Latent tuberculosis infection is associated with an enrichment of short chain fatty acid producing bacteria in the stool of women living with HIV

Background: Latent tuberculosis infection (LTBI) is common in people living with HIV (PLHIV) in high TB burden settings. Active TB is associated with specific stool taxa; however, little is known about the stool microbiota and LTBI, including in PLHIV. Method: Within a parent study that recruited adult females with HIV from Cape Town, South Africa into predefined age categories (18–25, 35–60 years), we characterised the stool microbiota of those with [interferon-γ release assay (IGRA)- and tuberculin skin test (TST)-positive] or without (IGRA- and TST- negative) LTBI (n=25 per group). 16S rRNA DNA sequences were analysed using QIIME2, Dirichlet Multinomial Mixtures, DESeq2 and PICRUSt2. Results: No α- or β-diversity differences occurred by LTBI status; however, LTBI-positives were Faecalibacterium-, Blautia-, Gemmiger-, Bacteroides-enriched and Moryella-, Atopobium-, Corynebacterium-, Streptococcus-depleted. Inferred metagenome data showed LTBI-negative-enriched pathways included several involved in methylglyoxal degradation, L-arginine, putrescine, 4-aminobutanoate degradation and L-arginine and ornithine degradation. Stool from LTBI-positives demonstrated differential taxa abundance based on a quantitative response to antigen stimulation (Acidaminococcus-enrichment and Megamonas-, Alistipes-, and Paraprevotella-depletion associated with higher IGRA or TST responses, respectively). In LTBI-positives, older people had different β-diversities than younger people whereas, in LTBI-negatives, no differences occurred across age groups. Conclusion: Amongst female PLHIV, those with LTBI had, vs. those without LTBI, Faecalibacterium, Blautia, Gemmiger, Bacteriodes-enriched, which are producers of short chain fatty acids. Taxonomic differences amongst people with LTBI occurred according to quantitative response to antigen stimulation and age. These data enhance our understanding of the microbiome’s potential role in LTBI.


Introduction
Tuberculosis (TB) is major cause of death, with 167 000 deaths among persons living with HIV (PLHIV) in 2022 1 .One strategy to prevent active TB is control latent tuberculosis infection (LTBI).LTBI is inferred from a positive tuberculin skin test (TST) or interferon-gamma release assay (IGRA).TB preventive treatment (TPT) strategies play a key role in TB prevention especially in vulnerable populations such as PLHIV.
PLHIV are at greater risk for Mycobacterium tuberculosis (MTB) infection and progression to active TB 2 .However, it is poorly understood why some individuals in TB-endemic countries are never infected despite high exposure and why a large proportion of infected individuals never progress.We need more information on correlates of infection and progression, which may have prognostic value.
The microbiome has important immunomodulating effects and the microbiome's role, including people with LTBI, is an emerging area of interest.For example, in the lung, Lactobacillus is enriched in people with LTBI compared to active pulmonary TB group and LTBI-negatives 3 .In the nasopharynx of LTBIpositives, Staphylococcus and Corynebacterium dominate the microbiome compared to healthy control and active TB cases 4 , and the nasopharyngeal microbiota of LTBI-positives has lower alpha-diversity than LTBI-negatives 5 .
The stool microbiota has potentially an important immunomodulatory role in respiratory disease, including active TB 6 .However, it is comparatively understudied in latent TB.One study among individuals with poorly-controlled diabetes, showed LTBI-positives people to be Bacteroides-, Alistipes-, and Blautiaenriched compared to LTBI-negatives 7 .
During LTBI infection, comparisons of TB cases, HIV-negative LTBI-positive individuals, LTBI-negative and active TB gut microbiomes showed trends of changes in Bacteriodes and Firmicutes, however no signi cant difference in composition of the stool microbiota 8 .In HIV-negative LTBI-positive individuals showed positive correlation between relative abundances of Coriobacteriaceae and IFN-gamma against MTB antigens more likely associated with of CD4 + T cell 9 .Those studies, however, did not include PLHIV where changes in the gut microbiome are a well-characterised complication of HIV 10 , commonly characterized by reductions in diversity in the stool microbiota 11 .If microbial dysbiosis is detected early in natural history of TB disease (after infection), it may be indicative of early microbial and immune dysregulation associated with subsequent risk of active TB.Tests of progression to active TB are a major public health priority, as is understanding the biological drivers of LTBI.To continue to strengthen our understanding of LTBI and the microbiome, we evaluated the stool microbiota of PLHIV with and without LTBI, nested within a larger study studying resistance to infection.

Recruitment
Participants (18-60 years) were recruited from community health care clinics in Cape Town, South Africa as part of a published parent study (ResisTB) 12 .This cohort was predominantly female, and age was a surrogate for TB exposure, resulting in two groups of 18-25 years and 35-60 years.All people had to be TB symptom screen negative, HIV-positive and stable on ART for ≥ 1 year.Study procedures were approved by the Stellenbosch University Human Research Ethics Committee (N16/03/033A) and each participant provided written informed consent.

De nitions
IGRA and TST positive (LTBI-positive) people was de ned by two QuantiFERON-TB Gold Plus-positive and a positive TST (> 0 mm).IGRA and TST negative (LTBI-negative) was de ned by two negative QuantiFERON-TB Gold Plus-negative and a TST (0 mm).

Microbiota specimen collection and processing
At TST administration, participants were provided with a home stool sampling kit containing an EasySampler (ALPCO, Salem, USA) and a receptacle containing DNA stabilization buffer (Stratec Biomedical, Birkenfeld, Germany).Generally, buffered stools were collected the night before TST reading and returned at TST reading.Upon receipt at the laboratory, buffered stool was frozen at -20°C until batched DNA extraction carried out using the PSP Spin Stool DNA Plus Kit (Stratec Biomedical, Birkenfeld, Germany).
16S rRNA gene sequencing and microbiota analysis V4 region sequencing of the bacterial 16S rRNA gene (150 bp read length, paired-end) was done using Illumina Miseq as described 6 .Sequences was analysed with Quantitative Insights into Microbial Ecology (QIIME2, version 2.0.8).Cluster analysis was carried out using Dirichlet-Multinomial Mixtures (DMM) 13 .
Alpha diversity was calculated by Shannon's diversity with Mann-Whitney testing using GraphPad Prism (v8; GraphPad Software, USA).Beta diversity was calculated using Bray-Curtis with permutational multivariate ANOVA (PERMANOVA) using R (v4.2.2; R Core Team).The functional metagenome was inferred from sequencing data using Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2; v2.0.0) 14 .Differentially abundant taxa and metabolic pathways were identi ed using DESeq2 (v1.22.2) 15 and Benjamini-Hochberg correction adjustment for multiple comparisons (signi cance level 0.20) 6,16 .For comparisons groups, taxa at higher relative abundance in one group were described as enriched (those at lower relative abundance were described as depleted).For the whole cohort, age and eld site correction was applied for DESeq2 analyses.Linear discriminant analysis (LDA) effect size (LEfSe) 17 was used to compare the clusters to each other.The proportions test was done using STATA (v18; StataCorp, USA) to determine whether a speci c variable was more frequent in different groups.Data and code are available upon publication.

Clinical and demographic characteristics across clusters
β-diversity differed by eld site (Supplementary table 1).There were no differences in the proportion of LTBI-positives per cluster (Table 2).C1 and C3 were more likely to be on INH prophylaxis than C2 and more likely to be from sites other than Khayelitsha (Site B) Youth.C2s were, compared to C3s, more likely to be from sites other than Khayelitsha (Site B) CHC Youth and Du Noon CDC.Within LTBI-positives, taxa are differentially enriched based on the magnitude of the response to antigen stimulation Taxonomic abundances between quantitative responses to antigen stimulation were compared overall and in LTBI-positives.No differential abundances were identi ed between the overall cohort above or vs the median average IGRA or TST quantitative response (Supplementary Fig. 5A and B).LTBI-positives with an IGRA response above the median were Acidaminococcus-enriched and Granulicatella-depleted (Fig. 6A) and LTBI-positives below the median TST response were Megamonas-, Alistipes-, and Paraprevotella-enriched (Fig. 6B).

Discussion
We compared, in PLHIV, the stool microbiota of LTBI-positive vs. LTBI-negative women.Our key ndings are: 1) stool from LTBI-positives differed from LTBI-negatives in terms of known short chain fatty acid (SCFA)-producing taxa and these were associated with a depletion of metabolite degradation pathways, 2) three taxonomic clusters occurred, characterised by high abundances of Bacteroides, Streptococcus and Prevotella, and clustering associated with INH prophylaxis and health facility location, 3) LTBIpositives with greater quantitative response to mycobacterial antigen stimulation were, vs. LTBI-positives with lesser responses, Acidaminococcus-enriched (IGRA readouts) and Megamonas-, Alistipes-, and Paraprevotella-depleted (TST), and 4) β-diversity differed by age group only in LTBI-positives but notnegatives.Our ndings help lay a foundation for understanding the microbiome's role in LTBI.
Stool from people with LTBI was Moryella-, Atopobium-, Corynebacterium-, Streptococcus-depleted and Faecalibacterium, Blautia, Gemmiger and Bacteriodes-enriched. Faecalibacterium and Gemmiger is a known producer of the butyrate 18 , which is a SCFA that increases incident TB risk 19 .Bacteroides produces SCFAs like acetate and propionate 20 .Blautia is enriched in people with active TB and independently predicts upregulation of pro-in ammatory pathways 6 .Although the role of Atopobium is unclear, Streptococcus, which we found to be depleted in LTBI-positives, produces acetate, which mitigates host in ammation 21 .
Three taxonomic clusters occurred [Bacteroides (C1), Streptococcus (C2), Prevotella-enriched (C3)], however, these were not associated with LTBI status (other studies have documented speci c clusters associated with active TB 16 ).People within each cluster were more likely to be recruited from different facilities, suggesting potential geographic associations to be considered in future studies.Additionally, C1 was more likely than C2 to be receiving INH prophylaxis.INH prophylaxis itself was associated with Blautia-enrichment and Moraxella-, Megamonasand Actinobacillus-depletion.Other studies have shown Clostridiales, Coprococcus, Lachnospiraceae, and Ruminococcaceae-enriched and Clostridium_XIVa, Romboutsia, and Roseburia-depleted stool to occur during rifamycin-based tuberculosis preventive therapy 22 .To our knowledge, our study is the rst to show Blautia-enriched and Moraxella-, Megamonasand Actinobacillus-depleted stool in humans on isoniazid TB preventive therapy.This association is interesting as isoniazid itself is a drug thought to have an extremely narrow antimicrobial spectrum (Mycobacteria only) 23 .
Additionally, LTBI-positives who had a larger quantitative response to antigen stimulation were, when IGRA readouts were used, Acidaminococcus-enriched and Granulicatella-depleted, and, when TST readouts was used, Megamonas-, Alistipes-, Paraprevotella-depleted. Acidaminococcus produces acetate and butyrate 24 and is primarily in uenced by diet 25 ; its enrichment likely re ects lifestyle differences within LTBI-positives.Paraprevotella (like Alistipes) is a SCFA-producer and generally considered bene cial 26 .This nding is notable because higher quantitative responses are associated with greater risk of incident TB 27 , suggesting such taxa may contribute to this risk, however, this requires prospective con rmation in controlled studies.
Younger people were more likely to be LTBI-positive than older people so, to adjust for age as a potential confounder, we dichotomised people (18-25, 35-60 years).Within LTBI-positives, older vs. younger individuals were enriched in Ochrobactrum (an opportunistic pathogen 28 ), Neisseria (role unclear), Mycoplasma (induces a proin ammatory cytokines 29 ) and depleted in Catenibacterium (enriched in PLHIV 30 and active TB 31 ), Alistipes (SCFA producer with potentially anti-in ammatory effects 32 ), and Methanobrevibacter (a methane producer 33 ).Older LTBI-negatives were also Methanobrevibacterdepleted and enriched in Actinobacillus (inversely associated with amino acid production 34 ) but did not show β-diversities differences.This could suggest LTBI results in greater age-related microbiome differences but requires further investigation.
Our study has strengths and limitations.This is a cross-sectional study that, to enhance feasibility, leveraged (but was constrained) the parent ResisTB study.We only evaluated women with HIV and other populations may result in different ndings, however, PLHIV do have elevated risk of incident TB.Although people were measured once, our study generates useful data to inform hypothesis-driven interventions to potentially modulate the microbiome.
In conclusion, amongst women living with HIV, those with LTBI were, vs. those without LTBI, primarily differentially abundant in SCFA-producing anaerobic bacteria.

Figures Figure 1
Figures

Figure 5 Three
Figure 5

Table 1
Demographic and clinical characteristics.People with LTBI were younger and more likely to from Khayelitsha (Site B) CHC.Abbreviations: LTBI: Latent TB infection; BMI: Body mass index; INH: Isoniazid; CHC: Community Health Centre; CDC: Community Day Centre.Data are median (IQR) or n (%).

Table 2
Demographic and clinical characteristics of the three clusters found in cohort.C1 was more likely than C2 to be on current INH prophylaxis and more likely to be recruited from Khayelitsha (Site B) CHC, Kraaifontein CHC, Site C Youth or Du Noon CDC.C2 was more likely than C3 to be recruited from Khayelitsha (Site B) CHC, Kraaifontein CHC or Site C Youth.Abbreviations: LTBI: Latent TB infection; BMI: Body mass index; INH: Isoniazid; CHC: Community Health Centre; CDC: Community Day Centre.Data are median (IQR) or n (%).
* Missing data: Current Tobacco smoker (n = 1) Taxonomic differences also occurred amongst people with LTBI by age group, suggesting the age-related microbiome perturbations are more pronounced in LTBI-positives.Longitudinal studies are needed to further delineate the microbiome's role in LTBI, which this work helps provide a justi cation for.
(N16/03/033A) and each participant provided written informed consent.HREC is committed to the ethical principles laid down in the South African and international guidelines, Declaration of Helsinki, Declaration of Taipei and The Belmont Report.Consent to participate