PWH on ART exhibit accelerated biological aging in the intestines, with rates differing from that in the blood. We collected colon and ileal biopsies, blood, and stool samples from 25 PWH on ART with a viral load of < 50 copies/ml and 23 HIV-negative controls matched by age, sex, ethnicity, and BMI (Table 1). Using the systems biology approach illustrated in Fig. 1A, we aimed to determine whether living with ART-suppressed HIV infection is associated with shifts in intestinal biological age, and whether microbial translocation and dysbiosis are linked to biological aging in PWH on ART.
Patterns of DNA methylation at specific CpG sites have been used to gauge both chronological and biological age across numerous cell types, tissues, and organs in humans and other mammals [39, 41, 42, 44, 56, 57]. Such distinct patterns of DNA methylation form the basis for multiple epigenetic clocks of aging, with different clocks using different sets of CpG sites for their calculations. We first gauged the blood biological age of PWH on ART and HIV-negative controls using six established DNA methylation epigenetic clocks of aging. Specifically, we used DNA from peripheral blood mononuclear cells (PBMCs) to estimate biological age with the following principal component-based epigenetic clocks [45]: Horvath's multi-tissue predictor DNAmAge (PCHorvath1) based on 353 CpG sites [39], Horvath's skin and blood clock (PCHorvath2) based on 391 CpG sites [56], Hannum's clock based on 71 CpG sites (PCHannum) [42], Levine’s DNAmPhenoAge based on 513 CpG sites (PCPhenoAge) [41], DNA methylation-based mortality risk assessment (PCGrimAge) [44], and Lu’s telomere length predictor (PCDNAmTL) [57]. For five of these clocks, a higher value indicates an older biological age. However, for the PCDNAmTL clock, a lower value indicates older age.
Despite having a similar chronological age (Fig. 1B), the blood of PWH on ART showed an older biological age than that of HIV-negative controls. The difference ranged from + 3.1 years (using PCGrimAge) to + 7.63 years (using PCHorvath2) (Supplementary Fig. 1A). We next calculated the acceleration of biological age by regressing the outputs of the clocks against chronological age. Larger indices imply faster biological aging, except for PCDNAmTL where a smaller index denotes accelerated aging. This analysis found that the biological age of PWH on ART was accelerated between 2.59 to 7.05 years (Fig. 1C). We also employed the DunedinPACE epigenetic clock, which estimates the pace of aging [46]. Higher values for this metric correlate with accelerated aging [46]. Consistently, the DunedinPACE estimate was markedly higher in PWH on ART than in the controls (Fig. 1C), echoing recent studies which suggest that PWH experience accelerated biological aging in blood [32–37].
Next, we applied the same epigenetic aging clocks to DNA isolated from the ileum and colon (Supplementary Fig. 1B-C). The ileum of PWH on ART showed accelerated biological aging by four of the seven epigenetic clocks (Horvath1, Horvath2, Hannum, and PCDNAmTL; Fig. 1D), compared to controls. Similarly, the colon of PWH on ART showed accelerated biological aging using two clocks (Horvath1 and DunedinPACE; Fig. 1E). That some clocks did not detect aging acceleration in the ileum and colon might be because many of the clocks were designed for use on blood samples. Since Horvath1 [39] was developed using tissues, we compared its age estimates and biological age acceleration across the blood, ileum, and colon samples (Fig. 1F-G). These data emphasize that the ileum, colon, and blood in PWH on ART all exhibit accelerated biological aging. However, the acceleration rate differs among tissues, suggesting that HIV accelerates aging in a tissue-specific manner.
Epigenetic clock estimates of biological age were validated using other established and emerging markers of aging. To support the results from the epigenetic clocks, we compared these to established and emerging biomarkers of aging. As telomere length (TL) is an established aging marker [58, 59], we evaluated TL in PBMCs via HT-Q-FISH. Median TL did not differ between PWH on ART and controls (Supplementary Fig. 2A-C); however, PWH on ART had a higher percentage of cells with shorter telomeres (and a lower percentage with longer telomeres) than controls (Supplementary Fig. 2D). We then determined the correlations between biological age, as estimated by the epigenetic clocks, and measures of TL (Fig. 2A). These correlations show that higher biological age estimated by the epigenetic clocks correlate strongly with shorter TL.
In addition, new metrics for biological age have recently emerged, including the deep learning based ‘inflammatory aging clock’ called iAge [60]. This metric is derived from the measurement of several inflammation markers in plasma, such as CXCL9 and eotaxin; these are incorporated into an inflammatory aging clock that can predict accelerated aging [60]. We measured the levels of some of the markers included in iAge, as well as other inflammatory indicators pertinent to HIV infection (e.g., IL4, IL-6, MIP-1α) [61, 62] in blood using multiple cytokine arrays. Levels of several markers, including CXCL9 and eotaxin, were elevated in PWH on ART compared to controls (Fig. 2B). Higher levels of these inflammation markers correlated with accelerated biological aging (derived from the epigenetic clocks as in Fig. 1C-G), especially in blood (Fig. 2C). We also examined correlations between accelerated aging (based on Horvath1) and the levels of cell-associated HIV DNA in PBMCs, ileum, and colon, and cell-associated HIV RNA in PBMCs, as surrogates for HIV persistence. Among these, the strongest association with epigenetic age acceleration was HIV DNA levels in the ileum (Supplementary Fig. 3). These findings validate the results obtained with the epigenetic clocks and support the conclusion that living with HIV, even with ART, accelerates biological aging in both tissues and blood, with the rate differing among them.
Intestinal permeability and microbial translocation link to accelerated biological aging. Microbial translocation and dysbiosis are increasingly hypothesized to drive systemic inflammation and thus promote inflammation-associated diseases of aging. Given that PWH on ART experience accelerated biological aging both systemically and within tissues, we explored the possibility that microbial translocation and microbial dysbiosis may drive this accelerated aging. First, we evaluated microbial translocation in PWH on ART and controls by assessing the levels of tight junction proteins (ZO-1 and occludin) in the ileum and colon using immunofluorescence and a scaling method described in Supplementary Fig. 4. Data in Fig. 3A-B show that intestinal integrity, as assessed by levels of ZO-1 and occludin, was significantly lower in PWH on ART compared to controls. This suggested that gut permeability was higher in PWH on ART. Consistently, markers of gut damage and microbial translocation in plasma were higher in PWH on ART compared to controls (Fig. 3C). The damage/translocation markers assessed were REG3α (intestinal stress marker [63]), I-FABP (enterocyte apoptosis marker [64]), Zonulin (tight junction permeability marker [65, 66]), LPS binding protein (bacterial translocation marker [67]), β-glucan (fungal translocation marker [68]), and sCD163 (microbe-triggered myeloid inflammation marker). Together, these data suggest that in PWH the intestinal integrity is compromised, resulting in enhanced microbial translocation.
Next, we investigated the relationships between the degree of intestinal integrity or microbial translocation and the two measures of biological aging. Specifically, we determined correlations between intestinal integrity (based on levels of tight junction proteins) or microbial translocation (based on levels of the damage/translocation markers) and either accelerated aging (calculated as in Fig. 1C-G using data from the epigenetic aging clocks in blood and tissues) or blood-based inflammatory aging markers (measured as in Fig. 2B). Correlation heat-maps (Fig. 3D) showed that intestinal integrity negatively correlated with accelerated biological aging and levels of inflammatory aging markers, while microbial translocation positively correlated with accelerated biological aging and levels of inflammatory aging markers. Moreover, the higher levels of HIV DNA and RNA in blood and/or tissues (as surrogates of HIV persistence) correlated with lower intestinal integrity (Fig. 3E-H). These findings highlight the connections between elevated intestinal permeability and microbial translocation, accelerated aging, greater inflammation, and greater HIV persistence in the blood and intestinal tissues of PWH on ART.
Living with HIV is linked to intestinal and fecal microbial dysbiosis, notably a decrease in butyrate-producing bacteria. As we described in the preceding sections, PWH on ART have compromised intestinal integrity which may lead to accelerated biological aging both systemically and in tissues. One plausible mechanism underlying this compromised intestinal integrity is microbial dysbiosis. Microbial dysbiosis can pave the way for an increase in bacteria that produce toxic metabolites, such as those involved in tryptophan catabolism [28, 69, 70]. It can also cause a decline in bacteria that generate metabolites considered beneficial, such as short-chain fatty acids (SCFAs) [71], notably butyrate, which are microbiome-derived metabolites known to bolster intestinal barrier integrity [72]. With this context in mind, we probed the microbiome in stool, ileum, and colon samples from PWH on ART and controls using 16S rRNA sequencing.
We found that microbial alpha diversity, a hallmark of a healthy microbiome [73] as measured by various models (Richness, Shannon, and Faith), was lower in the colon of PWH on ART compared to controls (Fig. 4A); smaller non-significant differences were observed in feces and ileum between the groups (Supplementary Fig. 5A-B). We then assessed the relative abundance of bacteria known to produce SCFA, particularly butyrate, and the relative abundance of bacteria considered pro-inflammatory (“pathobionts”; Supplementary Table 1). The relative abundance of butyrate-producing bacteria was lower in PWH on ART than in controls (Fig. 4B). PWH on ART also tended to have a more pro-inflammatory fecal microbiome and lesser SCFA-producing fecal bacteria, but trends were not statistically significant (Supplementary Fig. 5C-D).
When we examined specific bacterial genera in the feces, colon, and ileum, we found that the microbiome in these locations varied significantly (Fig. 4C; FDR < 0.05). Comparing PWH on ART with HIV-negative controls (Fig. 4D), we found that living with HIV on ART was associated with an enrichment of some bacterial genera and a depletion of others, in feces, colon, and/or ileum. Enriched bacterial genera include putatively pro-inflammatory bacterial genera [24] such as Catenibacterium, Prevotella 2, Allprevotella, Prevotella 9, and Enterobacteriaceae. Depleted genera included putatively anti-inflammatory bacteria [74, 75] and bacteria known for their ability to produce SCFAs such as Erysipelotrichaceae UCG − 003, Alistipes, Coprococcus 3, Peptostreptococcaceae, Romboutsia, Subdoligranulum, Bacteroidales, [Ruminococcus] gauvreauii group, and Faecalibacterium.) This reinforces findings from earlier studies [28, 71], suggesting an HIV-related microbial imbalance, characterized by higher levels of potentially pro-inflammatory bacterial genera and lower levels of potentially anti-inflammatory bacteria. This microbial imbalance may contribute to the previously observed decrease in intestinal integrity and consequently, the accelerated biological aging in PWH on ART.
A distinct mucosal microbial signature is linked to accelerated biological aging. Given the dysbiosis observed in PWH (Fig. 4D), we next asked if this dysbiosis was related to the accelerated biological aging we had observed in PWH. Our analyses in Fig. 5A revealed that specific bacterial genera that were enriched in colon tissue from PWH on ART (such as Catenibacterium, Prevotella 2, Allprevotella, and Prevotella 9) correlated strongly with greater accelerated aging (FDR < 10%). In contrast, other genera that were depleted in colon tissue from PWH on ART (like Erysipelotrichaceae UCG-003, Alistipes, Coprococcus 3, Romboutsia, and Subdoligranulum) correlated with slower accelerated aging. Notably, the correlations between the enriched bacteria and higher accelerated biological aging were driven by samples from PWH on ART, whereas the correlations between the depleted bacteria and slower accelerated biological aging were driven by samples from HIV-negative controls (Fig. 5B). Similar analyses using ileal (Fig. 5C), and fecal (Fig. 5D) samples did not yield any correlations with FDR < 10%, although some nominal P values were significant.
Beyond their associations with accelerated biological aging rates, taxa enriched in PWH on ART were linked to lower tight junction protein levels in tissues, elevated microbial translocation, and enhanced inflammation (Fig. 5E, top rows of each section). In contrast, taxa that were depleted in PWH on ART were associated with better intestinal integrity, lower microbial translocation, and lower inflammation (Fig. 5E, bottom rows of each section). Separate analyses revealed that the pro-inflammatory microbiome was associated with higher levels of HIV DNA and RNA in both blood and tissues. By contrast, the SCFA-producing bacteria, notably those producing butyrate, associated with lower levels of HIV DNA and RNA (Fig. 5F-G). These findings suggest that certain bacterial genera, especially those from the colon, may influence the pace of biological aging. Moreover, they shed light on the intricate relationship between microbial profiles, inflammation, HIV persistence, and the biological aging trajectory in PWH on ART.
Correlation networks reveal links between the mucosal microbiome, microbe-related metabolites, and accelerated biological aging. Building on our observations (Fig. 5) that SCFAs were associated with slower biological aging, we expanded our inquiry to other microbe-related metabolites. Recognizing that many effects of the microbiome are mediated by metabolites other than SCFAs, we conducted an untargeted metabolic analysis on stool and plasma samples from both PWH on ART and controls. Our goal was to identify additional metabolites that might bridge the microbial signature (Fig. 5) with the accelerated biological aging patterns observed.
First, we assessed a spectrum of microbiome- and gut-specific metabolites (Supplementary Table 2). PWH on ART had elevated levels of metabolites known to be detrimental, such as L-kynurenine and quinolinic acid, both by-products of tryptophan catabolism [69, 76]. We confirmed this by evaluating two common measures of tryptophan catabolism, the kynurenine to tryptophan (K/T) ratio and the quinolinic acid to tryptophan (Q/T) ratio [77]. Both ratios were indeed higher in PWH on ART than controls (Fig. 6A). PWH on ART also had lower levels of metabolites associated with microbial diversity and intestinal health, like hippuric acid [78], L-ergothioneine [79], and oleic acid [80] (Fig. 6A). The metabolites enriched in PWH on ART were associated with accelerated biological aging, compromised intestinal integrity, heightened microbial translocation, and greater inflammation (Fig. 6B). Conversely, metabolites that were less abundant in PWH on ART correlated with slower biological aging, greater intestinal integrity, lower microbial translocation, and lower inflammation (Fig. 6B).
To visualize these complex interactions, we performed a network analysis, which illustrated distinct three-way interactions among microbial genera enriched in PWH on ART, elevated tryptophan catabolism metabolites, diminished beneficial gut metabolites like hippuric acid and oleic acid, and accelerated biological aging (Fig. 6C). Conversely, the network analysis also identified distinct connections among microbial genera depleted in PWH on ART, diminished tryptophan catabolism metabolites, abundant protective gut metabolites, and slower biological aging (Fig. 6C). These intricate relationships were most pronounced in the colon, followed by the ileum, and then the feces, underscoring the tissue-specific microbial imprints of accelerated biological aging which were absent in the fecal microbiome.