Written by Baris Ozdinc
The human gut microbiome is intimately linked to human health. Genetics, lifestyle and environmental factors all shape the human gut microbiome. Disease also significantly alters the gut microbiome, but how is not known for all illnesses. The gut microbiome in Human Immunodeficiency Virus (HIV) infection has been examined in the past but results have been inconsistent. Noguera-Julien et al. (2016) sought to clarify the relationship between gut microbiome shifts and HIV by examining the fecal microbiota of 156 patients from Barcelona. They performed both 16S rRNA gene sequencing and whole genome sequencing on 129 HIV-1 positive (+) 27 HIV-1 negative (-) patients, and analyzed gut microbial community compositions of HIV – and + patients using 16S rRNA genes. We imported the whole genome shotgun (WGS) data from all 156 patients to the CosmosID-HUB in order to compare the gut microbiome in HIV+ and HIV- patients and to observe if these results are different from the original 16S results. We used total (rather than filtered) data for this analysis in order to preserve the ability to observe changes in rare taxa.
Plotting a heatmap of samples based on the relative abundances of the 50 most abundant bacterial species in HIV – and + patients’ gut microbiomes illustrates some clustering (Figure 1). However, the clustering does not appear to be by HIV status. On the right end of the heatmap is a distinct cluster of samples with high Bacteroides and Phocaeicola vulgatus relative abundance, but these are all HIV- patients (green). The HIV+ patients are mixed within the HIV-, suggesting that there is not a consistent effect of HIV on the gut microbiome composition, at least among the 50 most abundant taxa.
Both Simpson’s and Shannon’s alpha diversity reflect the lack of clustering of the HIV- samples. We found no significant difference in Simpson’s diversity index when we performed a Wilcoxon Rank Sum Test on the two cohorts (p-value = 0.221, Figure 2), implying that community compositions of the two cohorts are shared by similar numbers of dominant taxa. Furthermore, comparing the Shannon’s Diversity Index between the two cohorts resulted in no significant differences (p-value = 0.258 , Figure 3). Together, the two alpha diversity index comparisons indicate that HIV – and + cohorts gut microbiome data represent similar numbers of dominant and rare taxa, highlighting a variation between findings of WGS and 16S rRNA microbiome relative abundance analysis. The findings of Noguera-Julien et al. (2016) suggested that HIV + patients tend to be more species deficient when compared to HIV – patients in different 16S rRNA gene relative abundances.
After having observed no difference between species richness and evenness between the gut microbiomes of the two cohorts, a divergence from the findings of 16S rRNA gene analysis on the same data, we tested if the WGS data differed in beta diversity. The 16S rRNA gene data (Noguera-Julien et al. 2016) suggested that the samples did not cluster significantly by HIV infection status. However, using the Bray- Curtis method, WGS data suggests otherwise. Plotting the 3D principal coordinate analysis results of the Bray-Curtis dissimilarity matrix of the two HIV serostatus cohorts shows HIV infection drives gut community composition to be dissimilar to that of HIV negative patients. Performing a PERMANOVA test illustrates the significant dissimilarity of the gut microbiome compositions of HIV – and + cohorts (p-value = 0.003, Figure 4).
To identify differentially abundant taxa that may be responsible for the significant beta diversity results, we performed a linear discriminant analysis of effect size (LefSe) over the microbiome data of the two cohorts. We visualized only the taxa which were highly likely to be reproduced through filtering out the taxa with a linear discriminant analysis (LDA) effect size below 3.5 (Figure 5). The 7 differentially abundant taxa with a LDA effect size > 3.5 are three Prevotella species, associated with plant rich, Eastern and Mediterranean diets (Ley 2015), Butyrivibrio crossotus, which produces butyrate from breaking down plant fibers and cellulose (Bourassa et al. 2016), an unclassified Clostridium species that could break down plant fibers into short-chain fatty acids (Lopetuso et al. 2013), an unclassified Firmicutes species from the main butyrate producing phyla, and an unclassified Ruminococcaceae which may also break down plant fibers into butyrate (Venegas et al. 2019). In the HIV + cohort, species from Bacteroides and Parabacteroides families are differentially abundant. These families in are associated with inflammatory ulcerative colitis and Crohn’s diseases (Zhou and Zhi et al. 2016).
Key findings: In summary, we saw some differences in analyzing the gut microbiome by HIV status using WGS data as compared to the 16S data used in the original study. We did not see significant differences in species richness and evenness levels between HIV+ and HIV- patients. However, we did see significant differences by beta diversity, whereas the 16S data suggested there were no significant differences here. This could be due to the increased sensitivity and ability to detect a broader range of bacteria with WGS. Biologically, this difference could be due to the immunodeficiency caused by HIV in the + patients, as the immune system is an essential modulator of host-microbiome interactions. Performing LefSe analysis to illustrate which taxa drives the dissimilarity between HIV – and + cohorts illustrates that species associated with healthy diets, SCFA production and plant fiber consumption is differentially abundant in HIV – patients. These are general indications of a healthy gut microbiome. However, in the HIV + patients, taxa associated with inflammatory bowel diseases, which may decrease overall health, are enriched in HIV + patients. This could be due directly to the viral infection or to side effects from immunomodulary and anti-viral treatments.
