Antibiotics induced intestinal tight junction barrier dysfunction is associated with microbiota dysbiosis, activated NLRP3 inflammasome and autophagy

All experimental animal manipulations were conducted in accordance with the Animal Care and Use Committee of Daping Hospital, Army Medical University (Third Military Medical University) with the animal license SCXK(J)2007-017, and all the protocols were permitted by the Medical and Ethics Committee of Southwest Hospital, Army Medical University (Third Military Medical University), Chongqing, China. The project license number is NSFC81772081.

Healthy adult female C57BL/6 mice weighing 18-22g were used in this study. Mice were housed in wire-bottomed, wire-lid cages, allowed access to chow and water ad libitum, and acclimatized in a temperature-controlled room (25±2°C) with 12-hour light and dark cycles for 1 week before experiments. To induce intestinal microbiota dysbiosis, healthy mice were treated with broad-spectrum antibiotics as described previously [15]. Briefly, broad-spectrum antibiotics cocktail, consisting ampicillin (1g/L, Sigma), neomycin sulfate (1g/L, Amresco), metronidazole (1g/L, Sigma) and vancomycin (0.5g/L, Amresco), was administered to mice ad libitum via drinking water for 1, 3, 5, 7 or 14 days respectively. At each time point, mice were anesthetized for intestinal paracellular permeability assay, and then sacrificed for sample collections. Samples were used for the following experiments.

According to the methods we described previously [16–18], intestinal paracellular permeability was determined by measuring the appearance of a maker in blood, 4.4kDa fluorescein isothiocyanate-labeled dextran (FITC-dextran) (Sigma, St. Louis, MO). Briefly, a laparotomy was carried out under anesthesia before the mice were sacrificed at the end of experiment. A 5-cm segment of ileum was dissociated beginning 5cm proximal to the cecum, with well-protected superior mesenteric vessels. The bilateral end of the isolated ileum was ligated with 2–0 silk suture to prevent the leakage of FITC-dextran. Injecting 0.1ml of 20mg/ml FITC-dextran, which was dissolved in 0.1mol/L phosphate buffer saline (PBS, PH 7.2), into the lumen. After this procedure, the laparotomy was closed. After 30 min, blood samples were taken, and centrifuged at 3000g for 10 min. The plasma was collected and diluted at 1:10 with 0.1mol/L PBS. Then the fluorescence intensity of the diluted plasma was measured by using a microplate reader (Varioskan Flash, Thermo Electron Corporation, Vantaa, Finland) with an excitation wavelength of 480nm and an emission wavelength of 520nm. The plasma FITC-dextran concentrations were calculated from standard curves generated by serial dilution of FITC-dextran in PBS.

Frozen sections of ileal tissue were fixed with 4% paraformaldehyde for 10 min at room temperature, then washed thrice with PBS, permeabilized in 0.1% Triton X-100 in PBS for 5 min at room temperature. After washing with PBS, the nonspecific binding sites were blocked with 5% normal goat serum in PBS for 30 min. Then, the sections were incubated with monoclonal rabbit antibody against ZO-1 (1:200, Invitrogen), ASC (1:100, Cell Signaling), LC3B (1:100, Sigma) at 4°C overnight. After washing thrice in PBS, sections were incubated with secondary Alexa Fluor 594 conjugated goat anti-rabbit IgG antibody (1:200, Invitrogen) and DAPI (1:1000, Sigma) for 1 hour at room temperature. After washing thrice in PBS, sections were mounted using Slowfade reagents (Invitrogen). Images were obtained using a TCS SP5 laser confocal microscopy (Leica, Germany).

An appropriate ileal segment was taken to collect mucosa by a glass slide after the mice were sacrificed. The collected mucosa was homogenized with 10 volumes of ice-cold lysis buffer containing 100mM PMSF, protease and phosphatase inhibitor cocktails (KeyGEN BioTECH), followed by a brief sonication with a sonicator (Tomy Seiko, Tokyo, Japan). Thereafter, the homogenates were centrifuged at 4°C, 10000 rpm for 15 min, and the supernatants were collected to determine protein concentration using RC DC kit (Bio-Rad, Hercules, CA). Equal amounts of total protein extracted from the ileal mucosa were fractionated on 8%, 10% and 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel to analyze the expressions of tight junction proteins (ZO-1, occludin and claudin-1), autophagy proteins (Atg5, p62 and LC3) and NLRP3 inflammasome proteins (NLRP3, ASC, caspase 1, cleaved caspase 1, IL-1β, cleaved IL-1β and IL-18). Then the proteins were transferred onto polyvinylidene difluoride (PDVF) membrane (Millipore, Bedford, MA). After complete transfer, membranes were blocked with 5% nonfat milk dissolved in TBST buffer for 60 min at room temperature. After blocking, membranes were incubated with antibodies specific for ZO-1 (1:1000, Invitrogen), occludin (1:1000, Invitrogen), claudin-1 (1:1000, Invitrogen), Atg5 (1:1000, Sigma), p62 (1:1000, Sigma), LC3 (1:1000, Sigma), NLRP3 (1:1000, Adipogen), ASC (1:1000, Cell Signaling), caspase 1 (1:1000, Cell Signaling), cleaved caspase 1 (1:1000, Cell Signaling), IL-1β (1:1000, Cell Signaling), cleaved IL-1β (1:1000, Cell Signaling), IL-18 (1:1000, Cell Signaling) and β-actin (1:5000, Sigma) overnight at 4°C. After wash with TBST, membranes were incubated with appropriate peroxidase-conjugated secondary antibodies (KPL) at room temperature for 60 min. The blots were visualized with an enhanced chemiluminescence detection kit (GE Healthcare, Buckinghamshire, UK). The chemiluminescence signal was captured by a ChemiDox XRS system (Bio-Rad). The densities of the bands were quantified with Quantity One Image software (Bio-Rad).

Fresh ileocecal fecal matters were frozen immediately and stored at -80°C until processing. Weighed fecal matters were put into a 20ml headspace sampling bottle, followed by the addition of 0.95ml 6% H₃PO₄ and 0.05ml 2.49 mmol/L of 2-ethylbutyric acid (the internal standard, diluted by 6% H₃PO₄), and vibration for 1 min. Then samples were injected into the gas chromatography-mass spectrometry system (5975A/7890C Agilent) for analysis. Ultrapure Helium (1.0 mL/min) was used as the carrier gas. Headspace was maintained at 80°Cwith incubation time of 30min. The gas chromatography (DB-FFAP fused-silica capillary column: 30 m ×0. 25 mm ×0. 25 μm Agilent) stepwise thermal conditions were as follows: 50°C for 1 min, 10°C /min until 250°C for 2 min; injector temperature was 250°C, with split ratio of 5:1. Mass detector system was operated in electron impact ionization at 70 eV, set to scan mode for m/z 33–200, and in single ion monitoring (SIM) mode. Temperatures of the transfer line (interface), and source were maintained at 280°C and 250°C, respectively. The monitored ions were m/z 60 and 43.1 for acetic acid, m/z 74.1 and 28.1 for propionic acid, m/z 60 and 73.1 for butyric acid, m/z 60 and 87.1 for isovaleric acid, m/z 43.1 and 73.1 for isobutyric acid, and m/z 88.1 and 73.1 for 2-ethylbutyric. Data acquisition and analysis were done with Saturn GC/MS workstation software (Agilent).

Fresh ileocecal fecal matters were frozen immediately after collected and stored at -80°C until DNA isolation. Microbial DNA was extracted with the QIAamp® Fast DNA Stool Mini Kit (Qiagen, Valencia, CA) in accordance with manufacturer’s protocols, and then stored at -80°C for further processing.

The V3-V4 region of bacterial 16S rDNA gene was amplified using primers 338F () and 806R (). The PCR reactions (95°C for 3 min, followed by 27 cycles at 95°C for 30s, 55°C for 30s, and 72°C for 45s and a final extension at 72°C for 10 min) were performed in 20μL mixture containing 4μL of 5× FastPfu Buffer, 2μL of 2.5 mmol/L dNTPs, 0.8μL of primers (5μmol/L), 0.4μL of FastPfu Polymerase, and 10 ng of template DNA. Illumina MiSeq sequencing amplicons were extracted from 2% agarose gels, purified using AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA) according to manufacturer’s instructions, and quantified using QuantiFluor™-ST (Promega, USA). Purified amplicons were pooled in equimolar and paired-end sequenced (2 × 250) on an Illumina MiSeq platform according to the standard protocols. The raw reads were deposited into the NCBI Sequence Read Archive (SRA) database. 5’-ACTCCTACGGGAGGCAGCAG-3’ 5’-GGACTACHVGGGTWTCTAAT-3’

The raw fastq files were demultiplexed, quality-filtered using QIIME (version 1.9.1) with the following criteria: (i) the 300 bp reads were truncated at any site receiving an average quality score <20 over a 50 bp sliding window, discarding truncated reads that were shorter than 50bp. (ii) exact barcode matching, 2 nucleotide mismatch in primer matching, reads containing ambiguous characters were removed. (iii) only sequences that overlap longer than 10 bp were assembled according to their overlap sequence. The reads which could not be assembled were discarded. Operational Units (OTUs) were clustered with 97% similarity cutoff using UPARSE (version 7.1 http://drive5.com/uparse/↗) and chimeric sequences were identified and removed using UCHIME. The taxonomy of each 16S rRNA gene sequence was analyzed by RDP Classifier (http://rdp.cme.msu.edu/↗) against the silva (SSU123)16S rRNA database using confidence threshold of 70%.

Data are presented as means ± SEM. For data analysis, one-way ANOVA was performed by SPSS 13.0 statistical software. A p value of <0.05 was considered as minimum level of statistical significance in all cases. All reported significant levels represent two-tailed p values.

To figure out the alterations of intestinal microbiota after antibiotic treatment, we evaluated the composition and diversity through an Illumina high-throughput sequencing technique. We generated a dataset consisting of 1698557 filtered high-quality and classifiable 16S rDNA gene sequences, and the average number of sequences obtained for each individual was 36139.5 (range: from 30068 to 44125). All sequences were clustered with representative sequences, and a 97% sequence identification cut-off was used. Obtained total number of operational taxonomic units (OTUs) was 474. As shown in Fig 1A, all diversity indexes including Ace, Chao, Shannon and Sobs were significantly decreased after antibiotics treatment, except Shannon which was not statistical different among 3 day, 5 day, 7 day and 14 day groups. However, Simpson was evidently increased without statistical discrepancy among 3 day, 7 day and 14 day groups. Furthermore, bacterial composition structure was analyzed. As shown in Fig 1B, on phylum level, Bacteroidetes, Firmicutes and Proteobateria were the dominant bacteria in control mice. However, in antibiotics-treated mice, the proportions of Proteobateria and Tennericutes were obviously increased, whereas the percentages of Bacteroidetes and Firmicutes were evidently decreased. Thus, broad-spectrum antibiotics remarkably alter the diversity and composition of intestinal microbiota in healthy mice.

Based on the above results, we expected to further figure out the different species after antibiotics treatment, therefore, we re-divided these 47 samples into two groups, i.e. control group (C-1,C-2,C-3,C-4,C-5,C-6,C-7), antibiotics group containing allsamples treated by antibiotics for different durations. Then, we employed non-metric multidimensional scaling (NMDS) to reveal the dissimilarities in microbial composition on OTU level. As shown in Fig 2A, control group samples (green dots) accumulated on the lower right area, whereas most of antibiotics group samples (red triangles) were distributed on the lower left side, implying considerable discrepancy between two groups. Next, linear discriminant analysis effect size (LEfSe) method was employed to figure out the exact species with significant difference between two groups. As illustrated in Fig 2B, there were almost 168 species with significant variations. Specifically, p_Proteobacteria, c_Gammaproteobacteria, o_Enterobacteriales and other 32 species had statistical difference in antibiotics group. In addition, species including p_Bacteroidates, f_Prevotellaceae, f_Lachnospiraceae and other 130 species were the noticeable varied bacteria in control group. Taken together, broad-spectrum antibiotics are capable of disrupting intestinal microbiota composition.

Based on NMDS and LEfSe analysis, we further utilized wilcoxon rank sum test to figure out potential bio-markers for antibiotics treatment. As shown in Fig 3A, the specific differential bacteria in control group were norank_f_Bacteroidales_S24-7_group (p<0.001), Prevotellaceae_UCG-001 (p<0.001), Lachnospiraceae_NK4A136_group (p<0.001), Bacteroides (p<0.001), Desulfocibrio (p<0.001), Sutterella (p<0.001) and unclassified_f_Lachnospiraceae (p<0.001). However, the specific differential species in antibiotics group were Klebsiella (p<0.001), Parasutterella (p = 0.009), Ureaplasma (p = 0.03), Morganella (p = 0.002), unclassified_f_Enterobacteriaceae (p = 0.01), unclassified_f_Ruminococcaceae (p<0.001). Thus, broad-spectrum antibiotics could produce some potential biomarkers which may have important biological significance.

To address the correlation between specific bacteria species (top 50 species in abundance) and environmental factors including SCFAs’ concentrations and intestinal permeability, we employed Spearman correlation analysis on Phylum level. As shown in Fig 3B, intestinal permeability was negatively correlated with Verrucomicrobia, Deferribacteres, Bacteroidetes and Saccharibacteria. All SCFAs investigated in this study, including acetate, propionate and butyrate, were positively correlated with Deferribacteres, Bacteroidetes, Saccharibacteria, Cyanobacteria, Firmicutes and Spirochaetae, and also negatively correlated with Proteobacteria and Tenericutes. Besides, both acetate and butyrate were positively correlated with Verrucomicrobia. In brief, there are so many species show correlation with SCFAs, acting as the possible explanations for several phenomenon which will be described below.

Furthermore, we conducted function analysis of intestinal microbiota in order to figure out the possible functions of intestinal microbiota after antibiotics treatment. We employed cluster of orthologous groupe (COG) and Kyoto Encyclopedia of Genes and Genomes (KEGG) to analyze the possible functions and metabolic pathways respectively. As shown in Fig 4A, after antibiotics treatment, the principal functions of species were associated with transport and metabolism of both amino acids and carbohydrates, production and conversion of energy, and others. However, as illustrated in Fig 4B, when compared with that of control group, the relative abundance of species whose functions are associated with transcription, the transport and metabolism of amino acid, carbohydrate and inorganic ion were remarkably increased, whereas the relative abundance of species whose functions are related to cell membrane biogenesis, defense mechanisms, replication, recombination and repair were markedly decreased. Furthermore, as illustrated in Fig 4C, the abundance of species which function in biosynthesis increased to 2.57 folds of control. Meanwhile, the abundance of species associated with fatty acid metabolism also augmented to 5.75 folds of control. Taken together, antibiotics induce significant alterations in intestinal microbiota homeostasis.

It has been demonstrated that SCFAs are the principal byproducts of intestinal microbiota [19]. Therefore, we next investigated whether antibiotics alter intestinal SCFAs concentrations in healthy mice. As shown in Fig 4D, concentrations of all detected SCFAs, including acetate, propionate and butyrate, were significantly decreased after antibiotics treatment, especially on 14th day. Thus, broad-spectrum antibiotics could remarkably down-regulate intestinal SCFAs concentrations, which might be the potential contributor to pathological alterations induced by antibiotics.

It has been reported that intestinal microbiota and its derived SCFAs are crucial to intestinal barrier integrity [20, 21]. Thus, we further investigated the effect of antibiotics treatment on intestinal tight junction barrier in healthy mice. As shown in Fig 5A, when compared with that of control (0 day), intestinal permeability to 4.4kDa FITC-dextran was significantly increased after antibiotics treatment. Meanwhile, we evaluated the expressions of ZO-1, occludin, and claudin-1, which are important members of tight junction proteins. As illustrated in Fig 5B, when compared with those of control, the expressions of ZO-1, occludin and claudin-1 were gradually decreased after antibiotics treatment. Next, we assessed the morphological changes of ZO-1 by immunofluorescent assay. As shown in Fig 5C, in ileum of control mice, ZO-1 was localized to epithelial tight junction, which was appreciated as red spots lying at the apical compartment of cell-cell junctions. In contrast, this ordered appearance of ZO-1 was disrupted in ileum of antibiotics treated mice. Taken together, it is indicated that broad-spectrum antibiotics impairs intestinal tight junction barrier, as evidenced by increased intestinal paracellular permeability, reduced expressions of tight junction proteins and disrupted localization of ZO-1.

It has been reported that microbiota interact with NLRP3 inflammasome in intestine [22, 23], and that NLRP3 inflammasome could remodel intestinal microbiota [24]. Therefore, we next determined whether the intestinal microbiota dysbiosis induced by antibiotics was accompanied by the activation of NLRP3 inflammasome. Once activated, NLRP3 recruited the adaptor protein ASC and pro-caspase-1 to form a multiprotein complex termed inflammasome, and then stimulates the secretions of IL-1β and IL-18^[25^]. As illustrated in Fig 6A, expressions of NLRP3, ASC, caspase 1, cleaved caspase 1, IL-1β, cleaved IL-1β and IL-18 were significantly increased at different time points after antibiotics treatment, as compared with control. Consistent with these, immunofluorescence assay also revealed the amounts of red punctate, representing ASC protein, were evidently increased after antibiotics treatment, as shown in Fig 6B. Taken together, the results indicate that NLRP3 inflammasome is markedly activated by antibiotics treatment, which may contribute to antibiotics-induced intestinal barrier dysfunction.

Intestinal microbiota such as Fusobacterium nucleatum could regulate autophagy [26], therefore, we attempted to investigate whether autophagy was activated by intestinal microbiota dysbiosis caused by antibiotics. Microtubule-associated protein light chain (LC3), which is lipidated and incorporated into autophagosomal membrane, is widely used as a marker for autophagy [27]. Besides, autophagy substrates including P62 and Atg5 are also recognized as the markers for autophagy, because P62 is an adaptor protein which binds to LC3, and Atg5 facilitates the phosphatidylethanolamine conjunction of LC3^[28^,29^]. As shown in Fig 7A, both LC3-II/LC3-I ratio and Atg5 expression were remarkably increased after antibiotics treatment. Accordingly, P62 expression was evidently reduced. In parallel to these, fluorescent assay demonstrated that the amount of red punctate representing LC3B augmented markedly at different time points after antibiotics treatment, as revealed in Fig 7B. These results indicate that autophagy is activated by intestinal microbiota dysbiosis induced by antibiotics, which may contribute to the occurrence of intestinal barrier dysfunction in antibiotics treated mice.

All relevant data are within the manuscript.