Skip to main content
Download PDF

Causal relationships between gut microbiota, plasma metabolites, and HIV infection: insights from Mendelian randomization and mediation analysis

Virology Journal volume 21, Article number: 204 (2024) Cite this article

Abstract

Objective

Gut dysbiosis and metabolic abnormalities have been implicated in HIV infection. However, the exact causal relationships among the gut microbiota, metabolites, and HIV infection remain poorly understood. Our study involving Mendelian randomization (MR) and mediation analysis aims to unveil these causalities.

Methods

Genetic instrumental variables for the gut microbiota were retrieved from MiBioGen consortium (n = 18,340). Metabolism-related genetic variants were sourced from the CLSA cohort (n = 8299). GWAS summary statistics for symptomatic HIV infection were derived from the FinnGen study (n = 309,154), and the UK Biobank (n = 208,808). We performed the bidirectional two-sample MR to assess causalities with the inverse-variance weighted (IVW) method as the primary analysis. Moreover, we executed a mediation analysis using two-step MR methods.

Results

Compared to the causal effects of HIV infection on gut microbiota (or metabolites), those of gut microbiota (or plasma metabolites) on the risk of HIV infection were more substantial. Phylum Proteobacteria (OR: 2.114, 95% CI 1.042–4.288, P = 0.038), and genus Ruminococcaceae UCG013 (OR: 2.127, 95% CI 1.080–4.191, P = 0.029) exhibited an adverse causal effect on HIV infection, whereas genus Clostridium sensu stricto 1(OR: 0.491, 95% CI 0.252–0.956, P = 0.036) and family Erysipelotrichaceae (OR: 0.399, 95% CI 0.193–0.827, P = 0.013) acted as significant protective factors for HIV. The salicyluric glucuronide level (OR = 2.233, 95% CI 1.120–4.453, P = 0.023) exhibited a considerably adverse causal effect on HIV infection. Conversely, the salicylate-to-citrate ratio (OR: 0.417, 95% CI 0.253–0.688, P = 0.001) was identified as a protective factor for HIV. We identified only one bidirectional causality between 1-palmitoyl-GPI and HIV infection. Mechanistically, genus Haemophilus mediated the causal effects of three phospholipids on HIV infection risk: 1-palmitoyl-GPI (mediation proportion = 33.7%, P = 0.018), 1-palmitoyl-2-arachidonoyl-GPI (mediation proportion = 18.3%, P = 0.019), and 1-linoleoyl-2-linolenoyl-GPC (mediation proportion = 20.3%, P = 0.0216). Additionally, 5-Dodecenoylcarnitine (C12:1) mediated the causal effect of genus Sellimonas on the risk of HIV infection (mediation proportion = 13.7%, P = 0.0348).

Conclusion

Our study revealed that gut microbiota and metabolites causally influence HIV infection risk more substantially than the reverse. We identified the bidirectional causality between 1-palmitoyl-GPI (16:0) and HIV infection, and elucidated four mediation relationships. These findings provide genetic insights into prediction, prevention, and personalized medicine of HIV infection.

Introduction

Since the first cases of AIDS were reported more than 40 years ago, approximately 84 million people worldwide have been infected with HIV-1, 40 million of whom have died from AIDS-related illnesses [1]. The hallmark of HIV-1 infection is the depletion of CD4+ T cells, rendering the affected individuals susceptible to opportunistic infections and cancers [1]. HIV-1 infection activates both the mucosal and systemic immune systems, which, in turn, cause alterations and translocations of the gut microbiome [2,3,4]. Bacterial 16S ribosomal RNA gene sequencing of gut microbiota from individuals with HIV-1 infection revealed higher levels of Proteobacteria [2, 3] and lower levels of Firmicutes [2]. Additionally, the structure of the Bacteroidetes bacterial community was significantly altered [2, 4]. The intestinal dysbiosis observed in HIV-1 infection is associated with elevated levels of interleukin-1β, interferon-γ, tumor necrosis factor-α, sCD14 [3], and the activation ofT cells and myeloid dendritic cells [2]. Although there is some consistency in research findings as previously mentioned, conclusions about changes in gut microbiota abundance vary to some extent across studies. For instance, the abundance of Barnesiella was found to be elevated in HIV patients in the study by Dinh et al. [3], while it was found to be decreased in the study by Dillon et al. [2]. Notably, various studies on HIV infection have shown that gut microbiota is influenced by confounding factors such as age, race, diet, medications, and individual behaviors [5], resulting in considerable variations in research results.

The interplay between gut dysbiosis and metabolic abnormalities has been implicated in various pathological states. Gut microbiota-derived metabolites impact immune responses in infectious diseases [6], and participate in the development of diabetes mellitus [7] and metabolic syndrome [8]. Moreover, gut dysbiosis facilitates the development of some cardiovascular diseases [9, 10] via various metabolic pathways. Research on the effect of HIV infection on gut microbiota and metabolites is underway. Imidazole-propionate, a gut microbiota-derived metabolite, has been reported to be positively associated with carotid artery atherosclerosis in women with HIV infection [11]. However, the bidirectional crosstalk between gut microbiota and metabolites in the context of HIV infection is not yet fully understood.

Mendelian randomization (MR) is an inference approach to assess causal relationships between exposures and disease outcomes by using naturally occurring genetic variations as instrumental variables (IVs) for exposure factors [12, 13]. Mendelian randomization analysis are based on three core hypotheses: relevance, independence, and exclusion restriction [14]. It is assumed that the IVs are strongly related to the exposure factors (relevance), but not with any confounders (independence), and that the IVs are not directly correlated with the outcome via any pathways other than the exposure of interest (exclusion restriction). As genetic variations (single-nucleotide polymorphisms [SNPs]) are randomly allocated independent of environmental factors, MR can eliminate the influence of confounding factors [12, 15, 16]. Also, MR prevents reverse causality, as genetic variations precede clinical outcomes and disease progression [12, 16]. Compared to randomized clinical trials (RCTs), Mendelian randomization studies avoid issues such as high costs, extensive follow-up time, limited sample sizes, reverse causality, confounding and ethical approval [17]. Thus, MR affords practicality that cannot be attained by clinical research studies, and offers robust evidence of causal effects.

Currently, MR studies on HIV/AIDS are still in the early stages. So far, only two MR studies on HIV infection are available. One of them elucidates the mediating role of DNA methylation in the causal association between cocaine use and HIV severity [18], while the other explores the effects of protein biomarkers on cardiovascular diseases in individuals with HIV [19]. To comprehensively explore the causalities among gut microbiota, plasma metabolites, and HIV infection, we performed a bidirectional two-sample MR using genome-wide association study (GWAS) data. Furthermore, we conducted mediation analyses to clarify the crosstalk between gut microbiota and metabolites in the development of HIV infection.

Methods

Study design

This study used bidirectional two-sample MR to assess the causal relationships among gut microbiota, metabolites, and HIV infection. Sensitivity analysis was used to examine the robustness and credibility of causal estimates. A combined analysis of MR results was conducted using meta-analysis. Furthermore, a mediation model for metabolites (microbiota) to assess the causal effects of microbiota (metabolites) on HIV infection was developed. Figure 1 illustrates the overall design of this study.

Fig. 1
figure 1

Study design. CLSA, the Canadian Longitudinal Study on Aging Cohort; HIV, human immunodeficiency virus; LD, linkage disequilibrium; MAF, minor allele frequency; MR, Mendelian randomization; SNPs, single-nucleotide polymorphisms; UKB, the UK Biobank

Full size image

Data resources

Human gut microbiota-related SNPs were retrieved from the international consortium initiative MiBioGen (https://mibiogen.gcc.rug.nl/menu/main/home), which curated and analyzed data on 16S rRNA microbiome and genotypes from 18,340 European-dominated participants. [20]. Metabolism-related genetic variants were obtained from the GWAS conducted on the Canadian Longitudinal Study on Aging (CLSA) cohort (https://www.ebi.ac.uk/gwas/publications/36635386), which included 1091 plasma metabolites and 309 metabolite ratios analyzed in 8299 participants of European ancestry [21]. The only two publicly available GWAS summary statistics for symptomatic HIV infection were extracted from the FinnGen consortium R7 release data [22] (https://r7.risteys.finngen.fi/phenocode/AB1_HIV) and the UK Biobank (UKB) data [23, 24] (https://www.ebi.ac.uk/gwas/studies/GCST90041717). The FinnGen study included 309,154 samples from the Finnish population. The dataset from the UKB consists of 208,808 samples of European ancestry. This study has been conducted using publicly available GWAS summary statistics. Ethical approval and participant consent were obtained in the original studies [20,21,22,23,24].

IV selection

To select strong IVs for gut microbiota and HIV, given that very few SNPs (for gut microbiota) or no SNP (for HIV) association reached the genome-wide significant threshold (P < 5 × 10–8), a compromised significant level (P < 1 × 10–5 for gut microbiota [12, 25], P < 5 × 10–6 for HIV [18]) was used to extract IVs. European 1000 Genome Project was used to determine the linkage disequilibrium (LD) between all SNPs (clumping window size = 10,000 kb, r2 < 0.001) [12, 26, 27]. SNPs with minor allele frequency (MAF) ≤ 0.01 were excluded [12]. Harmonizing processes were conducted to exclude SNPs for being palindromic with intermediate allele frequencies. Approximated F-statistics were applied to assess the instrument strength and eliminate the bias originating from weak IVs using the following formula [12]:

$$F={R}^{2}\times (n-\kappa -1)/\kappa \times (1-{R}^{2})$$

R2 represents the exposure variance interpreted by the selected SNPs, n represents the sample size, and k represents the number of IVs. SNPs with F-statistics of > 10 were selected [27]. The Steiger filtering analysis was utilized to infer the direction of causality [28]. Prior to conducting the MR analysis, SNPs with a stronger association with the outcome than with the exposure were filtered out using the Steiger test [28].

To select strong IVs for plasma metabolites, we initially chose SNPs that strongly and independently (clumping window size = 10,000 kb, r2 < 0.001; F-statistics > 10; MAF > 0.01) predicted metabolites at genome-wide significance (P < 5 × 10−8). Out of the 1400 metabolites in the CLSA study, only 425 metabolites had strong IVs. As the majority of the metabolites had no or limited (< 3) SNPs at P < 5 × 10−8, a less stringent significance threshold (P < 5 × 10−6) was employed for selecting IVs [29, 30]. SNPs significantly associated with potential confounders were examined and excluded using the PhenoScanner GWAS database [31].

Bidirectional two-sample MR analysis

The pairwise causal effects among gut microbiota, metabolites, and HIV infection were assessed using inverse variance weighting (IVW) as the primary MR analysis method. The other four robust methods (MR-Egger, weighted median, simple mode, and weighted mode) were used for complementary analyses. The MR analyses were carried out by using the “TwoSampleMR” (version 0.5.6) package [32] in R software (version 4.3.1). Statistical significance for the causal effect was considered at P < 0.05.

Sensitivity analysis

To detect the degree of heterogeneity, Cochrane’s Q test was employed [30, 33]. P > 0.05 indicates the absence of significant heterogeneity among the estimates from each SNP. MR-PRESSO and MR-Egger regression were employed to examine the possible horizontal pleiotropy effect [30, 33] (P < 0.05 was judged significant). We further removed the MR results with pleiotropic effects and retained the remaining results for meta-analysis. We performed sensitivity analyses using the R package “MR-PRESSO” (version 1.0) [34] in R software (version 4.3.1).

Combined analysis

We conducted a meta-analysis to combine the MR results deriving from two HIV-related GWAS databases (FinnGen and the UKB). The selection of the meta-analysis model depends on the presence or absence of heterogeneity. In the absence of heterogeneity (P > 0.05 for the Q test and I2 ≤ 50%), the meta-analysis is conducted using a fixed effect model. Conversely, a random effect model is selected for the meta-analysis when the Q-value is significant (P < 0.05) or I2 > 50%, indicating the presence of heterogeneity across studies [35]. We performed the meta-analysis using the R package “meta” (version 6.5-0) [36]. P < 0.05 was considered statistically significant.

Mediation analysis

We conducted mediation analysis using two-step MR to estimate the indirect effect of metabolites (microbiota) as mediators in the causal effects of microbiota (metabolites) on HIV infection. Two-step MR involves the computation of two MR estimates: (1) the causal effect of the exposure on the mediator (β1), and (2) the causal effect of the mediator on the outcome (β2) [13]. The indirect effect of the mediator(s) can then be calculated by the product of these two estimates (β1 × β2) [13]. Here, β3 represents the total effect of exposure on the outcome based on the two-sample MR analysis. The mediation proportion was calculated by dividing the indirect effect by the total effect (\(\frac{\upbeta 1\times\upbeta 2}{\upbeta 3})\) [13, 37]. The Bootstrap method was used to estimate the significance of the product of coefficients [38]. P < 0.05 was considered statistically significant.

Results

Causality between HIV infection and gut microbiota (or plasma metabolites)

Bidirectional MR results on HIV infection and gut microbiota (or plasma metabolites) are reported in Figs. 2 and 3, Supplementary Figures S1–S6, and Supplementary Tables S1–S12. Cochran’s Q statistics revealed no significant heterogeneity across single instrument effects within each database (Supplementary Figures S1, S2, S4 and S6). After removing MR results with pleiotropic effects (P < 0.05 for MR-PRESSO or MR-Egger), we conducted meta-analyses to integrate the results of bidirectional MR analyses based on two HIV-related GWAS summary statistics from the FinnGen consortium and UKB datasets, as shown in Figs. 2 and 3, Supplementary Figures S3 and S5, and Supplementary Tables S3, S6, S9, and S12.

Fig. 2
figure 2

The meta-analysis combining the primary MR analyses (IVW) of the causal effects of HIV infection on gut microbiota deriving from FinnGen and UK Biobank datasets. CI, confidence intervals; meta_hetero_I2, I2 statistic assessing the heterogeneity of meta-analysis; meta_hetero_P, P value of Cochran's Q test of meta-analysis

Full size image
Fig. 3
figure 3

The meta-analysis combining the primary MR analyses (IVW) of the causal effects of gut microbiota on HIV infection deriving from FinnGen and UK Biobank datasets. CI, confidence intervals; meta_hetero_I2, I2 statistic assessing the heterogeneity of meta-analysis; meta_hetero_P, P value of Cochran's Q test of meta-analysis; OR, odds ratio

Full size image

Causal effects of HIV on gut microbiota

The meta-analysis of the primary MR analysis (IVW) yielded evidence supporting the causal effects of HIV infection on nine bacterial taxa (Fig. 2, Supplementary Figure S1, and Supplementary Tables S1–S3). HIV infection could increase the abundance of three phylum Bacteroidetes bacterial taxa, including: phylum Bacteroidetes (β: 0.017, 95% confidence interval [CI] 0.004–0.030, P = 0.01), class Bacteroidia (β: 0.016, 95% CI 0.003–0.029, P = 0.014), order Bacteroidales (β: 0.016, 95% CI 0.003–0.029, P = 0.014), Additionally, an increase was observed in one Proteobacteria phylum bacterial taxon, genus Sutterella (β: 0.026, 95% CI 0.009–0.043, P = 0.003). Conversely, HIV infection could decrease the population of five phylum Firmicutes bacterial taxa, including: class Bacilli (β: − 0.016, 95% CI − 0.029 to − 0.004, P = 0.013), order Lactobacillales (β: − 0.017, 95% CI − 0.030 to − 0.004, P = 0.01), family Lactobacillaceae (β: − 0.024, 95% CI − 0.046 to − 0.001, P = 0.037), family Streptococcaceae (β: − 0.018, 95% CI − 0.031 to − 0.004, P = 0.009), and genus Streptococcus (β: − 0.018, 95% CI − 0.032 to − 0.005, P = 0.006). Class Bacteroidia and order Bacteroidales exhibited identical IV and MR outcomes. We retained the lower bacterial taxonomic level (order Bacteroidales) for subsequent analysis. Overall, HIV infection was shown to increase the abundance of three Bacteroidetes taxa at the phylum (Bacteroidetes), class (Bacteroidia) and order (Bacteroidales) levels, as well as one Proteobacteria bacterial taxon at the genus (Sutterella) level, while decreasing the abundance of five Firmicutes taxa at the class (Bacilli), order (Lactobacillales), family (Lactobacillaceae, Streptococcaceae), and genus (Streptococcus) levels.

Causal effects of gut microbiota on HIV

Meta-analysis using the IVW method supported the causal effects of nine bacterial taxa on the risk of HIV infection (Fig. 3, Supplementary Figure S2, and Supplementary Tables S4–S6). A higher abundance of two phylum Proteobacteria bacterial taxa, specifically:

phylum Proteobacteria (odds ratio [OR]: 2.114, 95% CI 1.042–4.288, P = 0.038), genus Haemophilus (OR: 1.719, 95% CI 1.100–2.685, P = 0.017), as well as two phylum Firmicutes, class Clostridia bacterial taxa: genus Ruminococcaceae UCG013 (OR: 2.127, 95% CI 1.080–4.191, P = 0.029), and genus Victivallis (OR: 1.487, 95% CI 1.028–2.151, P = 0.035) exhibited causal effects on the increased susceptibility of HIV infection. Conversely, a higher abundance of three phylum Firmicutes, class Erysipelotrichia bacterial taxa: class Erysipelotrichia (OR: 0.399, 95% CI 0.193–0.827, P = 0.013), order Erysipelotrichales (OR: 0.399, 95% CI 0.193–0.827, P = 0.013), family Erysipelotrichaceae (OR: 0.399, 95% CI 0.193–0.827, P = 0.013), as well as two phylum Firmicutes, class Clostridia bacterial taxa: genus Clostridium sensu stricto1 (OR: 0.491, 95% CI 0.252–0.956, P = 0.036), and genus Sellimonas (OR: 0.656, 95% CI 0.473–0.908, P = 0.011) were causally associated with a lower risk of HIV infection. Class Erysipelotrichia, order Erysipelotrichales, and family Erysipelotrichaceae exhibited the same IVs and MR results. Thus, we used the lowest taxa (family Erysipelotrichaceae) for the subsequent analysis. Taken together, Proteobacteria was shown to increase the risk of HIV infection at the phylum (Proteobacteria) and genus (Haemophilus) levels, while Firmicutes demonstrated both risk-increasing and risk-reducing effects on HIV infection, with class Erysipelotrichia reducing the risk at the class (Erysipelotrichia), order (Erysipelotrichales), and family (Erysipelotrichaceae) levels, and class Clostridia showing both risk-increasing (Ruminococcaceae and Victivallis) and risk-reducing (Clostridium sensu stricto 1 and Sellimonas) effects on HIV infection.

Causal effects of HIV on plasma metabolites

The results from the meta-analysis regarding the causal effects of HIV infection on 53 plasma metabolites are shown in Supplementary Figure S3 and S4, and Supplementary Tables S7–S9. The findings of the combined IVW results revealed that symptomatic HIV infection was causally correlated with the elevated levels of three metabolite ratios and 18 metabolites. Of the 18 plasma metabolites, 13 had known identities across superpathways (i.e., amino acids [n = 7], vitamins [n = 1], and xenobiotics [n = 5]). The remaining five metabolites were categorized as unknown molecules. The beta of plasma metabolites ranged from − 0.019 (95% CI − 0.038 to − 0.000, P = 0.05) for Inosine to theophylline ratio to − 0.011 for Homoarginine (95% CI − 0.022 to − 0.000, P = 0.045), 1-stearoyl-GPI (18:0) (95% CI − 0.022 to 0.000, P = 0.047), and X-25420 (95% CI − 0.022 to − 0.001, P = 0.034) levels. HIV infection causally decreased the levels of five metabolite ratios and 27 metabolites. These metabolites were implicated in 22 superpathways, encompassing amino acids (n = 4), lipids (n = 11), nucleotides (n = 2), peptides (n = 1), and xenobiotics (n = 4), in addition to one partially characterized molecule and four unknown metabolites. The beta of plasma metabolites ranged from 0.011 (95% CI 0.000–0.022, P = 0.042) for Homostachydrine levels to 0.024 (95% CI 0.009–0.038, P = 0.001) for X-24801 levels.

Causal effects of plasma metabolites on HIV

The meta-analyses that combined the main MR (IVW) for the causality of metabolites on HIV are summarized in Supplementary Figure S5 and S6, and Supplementary Tables S10–S12. The levels of 15 metabolites and four metabolite ratios were found to causally increase the risk of HIV infection. These metabolites were implicated in 13 superpathways encompassing amino acids (n = 4), carbohydrates (n = 1), lipids (n = 6), nucleotides (n = 1), and xenobiotics (n = 1), in addition to two unknown metabolites. The OR ranged from 1.165 (95% CI 1.004–1.352), P = 0.044) for 5alpha-androstan-3alpha, 17beta-diol monosulfate (1) levels to 2.223 (95% CI 1.120–4.453, P = 0.023) for Salicyluric glucuronide levels. The levels of 23 metabolites and four metabolite ratios were found to causally reduce the risk of HIV infection. These metabolites were involved in 20 superpathways of amino acids (n = 4), lipids (n = 11), and xenobiotics (n = 5), as well as three unknown metabolites. The OR ranged from 0.417 (95% CI 0.253–0.688), P = 0.001) for Salicylate/citrate ratio to 0.806 (95% CI 0.652–0.996, P = 0.046) for Andro steroid monosulfate C19H28O6S (1) levels. Notably, only one bidirectional causal relationship was identified between 1-palmitoyl-GPI (16:0) and HIV infection (Supplementary Figure S3–S6). 1-Palmitoyl-GPI (16:0) is a protective factor for HIV infection (OR: 0.653, 95% CI 0.435–0.980, P = 0.04; converted β: − 0.426, 95% CI − 0.832 to − 0.02, P = 0.04, Supplementary Figure S5–S6). However, the causal effect of HIV on 1-palmitoyl-GPI (16:0) is quite small (β: − 0.012, 95% CI − 0.023 to − 0.001, P = 0.031, Supplementary Figure S3–S4).

Causality between gut microbiota and plasma metabolites

Causal effects of gut microbiota on plasma metabolites

Supplementary Figure S7 and Supplementary Table S13 present the 18 causal relationships of gut microbiota on plasma metabolites. The IVW technique outcomes revealed that family Erysipelotrichaceae was suggestively correlated with an elevated level of creatine (β: 0.167, 95% CI 0.001–0.332, P = 0.049) and the ratio of 3-methyl-2-oxovalerate to 4-methyl-2-oxopentanoate (β: 0.193, 95% CI 0.012–0.374, P = 0.036). Genus Clostridium sensu stricto 1causally decreased the levels of 5α-androstan-3α,17β-diol monosulfate (β: − 0.159, 95% CI − 0.317 to − 0.001, P = 0.048), and increased the levels of X-25957 (β: 0.174, 95% CI 0.005–0.343, P = 0.043). Genus Haemophilus demonstrated a causal correlation with the decline of N-acetyl-3-methylhistidine (β: − 0.126, 95% CI − 0.243 to − 0.009, P = 0.034). Genus Ruminococcaceae UCG013 was causally associated with a higher N-acetyl-3-methylhistidine level (β: 0.189, 95% CI 0.007–0.370, P = 0.042) and an increased N-acetyl-2-aminoadipate level (β: 0.183, 95% CI 0.022–0.345, P = 0.026), as well as an increased level of X-12407 level (β: 0.184, 95% CI 0.013–0.355, P = 0.035). Genus Sellimonas was causally linked to a higher 5-dodecenoylcarnitine (C12:1) level (β: 0.111, 95% CI 0.031–0.191, P = 0.006). Genus Victivallis exhibited a causal effect on the elevated levels of 4-hydroxy-2-oxoglutaric acid (β: 0.094, 95% CI 0.015–0.172, P = 0.019). Additionally, phylum Proteobacteria causally increased the levels of salicyluric glucuronide (β: 0.203, 95% CI 0.003–0.404, P = 0.047), N,N,N-trimethyl-5-aminovalerate (β: 0.182, 95% CI 0.023–0.342, P = 0.025), 1-palmitoyl-2-oleoyl-gpc (16:0/18:1) (β: 0.186, 95% CI 0.009–0.362, P = 0.039), and eicosapentaenoate (EPA; 20:5n3) (β: 0.218, 95% CI 0.061–0.376, P = 0.007). In summary, 16 positive causal relationships of gut microbiota on plasma metabolites were identified. Among these, the effect of genus Sellimonas on 5-dodecenoylcarnitine (C12:1) was the weakest (β: 0.111), while phylum Proteobacteria had the most pronounced effect on eicosapentaenoate (β: 0.218). Additionally, two negative causal relationships were found: genus Haemophilus on N-acetyl-3-methylhistidine (β: − 0.126) and genus Clostridium sensu stricto 1 on 5α-androstan-3α,17β-diol monosulfate (β: − 0.159).

Causal effects of plasma metabolites on gut microbiota

The results of IVW analyses revealed the 19 causal effects of plasma metabolites on gut microbiota, as shown in Supplementary Figure S8 and Supplementary Table S14. Andro steroid monosulfate C19H28O6S (1) and sphingomyelin (d18:2/24:1, d18:1/24:2) were positively associated with the abundance of family Erysipelotrichaceae (β: 0.067, 95% CI 0.001–0.134, P = 0.048; β: 0.089, 95% CI 0.003–0.175, P = 0.041). N-acetylthreonine, 1-stearoyl-2-oleoyl-GPE (18:0/18:1), and caprylate (8:0) were causally associated with a higher abundance of genus Clostridium sensu stricto 1 (β: 0.128, 95% CI 0.016–0.239, P = 0.025; β: 0.066, 95% CI 0.004–0.128, P = 0.038; β: 0.146, 95% CI 0.035–0.258, P = 0.01). In addition, (2 or 3)-decenoate (10:1n7 or n8) and the salicylate to citrate ratio were linked to an increase in genus Ruminococcaceae UCG013 (β: 0.110, 95% CI 0.011–0.208, P = 0.03; β: 0.143, 95% CI 0.009–0.278, P = 0.037). 1-Palmitoyl-GPI (16:0), 1-palmitoyl-2-arachidonoyl-GPI (16:0/20:4), and EPA (20:5n3) were negatively associated with the abundance of genus Haemophilus (β: − 0.268, 95% CI − 0.432 to − 0.103, P = 0.001; β: − 0.164, 95% CI − 0.272 to − 0.055; β: − 0.169, 95% CI − 0.325 to − 0.013, P = 0.034), while 1-linoleoyl-2-linolenoyl-GPC (18:2/18:3) was positively linked to the abundance of genus Haemophilus (β: 0.139, 95% CI 0.043–0.236, P = 0.005). 4-Allylphenol sulfate, dimethylglycine and 1-linoleoyl-2-linolenoyl-GPC (18:2/18:3) exhibited a negative correlation with the presence of phylum Proteobacteria (β: − 0.122, 95% CI − 0.224 to − 0.020, P = 0.019; β: − 0.051, 95% CI − 0.102 to − 0.001, P = 0.044; β: − 0.067, 95% CI − 0.131 to − 0.003, P = 0.041). Conversely, the 3-phosphoglycerate to adenosine 5'-diphosphate (ADP) ratio displayed a positive association with the abundance of phylum Proteobacteria (β: 0.106, 95% CI 0.014–0.199, P = 0.025). Overall, 13 positive causal relationships of plasma metabolites on gut microbiota were identified. Among these, the effect of 1-stearoyl-2-oleoyl-GPE (18:0/18:1) on genus Clostridium sensu stricto 1 was the weakest (β: 0.066), while Caprylate (8:0) had the most pronounced effect on genus Clostridium sensu stricto 1 (β: 0.146). Additionally, six negative causal relationships were found: beta from − 0.268 (for 1-palmitoyl-GPI (16:0) on genus Haemophilus) to − 0.051 (Dimethylglycine on phylum Proteobacteria).

Mediation analysis

Using metabolites as a mediator, a mediation relationship displaying statistical significance was established. Genus Sellimonas was found to enhance the plasma level of 5-dodecenoylcarnitine (C12:1) (β = 0.111, 95% CI 0.031–0.191, P = 0.006), which, in turn, increased the risk of HIV infection. The proportion of 5-dodecenoylcarnitine (C12:1) mediation was 13.7% (95% CI = 0.605–32.882%, P = 0.0348) (Fig. 4a, Table S16).

Fig. 4
figure 4

Mediation analysis. a 5-dodecenoylcarnitine (C12:1) mediated the causal effect of genus Sellimonas on HIV infection. b Genus Haemophilus mediated the causal effect of 1-palmitoyl-GPI (16:0) on HIV infection. c Genus Haemophilus mediated the causal effect of 1-palmitoyl-2-arachidonoyl-GPI (16:0/20:4) on HIV infection. d Genus Haemophilus mediated the causal effect of 1-linoleoyl-2-linolenoyl-GPC (18:2/18:3) on HIV infection

Full size image

Furthermore, we found that genus Haemophilus acted as a mediator in three mediation relationships (Fig. 4b–d). 1-Palmitoyl-GPI (16:0) causally decreased the abundance of genus Haemophilus (β = − 0.268, 95% CI − 0.432 to − 0.103, P = 0.001) and was found to be subsequently associated with a declined risk of HIV infection, with a mediation proportion of 33.7% (95% CI = 4.510–74.59%, P = 0.018) (Fig. 4b, Table S15). Likewise, 1-palmitoyl-2-arachidonoyl-GPI (16:0/20:4) reduced the abundance of genus Haemophilus (β = − 0.164, 95% CI − 0.272 to − 0.055, P = 0.003). This decrease was causally associated with a lowered risk of HIV infection, with a mediation proportion of 18.3% (95% CI = 2.162–41.229%, P = 0.019) (Fig. 4c, Table S15). Conversely, 1-linoleoyl-2-linolenoyl-GPC (18:2/18:3) was found to be causally linked to an elevation in the abundance of genus Haemophilus (β = 0.139, 95% CI 0.043–0.236, P = 0.005). This increase was associated with a higher risk of HIV infection, with a mediation proportion of 20.3% (95% CI = 2.079–46.320%, P = 0.0216) (Fig. 4d, Table S15).

Discussion

Causality between HIV infection and gut microbiota

In healthy individuals, the predominant gut microbiota primarily consists of two main phyla—the Bacteroidetes and Firmicutes. These two phyla, along with additional phyla, namely the Proteobacteria, Actinobacteria, Synergistetes, and Fusobacteria, encompass almost all of the bacterial species found in human gut microbiome [39, 40]. This complex assembly of microorganisms plays a crucial role in both host health and disease [41]. Numerous studies have established a correlation between HIV infection and microbiome dysbiosis [2,3,4,5, 11, 42,43,44,45,46,47,48,49,50]. However, due to confounding factors, the results of research show significant variability [5]. MR can eliminate the influence of confounding factors and does not require intervention on participants, affording a practicality unattainable by clinical research studies. In the present study, we first elucidated the causal relationships between symptomatic HIV infection and gut microbiota using MR.

Most of the available data on HIV-related gut dysbiosis indicates an increase in the population of potentially pathogenic phylum Proteobacteria and a decrease in that of beneficial commensals such as phyla Bacteroidetes and Firmicutes [2, 5, 44, 51]. The present study showed that HIV infection could increase the abundance of phylum Bacteroidetes (including class Bacteroidia and order Bacteroidales) and phylum Proteobacteria (including genus Sutterella) and decrease the abundance of phylum Firmicutes (including class Bacilli, order Lactobacillales, family Lactobacillaceae, family Streptococcaceae, and genus Streptococcus). The results regarding phyla Proteobacteria and Firmicutes were consistent with those of previous studies [2, 5, 44, 51]. Interestingly, alterations in the population of phylum Bacteroidetes in individuals with HIV were contrary to that reported in previous reports [2, 5, 51]. This change needs to be further investigated from the perspectives of diversity and composition ratios of gut bacteria.

The majority of current studies are focused on alterations in gut microbiota following HIV infection [2, 3, 11, 42, 44,45,46,47,48,49,50]. Only a few studies have shown that pre-existing pathogenic alteration in the gut microbiome were observed in individuals with HIV-1 infection, indicating the microbiome's influence on HIV susceptibility and the risk of developing AIDS [4, 5]. In the present study, we initially focused on the causal effects of HIV on gut microbiota. After converting the OR into β coefficients [β = ln (OR)], the absolute values of the β coefficients (|β|) for the causal effect of gut microbiota on HIV were found to vary from 0.397 to 0.919, whereas those for the causal effect of HIV on gut microbiota ranged from 0.016 to 0.026. Surprisingly, we found that the causal effects of gut microbiota on the risk of HIV infection are considerably more substantial. The MR results show that phylum Proteobacteria and genus Ruminococcaceae UCG013 exhibited a notable adverse causal effect on HIV infection (with an OR exceeding 2), whereas genus Clostridium sensu stricto 1 and family Erysipelotrichaceae acted as fairly significant protective factors for HIV (with an OR below 0.5). Family Erysipelotrichaceae has been observed to be significantly elevated prior to HIV-1 infection [4], aligning with our study results. However, the impact of the other three taxa on the risk of HIV infection has not been reported. Our study results enrich the current limited understanding of the role of gut microbiota dysbiosis in the risk and development of HIV infection.

Causality between HIV infection and plasma metabolites

The results of MR analyses reveal the causal effects of HIV infection on 53 plasma metabolites. The absolute values of β coefficients (|β|) ranged from 0.011 to 0.024. It is worth noting that |β| for the causal effect of 46 plasma metabolites on HIV infection ranged from 0.153 to 0.875 (after converting the OR into β coefficients). Therefore, the causal effects of plasma metabolites on the risk of HIV infection were more substantial than the impact of HIV infection on metabolites. Meta-analyses conducted using the IVW method showed that the salicyluric glucuronide level exhibited a considerably adverse causal effect on HIV infection with an OR exceeding 2. Conversely, the salicylate-to-citrate ratio was identified as a significant protective factor for HIV with an OR of < 0.5.

Aspirin (acetylsalicylic acid) is a famous drug containing salicylic acid. Salicyluric glucuronide is one of the major metabolites that salicylic acid produces [52]. No convincing evidence is yet available to show that aspirin can affect the risk of HIV infection. There is only one study indicating that low-dose aspirin in HIV-uninfected women may reduce T-cell activities, thus potentially preventing HIV infection [53]. The results of this study suggested that aspirin could be associated with the risk of HIV infection; however, we could not determine whether aspirin was an adverse or a protective factor. Salicylate and citrate share the same transporter using metabolite–protein associations recorded in the Human Metabolome Database (HMDB) [21]. Therefore, the salicylate-to-citrate ratio was calculated in the original research [21]. Citrate plays a significant role in energy metabolism. In the tricarboxylic acid cycle, citrate is metabolized into various metabolic products, releasing energy [54]. Therefore, we hypothesized that energy metabolism influences the metabolic process of salicylic acid, affecting the risk of HIV infection. The role of aspirin and its metabolite composition in HIV infection risk needs to be rigorously investigated for validation.

Only one bidirectional causal relationship has been identified between 1-palmitoyl-GPI (16:0) and HIV infection. 1-Palmitoyl-GPI, also known as 1-hexadecanoyl-sn-glycero-3-phospho-(1'-myo-inositol), is a specific form of glycerophosphoinositol molecules [55]. There have been no studies specifically reporting the correlation between 1-palmitoyl-GPI and HIV. Previous studies have reported that individuals who eventually acquire HIV have slightly higher levels of glycerophosphoinositol compared to matched controls prior to HIV infection [5]. This suggests that glycerophosphoinositol might play a role in the susceptibility or early process of HIV infection. Our findings, however, indicate that 1-palmitoyl-GPI is a protective factor. The discrepancy in these research results may be due to glycerophosphoinositols encompassing various molecular forms within a broader classification framework, some of which may have detrimental effects in HIV infection. Further research is needed to elucidate the role of glycerophosphoinositols in HIV infection.

Mediation relationship

Genus Sellimonas was found to enhance the plasma level of 5-dodecenoylcarnitine (C12:1), increasing the risk of HIV infection. 5-Dodecenoylcarnitine (C12:1) is involved in fatty acid metabolism [21]. Therefore, Genus Sellimonas may reduce the risk of HIV infection via lipid metabolism.

Genus Haemophilus acts as a mediator in three mediation relationships involving phospholipids. Mediation analysis indicated that 1-palmitoyl-GPI (16:0) and 1-palmitoyl-2-arachidonoyl-GPI (16:0/20:4) exhibit protective effects on the risk of HIV infection mediated by a reduction in the population of genus Haemophilus. Conversely, 1-linoleoyl-2-linolenoyl-GPC (18:2/18:3) is causally associated with a higher risk of HIV infection owing to an elevation in the abundance of Haemophilus. 1-palmitoyl-GPI (16:0) and 1-palmitoyl-2-arachidonoyl-GPI (16:0/20:4) are of the phosphatidylinositols (PIs) class; 1-linoleoyl-2-linolenoyl-GPC (18:2/18:3) is of the phosphatidylcholines (PCs) class. PIs and PCs are two major classes of glycerophospholipids (GPLs) that play significant roles in cell-membrane construction and signaling pathways [56]. In general, PCs make up 40–50% of the total phospholipids as a major constituent of the plasma membrane (PM) [57]. PIs account for 10% of the total phospholipids and are distributed broadly across most subcellular organelles, but lack at the PM [58]. Haemophilus influenzae, the most famous pathogen of the genus Haemophilus, is an opportunistic bacterial pathogen. [59]. However, when certain factors (such as viral infections and a weakened immune response) are present, Haemophilus influenzae can cause severe infections [60]. Its polysaccharide capsule and lipo-oligosaccharides are important virulence factors that contribute to resistance against complement-mediated phagocytosis [60], which is closely related to membrane phospholipid dynamics and signaling [61]. Our results indicated that the metabolism and function of certain types of cell-membrane phospholipids may alter the abundance of the Haemophilus genus, thereby influencing susceptibility to HIV infection.

Conclusion

Our study revealed that gut microbiota and metabolites exert causal influence on HIV infection risk more substantially than the reverse. We identified one bidirectional causality between 1-palmitoyl-GPI (16:0) and HIV infection, as well as four mediation relationships. Genus Haemophilus mediated the causal effects of 1-palmitoyl-GPI (16:0), 1-palmitoyl-2-arachidonoyl-GPI (16:0/20:4), and 1-linoleoyl-2-linolenoyl-GPC (18:2/18:3) on HIV-infection risk. Additionally, 5-Dodecenoylcarnitine (C12:1) mediated the causal effect of genus Sellimonas on the risk of HIV infection. These findings contribute to a deeper understanding of the pathophysiology of HIV infection and provide novel insights for primary prediction, targeted prevention, and personalized treatment of AIDS.

Availability of data and materials

The GWAS summary statistics used in this study are publicly available through the international consortium MiBioGen (https://mibiogen.gcc.rug.nl/menu/main/home), the Canadian Longitudinal Study on Aging (CLSA) cohort (https://www.ebi.ac.uk/gwas/publications/36635386), FinnGen consortium R7 release data (https://r7.risteys.finngen.fi/phenocode/AB1_HIV) and the UK Biobank (UKB) data (https://www.ebi.ac.uk/gwas/studies/GCST90041717). All data generated or analysed during this study are included in this published article.

References

  1. Bekker LG, Beyrer C, Mgodi N, Lewin SR, Delany-Moretlwe S, Taiwo B, et al. HIV infection. Nat Rev Dis Primers. 2023;9:42.

    Article  PubMed  Google Scholar 

  2. Dillon SM, Lee EJ, Kotter CV, Austin GL, Dong Z, Hecht DK, et al. An altered intestinal mucosal microbiome in HIV-1 infection is associated with mucosal and systemic immune activation and endotoxemia. Mucosal Immunol. 2014;7:983–94.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  3. Dinh DM, Volpe GE, Duffalo C, Bhalchandra S, Tai AK, Kane AV, et al. Intestinal microbiota, microbial translocation, and systemic inflammation in chronic HIV infection. J Infect Dis. 2015;211:19–27.

    Article  CAS  PubMed  Google Scholar 

  4. Chen Y, Lin H, Cole M, Morris A, Martinson J, Mckay H, et al. Signature changes in gut microbiome are associated with increased susceptibility to HIV-1 infection in MSM. Microbiome. 2021;9:237.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  5. Fulcher JA, Li F, Tobin NH, Zabih S, Elliott J, Clark JL, et al. Gut dysbiosis and inflammatory blood markers precede HIV with limited changes after early seroconversion. EBioMedicine. 2022;84:104286.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  6. Nagata N, Takeuchi T, Masuoka H, Aoki R, Ishikane M, Iwamoto N, et al. Human gut microbiota and its metabolites impact immune responses in COVID-19 and its complications. Gastroenterology. 2023;164:272–88.

    Article  CAS  PubMed  Google Scholar 

  7. Guo Z, Pan J, Zhu H, Chen Z. Metabolites of gut microbiota and possible implication in development of diabetes mellitus. J Agric Food Chem. 2022;70:5945–60.

    Article  CAS  PubMed  Google Scholar 

  8. Feng K, Zhang H, Chen C, Ho C, Kang M, Zhu S, et al. Heptamethoxyflavone alleviates metabolic syndrome in high-fat diet-fed mice by regulating the composition, function, and metabolism of gut microbiota. J Agric Food Chem. 2023;71:10050–64.

    Article  CAS  PubMed  Google Scholar 

  9. Moutsoglou D, Tatah J, Prisco S, Prins K, Staley C, Lopez S, et al. Pulmonary arterial hypertension patients have a proinflammatory gut microbiome and altered circulating microbial metabolites. Am J Respir Crit Care Med. 2023;207:740–56.

    Article  CAS  PubMed  Google Scholar 

  10. Nemet I, Li X, Haghikia A, Li L, Wilcox J, Romano K, et al. Atlas of gut microbe-derived products from aromatic amino acids and risk of cardiovascular morbidity and mortality. Eur Heart J. 2023;44:3085–96.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  11. Wang Z, Peters B, Bryant M, Hanna D, Schwartz T, Wang T, et al. Gut microbiota, circulating inflammatory markers and metabolites, and carotid artery atherosclerosis in HIV infection. Microbiome. 2023;11:119.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  12. Li P, Wang H, Guo L, Gou X, Chen G, Lin D, et al. Association between gut microbiota and preeclampsia-eclampsia: a two-sample Mendelian randomization study. BMC Med. 2022;20:443.

    Article  PubMed  PubMed Central  Google Scholar 

  13. Carter AR, Sanderson E, Hammerton G, Richmond RC, Davey Smith G, Heron J, et al. Mendelian randomisation for mediation analysis: current methods and challenges for implementation. Eur J Epidemiol. 2021;36:465–78.

    Article  PubMed  PubMed Central  Google Scholar 

  14. Emdin CA, Khera AV, Kathiresan S. Mendelian randomization. JAMA. 2017;318:1925–6.

    Article  PubMed  Google Scholar 

  15. Liu K, Cai Y, Song K, Yuan R, Zou J. Clarifying the effect of gut microbiota on allergic conjunctivitis risk is instrumental for predictive, preventive, and personalized medicine: a Mendelian randomization analysis. EPMA J. 2023;14:235–48.

    Article  PubMed  PubMed Central  Google Scholar 

  16. Ren F, Jin Q, Liu T, Ren X, Zhan Y. Causal effects between gut microbiota and IgA nephropathy: a bidirectional Mendelian randomization study. Front Cell Infect Microbiol. 2023;13:1171517.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Zhang Z, Li G, Yu L, Jiang J, Li R, Zhou S, et al. Causal relationships between potential risk factors and chronic rhinosinusitis: a bidirectional two-sample Mendelian randomization study. Eur Arch Otorhinolaryngol. 2023;280:2785–93.

    Article  PubMed  Google Scholar 

  18. Shu C, Justice AC, Zhang X, Wang Z, Hancock DB, Johnson EO, et al. DNA methylation mediates the effect of cocaine use on HIV severity. Clin Epigenet. 2020;12:140.

    Article  CAS  Google Scholar 

  19. Reilly CS, Borges ÁH, Baker JV, Safo SE, Sharma S, Polizzotto MN, et al. Investigation of causal effects of protein biomarkers on cardiovascular disease in persons with HIV. J Infect Dis. 2023;227:951–60.

    Article  CAS  PubMed  Google Scholar 

  20. Kurilshikov A, Medina-Gomez C, Bacigalupe R, Radjabzadeh D, Wang J, Demirkan A, et al. Large-scale association analyses identify host factors influencing human gut microbiome composition. Nat Genet. 2021;53:156–65.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  21. Chen Y, Lu T, Pettersson-Kymmer U, Stewart ID, Butler-Laporte G, Nakanishi T, et al. Genomic atlas of the plasma metabolome prioritizes metabolites implicated in human diseases. Nat Genet. 2023;55:44–53.

    Article  PubMed  PubMed Central  Google Scholar 

  22. Kurki MI, Karjalainen J, Palta P, Sipilä TP, Kristiansson K, Donner KM, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023;613:508–18.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  23. Jiang L, Zheng Z, Fang H, Yang J. A generalized linear mixed model association tool for biobank-scale data. Nat Genet. 2021;53:1616–21.

    Article  CAS  PubMed  Google Scholar 

  24. Sollis E, Mosaku A, Abid A, Buniello A, Cerezo M, Gil L, et al. The NHGRI-EBI GWAS Catalog: knowledgebase and deposition resource. Nucleic Acids Res. 2023;51:D977–85.

    Article  CAS  PubMed  Google Scholar 

  25. Sanna S, van Zuydam NR, Mahajan A, Kurilshikov A, Vich Vila A, Võsa U, et al. Causal relationships among the gut microbiome, short-chain fatty acids and metabolic diseases. Nat Genet. 2019;51:600–5.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Luo J, Xu Z, Noordam R, van Heemst D, Li-Gao R. Depression and inflammatory bowel disease: a bidirectional two-sample Mendelian randomization study. J Crohns Colitis. 2022;16:633–42.

    Article  PubMed  Google Scholar 

  27. Su D, Ai Y, Zhu G, Yang Y, Ma P. Genetically predicted circulating levels of cytokines and the risk of osteoarthritis: a Mendelian randomization study. Front Genet. 2023;14:1131198.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  28. Hemani G, Tilling K, Davey SG. Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLoS Genet. 2017;13:e1007081.

    Article  PubMed  PubMed Central  Google Scholar 

  29. Liu B, Lyu L, Zhou W, Song J, Ye D, Mao Y, et al. Associations of the circulating levels of cytokines with risk of amyotrophic lateral sclerosis: a Mendelian randomization study. BMC Med. 2023;21:39.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  30. Xiang M, Wang Y, Gao Z, Wang J, Chen Q, Sun Z, et al. Exploring causal correlations between inflammatory cytokines and systemic lupus erythematosus: a Mendelian randomization. Front Immunol. 2022;13:985729.

    Article  CAS  PubMed  Google Scholar 

  31. Kamat MA, Blackshaw JA, Young R, Surendran P, Burgess S, Danesh J, et al. PhenoScanner V2: an expanded tool for searching human genotype-phenotype associations. Bioinformatics. 2019;35:4851–3.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  32. Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-base platform supports systematic causal inference across the human phenome. Elife. 2018;7:e34408.

    Article  PubMed  PubMed Central  Google Scholar 

  33. Kintu C, Soremekun O, Kamiza AB, Kalungi A, Mayanja R, Kalyesubula R, et al. The causal effects of lipid traits on kidney function in Africans: bidirectional and multivariable Mendelian-randomization study. EBioMedicine. 2023;90:104537.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet. 2018;50:693–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. Wang R, Ye H, Zhao Y, Wei J, Wang Y, Zhang X, et al. Effect of sacubitril/valsartan and ACEI/ARB on glycaemia and the development of diabetes: a systematic review and meta-analysis of randomised controlled trials. BMC Med. 2022;20:487.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  36. Balduzzi S, Rücker G, Schwarzer G. How to perform a meta-analysis with R: a practical tutorial. Evid Based Ment Health. 2019;22:153–60.

    Article  PubMed  PubMed Central  Google Scholar 

  37. Dai H, Hou T, Wang Q, Hou Y, Wang T, Zheng J, et al. Causal relationships between the gut microbiome, blood lipids, and heart failure: a Mendelian randomization analysis. Eur J Prev Cardiol. 2023;30:1274–82.

    Article  PubMed  Google Scholar 

  38. Serang S, Jacobucci R. Exploratory mediation analysis of dichotomous outcomes via regularization. Multivar Behav Res. 2020;55:69–86.

    Article  Google Scholar 

  39. Hall A, Tolonen A, Xavier R. Human genetic variation and the gut microbiome in disease. Nat Rev Genet. 2017;18:690–9.

    Article  CAS  PubMed  Google Scholar 

  40. Browne H, Neville B, Forster S, Lawley T. Transmission of the gut microbiota: spreading of health. Nat Rev Microbiol. 2017;15:531–43.

    Article  PubMed  PubMed Central  Google Scholar 

  41. Woo A, Aguilar Ramos M, Narayan R, Richards-Corke K, Wang M, Sandoval-Espinola W, et al. Targeting the human gut microbiome with small-molecule inhibitors. Nat Rev Chem. 2023;7:319–39.

    Article  CAS  PubMed  Google Scholar 

  42. Borgognone A, Noguera-Julian M, Oriol B, Noël-Romas L, Ruiz-Riol M, Guillén Y, et al. Gut microbiome signatures linked to HIV-1 reservoir size and viremia control. Microbiome. 2022;10:59.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  43. Cook R, Fulcher J, Tobin N, Li F, Lee D, Javanbakht M, et al. Effects of HIV viremia on the gastrointestinal microbiome of young MSM. AIDS (Lond Engl). 2019;33:793–804.

    Article  CAS  Google Scholar 

  44. Gelpi M, Vestad B, Hansen S, Holm K, Drivsholm N, Goetz A, et al. Impact of human immunodeficiency virus-related gut microbiota alterations on metabolic comorbid conditions. Clin Infect Dis. 2020;71:e359–67.

    Article  CAS  PubMed  Google Scholar 

  45. Guillén Y, Noguera-Julian M, Rivera J, Casadellà M, Zevin A, Rocafort M, et al. Low nadir CD4+ T-cell counts predict gut dysbiosis in HIV-1 infection. Mucosal Immunol. 2019;12:232–46.

    Article  PubMed  Google Scholar 

  46. Mutlu EA, Keshavarzian A, Losurdo J, Swanson G, Siewe B, Forsyth C, et al. A compositional look at the human gastrointestinal microbiome and immune activation parameters in HIV infected subjects. PLoS Pathog. 2014;10:e1003829.

    Article  PubMed  PubMed Central  Google Scholar 

  47. Rocafort M, Noguera-Julian M, Rivera J, Pastor L, Guillén Y, Langhorst J, et al. Evolution of the gut microbiome following acute HIV-1 infection. Microbiome. 2019;7:73.

    Article  PubMed  PubMed Central  Google Scholar 

  48. Rubel M, Abbas A, Taylor L, Connell A, Tanes C, Bittinger K, et al. Lifestyle and the presence of helminths is associated with gut microbiome composition in Cameroonians. Genome Biol. 2020;21:122.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Shenoy M, Fadrosh D, Lin D, Worodria W, Byanyima P, Musisi E, et al. Gut microbiota in HIV-pneumonia patients is related to peripheral CD4 counts, lung microbiota, and in vitro macrophage dysfunction. Microbiome. 2019;7:37.

    Article  PubMed  PubMed Central  Google Scholar 

  50. Tuddenham S, Koay W, Zhao N, White J, Ghanem K, Sears C. The impact of human immunodeficiency virus infection on gut microbiota α-diversity: an individual-level meta-analysis. Clin Infect Dis. 2020;70:615–27.

    Article  PubMed  Google Scholar 

  51. Vujkovic-Cvijin I, Dunham RM, Iwai S, Maher MC, Albright RG, Broadhurst MJ, et al. Dysbiosis of the gut microbiota is associated with HIV disease progression and tryptophan catabolism. Sci Transl Med. 2013;5:193ra91.

    Article  PubMed  PubMed Central  Google Scholar 

  52. Li J, Guo J, Shang E, Zhu Z, Liu Y, Zhao B, et al. Quantitative determination of five metabolites of aspirin by UHPLC-MS/MS coupled with enzymatic reaction and its application to evaluate the effects of aspirin dosage on the metabolic profile. J Pharm Biomed Anal. 2017;138:109–17.

    Article  CAS  PubMed  Google Scholar 

  53. Lajoie J, Kowatsch M, Mwangi L, Boily-Larouche G, Oyugi J, Chen Y, et al. Low-dose acetylsalicylic acid reduces T cell immune activation: potential implications for HIV prevention. Front Immunol. 2021;12:778455.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  54. Alves T, Pongratz R, Zhao X, Yarborough O, Sereda S, Shirihai O, et al. Integrated, step-wise, mass-Isotopomeric flux analysis of the TCA cycle. Cell Metab. 2015;22:936–47.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  55. Ghatge M, Flora GD, Nayak MK, Chauhan AK. Platelet metabolic profiling reveals glycolytic and 1-carbon metabolites are essential for GP VI-stimulated human platelets-brief report. Arterioscler Thromb Vasc Biol. 2024;44:409–16.

    Article  CAS  PubMed  Google Scholar 

  56. Harayama T, Riezman H. Understanding the diversity of membrane lipid composition. Nat Rev Mol Cell Biol. 2018;19:281–96.

    Article  CAS  PubMed  Google Scholar 

  57. Gusdon A, Savarraj J, Redell J, Paz A, Hinds S, Burkett A, et al. Lysophospholipids are associated with outcomes in hospitalized patients with mild traumatic brain injury. J Neurotrauma. 2024;41:59–72.

    Article  PubMed  Google Scholar 

  58. Zewe J, Miller A, Sangappa S, Wills R, Goulden B, Hammond G. Probing the subcellular distribution of phosphatidylinositol reveals a surprising lack at the plasma membrane. J Cell Biol. 2020;219: e201906127.

  59. Eason M, Fan X. The role and regulation of catalase in respiratory tract opportunistic bacterial pathogens. Microb Pathog. 2014;74:50–8.

    Article  CAS  PubMed  Google Scholar 

  60. Ulanova M, Tsang R. Haemophilus influenzae serotype a as a cause of serious invasive infections. Lancet Infect Dis. 2014;14:70–82.

    Article  PubMed  Google Scholar 

  61. Saharan O, Kamat SS. Mapping lipid pathways during phagocytosis. Biochem Soc Trans. 2023;51:1279–87.

    Article  CAS  PubMed  Google Scholar 

Download references

Acknowledgements

The authors gratefully acknowledge the participants and investigators of MiBioGen consortium, CLSA, the FinnGen study, UK Biobank and GWAS catalog for providing the GWAS data.

Funding

This study was supported by the National Natural Science Foundation of China [Grant: 82200036] and 345 Talent Project of Shengjing Hospital of China Medical University funding (M1340).

Author information

Author notes

  1. Jiapeng Hu and Jinxin Hu contributed equally to this manuscript.

Authors and Affiliations

  1. Department of Pediatrics, Shengjing Hospital of China Medical University, Shenyang, China

    Jiapeng Hu & Jinxin Hu

  2. Medical Research Center, Liaoning Key Laboratory of Research and Application of Animal Models for Environmental and Metabolic Diseases, Shengjing Hospital of China Medical University, Shenyang, China

    Jinxin Hu

  3. Department of Neonatology, The First Hospital of China Medical University, Shenyang, China

    Dan Han

Authors
  1. Jiapeng Hu
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Jinxin Hu
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Dan Han
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

JH: data curation; funding acquisition; formal analysis; writing—review and editing. JH: writing—original draft preparation; writing—review and editing. DH: conceptualization; data curation; formal analysis; methodology; project administration; writing—review and editing.

Corresponding author

Correspondence to Dan Han.

Ethics declarations

Ethics approval and consent to participate

Ethical approval and participant consent were obtained in the original studies (MiBioGen, CLSA, FinnGen and UKB).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1

: Figure S1. Forest plot illustrating the causal effects of HIV infection on gut microbiota (including MR results from the FinnGen and UK Biobank datasets, and a meta-analysis of both above). CI, confidence intervals; hetero_P, P-value of Cochrane Q test assessing heterogeneity of MR analysis; meta_hetero_I2, I2 statistic assessing the heterogeneity of meta-analysis; meta_hetero_P, P-value of Cochran's Q test of meta-analysis; presso_P, P-value of MR-PRESSO evaluating horizontal pleiotropy effect of MR results; nsnp, the number of SNPs; P, P-value of MR analysis (IVW) or the meta-analysis; pleio_P, P-value of MR-Egger regression examining horizontal pleiotropy effect of MR results.

Additional file 2

: Figure S2. Forest plot illustrating the causal effects of gut microbiota on HIV infection (including MR results from the FinnGen and UK Biobank datasets, and a meta-analysis of both above). CI, confidence intervals; hetero_P, p-value of Cochrane Q test assessing heterogeneity of MR analysis; meta_hetero_I2, I2 statistic assessing the heterogeneity of meta-analysis; meta_hetero_P, P-value of Cochran's Q test of meta-analysis; presso_P, P-value of MR-PRESSO evaluating horizontal pleiotropy effect of MR results; nsnp, the number of SNPs; OR, odds ratio; P, P-value of MR analysis (IVW) or the meta-analysis; pleio_P, P-value of MR-Egger regression examining horizontal pleiotropy effect of MR results.

Additional file 3

: Figure S3. The meta-analysis combining the primary MR analyses (IVW) of the causal effects of HIV infection on plasma metabolites deriving from FinnGen and UK Biobank datasets. CI, confidence intervals; meta_hetero_I2, I2 statistic assessing the heterogeneity of meta-analysis; meta_hetero_P, P-value of Cochran's Q test of meta-analysis.

Additional file 4

: Figure S4. Forest plot illustrating the causal effects of HIV infection on plasma metabolites (including MR results from the FinnGen and UK Biobank datasets, and a meta-analysis of both above). CI, confidence intervals; hetero_P, P-value of Cochrane Q test assessing heterogeneity of MR analysis; meta_hetero_I2, I2 statistic assessing the heterogeneity of meta-analysis; meta_hetero_P, P-value of Cochran's Q test of meta-analysis; presso_P, P-value of MR-PRESSO evaluating horizontal pleiotropy effect of MR results; nsnp, the number of SNPs; P, P-value of MR analysis (IVW) or the meta-analysis; pleio_P, P-value of MR-Egger regression examining horizontal pleiotropy effect of MR results.

Additional file 5

: Figure S5. The meta-analysis combining the primary MR analyses (IVW) of the causal effects of plasma metabolites on HIV infection deriving from FinnGen and UK Biobank datasets. CI, confidence intervals; meta_hetero_I2, I2 statistic assessing the heterogeneity of meta-analysis; meta_hetero_P, P-value of Cochran's Q test of meta-analysis; P, P-value of meta-analysis; OR, odds ratio.

Additional file 6

: Figure S6. Forest plot illustrating the causal effects of plasma metabolites on HIV infection (including MR results from the FinnGen and UK Biobank datasets, and a meta-analysis of both above). CI, confidence intervals; hetero_P, P-value of Cochrane Q test assessing heterogeneity of MR analysis; meta_hetero_I2, I2 statistic assessing the heterogeneity of meta-analysis; meta_hetero_P, P-value of Cochran's Q test of meta-analysis; presso_P, P-value of MR-PRESSO evaluating horizontal pleiotropy effect of MR results; nsnp, the number of SNPs; OR, odds ratio; P, P-value of MR analysis (IVW) or the meta-analysis; pleio_P, P-value of MR-Egger regression examining horizontal pleiotropy effect of MR results.

Additional file 7

: Figure S7. Forest plot illustrating the causal effects of gut microbiota on plasma metabolites using IVW method. CI, confidence intervals; hetero_P, P-value of Cochrane Q test assessing heterogeneity of MR analysis; presso_P, P-value of MR-PRESSO evaluating horizontal pleiotropy effect of MR results; P, P-value of MR analysis using IVW method; pleio_P, P-value of MR-Egger regression examining horizontal pleiotropy effect of MR results.

Additional file 8

: Figure S8. Forest plot illustrating the causal effects of plasma metabolites on gut microbiota using IVW method. CI, confidence intervals; hetero_P, P-value of Cochrane Q test assessing heterogeneity of MR analysis; presso_P, P-value of MR-PRESSO evaluating horizontal pleiotropy effect of MR results; P, P-value of MR analysis using IVW method; pleio_P, P-value of MR-Egger regression examining horizontal pleiotropy effect of MR results.

Additional file 9

: Supplementary Table S1-S16.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Hu, J., Hu, J. & Han, D. Causal relationships between gut microbiota, plasma metabolites, and HIV infection: insights from Mendelian randomization and mediation analysis. Virol J 21, 204 (2024). https://doi.org/10.1186/s12985-024-02480-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12985-024-02480-1

Keywords

Virology Journal

ISSN: 1743-422X