Clinical characteristics and pyrosequencing data summary
The characteristics of all 28 IR patients, 30 INR patients and 36 healthy controls, including demographics, clinical characteristics, and pyrosequencing results are presented in Table 1. There is no significant difference between the rate of the transmission route in the IR and INR groups (p=0.779). The rate of homosexual (MSM) transmission route is 57.1% vs 51.7% in the INR and IR groups, whereas the rate of the heterosexual transmission route is 20.7% vs 21.4%, and other rates are missing from their records. The viral load of all HIV-infected individuals with ART is not detected. Nadir and current CD4+ T cell counts are significantly higher in the IR group than the INR group (Table 2). No differences in the duration of ART and ongoing ART medications are observed between the IR and the INR groups. Other characteristics such as gender, age and body mass index (BMI) are generally matched among the IR, INR and healthy controls. 3,549,077 high-quality sequences in total were obtained (average sequence length 440 bp) from 94 participants. 38,849 sequences per sample on average were obtained from the healthy controls, while 35,947 and 38,134 sequences per sample were obtained respectively from the IR and INR patients. Rarefaction was conducted on the OTU (Operational taxonomic unit) table to 30,174 reads per sample to avoid methodological artefacts. Specifically, 609 OTUs are defined in the healthy controls, while 486 OTUs and 567 OTUs in the IR and INR groups are defined relatively at a 97% similarity level. Significant differences of bacterial diversity (Shannon, Simpson, and Sobs), richness (ACE, Chao1) and Good’s coverage are observed among the three groups, while no significant difference are found between the IR and the INR groups. A summary is shown in Table 2.
Compositional analysis of fecal microbiota
Principal coordinate analysis (PCoA) by weighted UniFrac matrices shows obvious differentiation of bacterial communities between the IR and the healthy controls (PERMANOVA, pseudo-F: 8.99, R2=0.13, P =0.001, Fig.1a), the INR and the healthy controls (PERMANOVA, pseudo-F: 8.77, R2=0.12, P =0.001, Fig.1b), while no significant differences are observed between the IR and INR groups (PERMANOVA, pseudo-F: 0.80, R2=0.01, P =0.71, Fig. 1c).
The data of average relative abundance of each bacterial phylum and genus in patients and the healthy controls are showed respectively in the figures (Fig. 2; S1, S2 and S3 Fig) and table (S1 Table). The Wilcoxon rank sum test was used to detect taxa with significant differences in relative abundances among groups (confidence interval method). At the phylum level, Bacteroidetes, Actinobacteria, Tenericutes and Lentisphaerae are more abundant in the healthy controls than in the IR group. The relative abundances of Proteobacteria, Fusobacteria and Saccharibacteria are significantly higher in the IR group (S1 Fig. 2a) than those in the healthy controls. The relative abundances of 11 families are significantly different between the IR and the healthy controls. The relative abundances of 93 genera, including 15 predominant (>1% of the total sequences in either group) and 78 less-predominant genera, are significantly different between the healthy controls and the IR groups. Among the different predominant genera, Lachnoclostridium, Megasphaera, Escherichia-Shigella, Veillonella, Streptococcus, Fusobacterium, and Ruminococcus gnavus are found to be overrepresented in the IR group. The relative abundances of Faecalibacterium, Eubacterium rectale, Alistipes, Subdoligranulum, Bifidobacterium, Roseburia, Ruminococcaceae and Parasutterella are higher in the healthy controls (Fig. 2 and S1 Fig. 2b) than in the IR group. A taxonomy-based bacterial comparison was conducted to define the differences between the healthy controls and the INR groups. At the phylum level, Bacteroidetes, Actinobacteria, Lentisphaerae are more abundant in the healthy controls than those in the INR group, while Proteobacteria, Fusobacteria, Tenericutes, Saccharibacteria and unclassified k norank are more abundant in the INR group than those in the healthy controls (Fig. 2 and S1 Fig. 2c). At the genus level, the relative abundances of 83 genera (including 11 predominant genera) are different between the healthy controls and the INR groups. The relative proportions of Faecalibacterium, Eubacterium rectale, Alistipes, Bifidobacterium, Blautia, Roseburia and Ruminococcaceae are more abundant in the healthy controls than those in the INR group. Parasutterella, Megasphaera, Fusobacterium, and Ruminococcus gnavus are found to be overrepresented in the INR group (Fig. 2 and S1 Fig. 2d). Although there is no significant difference between the IR and the INR group at the phylum level (S1 Fig. 2e), the abundances of 12 genera (including 2 predominant genera) are different between the IR and INR groups. The abundances of the two predominant genera Escherichia-Shigella and Blautia are significantly higher in the IR than those in the INR group (Fig. 2 and S1 Fig. 2f).
In order to identify the key phylotypes responsible for the difference found in distinguishing fecal microbiota of different groups, linear discriminant analysis (LDA) effect size (LEfSe) was performed and a threshold of three on effect size was used. Taxonomic cladograms which represents the microbiota structure and predominant bacteria in the three groups is presented, and the biggest differences between the two communities are presented in S4 Fig.
Comparison of the T-cell activation in the IR and INR groups
As expected, nadir CD4+ T-cell, current CD4+ T-cell counts and CD4/CD8 ratio are lower in the INR group than those in the IR group (p<0.0001). The proportion of CD8+CD57+ T-cell in the INR group is significantly lower than those in the IR group (p<0.001). The proportion level of CD4+ and CD8+ T-cells immune activation (CD4/8+ T-cell by the expression of CD25+, HLA-DR+, and HLA-DR+/CD38+) is similar in the INR and IR groups (Table 1).
Comparison of the bacterial translocation markers and inflammation profiles
Lipopolysaccharide (LPS), which translocates from the gut to the blood stream, is commonly used as the major antigens to drive the chronic immune activation. The level of LPS is significantly higher in the INR group compared with other groups(p<0.0001). However, the soluble immune activation marker sCD14 shows no difference between the groups. Of the 13 markers studied, the level of IL-13 shows no difference among groups while the other twelve markers (IL-2, IL-4, IL-5, IL-6, IL-9, IL10, IL-17A, IL-17F, IL-21, IL-22, IFN-γ and TNF-α) appear to be significantly higher in the INR and IR groups when compared with the healthy controls, but there is no significant difference between the INR and IR groups (Table 1).
Association between fecal microbiota and immune activation
Spearman correlations of the relative abundances of bacteria genera and levels of T-cell activation, inflammation or translocation biomarkers are evaluated (Fig. 3). Interestingly, nadir CD4+ T-cell counts are positively correlated with the abundances of Ruminococcaceae and Alistipes, while current CD4+ T-cell counts are strongly positively correlated with the abundances of Ruminococcaceae and Subdoligranulum. The genus Fusobacterium is negatively correlated with nadir and current CD4+ T-cell. The CD4/CD8 ratio is positively correlated with the genus Faecalibacterium and Ruminococcaceae, but negatively correlates with Escherichia-Shigella. Moreover, the CD8+CD57+ T-cell counts is positively correlated with Escherichia-Shigella but negatively correlates with the genus Ruminococcaceae and Alistipes. The genus Roseburia and Blautia are negatively associated with nadir CD4+ T-cell and positively associated with CD8+CD57+ T-cell counts. Inflammation markers and LPS are positively correlated with the Ruminococcus and Fusobacterium but negatively correlates with the genus Faecalibacterium (Fig. 3).