2.1 Subject recruitment

Thirty-three RA patients that fulfilled the inclusion criteria of the American College of Rheumatology (ACR) 2010 were recruited in the Traditional Chinese Medicine Hospital of Ili Kazak Autonomous Prefecture in Yi’ning, China. Thirty-two healthy controls (HCs) were also enrolled. Patients with long disease duration (12–48 months) and routinely received methotrexate treatment were included. Any participants having a personal history of other autoimmune illness, or taking antibiotics, probiotics and prebiotics within three months were excluded. All the collected fecal samples were for public health purposes and were the only human materials used in the present study, and fecal sample collection involved no risk of predictable harm or discomfort to the participants. Written informed consent regarding fecal sample use was obtained from all participants. All procedures performed in study involving human participants was approved by the Ethics Committee in Jiangnan University, China (SYXK 2012–0002).

2.2 Animal experimental designed

Six-week-old female Wistar rats (Beijing Vital River Laboratory Animal Ltd, Beijing, China) were kept in SPF conditions. After one week adaptation, rats were randomly divided into control, model, propionate and butyrate groups (n = 8 per group). The rats in propionate or butyrate groups were respectively given 150 mM propionate or butyrate in drinking water (freshen every day), and rats in control and model groups were given sodium-matched water. The CIA induction was according to the previous described method (Brand et al., 2007). Briefly, the rats were subcutaneously injected in the base of tail with 150 μg bovine type II collagen (Chondrex, Redmond, WA, USA) emulsified in incomplete Freund adjuvant (Chondrex, Redmond, WA, USA) at day 14 and 21, respectively. The thickness of hind paws was evaluated for swelling, and each paw was scored according to the previous description (Brand et al., 2007). Arthritis assessments were performed every three days until no more severe swollen throughout the trials. The animal experiment was approved by the Ethics Committee of Jiangnan University (JN. No. 20191115 W0640201 [313]).

2.3 Fecal sample collection, DNA extraction and sequencing

Fecal samples from participants and rats were collected and stored at −80 °C until further analysis. Total DNA was extracted using the FastDNA Spin Kit for Feces (MP Biomedical, Irvine, CA, USA). The V3-V4 region of the 16S rRNA gene, Bifidobacterium and Lactobacillus groEL genes was amplified as previously described (Hu et al., 2017; Yang et al., 2018; Xie et al., 2019). Amplifications were purified and sequenced on the Illumina MiSeq platform (Illumina, San Diego, CA, USA) according to the manufacturer’s instructions.

2.4 Bioinformatic analysis

The sequence reads were processed as previously described (Hu et al., 2017). Taxonomic identification of OTUs for Lactobacillus and Bifidobacterium groEL genes was performed as previously described (Hu et al., 2017; Xie et al., 2019). To overcome the random sequencing errors, high sequencing coverage depth threshold for V3-V4, Bifidobacterium_groEL and Lactobacillus_groEL genes were 10000, 4500 and 4500 reads, respectively. The 16S rRNA gene sequencing data of humans and rats were respectively deposited in SRA database: PRJNA646930 and PRJNA647234.

2.5 Fecal SCFAs measurement

The SCFAs in dried fecal samples were extracted, and then measured using gas chromatography (GC-2010 Plus, Shimadzu Corporation, Japan) as previously described (Wang et al., 2017).

2.6 Statistical analyses

Data were presented as means ± SEM. Statistical analyses were performed using GraphPad Prism 8 (Graphpad Software Inc., San Diego, CA). Mann-Whitney test was used to compare the differences between two groups, and a one-way ANOVA with Turkey post-hoc test was performed between four groups. Permutational multivariate analysis of variance (PERMANOVA) using 999 permutations was performed to determine the significance in PCoA using R software (v 3.5.6). Spearman correlated analysis was used to evaluate the relationship between SCFAs and gut microbiota. Statistical significance was defined as *p < 0.05, **p<0.01, ***p<0.001. All data were visualized in R (v 3.5.6).

3.1 The diversity and compositions of gut microbiota in MTX-RA patients

Based on the 16S sequencing results, there were significant differences in beta diversity when calculated unweighted UniFrac distance among HCs and MTX-RA patients (Fig. 1A-B). Bacteroidetes, Firmicutes, Proteobacteria and Actinobacteria were the main phylum (Fig. 1C). Specifically, the relative abundance of Firmicutes tended to be depleted (p = 0.06) while Actinobacteria was significantly enriched (p = 0.017) within the MTX-RA cohort. Phylum-specific Bray-Curtis distances suggest that the taxonomic structure of Actinobacteria and Proteobacteria in MTX-RA patients differed from that in the HCs group (Fig. 1D). And unclassified Ruminococcaceae, unclassified Enterobacteriaceae, unclassified Lachnospiraceae, Blautia, Bifidobacterium and Streptococcus were the major genera in the two groups (Fig. 1E). Compared with the HCs group, five genera exhibited significant differences and Haemophilus and Megamonas tended to be decreased in MTX-RA patients (p = 0.058 and p = 0.076, respectively) (Fig. 1F).

Close

Fig. 1

The differences of gut microbiota in MTX-RA and HCs. The alpha-diversity (A) and beta-diversity (B) illustrated by PCoA on weighted (left panel) and unweighted (right panel) UniFrac distances. The average relative abundances of predominant phyla (C) and the top fifteen classified genus (E). (D) PCoA based on the Bray-Curtis distance using OTUs from Firmicutes, Bacteroides, Actinobacteria and Proteobacteria. The selected significantly different genera (F) and differentially abundant taxa (LDA score > 2 and p < 0.05) based on LDA analysis (G). (H) The significantly discriminative predicted metabolic pathways on 16S rRNA gene (LDA score > 2 and p < 0.05). Statistical significance was considered at *p < 0.05.

Export to PPT

Logarithm linear discriminant analysis (LDA) was performed and identified a total of 12 feature taxa responsible for microbial differences among individuals from HC and MTX-RA groups (LDA score > 2, p < 0.05) (Fig. 1G). The MTX-RA patients showed significantly increased feature taxa including Bifidobacteriaceae alongside its genus Bifidobacterium, Christensenellaceae alongside its unclassified genus, and Ruminococcaceae alongside its unclassified genus and Ruminococcus, while no difference was observed for genus Lactobacillus. PICRUSTs was performed to infer the functional genes of the gut microbiota. Three KEGG pathways were significantly enriched in MTX-RA patients including nicotinate and nicotinamide metabolism, folate biosynthesis and fatty acid metabolism pathways (Fig. 1H).

3.2 Diversity of bifidobacterial and lactobacilli communities in HCs and MTX-RA patients

After filtering coverage depth, 61 sequenced samples were used to evaluate the bifidobacterial community profiles in individuals. A total of 1,502,045 reads were acquired with an average of 26,429 and 21,876 reads per sample in MTX-RA patients and HCs, respectively. There were no significant differences in alpha and beta diversity (Fig. 2A,B). The bifidobacterial community was dominated by three typical species (B. longum, B. adolescentis and B. pseudocatenulatum) in both groups (Fig. 2C). No bifidobacterial species changed significantly between MTX-RA and HC groups, as well as among gender-specific groups (Fig. 2D).

Close

Fig. 2

The bifidobacterial and lactobacilli communities in MTX-RA and HCs. The alpha diversity and beta diversity in bifidobacterial community (A,B) and lactobacilli community (E,F). PCoA were based on weighted (left panel) and unweighted (right panel) UniFrac distances (B and E). The average relative abundances of major Bifidobacterium (C,D) and Lactobacillus (G–I) species in MTX-RA and HCs.

Export to PPT

There were 54 qualified sequenced samples after filtering coverage depth at 4500 targeting the Lactobacillus groEL gene. From those samples, 1,020,972 high-quality sequenced reads were identified with an average of 18,702 and 19,144 reads per sample in MTX-RA patients and HCs group, respectively. MTX-RA samples tended to have higher alpha diversity (Fig. 2E). In line with the differences of PCoA on weighted and unweighted UniFrac distance, the major species were L. delbrueckii (20.54%), L. mucosae (17.00%) and L. plantarum (11.92%) in HCs, however, in MTX-RA patients it was dominated by L. mucosae (27.47%), L. salivarius (19.02%) and L. ruminis (12.24%) (Fig. 2F–G). In addition, L. delbrueckii, L. manihotivorans and L. intestinalis were under-represented in the MTX-RA group (Fig. 2H), conversely, L. mucosae and L. salivarius expanded in some MTX-RA patients (Fig. 2I).

3.3 Fecal SCFAs in HCs and MTX-RA patients

Acetate, propionate and butyrate were the major fecal SCFAs in both groups, without significant difference between the two groups (Table 1), though 16S rRNA gene-based predictions indicated that fatty acid metabolism were upregulated in MTX-RA patients (Fig. 1H). However, a significant decline of the propionate/butyrate ratio was observed in the MTX-RA group compared with HCs.

Spearman correlation analyses were performed to identify correlations between fecal SCFAs and the taxa. The results revealed that the abundances of Adlercreutzia and Turicibacter had significant positive correlations with acetate, propionate and butyrate in MTX-RA patients (Fig. 3A). Furthermore, [Ruminococcus] and Bifidobacterium, both of which promote SCFAs production, were inversely correlated with all the SCFAs. The microbe-SCFAs interaction in male MTX-RA patients was distinct from that in female patients. The interactions of some genera with SCFAs in male patients were reversed in the female group, such as Enterococcus, Coprococcus, Dorea, Megamonas and unclassified Christensenellaceae. In line with the insignificant undiminished total SCFAs, no obvious differences were found in SCFA-producing bacteria Lachnospiraceae, Lactobacillaceae, Ruminococcaceae and Bacterioidaceas (Fig. 3B).

Close

Fig. 3

Spearman correlation between SCFAs and gut microbiota. (A) The correlation between the genus (the average abundance > 0.1%) and SCFAs. (B) The abundance of SCFA-producing bacteria at family level. Statistical significance was considered at *p < 0.05, **p < 0.01, ***p < 0.001.

Export to PPT

3.4 Propionate protected arthritis and rebalanced fecal SCFAs in CIA rats

To further address the effect of SCFAs on gut microbiota under arthritis, the CIA rats were orally administered with propionate or butyrate. The swollen symptoms in joint and paws appeared at 14 days after the first immunization, and the three most abundant SCFAs (acetate, propionate and butyrate) were lower in CIA rats. Supplementation with butyrate or propionate showed no differences in remitting swelling and slowing inflammatory progress (Fig. 4A). Administrating of propionate upregulated SCFAs concentration of which butyrate reach a significant level, whereas butyrate supplementing didn’t change any types of SCFAs (Fig. 4B). Comparing with CIA rats, the propionate/butyrate ratio was lower in propionate (p = 0.06) and butyrate (p = 0.14) treated rats.

Close

Fig. 4

The effects of propionate on CIA rats. Thickness of paw and arthritis score (A) and the distribution of fecal SCFAs (B) in rats. Propionate changed the gut microbiota of CIA rats in alpha diversity (C) and beta diversity (D). PCoA were analysed on weighted (left panel) and unweighted (right panel) UniFrac distances (D). Differentially microbiota taxa were shown as cladogram (E) and bar charts (F). Statistical significance was considered at *p < 0.05, **p < 0.01, ***p < 0.001.

Export to PPT

3.5 Propionate modulated gut microbiota in CIA rats

The gut microbiota were analyzed to evaluate the effects of propionate or butyrate on gut microbiota in arthritic rats. Alpha and beta diversity recovered in propionate and butyrate treated rats (Fig. 4C,D). The LEfSe analysis further accounted that Bacteroidetes, containing Prevotella, [Prevotella], Parabacteroides, unclassified [Barnesiellaceae], Rikenellaceae and S24-7, clustered differently, and Bifidobacterium, Allobaculum, unclassified Enterococcaceae and Erysipelotrichacea were enriched in feces of CIA rats compared with other groups (Fig. 4E). Propionate significantly restored the abundance of Bacteroidetes along with genus Parabacteroides and unclassified [Barnesiellaceae], Rikenellaceae and S24-7 (Fig. 4F).

3.6 The modulation of propionate on lactobacilli and bifidobacterial communities in CIA rats

Lactobacillus and Bifidobacterium were enriched in the gut microbiota in CIA rats, therefore, whether propionate or butyrate can regulate the two communities at species level were investigated. The alpha and beta diversity in Lactobacillus community changed slightly (Fig. 5A-B). Comparing with control rats, the abundances of L. taiwanensis (p = 0.03), L. johnsonii (p = 0.56), L. acidophilus (p = 0.22), L. intestinalis (p = 0.24) and L. amylovorus (p = 0.22) were lower in CIA rats (Fig. 5C). Administration of propionate could upregulate the last three species’s abundance, however, butyrate treatment inversely exacerbated the depletion. The decreased abundance of L. taiwanensis and L. johnsonii evoked by arthritis can not be recovered after neither propionate nor butyrate treatment. Simpson index of bifidobacterial community was significantly increased in propionate-treated rats (Fig. 5D). In addition, B. pseudolongum was the dominant bifidobacterial species which accounted for over 99% of bifidobacterial community in all rats (Fig. 5E).

Close

Fig. 5

The effects of propionate on CIA rats on lactobacilli and bifidobacterial community. The indexes of alpha and beta diversity of lactobacilli (A,B) and bifidobacterial community (D) in rats. PCoA based on weighted (left panel) and unweighted (right panel) UniFrac distances (B). The relative abundances of Lactobacillus (C) and Bifidobacterium (E) species in rats. Statistical significance was considered at *p < 0.05, **p < 0.01, ***p < 0.001.

Export to PPT