Background

Human immunodeficiency virus (HIV) was firstly isolated from lymph node in 1983, which caused acquired immune deficiency syndrome [1]. At present, HIV infection has been globally pandemic, with near 39 million people living with HIV [2]. In China, the prevalence of new HIV infection was increasing [3], especially among men who have sex with men (MSM) [4]. Mechanically, HIV interacts with receptor CD4 and coreceptor CCR5 to entry T cells, allowing the virus to replicate [5]. Exposure to the semen or mucosal surface were the most common route of HIV transmission. However, as reported, the probability of HIV transmission was only 0.01%-0.4% per sexual contact [6], suggesting that HIV infection also depends on the interaction between the virus and host. Host immune activation might facilitate HIV transmission and infection. Research showed that higher levels of inflammatory factors in male penile, including IL-6 and IL-10, contributed to HIV virus shedding [7]. Besides, a nested case-control study indicated that the rate of HIV seroconversion was significantly higher in women with higher genital inflammatory cytokines [8]. In MSM, HIV mainly transmitted through anal intercourse, thus the immune barrier of rectal mucosa plays important role in the early process of HIV infection.

Gut microbiome takes part in constituting gut immune barrier and could regulate functions of small intestinal innate lymphoid cells [9]. Recent studies found that HIV infection was associated with gut microbiome dysbiosis [10,11,12], with decreased α-diversity, enrichment of genus Prevotella and depletion of Bacteroides in HIV patients. Furthermore, in a prospective study gut microbiome dysbiosis was demonstrated to contribute to increased HIV susceptibility in MSM [13]. The relative abundance of family Succinivibrionaceae, Erysipelotrichaceae and Coriobacteriaceae were significantly higher while Bacteroidaceae and Rikenellaceae were significantly depleted in HIV sero-converters. The gut microbiome dysbiosis in HIV patients were associated with immunologic response to antiretroviral therapy [14], activation of T cells [15] and increased inflammatory factors [16]. However, the association between HIV infection and gut microbiome remains controversial. After stratifying sexual preference, HIV status was reported to have subtle effects on gut microbiome [17, 18]. Meanwhile it’s also unclear whether gut microbiome dysbiosis was induced by HIV infection or prompting to the virus infection [19]. The causal relationship between gut microbiome and HIV infection was remained to be elucidated.

Mendelian randomization (MR) is an analytic approach that utilizes genetic variants as instrumental variations to assess the causality between exposure and outcome [20]. With the help of genome-wide association studies (GWAS), MR analysis is widely used to explain the causal effects of an interested exposure on outcomes, which overcomes the potential influence of confounding factors in observational studies [21]. To our knowledge, no studies have documented the causal effects of gut microbiome on HIV infection before. Hence, this study aimed to explore the causal associations between gut microbiome and HIV infection through a two-sample mendelian randomization analysis.

Methods

GWAS summary data source

The GWAS summary data of gut microbiome was retrieved from the MiBioGen consortium. MiBioGen is a large-scale genome-wide meta-analysis to study the association between gut microbiome and human genetic variants, which contains 18 340 samples from 24 cohorts from the USA, Canada, Israel, South Korea, Germany, Denmark, the Netherlands, Belgium, Sweden, Finland and the UK [22]. The microbiome data were generated by 16S rRNA sequencing using an Illumina sequencing platform. Multiple hyper-variable regions of the 16S rRNA gene was sequenced, including V1-V2, V3-V4 and V4 region. After adjustment for age, sex, technical covariates and genetic principal components, 211 gut microbiome taxa had GWAS summary statistics, including 9 phyla, 16 classes, 20 orders, 35 families and 131 genera. Fifteen unknown bacteria were excluded, and 196 gut microbiotas were enrolled for the final analysis. The GWAS summary data for HIV infection were collected from the 5th release of the FinnGen consortium in May 2021, composed of 357 HIV infected cases and 218 435 controls [23]. We selected the gut microbiome as the exposure and the HIV infection as the outcome.

Instrumental variables selection

To indicate causal association between gut microbiome and HIV infection, we selected SNPs that were associated with gut microbiome at the level of P < 1× 10 -5 as instrument variables (IVs). This value was identified as the optimal threshold in many gut microbiota-related MR research to increase the amount of genetic variance explained by the genetic predictors [24].The linkage disequilibrium (LD) across SNPs were calculated in a 10,000-kb window and the LD threshold for R2 was 0.001. In the condition of no SNPs shared between gut microbiome and HIV infection, proxy SNPs with r2 > 0.8 were selected to replace the original SNPs. Then F-statistics was calculated for each SNP with the following formula: F = β2/Se2, whereas β and Se are the coefficient and stand error of exposure respectively. SNPs with F-statistics < 10 were regarded as weak IVs and were discarded in the following analysis. MR pleiotropy residual sum and outlier (MR‐PRESSO) tests was also conducted to remove outlier SNPs.

Mendelian randomization analysis

Five methods were used to perform the MR analysis, including inverse variance weighted (IVW), weighted mode, MR-egger, weighted median, and simple mode. IVW uses a meta-analysis approach combined with the Wald estimates for each SNP to obtain an overall estimate, which assumes that all SNPs are valid instruments [25]. The weighted mode produces a robust causal effects estimate when the horizontal pleiotropy existed [26]. ME-egger could provide a valid effect on the condition that all SNPs are invalid [27]. The estimates based on weighted median are consistent even when more than 50% of the information comes from invalid instrumental variables [28]. Simple mode is an unweighted causal estimation [29]. The primary method was IVW, and only when the result of IVW analysis was significant, the causal effects of exposure on outcome presented. Other four MR methods were used as supplement. Additionally, false discovery rate (FDR) was calculated based on the Benjamini-Hochberg (BH) method to adjust the multiple tests. FDR < 0.1 were considered significant [30].

Sensitivity analysis was conducted to evaluate the stability of the results. Cochran’s Q statistic was used to test the heterogeneity. MR egger intercept was used to assess the horizontal pleiotropy. P value < 0.05 indicated no significant heterogeneity or horizontal pleiotropy. Besides, leave-one out analysis was conducted to identify whether the overall estimates were influenced by one single SNP. To further exclude the impact of confounding factors, such as behavior and environment, we searched the Phenoscanner database (http://www.phenoscanner.medschl.cam.ac.uk/) to explore the reported traits that were associated with our selected IVs. The study flowchart was stated in Fig. 1.

Fig. 1
figure 1

The flowchart of this mendelian randomization study

Full size image

All the analysis were performed by R software (version: R 4.3.0). R packages “TwosampleMR” was used to conduct the two-sample mendelian randomization analysis. And R package “MRPRESSO” was used to identify the outlier SNPs.

Results

Causal effects of gut microbiome on HIV infection

At the level of p < 1 x 10-5, 2 774 SNPs were associated with gut microbiota at phylum, class, order, family, and genus levels. Fifty-three SNPs were excluded due to the F-statistic < 10. Thus, a total of 2 721 SNPs were enrolled for MR analysis. Fifteen bacteria were found to have causal effects on HIV, including 1 class, 2 orders, 3 families and 9 genera (Table 1). One genus of phylum Actinobacteria was negatively related to HIV infection. One class, 2 orders, 3 families and 6 genera of phylum Firmicutes were significantly associated with HIV infection. One genus of phylum Lentisphaerae and 1 genus of phylum Methanomada were also showed causal effects on HIV infection.

Table1 Causal effects of gut microbiome on HIV infection
Full size table

At the class level, 13 SNPs were selected for genetically predicting Erysipelotrichia. Based on IVW analysis and weighted median analysis, Erysipelotrichia was protective factor of HIV infection (IVW: OR: 0.278[0.106, 0.731], P < 0.01; weighted median: OR: 0.161[0.045, 0.575], P<0.01). Order Bacillales and Erysipelotrichales, both belonged to Firmicutes, were significantly associated with HIV. The results of IVW analysis indicated that both Bacillales and Erysipelotrichales were protective factors of HIV (P<0.05). Three families of Firmicutes, Defluviitaleaceae, Erysipelotrichaceae and Ruminococcaceae had causal effects on HIV infection. Defluviitaleaceae and Erysipelotrichaceae were significantly associated with decreased risk of HIV infection, while Ruminococcaceae was identified as risk factor of HIV infection in IVW and weighted median analysis (IVW: OR:2.468[1.043,5.842], P=0.04; Weighted median: OR:3.365[1.045, 10.836], P=0.04.). At the genus level, Eggerthella, Anaerofilum, Anaerotruncus, Clostridium, Coprococcus2, Fusicatenibacter and Methanobrevibacter were protective factors. Hungatella, Ruminococcaceae UCG005, Subdoligranulum and Victivallis were risk factors of HIV infection. After adjusting for the multiple testing, genus Eggerthella and genus Subdoligranulum were found to significantly decrease the risk of HIV acquirement (FDR < 0.1). All these results indicated that gut microbiome had significant causal effects on HIV infection.

Among all these fifteen gut microbiotas, the results of Cochran’s Q test indicated no significant heterogeneity existed. However, horizontal pleiotropy was found for genus Fusicatenibacter (P<0.05) according to the MR-Egger regression intercept analysis. Leave-one out sensitivity analysis suggested that the causal association between gut microbiome and HIV infection were not driven by one single SNP (Fig. 2, Figure S1). Besides, we searched for the Phenoscanner database to explore whether potential confounding factors existed. The results (Table S1) showed that our selected IVs were associated with a few physical indicators, such as impedance of arms, body mass index, arm fat mass, weight, height and so on. These factors were known as influencing factors of gut microbiome composition, but were not reported to affect HIV infection, which demonstrated that the causal association between gut microbiome and HIV was robust.

Fig. 2
figure 2

Leave-one-out analysis of the significant causal effects of gut microbiome on HIV infection. The X-axis indicates the estimated β value. In each panel, the red line stands for the overall estimates, and each blank line indicates the overall estimate after excluding the left SNP. A: class Erysipelotrichia, B: family Erysipelotrichaceae, C: family Ruminococcaceae, D: family Defluviitaleaceae, E: order Erysipelotrichales, F: order Bacillales, G: genus Ruminococcaceae UCG005, H: genus Methanobrevibacter, I: genus Eggerthella

Full size image

Discussion

The present study was the first MR analysis to evaluate the causal relationships between gut microbiome and HIV infection. We demonstrated causal effects of 15 gut microbiotas on HIV infection. In our study, we found that family Ruminococcaceae, and genus Hungatella, Ruminococcaceae UCG005, Subdoligranulum and Victivallis were significantly associated with HIV infection, which might facilitate HIV infection. Genus Clostridium, Coprococcus2 and Fusicatenibacter were identified as protective factors against HIV infection. Typically, most of our reported gut microbiotas were belonged to Firmicutes phylum. Importantly our results stated causality between HIV infection and family Erysipelotrichaceae and genus Methanobrevibacter, which was consistent with one prior study [13], further demonstrating that gut microbiome dysbiosis had major impact on HIV incidence. After adjusting for the multiple testing, genus Eggerthella and genus Subdoligranulum were found to significantly decrease the risk of HIV acquirement, indicating robust causal effects of gut microbiome on HIV infection.

In the field of HIV infection, gut microbiome has aroused researchers’ interests. Researchers found that the composition of gut microbiome significantly changed in HIV-infected patients [18]. To explain the mechanism how gut microbiome affect HIV infection, researches have focused on exploring the association between gut microbiome and immune. In MSM, the high-risk population of HIV infection, gut microbiome can influence the CCR5 expression and influx of HIV targeted cells to colon through modulation of integrin and chemokines expression on T cells [31]. Using gnotobiotic mice model, researchers found that gut microbiome from high-risk MSM activated T cells, with higher frequency of CD69+ and CD103+ T cells [14]. In vitro assays, gut microbiome from HIV-infected patients could induce significantly higher levels of activated monocytes and T cells, which were mediated by TNF-α and TLR (toll like receptor) [16]. Further study revealed that Holdemanella biformis was the main signature that was responsible for elevated frequency of CCR5+ CD4+ T cells in the vitro stimulation assay of Lamina Propria Mononuclear Cells with fecal bacterial communities [32]. All these evidences demonstrated gut microbiome had the potential of increasing HIV transmission through activating immune reaction.

In the present study, family Erysipelotrichaceae decreased risk of HIV infection (OR: 0.28, P < 0.05), which means that higher the relative abundance of Erysipelotrichaceae protects patients against HIV invasion. As reported members of this bacterial were highly immunogenic, for example Erysipelotrichi was reported to correlate with TNF in chronic HIV infection patients [33]. Besides, Erysipelotrichaceae has been found to associate with inflammatory bowel diseases (IBD) and colon cancers [34]. Actually, Erysipelotrichaceae is a kind of butyrate-producing bacteria [35]. Butyrate is the primary energy source of colon epithelial cells and plays important role on ameliorating mucosal inflammation and oxidative status, reinforcing the epithelial defense barrier to maintain intestinal health [36]. Moreover, butyrate is a well-known histone deacetylase inhibitor and could regulate gene expression through modulating chromatin structure, indicating its potential therapeutic implication for clinical use [37]. Studies documented that oral butyrate supplementation in patients with IBD activated epithelial peroxisome proliferator-activated receptor-γ (PPAR-γ) signaling and promoted generation of Treg cells in colonic lamina propria [38]. Animal experiment also demonstrated that oral butyrate downregulated inflammation in mice implanted with polyether-polyurethane sponge, presented with decreased neutrophil infiltration and reduced levels of TNF-α, IL-10 and TGF-β1 [39]. All these evidences shed light on the hypothesis that Erysipelotrichaceae decreased the risk of HIV infection through enhancing the production of butyrate which inhibited excessive activation of immune system. Meanwhile, Methanobrevibacter was also showed protective effects on HIV infection in our analysis. Methanobrevibacter belongs to methanogens and plays important physiological role in human health by producing methane from hydrogen [40]. Study indicated that Methanobrevibacter smithii colonized in the human gastric mucosa just early after birth [41]. Methanobrevibacter was documented to be associated with irritable bowel diseases and possibly reversed the susceptibility of irritable bowel diseases [42, 43]. However, further research is needed on the specific relationship between Methanobrevibacter and host health.

Moreover, we also found Subdoligranulum and Victivallis were positively related to HIV infection. Consistent with previous studies, Subdoligranulum and Victivallis were enriched in HIV infected patients, which further supporting our results [11, 44]. Interestingly, contrary to Erysipelotrichaceae, butyrate-producing bacteria Ruminococcaceae was demonstrated to increase the risk of HIV infection in our study. This conclusion seems to be contradictory, but in fact it implies the complex interaction between gut microbiome and suggests that the effects of gut microbiome on HIV infection are not simply caused by single microbiota. This also indicates that the structure of gut microbiome is in dynamic balance, and only when the balance is broken will it affects health of the host.

What is worth to note is that genus Prevotella was not statistically correlated with HIV infection in this MR analysis. However, Prevotella was the most frequently reported HIV associated genus and might promote to virus infection. For example, researchers found that the abundance of Prevotella was higher in women with high-risk HPV infection and associated with NF-KB signaling, suggesting its role in promoting virus infection by altering immune regulators [45]. Besides, the abundance of Prevotella was associated with immune response, which could drive Th17-mediated mucosal inflammation and stimulate epithelial cells to produce inflammatory factors, such as IL-8 and IL-6 [46]. The impact of genus Prevotella on HIV infection remains to be further elucidated.

To be honest, there are limitations in this study. Firstly, the results of reverse MR analysis exploring causal effects of HIV infection on gut microbiome was not stated here. In fact, we indeed conducted the reverse MR analysis and attempted to explore the causal effects of HIV infection on gut microbiome. However, there are no SNPs available as instrumental variables in the currently used summary GWAS data of HIV infection. On the other hand, we also tried using the latest release HIV GWAS summary data of FinnGen consortium in May 2023, but still lack of available instrumental variables. Secondly, we recommend to treat the significant results carefully. We adjusted the P values based on BH method to avoid false discovery and only two taxa passed the BH correction (FDR < 0.1). More studies are expected to elucidate the role of gut microbiome in HIV infection.

Conclusions

In conclusion, this study analyzed the causal association between gut microbiome and HIV infection by a two-sample MR analysis. Our results demonstrate that several gut taxa increase the risk of HIV acquirement, such as Ruminococcaceae, Subdoligranulum and Victivallis. While Erysipelotrichaceae and Methanobrevibacter significantly reduce the risk of HIV infection. We hold the concept that the risk of HIV infection may not be significantly increased by one single microbiome, it is the results of complex interaction of variable gut microbiome. These findings facilitate future studies to develop better strategies for HIV prophylaxis through gut microbiome regulation. Further explorations are also warranted to dissect the mechanism of how gut microbiome affects HIV susceptibility.