SCFAs (acetic acid, propionic acid, pentanoic acid,butyric acid,and isobutyric acid)in the colonic contents were determined by gas chromatography-coupled mass spectrometry(GC-MS), as previously described [38].

2.9. Gut microbiome of mice

The DNA of bacteria in fecal samples was extracted with a fast DNA spin kit for feces (MP Biomedicals, USA) and amplification of the 16S ribosomal RNA (rRNA) gene sequences (V3–V4 regions)was performed as previously reported [39]. The DNA amplicons were sequenced by Illumina MiSeq platform and clustered into operational taxonomic units (OTUs) with 97% similarity using UCLUST v11.

Table 1 Primers used for qPCR.

2.10. Species-specific qPCR

To examine the total amount of L. fermentum strains in fecal samples from mice, a species-specific qPCR method was used[40].A standard curve was generally used to react the interactions of cycling threshold (Ct) value and cell counts of L. fermentum. A total of 0.04 g fecal sample of each mice was weighted and fecal DNA was extracted as mentioned in Section 2.9 [41].

2.11. Resistance of L. fermentum strains to gastrointestinal environment

The tolerance of L. fermentum strains to acid and bile salt was studied [42]. Bacterial culture solution was harvested (6000 revolutions per minute (rpm), 2 min) and washed with sterile saline solution. Cell pellets were exposed to simulated gastric juice (3 g?L-1pepsin in sterile saline, pH 3.0) for 3 h and subsequently to simulated small intestinal juice (trypsin (1 g?L-1) and bile salts (3 g?L-1) in sterile saline, pH 8.0) for 4 h. The viable bacterial count was determined at 0, 3, and 7 h.

2.12. Lactate production by L. fermentum in vitro

Quantification of lactate production was determined as described [43]. L. fermentum strains were pre-cultivated twice using MRS broth at 37 °C for 48 h. The concentration of lactate in the supernatant was quantitatively determined by the ELISA kit for lactate (MLBio, China).

2.13. Statistics

All data were described as mean±standard error of mean(SEM).Difference analyses of DAI values and body weight of mice were carried out using one-way analysis of variance (ANOVA) followed by Dunnett’s test.The difference analysis of other biomarkers was analyzed using one-way analysis of variance (ANOVA) followed by Tukey’s test (P < 0.05, P < 0.01, and P < 0.001). Statistical analyses in this study were performed using the GraphPad Prism v6.0.

3. Results

3.1. Genetic diversity and evolution of L. fermentum strains

The genomic information and detailed sources of 105 L.fermentum strains isolated from human fecal samples of healthy Chinese subjects(41 male,62 female,2 unknown;aged 0–100 years)were provided in Fig.1(a)and Appendix A Table S2.Taking the four additional type strains (3872, CECT5716, IFO3956, and VRI003) into consideration,the genome size of L.fermentum was approximately 2.0 Mbp (G + C content ranged from 50.56% to 52.50%). Pangenome analysis reveals 11 579 gene families corresponding to 109 L. fermentum genomes and the pan-genome curve increased sharply as the number of genomes increases.Conversely,the number of core genes decreased gradually as the number of genomes increased,reaching 1179 for 109 genomes(Fig. 1(b)). The ANI values of whole genomes between homologous regions shared by any two genomes are generally greater than 97% (Appendix A Fig. S1).

Fig.1. The genomic and genetic characteristics of L.fermentum strains isolated from fecal samples of populations in some regions of China:(a)number of strains isolated from different regions;(b)pan-genome and core genome(the curve represents the pan-genome and core-genome plotted against the number of genomes of L.fermentum);and(c)Venn diagram of the homologous clusters shared among the core genes (the number in the outer ring means counts of specific genes in each strain).

The 109 L. fermentum strains shared 1303 orthologous genes(Fig. 1(c)). Phylogenetic analysis of L. fermentum strains based on 11 housekeeping genes was annotated with different shapes and colors representing several factors, including the geographical location,sex,ethnicity,and age.L.fermentum strains with common characteristics did not cluster together and this revealed that no direct relationships existed between the genomes of L. fermentum strains and the geographical location, sex, ethnicity, and age of the hosts (Figs. 2(a) and (b)). Phylogenetic analysis based on 1303 core genes of the 109 strains indicated that while all L. fermentum isolates were divided into three distinct branches, no single selective pressure factor significantly affected the evolution of L.fermentum strains (Fig. 2(c)). Three L. fermentum strains (FWXBH115,FGDLZR121, and FXJCJ61, abbreviated as WX115, GD121, and XC61 respectively in Figs. 1 and 2) from different branches and the type strain L. fermentum CECT5716 (marked with black maple leaf in Fig. 2(c)) were selected to study their effects on DSSinduced colitis in BALB/c mice.

3.2. Effects of L. fermentum supplementation on body weight, DAI value, and colonic histopathology of DSS-treated mice

Symptoms such as loss of body weight(Fig.3(a)),apparent diarrhea and rectal bleeding were induced by DSS treatment. DSS treatment also significantly raised the DAI scores (Fig. 3(b)) and shortened the colon (Fig. 3(c)). Oral gavage with L. fermentum FXJCJ61 and the type strain L. fermentum CECT5716 significantly alleviated these symptoms. And compared with L. fermentum CECT5716, L. fermentum FWXBH115, and FGDLZR121 had little effect on DSS-induced symptoms in mice.

A decreased number of crypts and marked infiltration of inflammatory cells were observed in the DSS group(Fig.3(e)).Treatment with L. fermentum FXJCJ61 and the type train L. fermentum CECT5716 significantly alleviated these pathological damages in DSS-treated mice,while the other two strains failed to offer similar protection (Fig. 3(f)).

3.3. Effects of L. fermentum supplementation on inflammatory cytokines and NF-κB signaling in colon of DSS-treated mice

Fig.2. Phylogenetic analysis of 109 L.fermentum strains:(a)maximum likelihood tree based on 11 housekeeping genes(each label annotated with various shapes and colors corresponds to certain characteristics; bootstrap confidence values were marked in the branches of the tree; I, II, and III represent three separate phylogenetic clades);(b) information on 11 housekeeping genes; and (c) maximum likelihood phylogeny derived from 1303 core genes across 109 strains. ATP: adenosine triphosphate; UTP:uridine triphosphate; CTP: cytidine triphosphate.

Compared with the DSS group, L. fermentum FXJCJ61 and CECT5716 supplementation significantly reduced the colonic concentration of TNF-α but did not affect the concentrations of IL-6,IL-4, and IL-1β (Fig. 4). Unlike the type strain L. fermentum CECT5716, L. fermentum FWXBH115 and FGDLZR121 had no regulatory effect on TNF-α and IL-1β but increased the colonic concentration of IL-6 (Fig. 4). Oral administration of all of the four L.fermentum strains triggered rapid upregulation in the expression of anti-inflammatory cytokines, such as IL-10, compared with the DSS group (Fig. 4(d)).

The expression of p-p65 in colon tissue increased significantly after treatment with DSS when compared with the normal group.Compared with the DSS group, L. fermentum FXJCJ61 and the type strain L. fermentum CECT5716 markedly decreased the level of pp65 in the colon of DSS-treated mice. However, L. fermentum FWXBH115 and L. fermentum FGDLZR121 did not have an obvious effect on the expression of p-p65 protein in DSS-treated mice(Figs. 4(f) and (h)).

3.4. Effect of L. fermentum supplementation on the concentrations of intestinal SCFAs in DSS-treated mice

The effect of L.fermentum supplementation on SCFA concentrations in DSS-induced colitis mice was shown in Fig. 5. DSS treatment significantly reduced the production of propionic acid and pentanoic acid compared with the normal group. All groups of L.fermentum strains significantly increased (P < 0.05) the concentrations of acetic acid and propionic acid compared with the DSS group(Figs.5(a)and(b)).Besides,L.fermentum FXJCJ61,the strain with the most similar regulatory effects to the type strain L.fermentum CECT5716,was the most effective orally administered strain for improving the concentrations of pentanoic acid(Fig.5(c))and butyric acid (Fig. 5(d)).

Fig.3. Effects of L.fermentum supplementation on DSS-treated mice(n=8):(a)weight changes of mice during the experiment;(b)disease activity index(DAI)of mice during the experiment (DAI was evaluated based on body-weight changes, rectal bleeding, and diarrhea); (c) colon length of mice; (d) representative images of the colons;(e) pathological images of the distal colon ((i) normal; (ii) DSS-treated; (iii) L. fermentum FWXBH115-treated; (iv) FGDLZR121-treated; (v) FXJCJ61-treated; and(vi) CECT5716-treated (the images are shown at 20× magnification and scale bars are 500 μm)); and (f) histological scores of the colon. All data are expressed as the mean±SEM.*and #indicate significant differences when compared with normal or DSS groups,respectively(*P<0.05,***P<0.001, #P<0.05, ##P<0.01,and ###P<0.001).

3.5.Effect of L.fermentum supplementation on message RNA(mRNA)expression of occludin,claudin-1,ZO-1,and ZO-2 in DSS-treated mice

Compared with the DSS group,L.fermentum FXJCJ61 and the type strain L. fermentum CECT5716 enabled substantial recovery of the expression of occludin,claudin-1, and ZO-1 in DSS-treated mice.In contrast, L. fermentum FWXBH115 and L. fermentum FGDLZR121 had no significant effects on the expression of ZO-1 and claudin-1 in the colon of DSS-treated mice.Oral administration of L.fermentum FXJCJ61 increased the expression of claudin-1 and ZO-1 in the colon ofDSS-induced micemorethanL.fermentumCECT5716.NoL.fermentum strains had significant effects on the expression of ZO-2(Fig.6).

Fig.4. Effect of L.fermentum supplementation on cytokines (n=8) and NF-κB signaling in colon of colitis mice: the concentration of cytokines (a)TNF-α, (b)IL-6, (c) IL-1β,(d)IL-10,and(e)IL-4;(f)detection of p65 and p-p65 by western blot analysis;(g)expression of p65;and(h)expression of p-p65.Densitometric quantification normalized to actin was calculated as change from the normal group;OD:optical density.For box and whiskers plot in(a)–(e),the middle line represents the median;the boxes represent the interquartile range(IQR);the error bars represent min to max values.All data in(g)and(h)are expressed as the mean±SEM.*and #indicate significant differences when compared with the normal or DSS-treated group, respectively (*P < 0.05, **P < 0.01, ***P < 0.001, #P < 0.05, ##P < 0.01, and ###P < 0.001).

Fig. 5. Effect of L. fermentum supplementation on the intestinal concentrations of SCFAs (a) acetic acid, (b) propionic acid, (c) pentanoic acid, (d) butyric acid, and(e) isobutyric acid in DSS-treated mice (n = 8).

3.6. Effects of L. fermentum on fecal microbiota in DSS-treated mice

The Shannon index indicated that alpha diversity for the DSS and L. fermentum groups increased significantly compared with the normal group(Fig.7(a)).Though there was no obvious separation trend among normal, DSS-treated and L. fermentum supplementation groups in the principal component analysis (PCA) plot,permutational multivariate analysis of variance (PERMANOVA)and pairwise comparison results showed that treatment with L.fermentum FWXBH115,FXJCJ61,and CECT5716 changed the structure of the gut microbiota compared with the DSS group (Fig. 7(b) and Appendix A Table S3). Linear discriminant analysis effect size(LEfSe) analysis and a column diagram show that there are 33 dominant OTUs from the six groups (Fig. 7(c)). As can be seen, at genus level, L. fermentum FXJCJ61 treatment markedly increased the level of Lactobacillus and Bacteroides compared with DSS groups(Fig. 7(d)).

3.7. Comparative genomic analysis of the specific genes in the four L.fermentum strains

In Fig.8(a),the number of genes shared between the four L.fermentum isolates ranges from 1635/3160 (51.74%) to 1747/3160(55.28%), and 1497 core genes are shared in the genome of these strains. The genomes of the four L. fermentum CECT5716 strains displayed an average of 2 055 461 bp and the G + C content is 51.72% (Fig. 8(b)). The L. fermentum genomes were predicted by comparison with a protein database(COG),and the results showed that 11 COG families were present only in a subset of the examined L. fermentum FXJCJ61 and CECT5716 genomes and 17 additional COGs were unique to a single strain of L. fermentum FXJCJ61(Figs. 8(c)–(e) and 8 Appendix A Table S4). Except for an average of 182 genes assigned to the COG ‘‘general function prediction only”, the genes were predicted to be mainly associated with amino acid metabolism, translation, replication and repair. Based on carbohydrate active enzymes database, our research showed that relatively large concentrations of glycoside hydrolases (GH2,GH25, GH43, and GH78) and glycosyl transferases (GT14 and GT83) existed in L. fermentum FXJCJ61 (Appendix A Fig. S2).

3.8. Detection of L. fermentum in fecal samples in mice

The detection limit of L. fermentum was considered to be 104CFU?mL-1, as estimated from the standard curve (Appendix A Fig. S3). In the DSS, normal and L. fermentum FWXBH115 groups,L. fermentum strains were not detected. L. fermentum strains were detected in the FXJCJ61, FGDLZR121, and CECT5716 groups. And the concentration of L. fermentum strains in feces of mice in the FXJCJ61-treated group was significantly greater compared with all other groups (Table 2).

3.9. Survival of L. fermentum in simulated gastric juices and simulated small intestinal juices

L. fermentum FWXBH115, FGDLZR121, FXJCJ61, and CECT5716 were successively exposed to simulated gastric juices and simulated small intestinal juices. The viable cell count was measured by plate counting and the survival rate of the strains was calculated by comparison of the initial counts with those after exposure to simulated gastric juices and simulated small intestinal juices.The data in Table 3 show that all L.fermentum strains were tolerant to simulated gastric juices(survival rate was greater than 85%)and L. fermentum FXJCJ61 had the highest survival rate of 98.33% ±3.51%. In the simulated small intestinal juices, L. fermentum FXJCJ61 also showed significantly higher tolerance to 3 g?L-1bile salts compared with the other strains. The survival rate of FGDLZR121 in simulated small intestinal juices was significantly higher than L. fermentum CECT5716 (Table 3).

3.10. Quantification of lactate production by L. fermentum in vitro

The production of lactate by four L. fermentum strains was determined in vitro and L.fermentum FWXBH115 showed the least ability to produce lactate (Fig. 9). Compared with L. fermentum FWXBH115, L. fermentum FGDLZR121 and CECT5716 produced higher concentrations of lactate (> 65 μg?L-1MRS). The highest concentration of lactate was 75 μg?L-1MRS,which was fermented by L. fermentum FXJCJ61.

Fig. 6. Effect of L. fermentum supplementation on mRNA expression of four tight junction proteins (a) ZO-1, (b) ZO-2, (c) claudin-1, and (d) occludin in the colon of DSStreated mice (n = 8). All data are expressed as the mean ± SEM.

4. Discussion

In this study,we analyzed the population genetic structure and phylogeny of 109 L. fermentum strains (105 isolated from fecal samples of the Chinese population, and four reference strains downloaded from NCBI), based on both 1303 core genes and 11 housekeeping gene fragments. Phylogenetic analyses showed that the evolution of L.fermentum lineages was not associated with geographical location, sex, age or ethnicity of fecal donors. The unequal sample numbers from different geographical location may not be the cause of the deviation in our results,since the samples from a same region (like Jiangsu, Guangdong, or Xinjiang)showed no obvious clustering trend in the phylogenetic tree.A previous population genetic structure and phylogeny analysis of L.reuteri strains by Oh et al.[14]indicated that evolution of L.reuteri lineages was adaptive for the different host species, although the sample numbers from different host were unequal (humans(n = 35), mice (n = 35), rats (n = 26), pigs (n = 41), chickens(n = 26), and turkeys (n = 5)). This was consistent with a previous report indicating that L. fermentum strains isolated from food sources did not exhibit a co-evolutionary trend with geographical location or type of food [15]. Previous studies showed that only a small number of gut microbiomes (such as those of Helicobacter pylori, L. reuteri, L. johnsonii, and L. acidophilus) evolved in parallel with their hosts, while others (e.g., L. plantarum, L. casei, and L.rhamnosus) did not [44,45].

Our results showed that all these isolates were clearly categorizable into three sub-types. Phylogenetic and functional analyses previously revealed that Lactobacillus species belonging to separate phylogenetic clades exhibited distinct types of metabolism,ecological niches and lifestyles[46].Genetical difference has been shown to be associated with certain characteristics of a strain and to result in diverse phenotypes [31]. For example, phylogenetic analysis of 39 Akkermansia muciniphila strains derived from human and other mammalian gut microbiotas showed that the genomes were assignable to three phylogroups, and two of these phylogroups were associated with obesity in their hosts [47]. Ottman et al.[48]elucidated that the protein Amuc_1100 of A.muciniphila annotated as a hypothetical protein was an outer membrane pili-like protein, which could regulate the immune and metabolic responses by inducing the production of IL-8, IL-1β, IL-6, and IL-10 through activation of Toll-like receptor (TLR) 2 and TLR4.

In this study,L.fermentum FWXBH115,FGDLZR121,and FXJCJ61 were distributed in different evolutionary branches by phylogenetic analysis derived from 1303 core genes and were randomly selected for further study in a mouse model of DSS-induced colitis.Oral administration of L.fermentum FXJCJ61 and CECT5716 showed a protective effect against DSS-induced intestinal inflammation in mice, in the form of reduced disease activity scores, alleviated histopathology symptoms (Fig. 3), regulated relevant inflammatory factors and inhibition of the expression of NF-κB(Fig.4).Interestingly, L. fermentum FXJCJ61 and CECT5716 belonging to the same evolutionary branch in the cladogram based on core genes had similar protective effects against DSS-induced colitis,but these two strains did not cluster in the maximum likelihood tree based on 11 housekeeping genes (Fig. 2). This may be due to the more comprehensive and representative phylogenetic analysis based on the whole core genes.

Fig. 7. Effect of L. fermentum supplementation on the gut microbiota composition of DSS-treated mice (n = 8): (a) Shannon diversity index; (b) PCA for gut microbiota;(c)column diagram of microbial composition(Lefse analysis);and(d)effects of L.fermentum on the structure of fecal microbiota at genus level(Bacteroides and Lactobacillus)in DSS-treated mice. * indicates a significant difference (*P < 0.05, **P < 0.01, and ***P < 0.001); ns: not significant.

It has been pointed out that the reduction of proinflammatory cytokines (TNF-α, IL-6, IL-1β, and IL-4) in mice with colitis could be a logical target for UC therapy [49]. To this end, L. fermentum FXJCJ61 and CECT5716 greatly decreased the concentrations of proinflammatory cytokines (TNF-α) and increased the concentrations of anti-inflammatory cytokines (IL-10) (Fig. 4). T cellderived cytokines, such as interferon gamma (IFN-γ) and TNF-α,were found to upregulate intestinal epithelial TLR4 expression,mediate epithelial cell death and correlate to immunopathology in inflammatory bowel disease (IBD) colitis [50]. Another report indicated that the downregulation of the translocation and transcriptional activity of the nuclear factor of activated T cells could block the production of TNF-α, which could be a therapeutic approach for treating bowel diseases [51]. In addition, IL-10 produced by T cells,dendritic cells,macrophages and B cells has been shown to suppress the expression of TNF-α in immune regulatory processes [52]. It has also been demonstrated that persistent NFκB activation in epithelial cells can contribute to the development of inflammation and TNF-α could activated the NF-κB pathway[53].This may suggest that these strains could also play a potential role against a range of inflammatory diseases.

In our work, L. fermentum FXJCJ61 and CECT5716-treated groups displayed downregulation of the expression of NF-κB(Fig. 4), which may partly explain the regulatory mechanisms of these strains in relieving DSS-induced colitis. Taking all of the above mentioned in vivo biological indicators into consideration,L. fermentum FXJCJ61 and CECT5716 could more significantly reduce DSS-induced colitis than the other tested strains. These findings were instrumental for the clinical application of these strains in the future.Our findings showed the potential of the clinical translation of L. fermentum FXJCJ61 in the treatment of colitis.

Comparative genomic analysis has become a routine method with which to explore specific functions of genes, such as those coding for transporter proteins and catabolic enzymes of bacteria[54]. A comparative genomic analysis of 43 uropathogenic Escherichia coli strains isolated from a cohort of 14 women with urinary tract infections revealed that strains from distinct clades of the phylogenetic tree possessed different pathogenicities [55].

Fig. 8. Comparative genomic analyses of L. fermentum: (a) Venn diagram of core, distributed and unique gene numbers between FWXBH115, FGDLZR121, FXJCJ61, and CECT5716;(b)genomic information of FWXBH115,FGDLZR121,FXJCJ61,and CECT5716(CDS:coding sequence;rRNA:ribosomal RNA;tRNA:transfer RNA;CRISPR:clustered regularly interspaced short palindromic repeats); (c) functional annotation based on COG database; (d) functional annotation of FWXBH115, FGDLZR121, FXJCJ61, and CECT5716 based on COG database (number of genes associated with COG functional categories for all three sequenced strains and CECT5716 is shown); and (e) COG categories that are contained only in FXJCJ61 and CECT5716.

Table 2 Qualitative detection of L. fermentum in fecal samples in mice (n = 5).

Table 3 Survival rate of L. fermentum strains under simulated gastric juices and simulated small intestinal juicesa.

Fig.9. Quantification of lactate production by L.fermentum in vitro.The superscript letters a, b, and c indicate statistically significant differences at P values less than 0.05 among the different groups(one way ANOVA and Tukey’s test).Duplicate samples were included for each treatment and the experiment was repeated three times. All data are described as mean ± SEM.

In our work,we compared the genomes of L.fermentum FXJCJ61 and CECT5716, two strains with significant anti-inflammatory abilities, with those of two inactive strains, L. fermentum FWXBH115 and FGDLZR121.As shown in Fig.8,there are 11 specific genes existing only in the genomes of L.fermentum FXJCJ61 and CECT5716.These specific genes were characterized and assigned to specific pathways. First, three genes associated with the production of SCFAs and lactate were analyzed. Beta-xylosidase(COG3507) is the main enzyme for the production of monomeric xylose and Yang et al. [56] showed that fecal SCFA production was increased in mice fed with fiber xylooligosaccharide. Alcohol dehydrogenase (COG1454) and aldehyde dehydrogenase (ALDH)are involved in the metabolism of alcohol to acetic acid [57]. Supporting these analyses, our work suggests that L. fermentum FXJCJ61 and CECT5716 can effectively trigger the recovery of normal concentrations of SCFAs in the colon of mice (Fig. 5). Since the SCFAs may also play a pivotal role in regulating host metabolism and reduce the risk of cancer and diabetes [58,59], L. fermentum FXJCJ61 and CECT5716 were expected to exhibit protective effects against these diseases.

Mechanistically, the SCFAs produced by bacterial fermentation were shown to regulate intestinal inflammation by increasing cluster of differentiation 4 positive(CD4+)T cell frequency and number and IL-10 expression [60]. Butyrate, acetate and propionate have been shown to be able to protect the intestinal epithelium from inflammatory damage by increasing tight junctions and reducing intestinal epithelial permeability[61–63].In our experiments,both L. fermentum FXJCJ61 and CECT5716 increased the expression of occludin, claudin-1, and ZO-1 mRNA compared with the levels in DSS-induced colitis mice, which might correlate with the ability of these strains to increase SCFAs production (Fig. 6). Whether the L. fermentum strains can also increase the expression of tight junction protein in normal mice was unknown and need to be evaluated in the future study. Tight junction proteins in the gut were also associated with irritable bowel syndrome (IBS) and arthritis and this indicated that these strains may have a potential clinical application prospect in the future [64,65].

Candidate genes of malate/L-lactate dehydrogenases(COG2055)have been shown to play a vital role in lactate synthesis and generating the lactate secreted from lactic-acid-producing bacteria (LAB). Lactate was reported to have a protective effect in response to gut injury [66] and it is also one of the precursors in the production of acetate, propionate or butyrate. Moreover, lactate can regulate the proinflammatory cytokines produced by intestinal epithelial cells through TLR 4 and TLR 5 [67]. Our in vitro experiment confirmed that L. fermentum FXJCJ61 had a greater ability to produce lactate than other strains (Fig. 9).

Specific genes (denoted COG0553, COG1468, COG2189, and COG3587) belonging to L. fermentum FXJCJ61 and L. fermentum CECT5716 have been found to be responsible for defense mechanisms,replication,and recombination and repair of DNA,and were predicted to enhance cell survival [68]. Furthermore, genes denoted isopropyl-malate/homocitrate/citramalate synthase(COG0119) and 5,10-methylenetetrahydrofolate reductase(COG0685) belonging to L. fermentum FXJCJ61 were found to be involved in amino acid transport and metabolism (Table S4).Additionally, COG0685 may enable the strains to produce methionine,and COG0119 was demonstrated to be involved in the biosynthesis of lysine [69]. Research also showed that L. casei can synthesize cysteine from methionine and the related metabolism may be associated with the resistance of bacteria to bile-salt stress[70].The production of lysine by gut microbes has been reported to be associated with their microbial competitive capacities in the gut[71]. Enhancement of stress tolerance and good gut fitness are the foremost criteria for probiotics to survive in the host intestinal environment and play a beneficial role in health [72]. Thus, these specific genes may guarantee the anti-inflammatory effect of L.fermentum FXJCJ61 and CECT5716.

Supporting these analyses,qualitative detection of L.fermentum in fecal samples in mice indicated that L.fermentum FXJCJ61 had a greater intestinal colonization ability than the type strain L. fermentum CECT5716(Table 2).This also correlates with the high tolerance of L. fermentum FXJCJ61 to a simulated gastrointestinal environment (Table 3).

Utilization of oligosaccharides (such as xylan) by bacteria can contribute to anti-oxidant and immune-modulatory activities in their host,and the ability to utilize more carbon sources was found to be involved in the adaptive ability of the microbiota[73].Several genes involving the synthesis of specific carbohydrate active enzymes were found in the genome of L. fermentum FXJCJ61(Fig. S2). Moreover, genes belonging to COG5416 and COG5523 were found only in L. fermentum FXJCJ61 (Table S4), these were responsible for producing integral membrane proteins and their functions need to be further studied.

In conclusion, the present study showed the strain-specific properties of L. fermentum strains selected from different clades of the phylogenetic tree against DSS-induced intestinal inflammation.L.fermentum FXJCJ61 and CECT5716 offered a more significant protective effect than the other two control strains. Inhibition of NF-κB signaling pathways may have contributed to the reduction of inflammatory cytokine secretions and alleviation of disease symptoms in a UC mouse model.Our study further suggested that specific genes existing only in the genome of L.fermentum FXJCJ61 and CECT5716 were responsible for their colonic SCFAs production and tolerance to the gastrointestinal environment. This may account for the protective effect of L. fermentum FXJCJ61 and CECT5716 in a DSS-induced colitis mouse model.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (31820103010, 31530056, and 31871773);National Key Research and Development Project(2018YFC1604206);Projects of Innovation and Development Pillar Program for Key Industries in Southern Xinjiang of Xinjiang Production and Construction Corps (2018DB002); National First-Class Discipline Program of Food Science and Technology(JUFSTR20180102); BBSRC Newton Fund Joint Centre Award (BB/J004529/1); and Collaborative Innovation Center of Food Safety and Quality Control in Jiangsu Province.

Compliance with ethics guidelines

Yan Zhao, Chengcheng Zhang, Leilei Yu, Fengwei Tian, Jianxin Zhao,Hao Zhang,Wei Chen,and Qixiao Zhai declare that they have no conflict of interest or financial conflicts to disclose.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.eng.2020.09.016.