Host genetic and immune factors drive evasion of HIV-1 pathogenesis in viremic non-progressors

Summary:

Background:

Viremic non-progressors (VNPs) represent an exceptional and uncommon subset of people with HIV-1, characterized by the remarkable preservation of normal CD4⁺ T-cell counts despite uncontrolled viral replication — a trait reminiscent of natural hosts of simian immunodeficiency virus. The mechanisms orchestrating evasion from HIV-1 pathogenesis in human VNPs remain elusive, primarily due to the absence of integrative studies.

Methods:

We implemented a novel single-cell and multiomics approach to comprehensively characterize viral, genomic, transcriptomic, and metabolomic factors driving this exceedingly rare disease phenotype in 16 VNPs and 29 HIV+ progressors.

Findings:

Genetic predisposition to the VNP phenotype was evidenced by a higher prevalence of CCR5Δ32 heterozygosity, which was associated with lower levels of CCR5 expression and lower frequency of infected cells in peripheral circulation. We also observed reduced levels of plasma markers of intestinal disruption and attenuated interferon responses in VNPs. These factors potentially drive the other phenotypic traits of immune preservation in this population, including the unaltered tryptophan metabolic profile, reduced activation of cytotoxic lymphocytes, and reduced bystander CD4⁺ T-cell apoptosis.

Conclusions:

In summary, our comprehensive analysis identified intricate factors collectively associated with the unique immunovirological equilibrium in VNPs, shedding light on potential avenues for therapeutic exploration in managing HIV pathogenesis.

Graphical Abstract

eToc blurb: two-sentence scientific summary

The mechanisms enabling evasion from HIV-1 pathogenesis in VNPs are poorly understood. Here, Bayón-Gil et al. perform a comprehensive multiomics characterization which reveals an intricate multifactorial mechanism, including genetic predisposition to lower infection rate, moderation of chronic immune activation and preservation of intestinal homeostasis, jointly contributing to pathogenesis resistance.

INTRODUCTION:

Human immunodeficiency virus type-1 (HIV-1) targets CD4⁺ T cells to establish chronic infection, which is characterized by persistent viremia and progressive decline in CD4⁺ T-cell count ultimately leading to AIDS. While antiretroviral therapy (ART) efficiently suppresses viral replication and stops disease progression, the infection is never cured owing to the integration of the viral genetic material in the genome of infected cells and the subsequent establishment of a long-term persistent viral reservoir of latently infected cells¹. Despite ART-mediated viral suppression, chronic immune activation often prevails², and a significant number of individuals do not recover their original CD4⁺ T-cell counts, leading to sustained immune dysfunction³. New therapeutic strategies are needed to tackle these yet unsolved adverse events of HIV-1 infection.

Extreme phenotypes of HIV-1 infection with non-pathogenic progression profiles in the absence of ART have long been described. Understanding the biological mechanisms underlying these phenotypes can potentially unravel valuable new targets for therapy. For example, elite controllers (ECs) naturally maintain viremia at undetectable levels in sensitive monitoring assays. Despite the low frequency of ECs among the global population of people with HIV (PWH), numerous studies have been performed to identify the immune correlates involved in natural control of viremia in these individuals. Host factors such as protective HLA-I alleles⁴, enhanced effector functions of HIV-1–specific CD8⁺ T cells^5,6, and the selective persistence of proviruses integrated in silenced genomic regions⁷ have been associated with the EC phenotype. These observations provide a solid basis for studies seeking to emulate functional cure of HIV-1 infection through diverse immunological approaches^8,9, such as therapeutic vaccination, broadly neutralizing antibodies, and immune checkpoint blockers.

In contrast with ECs, viremic non-progressors (VNPs) constitute another extreme phenotype of HIV-1 infection, which has been only partially studied. Remarkably, VNPs can maintain persistently high CD4⁺ T-cell counts despite uncontrolled HIV-1 replication. The VNP phenotype is extremely rare among adult PWH, accounting for less than 0.1% of cases¹⁰, although it is relatively more frequent among the pediatric population (5–10%)¹¹. Interestingly, VNPs have a similar profile to that of natural hosts of simian immunodeficiency virus (SIV), such as sooty mangabeys and African green monkeys, whose lack of pathogenicity is probably due to mutual virus-host adaptation over thousands of years of coevolution^12–15. Studies of both human and non-human primate (NHP) models of VNPs have supported a more important contribution of host-related factors than virus-related factors^16–19. Nonetheless, various limitations have hampered a complete understanding of the VNP phenotype, as these studies usually assess a low number of individuals, lack a complete evaluation of the different components of the immune system, and do not include samples from ART-treated VNPs.

Based on previous studies, we hypothesized that the immune system of VNPs shows specific features that are responsible for the absence of HIV-1 pathogenicity. Understanding how VNPs preserve normal CD4⁺ T-cell counts can potentially benefit ART-treated people whose immune recovery is incomplete. In this study, we characterize one of the largest populations of adult VNPs to date using, for the first time, a multiomics approach to generate a comprehensive description of this extremely rare and valuable phenotype. We found that VNPs display distinctive immune characteristics that may limit the pathogenicity of HIV-1, revealing important similarities with natural hosts of SIV.

RESULTS:

Clinical characteristics of individuals and samples

We selected a cohort of 16 VNPs and 29 progressors with extensive follow-up of viral load and CD4⁺ T-cell counts (Fig.1A). VNPs were defined as individuals with a high viral load setpoint (mean viral load > 10,000 HIV RNA copies/ml) and slow CD4⁺ T-cell decay rates (< 10% CD4⁺ T-cell count loss annually) for a minimum period of 4 years (median, 7.4; range, 4.3–17.8 years) since diagnosis of HIV-1 infection or the estimated seroconversion date (Fig.1B). Individuals in the comparative group of progressors were selected according to their characteristics of viral load and CD4⁺ T-cell counts. Specifically, progressors had a comparable viral load setpoint (mean viral load > 10,000 HIV RNA copies/ml) but quicker CD4⁺ T-cell decay rates (> 10% CD4⁺ T-cell count loss annually) (Fig.1B). VNPs remained off ART for longer time than progressors owing to the absence of disease progression (Fig.1B). However, most individuals in both groups eventually started ART regardless of CD4⁺ T-cell counts, as recommended by current clinical guidelines. Demographic and epidemiological information for all the study participants is reported in (Table S1).

Figure 1. Clinical characteristics of individuals and samples.

(A) Longitudinal viral load and CD4+ T-cell count determinations for all individuals. (B) For all individuals, we report viral load setpoint (mean viral load), CD4+ T-cell decay rate (percentage of mean CD4+ T-cell count that is lost per year of untreated infection) and time from estimated seroconversion or diagnose to treatment initiation. To calculate mean viral load and mean CD4+ T-cell count, we considered all timepoints from 6 months after diagnose onwards to ensure the chronic phase had been reached. (C) For pre-ART samples included in the study, we report viral load and CD4+ T-cell counts. VNP and progressors are shown in blue and red, respectively. Median, IQR and range are presented in boxplots. Two-tailed Mann-Whitney U tests were used to compare groups. p≥0.05 not significant (ns), p<0.01 (**), p<0.0001 (****).

For each individual, we selected a pre-ART sample and an on-ART sample. We selected the earliest pre-ART sample available after acute infection to unravel biological factors intrinsic to VNPs that may predict their progression profile in an early stage of infection and to avoid possible confounding effects arising at more advanced stages. Plasma viremia was similar between groups, and, although CD4⁺ T-cell counts were significantly higher in VNPs, most individuals were still within the normal range (>500 cells/μl) (Fig.1C). We selected on-ART samples after at least 1 year of successful viral suppression to ensure that the levels of persistently infected cells were relatively stable. Clinical data on selected samples are presented in (Table S2).

VNPs have lower levels of proviral HIV-1 DNA than progressors before initiating ART

We first examined whether CD4⁺ T-cell preservation in VNPs could be explained by protection of certain cell subsets from infection. We therefore isolated 5 different CD3⁺CD8⁻ T-cell subsets using FACS and, together with whole peripheral blood mononuclear cells (PBMCs), quantified different parameters of infection using droplet digital PCR.

We observed significantly lower levels of total HIV-1 DNA (Fig.2A) and intact HIV-1 DNA (Fig.2B) in VNPs than in progressors in PBMCs during untreated infection. Next, we confirmed the same pattern across T-cell subsets; during untreated infection, VNPs had significantly lower levels of both total and intact HIV-1 DNA in the effector memory (Tem) compartment, but a similar trend was observed in all subsets analyzed (Fig.2A–2B). As expected, ART reduced total HIV-1 DNA levels, although the size of the HIV reservoir was similar between VNPs and progressors in all compartments analyzed (Fig.2C). Quantification of intact HIV-1 DNA after ART initiation was hampered because of a high number of samples below the limit of quantification (data not shown). We observed that all subsets make a similar contribution to the total pool of infected cells in VNPs and progressors, with central memory (Tcm) and transitional memory (Ttm) being the most abundant cell subpopulations both before and after initiation of ART (Fig.2D).

Figure 2. VNPs have lower levels of proviral HIV-1 DNA than progressors before initiating ART.

(A) Total HIV-1 DNA levels and (B) intact HIV-1 DNA levels in PBMCs and CD4⁺ T-cell subsets pre-ART, together with (C) total HIV-1 DNA levels on-ART were quantified by droplet digital PCR. We also report (D) contribution of each CD4⁺ T-cell subset to the pool of infected cells by total HIV-1 DNA. VNP and progressors are shown in blue and red, respectively. Median, IQR and range are presented in boxplots. Values below the limit of quantification (LOQ) are depicted in white. LOQ/2 is represented and considered for calculations of statistical significance. Two-tailed Mann-Whitney U tests were used to compare between VNPs and Progressors. Wilcoxon signed-rank tests were used to compare between pre-ART and on-ART samples of the same individuals. p<0.05 (*), p<0.01 (**). Tn: naïve; Tscm: stem cell memory; Tcm: central memory; Ttm: transitional memory; Tem: effector memory.

Together, these results suggest that cells from VNPs are partially protected against HIV-1 infection compared to cells from progressors. However, viral reservoir composition in VNPs resemble that of progressors, and viral reservoir size after ART-mediated viral suppression becomes similar in both groups.

CCR5Δ32 heterozygosity is more frequent in VNPs than in progressors and is associated with lower HIV-1 infection rates

We next determined whether lower rates of cellular infection could be explained by lower expression of the viral coreceptor CCR5, as observed in previous studies on pediatric VNPs and non-human primate models^20,21. We analyzed the expression of CCR5 on the cell surface by flow cytometry in peripheral cells and observed that expression of CCR5 was significantly lower in VNPs than in progressors in total CD4⁺ T cells (Fig.3A). When analyzed by CD4⁺ T-cell subset, CCR5 expression was significantly lower in the more differentiated memory subsets: Ttm, terminally differentiated (Temra), and Tem cells (Fig.3A).

Figure 3. CCR5Δ32 heterozygosity is more frequent in VNPs than in progressors and is associated with lower HIV-1 infection rates.

(A) Percentage of CCR5⁺ cells in total CD4⁺ T cells and CD4⁺ T-cell subsets by flow cytometry in samples pre-ART. VNP and progressors are shown in blue and red, respectively. Genotypic information from whole exome sequencing allowed the determination of the (B) percentage of CCR5⁺ cells by CCR5 genotype group (also represented per study group), (C) total HIV-1 DNA in PBMCs by CCR5 genotype group (also represented per study group). Median, IQR and range are presented in boxplots. In (B) and (C), individuals with CCR5Δ32 heterozygosity and full-length CCR5 are depicted with grey and black dots, respectively. Two-tailed Mann-Whitney U tests were used to compare between groups. p<0.05 (*), p<0.01 (**). Non-significant comparisons with p<0.1 are represented numerically. Tn: naïve; Tscm: stem cell memory; Tcm: central memory; Ttm: transitional memory; Tem: effector memory; Temra: terminally-differentiated.

To investigate the possibility that genetic polymorphisms account for the VNP phenotype, we performed whole-exome sequencing and confirmed that deletion CCR5Δ32 might play a relevant role. We found that 7/13 VNPs (53.8%) but only 4/25 progressors (16.0%) carried a heterozygous CCR5Δ32 deletion (p-value=0.015; chi-square test; Table S1), with the remaining individuals being homozygous for the full-length variant of CCR5. Accordingly, significantly lower expression of surface CCR5 protein (Fig.3B) and lower rates of infection (Fig.3C) were observed in participants with CCR5Δ32 heterozygosity, regardless of their status as VNPs or progressors.

These data suggest that CCR5Δ32 heterozygosity is associated with lower infection rates in target cells and may play an important role in the absence of pathogenicity in some VNPs. However, since CCR5Δ32 heterozygosity was neither common to all VNPs nor restricted to individuals from this group, we explored other immune-associated factors that may explain this likely heterogenous phenotype.

VNPs are characterized by lower differentiation and activation in the CD8⁺ T-cell compartment and lower levels of apoptosis in CD4⁺ T cells

To obtain a full picture of the immune status of the individuals we studied, we performed flow cytometry with peripheral T, B, NK, and myeloid cells (Fig.S1). Given the preferential targeting of CD4⁺ T cells in HIV-1 infection and the important role that CD8⁺ T cells play in other non-pathogenic HIV-1 phenotypes, we were particularly interested in the characterization of the T-cell compartment. In pre-ART samples, we found a similar CD4⁺ T-cell subset distribution in VNPs and progressors (Fig.4A), with no significant differences in the proportion of activated HLA-DR⁺CD38⁺ cells (Fig.4B). However, findings for the CD8⁺ T-cell compartment were strikingly different: VNPs had a significantly higher proportion of naïve CD8⁺ T cells (Fig.4C) and fewer activated memory CD8⁺ T cells across multiple subsets than progressors (Fig.4D). Initiation of ART promoted a slight contraction of CD8⁺ memory T cells and reduced the proportion of activated cells as expected, with a similar impact in both VNPs and progressors (Fig.S2A–S2C).

Figure 4. VNPs are characterized by lower differentiation and activation in the CD8+ T-cell compartment and lower levels of apoptosis in CD4+ T cells.

Flow cytometry was used to determine the (A) distribution and (B) activation levels of CD4⁺ T-cell subsets, the (C) distribution and (D) activation levels of CD8⁺ T-cell subsets, and (E-F) levels of T-cell death among CD3⁺CD8⁻ cells in samples pre-ART. CD3⁺CD8⁻ T cells were analyzed instead of CD4⁺ T cells to ensure that productively-infected cells with downregulated expression of CD4 are also included. Median and IQR are presented in (B,D). Median, IQR and range are presented in (E,F). HIV-, VNP and progressors are shown in grey, blue and red, respectively. Two-tailed Mann-Whitney U tests to compare between VNPs and Progressors in (A) and (B). Kruskal-Wallis and Conover-Iman test to compare between HIV-, VNPs and Progressors in (E,F). p<0.05 (*), p<0.01 (**), p<0.001 (***). Non-significant comparisons with p<0.1 are represented numerically. Tn: naïve; Tscm: stem cell memory; Tcm: central memory; Ttm: transitional memory; Tem: effector memory; Temra: terminally-differentiated.

Death of uninfected bystander CD4⁺ T cells explains most of the CD4⁺ T-cell decay during HIV-1 infection. Both Casp3/7-mediated apoptosis and Casp1-mediated pyroptosis have been associated with CD4⁺ T-cell decay^22,23. Given the radical difference in CD4⁺ T-cell preservation between VNPs and progressors, we wondered whether cell death pathways were differentially activated in these individuals before CD4⁺ T-cell count decay became evident. We observed that progressors have a significantly higher proportion of apoptotic active Casp3/7⁺ CD4 T cells than VNPs (Fig.4E). By contrast, VNPs and progressors presented similar levels of pyroptotic active Casp1⁺ CD4 T cells, which were higher than in HIV-1 seronegative controls (HIV-) (Fig.4F).

Among innate immune cells, we observed that the relative proportion of pre-ART CD3⁻ CD56⁺ NK cells was lower in both HIV+ groups than in HIV- (Fig.S2D). Moreover, the frequency of CD14⁺ monocytes and plasmacytoid dendritic cells (pDCs) tended to be lower among HIV+, with no significant differences between progressors and VNPs.

Taken as a whole, these results indicate that, in peripheral blood, VNPs are characterized by lower activation of Casp3/7-mediated apoptosis in bystander CD4⁺ T cells and a less differentiated and less activated CD8⁺ T-cell compartment than progressors.

Single-cell transcriptomics of PBMCs revealed profound downregulation of chronic IFN response in VNPs compared to progressors

Single-cell RNA sequencing (scRNAseq) offers insightful information on a wider range of cellular and molecular mechanisms. Therefore, we selected a subset of 7 VNPs and 7 progressors with matched demographic and clinical characteristics including viral load, CD4⁺ T-cell counts, sex, age, and geographical region (Table S1-S2) to perform a scRNAseq study on total PBMCs from pre-ART samples. Importantly, the VNPs and progressors selected were representative of their whole groups in terms of viral load, CD4⁺ T-cell count, and stage of disease progression, as indicated by the CD4/CD8 ratio at the sampling timepoint (Fig.S3A).

We obtained the single-cell transcriptome from more than 70,000 single cells belonging to multiple cell types. To better define the clusters within each cell type, we segregated single cells into 4 major groups based on canonical markers (Fig.S3B): myeloid cells, CD4 T cells, CD8 T/NK cells, and B cells. The proportion of these cell types with respect to all the cells analyzed was similar in VNPs and progressors (data not shown). Subsequent unsupervised clustering enabled the identification of 35 clusters (Fig.5A), which were annotated according to the expression of defining genes (Fig.S3C). The relative abundance of most clusters was similar in both VNPs and progressors (Fig.5A). However, in line with previous observations from flow cytometry, we observed that naïve cells were more abundant and effector cells less abundant in the cytotoxic compartment in VNPs than in progressors. In addition, frequencies of memory-unswitched B cells and gamma delta T cells were higher in VNPs, whereas frequencies of transitional B cells, memory unswitched CD11c+ B cells, and non-classical monocytes were lower.

Figure 5. Single-cell transcriptomics of PBMCs revealed profound downregulation of chronic IFN response in VNPs compared to progressors.

The transcriptome of more than 70,000 single-cells from 7 VNPs and 7 Progressors was analyzed. (A) UMAP representation and abundance of clusters within each of these cell types. The color code for the UMAP is indicated in the X axis of the boxplots. (B) Selection of GO terms differentially regulated in VNPs compared to Progressors across clusters. Relative expression of (C) Interferon Stimulated Genes, (D) genes involved in IFN pathway, (E) genes related with maintenance and functionality of the cytotoxic compartment, and (F) tissue fibrosis. Median, IQR and range are presented in boxplots. Blue dots represent VNPs; red dots represent Progressors. Two-tailed Mann-Whitney U tests were used to compare between VNPs and Progressors in (A). p<0.05 (*), p<0.01 (**). Non-significant comparisons with p<0.1 are represented numerically. Only GO terms and genes with p<0.05 are represented in (B-F).

We next assessed biological pathways that were differentially enriched between groups. Compared to progressors, VNPs showed profound downregulation of Gene Ontology terms related to antiviral defence and interferon (IFN) response (Fig.5B). Importantly, this difference was not limited to specific clusters, but rather was widely spread across multiple cell types, especially within myeloid and cytotoxic compartments. Analysis of individual genes revealed that a wide range of IFN-stimulated genes (ISGs), whose expression is triggered in response to stimulation of target cells by IFN, were downregulated in VNPs (Fig.5C). Remarkably, other genes participating in previous steps of the IFN signalling pathway (STAT1, IRF9) and regulating IFN production (IRF7) were also downregulated in VNPs (Fig.5D).

Besides general pathways showing differential activation levels between study groups, we looked further into specific genes that were differentially expressed in VNPs. Consistent with our previous flow cytometry findings, scRNAseq revealed lower expression of genes related to cytotoxic functions (eg, GZMA, GZMB, and NKG7) (Fig.5E) in the CD8 T-/NK-cell compartment. By contrast, expression was higher among VNPs for GZMK and genes involved in cell proliferation and survival such as IL7R and TCF7 in CD8⁺ T-cell clusters and inhibitory receptor KLRC1 in NK subsets.

Interestingly, genes with an established role in lymph node homeostasis and fibrosis, such as TGFB1 and LTB, are also downregulated in CD4 T cell clusters from VNPs (Fig.5F). These findings indicate that, as previous studies on human and NHP models have suggested, control of lymph node fibrosis in VNPs might help preserve CD4⁺ T-cell counts^24–27.

In summary, we confirmed reduced levels of activation of cytotoxic cell populations in VNPs. Of note, downregulation of the IFN response in multiple cell types is the most characteristic hallmark of VNPs compared to progressors, despite similar levels of viral load.

VNPs have lower plasma levels of IFNγ than progressors and preserved lipopolysaccharide responsiveness

Results from scRNAseq highlighted that moderation of the chronic IFN response played a key role in the VNP phenotype. The antiviral type-I IFN pathway and proinflammatory type-II IFN pathway differ in terms of the cell types and molecular mechanisms involved, although they eventually induce expression of a set of ISGs that largely overlap²⁸. Therefore, the weaker IFN response we observed in VNPs during chronic infection could potentially be the result of differences in the activation level among any of those pathways. Thus, we aimed to quantify IFNα and IFNγ in plasma samples from the same pre-ART timepoints using single-molecule array (SIMOA) and ELISA, respectively. Levels of IFNγ were higher among progressors than among HIV– controls, although an intermediate state between HIV– controls and progressors was observed in VNPs (Fig.6A). A similar trend was observed for IFNα. However, many samples were below the limit of quantification of the technique, thus hindering our interpretation (Fig.6B). We therefore measured expression of Siglec-1, a marker of IFNα response²⁹, in peripheral CD14⁺ monocytes using flow cytometry of pre-ART samples and confirmed that VNPs express lower levels of Siglec-1 protein than progressors, although these are still higher than in HIV– individuals (Fig.6C). These results suggest that the difference in chronic activation of IFN response between VNPs and progressors could result from a differential production capacity for IFNγ, although the potential role of IFNα should not be dismissed.

Figure 6. VNPs have lower plasma levels of IFNγ than progressors and preserved lipopolysaccharide (LPS) responsiveness.

Quantification of plasma (A) IFNγ and (B) IFNα by ELISA and SIMOA, respectively, and (C) flow cytometry measurement of Siglec-1 expression on CD14⁺CD11c⁺ myeloid cells. Determination of (D) IFNα production with or without overnight (o/n) stimulation with TLR7 agonist Imiquimod (5μg/ml) or TLR9 agonist ODN 2216 (1μM), and (E) determination of IFNγ and IL-18 production, and IFNγ/IL-18 ratio after o/n stimulation with TLR4 agonist LPS (100ng/ml). A schematic representation of the biological process being assessed accompanies each graph. Median, IQR and range are presented in boxplots. Quantifications below the LOQ of the assay are depicted in white. The LOQ for IFNα determination by SIMOA (B) was defined as 0.071 pg/ml according to manufacturer’s indications and internal calculations of blank measurement variability. The LOQ for IFNα determination by proximity-based amplification (D) was defined as the concentration of the most-diluted standard in each assay plate. HIV–, VNP and progressors are shown in grey, blue and red, respectively. Kruskal-Wallis and Conover-Iman test were used to compare between HIV- , VNPs and Progressors in (A,C,E). IFNα production in (D), was evaluated by non-parametric Peto left-censored test adjusted by the Holm method for multiple comparisons was used to consider the values below LOQ. p<0.05 (*), p<0.01 (**), p<0.0001 (****). Non-significant comparisons with p<0.1 are represented numerically.

We performed a series of functional assays to test whether the different level of chronic IFN response in VNPs and progressors is due to intrinsic differences in IFN production capacity. To evaluate IFN production capacity after in vitro stimulation of molecular pathways triggered by sensing of diverse pathogens (of either viral or bacterial origin), we cultured PBMCs from early chronic infection of VNPs and progressors as well as HIV– controls in the presence of the TLR7 agonist imiquimod, the TLR9 agonist ODN 2216, and the TLR4 agonist lipopolysaccharide (LPS) and quantified IFNα and IFNγ production in the supernatants after 18–20 h (Fig.6D–E).

Regarding IFNα production, we observed that PBMCs from HIV+ individuals produce lower amounts of IFNα upon stimulation of TLR7 or TLR9, although no differences were observed between VNPs and progressors (Fig.6D). This lower responsiveness is partially due to the lower percentage of pDCs in HIV+ samples, as suggested by the strong positive correlation between those parameters (Fig.S4A). However, normalization of IFNα levels by percentage of pDCs shows that HIV-1 infection indeed reduces the in vitro response of pDCs in both groups (Fig.S4B).

After exposure to TLR7 or TLR9 agonists we also measured IFNγ production, again observing that induction of IFNγ is weaker in cells from HIV+ individuals than in HIV– controls (Fig.S4C). Of note, this lower responsiveness was also observed upon stimulation of TLR4 in progressors, but not in VNPs (Fig.6E). To further explore whether this difference between VNPs and progressors was due to distinct functionality of IFNγ-producing cells or whether it originated at a previous step in the IFNγ pathway, at the level of myeloid cells producing pro-inflammatory cytokines that stimulate IFNγ production, we quantified the intermediary IL-18 cytokine in the same cultures, although we did not find differences across groups (Fig.6E). However, the IFNγ/IL-18 ratio (Fig.6E) was reduced in progressors, suggesting decreased responsiveness of IFNγ-producing cells, in contrast with preserved functionality in VNPs.

In conclusion, our results demonstrated that the ability to produce IFNα and IFNγ in response to innate sensing was altered during chronic HIV-1 infection. Both HIV+ groups showed reduced induction of IFN upon in vitro stimulation of TLR7/9 than HIV– controls, probably owing to the lower abundance and productivity of IFNα-producing pDCs resulting from chronic in vivo exposure to viral components. Conversely, progressors showed reduced capacity to produce IFNγ upon stimulation of TLR4 with LPS, while VNPs resembled HIV– controls.

VNPs have lower circulating levels of gut disruption markers and alterations in tryptophan degradation

Since our results from scRNAseq analysis suggested that a low degree of chronic immune activation may contribute to the VNP phenotype, we measured the concentration of different markers of gut integrity (zonulin), microbial translocation (LPS-binding protein [LBP] and β-glucan), and chronic inflammation (sCD14, sCD163, IL-6, IL-18, and TNFα) in pre-ART plasma samples from VNPs and progressors and compared them with plasma samples from HIV– controls. As expected, levels of many biomarkers—including TNFα and sCD163—were higher in samples from HIV+ individuals than HIV– controls (Fig.7A). Data from bacterial and fungal translocation markers did not reveal conclusive results. Remarkably, and in contrast to progressors, levels of circulating zonulin were similar in VNPs and HIV– individuals, suggesting better integrity of gut mucosa in VNPs than in progressors.

Figure 7. VNPs have lower circulating levels of gut disruption markers and alterations in tryptophan degradation.

(A) Quantification of plasma markers indicative of gut disruption, microbial translocation and chronic inflammation. HIV–, VNP and progressors are shown in grey, blue and red, respectively. (B) Heatmap and PCA representation with the relative expression of 17 plasma metabolites with significantly different expression between the 3 groups. Metabolites with different concentration between VNPs and Progressors are highlighted in bold and with a red arrow. (C) Quantification of plasma levels of the Trp catabolite anthranilic acid. HIV–, VNP and progressors are shown in grey, blue and red, respectively. (D) Correlogram showing correlations between plasma markers of gut disruption, microbial translocation, inflammation, and Trp degradation. Median, IQR and range are presented in boxplots. Kruskal-Wallis and Conover-Iman test were used to compare between HIV-, VNPs and Progressors in (A) and (C). Metabolites with significantly different concentration between groups in (B) were selected upon Kruskal-Wallis or ANOVA test with p<0.05. Spearman correlation was used in (D), the correlation coefficient is color-coded, and significant p-values are represented with asterisks. p<0.05 (*), p<0.01 (**), p<0.001 (***).

To explore whether additional specific metabolites play a role in immune modulation and absence of pathogenicity in VNPs, we performed an untargeted metabolomic analysis by mass spectrometry in pre-ART plasma samples from VNPs and progressors and compared them with those of HIV– controls. We quantified the relative abundance of 190 metabolites, of which 17 had significantly different levels between the 3 groups (Fig.7B). As shown by principal component analysis, these metabolites mainly differentiated HIV– controls from HIV+ individuals. Only 3 metabolites differed between VNPs and progressors, namely, 3-methylphenylacetic acid (a fungal xenobiotic), cholic acid (a bile acid related to cholesterol metabolism), and anthranilic acid. In particular, the levels of anthranilic acid were similar in HIV– controls and VNPs but significantly higher in progressors (Fig.7C). Of note, anthranilic acid is an intermediate degradation product in the tryptophan (Trp) catabolism pathway, which has previously been associated with progression of HIV-1 infection^30,31. Indeed, in our setting, levels of Trp catabolites correlated positively with multiple immune mediators, especially IFNγ, as a hallmark of HIV-1 infection (Fig.7D). We hypothesize that reduced chronic levels of IFNγ in VNPs might be associated with lower levels of Trp degradation and, in particular, anthranilic acid, with a possible impact on disease progression.

In conclusion, despite important similarities between both HIV+ groups, VNPs resemble HIV– controls in levels of zonulin and anthranilic acid, which are markers of gut integrity and the Trp catabolism pathway, respectively.

DISCUSSION:

VNPs are an extremely infrequent phenotype of HIV-1 infection characterized by relatively stable CD4⁺ T-cell counts over prolonged periods of time despite uncontrolled viral replication. Understanding the mechanisms behind this spontaneous resistance to HIV-1 pathogenicity might unravel novel targets with potential therapeutic application for the management of chronic immune activation and insufficient CD4⁺ T-cell recovery in the ART-treated population. The present study analyzes one of the largest VNP cohorts and takes advantage of novel single-cell and multiomics technologies to generate a comprehensive description of the specific biological features that characterize VNPs.

We observed that the CCR5Δ32 variant in heterozygosity is more frequent among VNPs than progressors and is associated with reduced levels of CCR5 protein expression and partial protection from infection. Interestingly, low expression of CCR5 has been reported in multiple natural hosts of SIV infection^32,33 and pediatric VNPs¹¹. Specifically, the CD4⁺ Tcm subset from natural SIV hosts failed to upregulate CCR5 expression upon in vitro activation²⁰, and lower levels of infection in this subset were found in vivo^11,20,21. In addition to analyzing infection levels at pre-ART timepoints, we examined, for the first time, the specific characteristics of the latent HIV-1 reservoir in VNPs on ART. We found that the size and composition of the reservoir in VNPs and progressors was very similar after 1 year of viral suppression, despite lower levels of infection in VNPs in the early chronic phase. The exceedingly longer time that VNPs remained off ART while CD4⁺ T-cell counts were still high may have favored the seeding of the viral reservoir. This fact may compensate for their partial protection from infection and lead them to eventually reach similarly high levels of HIV-1 DNA than progressors, thus explaining their similarities during ART. Finally, lower expression of CCR5 protects from infection with R5-tropic viruses, but not against X4- or dual-tropic viruses. Remarkably, a previous study³⁴ on 2 untreated VNPs who lost their status after more than 17 years associated the sudden sharp increase in viral load and accelerated decay of CD4⁺ T-cell counts with a shift in tropism to X4-tropic viruses, thus highlighting the major role that CCR5 may play in these VNPs. However, the absence of pathogenicity in VNPs carrying the full-length CCR5 allele (i.e., approximately half of all VNPs) might be explained by alternative factors, such as modulation of CCR5 expression by epigenetic mechanisms, or be independent of this viral coreceptor.

A robust observation from our study is that VNPs show different signs of lower immune activation than progressors. Studies on natural SIV hosts have established a link between a preserved mucosal Th17 compartment^35,36 and gut epithelial barrier function³⁷, reduced levels of microbial translocation³⁸, and low chronic immune activation²⁴ as the main contributors to absence of pathogenicity. Similar traits have been observed in pediatric VNPs^11,39, although the degree of chronic immune activation in adult VNPs remains controversial^10,21,40. Given the preferential abundance of memory CD4⁺CCR5⁺ T cells in gut-associated lymphoid tissue^41,42, lower levels of infection in this anatomical compartment may have contributed to preserved gut homeostasis, as suggested by lower levels of zonulin, a marker of gut disruption, in VNPs. Alternatively, reduced tryptophan catabolism in VNPs might also maintain the Th17/Treg balance in favor of increased Th17, therefore preserving gut immunity and hindering gut mucosal disruption and microbial translocation⁴³. While we did not detect differential levels of plasma markers of microbial translocation, this might have been hampered by a lack of sensitivity in our analyses or the kinetic nature of these markers, as we were not able to detect differences, even between HIV– controls and progressors. Altogether, these observations show the potential relevance of preservation of mucosal immunity in VNPs and reinforce the need for further investigation in this anatomical compartment.

Adaptive immunity, in particular the HIV-1–specific CD8⁺ T-cell response, is thought to play a minor role in the VNP phenotype, as suggested by the high levels of viral replication observed. Previous studies have shown minimal alteration of chronic viral load or CD4⁺ T-cell counts after CD8⁺ T-cell depletion in natural SIV hosts^44,45 and suggested a limited contribution of HIV-1–specific cytotoxic T-cell responses in adult VNPs^46,47. Our results show that the global compartment of CD8⁺ T cells in VNPs is less activated and less differentiated and expresses lower levels of cytotoxic molecules at the transcriptomic level than in progressors. Instead, these cells express higher levels of IL7R and TCF7 encoding IL7Rα and TCF1, which are related to homeostatic proliferation, long-term survival, and stem-like properties. Interestingly, CD8⁺TCF1⁺PD-1⁺ T cells have been associated with recall ability in chronic viral infections⁴⁸, and a recent study revealed enrichment of HIV-1– specific CD8⁺TCF1⁺PD-1⁺ T cells with high proliferative capacity and effector functionality in post-treatment pediatric VNPs⁴⁹. We speculate that these cells may not contribute to control of HIV-1 viremia, given the persistently high level of viral replication in adult VNPs, but instead may serve as a memory backup ready to quickly respond against reactivation of other chronic viral coinfections such as those caused by HBV, HCV, and CMV, thus helping to suppress a potential source of increased immune activation.

Natural SIV hosts mount an early, strong, but transient IFN response during acute infection that resolves rapidly, in contrast with the persistently high long-term IFN responses seen in pathogenic SIV infection^50–55. A parallelism with VNP and progressor HIV-1 phenotypes, respectively, has been suggested in transcriptional studies¹⁰. However, the molecular mechanisms behind the differential chronic IFN response remain unclear. In our study, we report in unprecedented detail that compared to progressors, VNPs are characterized by marked downregulation of chronic IFN responses spanning across multiple cell types, with potential links to lower levels of immune activation. Type-I and type-II IFN pathways are tightly interconnected and could both potentially explain our observations. In our cohort, we show that VNPs have significantly lower plasma levels of IFNγ than progressors, suggesting an important role for type-II IFNs in maintaining the IFN response during chronic HIV-1 infection. Although we must not rule out the role of IFNα in our setting, previous studies have also shown that IFNγ contributes markedly to maintaining upregulation of ISGs in the context of chronic SIV infection⁵⁶.

IFNγ production is primarily induced in lymphoid cells (CD4⁺ Th1, CD8⁺ T, NK cells) by proinflammatory cytokines secreted by myeloid cells that become activated upon recognition of microbial products such as LPS. Therefore, a possible explanation for the lower levels of plasma IFNγ in VNPs would be diminished responsiveness of any of these cells to their activating stimuli. For instance, studies on sooty mangabeys showed reduced production of proinflammatory cytokines by myeloid cells in response to LPS⁵⁷, which was later associated with blunting of the response by genetic polymorphisms in TLR4⁵⁸. However, we did not find a similar effect in our cohort. Although we only considered TLR-mediated stimulation, and activation by other pathogen receptors should also be assessed, we clearly showed that the production of IFNα and IFNγ upon equivalent TLR stimulation is not lower in VNP than in progressors. Indeed, stimulation by LPS resulted in a higher IFNγ/IL-18 ratio in VNPs, interestingly suggesting, more pronounced responsiveness of IFNγ-producing cells to proinflammatory stimuli in this group. Such preserved functionality might indicate a less exhausted phenotype derived from reduced exposure to activating stimuli in vivo (i.e., reduced microbial translocation, antigen presentation, and inflammation) through the historical course of infection. Alternatively, this could indicate a more tolerogenic response of IFNγ-producing cells to elevated chronic stimulation. Any of these hypotheses would explain why VNPs have lower in vivo levels of IFNγ despite higher responsiveness to our strong in vitro stimulation and would be in line with the lower degree of differentiation and activation in the cytotoxic compartment, a key contributor to IFNγ production, that we observe during chronic infection in VNPs.

Microbial products (e.g., LPS), as well as IFNγ, have proved to promote the Trp catabolism pathway, which is in turn associated with progression of HIV-1 infection^30,31. Importantly, in vitro experiments have shown that IFNγ can induce Trp degradation and that Trp catabolites can, in turn, induce CD4⁺ T-cell apoptosis⁵⁹. Indeed, apart from the elimination of infected cells by direct viral cytopathic effects and immune-mediated clearance, CD4⁺ T-cell count decay, importantly, relies on the death of abundant uninfected bystander CD4⁺ T cells by activation-induced mechanisms involving both apoptosis and pyroptosis^22,60. Here, we have shown that bystander CD4⁺ T cells in VNPs are characterized by lower levels of apoptosis, but not pyroptosis, than progressors. Similarly, lower levels of CD4⁺ T-cell apoptosis have been observed in natural SIV hosts than in pathogenic non-human primate models^24,26,61,62. We observed that levels of anthranilic acid, a Trp degradation product, are lower in VNPs than in progressors. The biological role of anthranilic acid is less well characterized than that of other Trp catabolites, such as kynurenine or quinolinic acid. Previous studies suggest that this metabolite may act as a counterbalance mechanism to limit excessive fibrosis derived from inflammatory conditions^63–65, perhaps suggesting increased fibrosis during chronic infection in progressors, but not in VNPs. Indeed, scRNAseq has shown reduced expression of TGFB1 and LTB, which encode molecules involved in lymphoid tissue homeostasis and fibrosis in CD4⁺ T cells from VNPs. Also, the composition of the B cell compartment, with moderated frequencies of transitional B cells and more abundant memory unswitched population in VNPs, may be indicative of a better preservation of the lymphoid tissue architecture. Although we should be cautious about inferring tissue events from peripheral markers, these might be indicators of limited lymphoid tissue fibrosis. It has been shown that lymph node fibrosis disrupts tissue architecture and prevents the access of CD4⁺ T cells to essential survival factors, thus contributing to increased cell death; notably, this process is absent in natural SIV hosts^25,27. Therefore, we suggest that lower IFNγ levels in VNPs might help to reduce the concentration of detrimental Trp catabolites and that, together with preservation of lymphoid tissue homeostasis, this might contribute to maintaining CD4⁺ T-cell counts.

To conclude, we report the results of one of the most extensive and comprehensive analyses of the VNP phenotype of HIV-1 infection to date. We showed that protection against CD4⁺ T-cell death is the result of a complex and heterogenous phenotype driven by low expression of the CCR5 viral coreceptor, reduced levels of immune activation and chronic IFN responses, probably associated with preservation of lymphoid tissue and gut barrier integrity and altered tryptophan catabolism. The extensive knowledge generated in this study will serve as a source for the validation of novel biomarkers of good prognosis, as well as the basis for the design of targeted interventions to restore immune status in PWH with incomplete CD4⁺ T-cell count recovery

Limitations of the study

Our study is subject to a series of limitations. The number of individuals recruited is relatively low. However, it remains one of the largest and most comprehensive studies on adult VNPs performed to date and the only one comparing longitudinal samples before and after initiation of ART in this population. Many of the observations made in our study are descriptive; therefore, further research is needed to dissect the molecular and cellular bases of the alterations introduced here, including validation of observations in functional assays. In this regard, we generated a considerable amount of genomic, transcriptomic, metabolomic, and phenotypic data that might be of considerable value for future studies in the field. Universal access to ART, which carries invaluable benefits for PWH, including VNPs, strongly restricts access to pre-ART samples from these uncommon individuals. In our study, the limited amount of available cryopreserved PBMCs has hindered our ability to precisely assess HIV-1 infection rates in less abundant circulating CD4⁺ T-cell subsets, like Tfh cells. Alternatively, pre-ART tissue samples from relevant anatomical locations such as gut and lymph nodes would offer extremely interesting information on immune features of VNPs. Unfortunately, they are unavailable in our cohort. Future studies on this phenotype should therefore consider identifying VNPs in settings with delayed implementation of ART, including results from non-human primate models with similar non-pathogenic characteristics, and reframing hypotheses so that they can be successfully addressed in samples from ART-treated individuals.

STAR METHODS:

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Maria C. Puertas (mcpuertas@irsicaixa.es).

Materials availability

This study did not generate new unique reagents.

Data and code availability

Data.

Raw data files from exome sequencing have been deposited at the European Genome-phenome Archive (EGA) database and are available upon request to the Data Access Committee (mcpuertas@irsicaixa.es, lgarrido@irsicaixa.es) as of the day of publication (EGA: EGAS50000000079). Raw data files from single-cell RNA sequencing have been deposited at the European Nucleotide Archive (ENA) and are publicly available as of the day of publication (ENA: PRJEB68223). Metabolomic data has been deposited at the Metabolomics Workbench and is publicly available as of the day of publication (Metabolomics Workbench: PR001838).

Code.

All original code is deposited at GitHub (https://github.com/irsi-grec/2024vnps.git) and is publicly available as of the day of publication.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Study participants

The following inclusion criteria were considered to select HIV⁺ individuals in this study.

• Viremic Non-Progressors (VNPs; n=16). Mean VL (viral load) > 10,000 HIV RNA copies/ml during chronic infection, starting six months after date of seroconversion / diagnose. CD4⁺ T-cell count decay rate < 10% cells/μl/year for at least 4 years.

• Progressors (n=29). Mean VL > 10,000 HIV RNA copies/ml during chronic infection. CD4⁺ T-cell count decay rate > 10% cells/μl/year.

Individuals were selected upon retrospective examination of clinical recordings of HIV⁺ individuals followed up at University Hospital “Germans Trias i Pujol” (Badalona, Spain). Additional information on these individuals, retrieved from their clinical records, is included in Table S1. Cryopreserved PBMCs and plasma+EDTA samples from early chronic phase of untreated infection (pre-ART samples) and after at least 1 year of ART-mediated virologic suppression (on-ART samples) were selected upon availability. All individuals with ongoing follow-up except 1 VNP had started ART by the end of 2016 regardless of their CD4⁺ T-cell counts, as recommended by current guidelines. Additional information on the samples is included in Table S2.

Samples from different HIV seronegative controls were included in different parts of the study. To perform flow cytometry Panels 2–4 and functional assays, 10 cryopreserved PBMC samples from HIV seronegative individuals were included. Samples were obtained by Ficoll extraction of seronegative blood purchased to Banc de Sang i Teixits (Barcelona, Spain). To perform metabolomics experiments, as well as ELISA, MSD and LAL assay determinations, 6 cryopreserved plasma+EDTA samples from HIV seronegative male individuals aged between 21 and 45 years were included. Samples were donated and processed at the phlebotomy service at The Wistar Institute (Philadelphia, US). To perform IFNα determinations, 5 cryopreserved plasma+EDTA samples from HIV seronegative individuals were included. Samples were donated and processed at IrsiCaixa AIDS Research Institute (Badalona, Spain).

Written informed consent was obtained from all study participants (IrsiCaixa’s VNPs Cohort and healthy donors), and the study followed all bioethical and legal requirements. This retrospective observational study was approved by the Ethics Committee of the University Hospital “Germans Trias i Pujol” (ethics committee approval number PI-19-273). All clinical investigations were conducted according to the standards indicated by the Declaration of Helsinki.

METHOD DETAILS

Flow cytometry

A total of 4 different flow cytometry panels were carried out in this study (Panels 1–4).

For Panel 1, cryopreserved PBMCs were quickly thawed at 37°C, washed and resuspended in RPMI medium (Gibco Sciences) containing 20% (v/v) of Fetal Bovine Serum (FBS; Gibco Sciences). Cells were counted at NucleoCounter NC-3000 (ChemoMetec). Cells were then resuspended in PBS+0.5%FBS and stained with antibodies against CD3 (clone SK7; PerCP; #345766; BD Biosciences), CD4 (clone RPA-T4; APC-R700; #564975; BD Biosciences), CD8 (clone SK1; V500; #561617; BD Biosciences), TCRγδ (clone B1; APC Fire 750; #331228; BioLegend), CD45RA (clone HI100; AF647; #304154; BioLegend), CD95 (clone DX2; PE; #305608; BioLegend), CCR7 (clone G043H7; PE/DAZZLE 594; #353236; BioLegend), CD27 (clone M-T271; FITC; #555440; BD Biosciences), CCR5 (clone 2D7; BV421; #562576; BD Biosciences), HLADR (clone L243; PE/Cy7; #307616; BioLegend), CD38 (clone HIT2; BV605; #303532; BioLegend), and LIVE/DEAD Fixable Near-IR Dead Cell Stain kit (Invitrogen; #L10119; Invitrogen). Immunophenotypic data was generated upon acquisition at a LSRFortessa (BD Biosciences) and CD3⁺CD8⁻ T-cell subsets were isolated by FACS at a FACSAria Cell Sorter (BD Biosciences).

For Panels 2–4, cryopreserved PBMCs were thawed at 37°C, washed and resuspended in RPMI medium containing 20% of FBS, and cells were counted at NucleoCounter NC-3000 (ChemoMetec). A total of 1 million cells were used in each panel for Panels 2–4.

• In Panel 2, cells were stained with antibodies against CD3 (clone SK7; PerCP; #345766; BD Biosciences), CD19 (clone 4G7; PerCP; #345778; BD Biosciences), CD14 (clone M5E2; FITC; #555397; BD Biosciences), CD11c (clone B-LY6; APC; #559877; BD Biosciences), HLADR (clone G46–6; BV786; #564041; BD Biosciences), CD123 (clone 6H6; PE/Cy7; #25–1239-42; eBiosciences), CD303 (clone V24–785; BV605; #748004; BD Biosciences), CD169 (clone 7–239; PE; #346004; BioLegend), and LIVE/DEAD Fixable Near-IR Dead Cell Stain kit (#L10119; Invitrogen).

• In Panel 3, cells were stained with antibodies against CD3 (clone HIT3a; PE; #300308; BioLegend), CD4 (clone RPA-T4; BV605; #562658; BD Biosciences), CD8 (clone SK1; BV510; #344732; BioLegend), CD56 (clone NCAM16.2; BV786; #564048; BD Biosciences), LIVE/DEAD Fixable Near-IR Dead Cell Stain kit (#L10119; Invitrogen), and FAM FLICA Caspase 3/7 kit (#ICT094; Bio-Rad).

• In Panel 4, cells were stained with antibodies against CD3 (clone HIT3a; PE; #300308; BioLegend), CD4 (clone RPA-T4; BV605; #562658; BD Biosciences), CD8 (clone SK1; BV510; #344732; BioLegend), LIVE/DEAD Fixable Near-IR Dead Cell Stain kit (#L10119; Invitrogen), and FLICA 660 Caspase 1 kit (#ICT9122; Bio-Rad).

• Samples from Panels 2–4 were resuspended in PFA 1% and acquired at a LSRII (BD Biosciences).

Reservoir studies

The proportion of PBMCs and FACS-sorted CD3⁺CD8⁻ T-cell subsets harboring total HIV DNA and intact HIV DNA was measured by ddPCR using a QX100 Droplet Digital PCR System (Bio-Rad), as previously described^66,67.

PBMCs were lysed in lysis buffer for DNA extraction. FACS-sorted CD3⁺CD8⁻ T-cell subsets underwent DNA extraction using RNA/DNA Purification Micro Kit (#50300; Norgen Biotek).

Briefly, for total HIV DNA, lysed extracts from PBMCs were screened to quantify the frequency of proviral DNA based on primer/probe sets annealing at the 5′ long terminal repeat (LTR) or GAG. The HIV primer/probe set yielding the highest proviral quantification in the screening was assigned to each individual and subsequently used to quantify total HIV DNA in T-cell subsets. The cellular single-copy RPP30 gene was measured in parallel to normalize sample input in HIV DNA quantifications.

To measure the frequency of intact provirus, duplex ddPCR was performed in lysed extracts from PBMCs using the packaging signal (ψ) and nonhypermutated Env primer/probe sets, as indicated in the original IPDA protocol⁶⁶. For those samples failing Env detection, a secondary prime/probe set targeting Env was used to rescue intact provirus quantification⁶⁸. For those samples failing (ψ) detection, a secondary prime/probe set targeting 5’LTR was used to rescue intact provirus quantification. To normalize data and to correct for DNA shearing in each sample, 2 primer/probe sets targeting the RPP30 gene were used in a duplex ddPCR simultaneous run. Once a primer/probe set was assigned to each individual, it was subsequently used to quantify intact HIV DNA in T cell subsets.

All probes were double-quenched 6-carboxyfluorescein (FAM)/6-carboxy-2,4,4,5,7,7-hexachlorofluorescein (HEX)-ZEN-Iowa Black FQ and were purchased from IDT (Integrated DNA Technologies, Belgium). Sequences of all primer/probes are shown in Table S3. The following settings were used for amplification for total HIV-1 DNA (enzyme activation 10’ 95°C; amplification 40 cycles 30” 94°C + 1’ 57°C; enzyme inactivation 10’ 98°C) and intact HIV-1 DNA (enzyme activation 10’ 95°C; amplification 45 cycles 30” 94°C + 1’ 53°C; enzyme inactivation 10’ 98°C).

Determination of viral tropism

Viremic plasma samples (pre-ART samples) and cryopreserved PBMCs (on-ART samples) were used to determine viral tropism.

For pre-ART samples, HIV RNA was extracted from plasma samples using QIAamp Viral RNA Mini kit (#52904; Qiagen). Viral RNA was retrotranscribed and amplified by RT-PCR and nested PCR using SuperScript^™ One-Step RT-PCR System with Platinum^™ Taq DNA Polymerase (#12574; Invitrogen) and Platinum^™ Taq DNA Polymerase High Fidelity (#11304; Invitrogen), according to manufacturers’ instructions. Amplification was confirmed by electrophoresis. Conditions for RT-PCR were as follows: reverse transcription (30' 52°C, 2' 94°’); amplification 35 cycles (30” 94°C, 30” 55°C, 270” 68°C); termination (5’ 68°C). Conditions for nested PCR were as follows: denaturation (2’ 94°C), amplification 30 cycles (30” 94°C, 30” 55°C; 4’ 68°C), termination (5’ 68°C).

For on-ART samples, cryopreserved PBMCs were thawed at 37°C, washed and resupended in RPMI medium (Gibco Sciences) containing 20% (v/v) of Fetal Bovine Serum (FBS; Gibco Sciences). Cells were counted at NucleoCounter NC-3000 (ChemoMetec). Five million cells were processed for DNA extraction using QIAamp DNA Blood Mini Kit (#51104; QiaGen). Proviral DNA was amplified by nested PCR using Platinum^™ Taq DNA Polymerase High Fidelity (#11304; Invitrogen), according to manufacturers’ instructions. Amplification was confirmed by electrophoresis. Conditions for the first PCR were as follows: denaturation (2’ 94°C), amplification 40 cycles (15” 94°C, 30” 55°C, 4’ 68°C), termination (5’ 68°C). Conditions for nested PCR were as follows: denaturation (2’ 94°C), amplification 30 cycles (30” 94°C, 30” 55°C, 4’ 68°C), termination (5’ 68°C).

Amplification products from both pre-ART and on-ART samples were purified with ExoSAP-IT (#78200; Applied Biosystems) by mixing 10μl of sample with 4μl of ExoSAP-IT and incubation at 37°C for 15’ followed by 80°C for 15’. A plate was prepared by loading 5μl of purified sample and 5μl of sequencing primer at 5–10μM and sent to Macrogen for Env sequencing.

Single-cell RNA sequencing

Single-cell transcriptomic data from total PBMCs was obtained for a subset of 14 samples (Table S2). Cryopreserved PBMCs were thawed at 37°C, washed and resuspended in RPMI medium (Gibco Sciences) containing 20% (v/v) of Fetal Bovine Serum (FBS; Gibco Sciences). Cells were counted at NucleoCounter NC-3000 (ChemoMetec). One million cells were processed according to manufacturers’ instructions to ensure virus inactivation (‘Methanol Fixation of Cells for Single Cell RNA Sequencing’; 10X Genomics). Briefly, cells were washed in pre-chilled 1X DPBS (Gibco Sciences), resuspended in pre-chilled 100% methanol drop-by-drop and incubated for 30’ at −20°C. Fixed samples were sent to National Center of Genomic Analysis (CNAG-CRG, Barcelona, Spain), where single-cell RNA sequencing was performed. Briefly, fixed cells were equilibrated to 4°C, centrifuged at 1,000 rcf for 5’ and resuspended in 3X SSC Buffer (Sigma Aldrich) containing 0.04% BSA (Miltenyi Biotec), 1mM DTT (Sigma Aldrich) and 0.2 U/ul RNase Inhibitor (Applied Biosystems). Cell concentration and viability (<2%) were verified by counting with a TC20^™ Automated Cell Counter (Bio-Rad) after staining cells with Trypan blue (Thermo Fisher Scientific). Each sample was loaded for a target recovery of 5000 cells on the Chromium Controller system (10X Genomics)_using the Next GEM Single Cell 3’ Reagent Kits v3 (#PN-1000075, 10X Genomics). Corresponding cDNA sequencing libraries were prepared according to manufacturer’s instructions (‘Chromium Single Cell 3ʹ Reagent Kits v3 User Guide’, #CG000183), and their size distribution and concentration were verified on an Agilent Bioanalyzer High Sensitivity chip (Agilent Technologies). Finally, library sequencing was carried out on a NovaSeq 6000 sequencer (Illumina) using the following sequencing conditions: 28 bp (Read 1) + 8 bp (i7 index)_+ 0 bp (i5 index) + 91 bp (Read 2), to obtain >20,000 paired-end reads per cell.

Whole-exome sequencing

Cryopreserved PBMCs were thawed at 37°C, washed and resuspended in RPMI medium (Gibco Sciences) containing 20% (v/v) of Fetal Bovine Serum (FBS; Gibco Sciences). Cells were counted at NucleoCounter NC-3000 (ChemoMetec). Five million cells were processed for DNA extraction using QIAamp DNA Blood Mini Kit (#51104; QiaGen). Extracted DNA was sent to National Center of Genomic Analysis (CNAG-CRG, Barcelona, Spain), where whole-exome sequencing was performed. Paired-end multiplex libraries were prepared according to manufacturer's instructions and enriched with the Kapa HyperExome DNA genome design exome kit (#09062564001; Roche Diagnostics). Libraries were loaded to Illumina flowcells for cluster generation prior to producing 150 base read pairs on a NovaSeq6000 instrument following the Illumina protocol. Image analysis, base calling and quality scoring of the run were processed using the manufacturer’s software Real Time Analysis and followed by generation of FASTQ sequence files.

Plasma untargeted metabolomics

Metabolomics analysis was performed as described previously⁶⁹. Briefly, polar metabolites were extracted from 50 μl plasma samples with 500 μl of ice-cold 80:20 (v/v) methanol/water spiked with 1.5 μM heavy-labeled amino acid internal standard mix. Deproteinated supernatants were stored at −80 °C prior to analysis. A quality control (QC) pool sample was made by pooling a small aliquot from each sample extract. The metabolomics analysis was performed at The Wistar Institute Proteomics and Metabolomics Shared Resource on a Thermo Q Exactive HF-X mass spectrometer in-line with a Thermo Vanquish Horizon UHPLC. Samples were analyzed by LC-MS/MS in a randomized order, and 4 μl of each sample was injected per run. The QC pool was run periodically throughout the analysis. LC separation was performed under HILIC, pH 9 condition using a ZIC-pHILIC column (2.1 x 150 mm, EMD Millipore). Data were acquired with positive and negative polarity switching on the mass spectrometer. Raw MS data were analyzed using Compound Discoverer 3.3 SP1 (Thermo Fisher Scientific). Features were detected by [M+H]¹⁺ and [M-H]¹⁻ adducts, and relative quantification was based on extracted peak areas. Metabolites were annotated and identified by accurate mass and either retention time from an in-house library generated from authentic standards or MS/MS fragmentation by querying the mzCloud database. mzCloud matches that did not have corresponding mass list matches were required to have a minimum score of 50 in either the Reference or Autoprocessed databases. Compounds were identified in either positive and negative polarities. For compounds identified in both polarities, a single polarity was selected based on peak area and CVs. Compounds were filtered if they corresponded to in-source fragmentations or had lower quality annotations or quantifications. Peak areas were normalized to the QC pool sample runs to correct for instrument drift and to the total signal from filtered metabolites after removing background compounds, internal standards, drugs, and the top 10% of metabolites by maximum area. A final list of 190 annotated and quantified metabolites was generated and used in further analysis.

Plasma marker determination

Plasma levels of IFNα were measured by SIMOA using the Interferon-alpha (IFNα) Assay (Cat. #100860, Quanterix) and SR-X Biomarker Detection System (Quanterix). Plasma levels of zonulin (Cat. #MBS167049; MyBioSource), LBP (Cat. #DY870–05; R&D Systems), sCD14 (Cat. #DY383–05; R&D Systems), and sCD163 (Cat. #DY1607–05; R&D Systems) were measured by ELISA kits. β-glucan detection in plasma was performed using Limulus Amebocyte Lysate (LAL) assay (Glucatell Kit; Cat. #GT003; CapeCod). Plasma levels of IL-6, IL-18, TNFα, and IFNγ were determined using customized MSD U-PLEX multiplex assay (#K15067L-2; Meso Scale Diagnostic).

Functional assays to evaluate IFN production

Cryopreserved PBMCs were thawed at 37°C, washed and resuspended in RPMI medium (Gibco Sciences) containing 20% (v/v) of Fetal Bovine Serum (FBS; Gibco Sciences). Cells were counted at NucleoCounter NC-3000 (ChemoMetec). 500,000 cells were seeded per well in U-bottom Nunc^™ 96-Well Polypropylene MicroWell^™ Plates (# 267334; Sigma Aldrich) and cultured at 37°C for 18–20h in RPMI medium (Gibco Sciences) containing 10% (v/v) of Fetal Bovine Serum (FBS; Gibco Sciences) in the presence of either TLR7 agonist Imiquimod (5μg/ml; #tlrl-imqs; Invivogen), TLR9 agonist ODN 2216 (1μM; #tlrl-2216; Invivogen), or TLR4 agonist LPS (100ng/ml; #L4391; Sigma Aldrich). Supernatants were collected to quantify IFNα production with ProQuantum^™ Human IFNα Immunoassay Kit (#A428975; Invitrogen) and 7500 Fast Real-Time PCR Instrument (ThermoFisher Scientific) and IFNγ production by ELISA (#ab46025; Abcam) using the EnSight MicroPlate Reader (Perkin Elmer).

QUANTIFICATION AND STATISTICAL ANALYSIS

Flow cytometry

FCS files were generated after flow cytometry processing. FlowJo v10.7.1 (BD Biosciences) was used to analyse the files. Gating strategy for each Panel is shown in Fig.S1. Two-tailed Mann-Whitney U tests or Kruskal-Wallis and Conover-Iman test were used when 2 or 3 groups were considered, respectively, to assess the presence of statistical differences in cell type proportions, expression of CCR5 coreceptor, activation markers, CD169, and cell death markers.

Reservoir studies

QLP files were generated after QX100 droplet reader processing. QuantaSoft software version 1.6.6.0320 (Bio-Rad) was used to analyse the files. Replicate wells were merged together for quantification. Thresholds were set manually based on amplitude of negative and positive populations. Results were normalized to the number of cells by considering genomic RPP30 or TBP expression. IPDA quantifications are corrected by RPP30 Shearing index, as reported in the original publication⁶⁶. In those samples lacking detection of positive droplets, LOQ was calculated as the expected result if one single droplet was positive, and LOQ/2 was used as input value for analysis. Two-tailed Mann-Whitney U tests were used to assess differences between VNPs and Progressors; Wilcoxon signed-rank tests were used to compare pre-ART and on-ART samples.

Determination of viral tropism

AB1 files were generated after viral sequencing. Sequencher v5.3 (Gene Codes Corporation) was used to evaluate base quality, and FASTA files with query sequences were created. Geno2Pheno online tool was used to predict viral tropism with 5% FPR.

Single-cell RNA sequencing

Single-cell data analysis

All data was aligned to the GRCh38 reference genome using Cell Ranger v3.1.0 (10x Genomics). Single-cell data analysis was performed using Seurat (version 3.2.0)⁷⁰. Low quality cells were filtered based on mitochondrial RNA percentage (>5%), number of counts (<1000) and number of features (<500), based on the bimodal distribution of features in marginal plots. From the filtered count matrices, we performed integration of all patients and donors with the SCTransform Seurat v3 integration workflow. SCTransform normalization was performed separately for each dataset, removing mitochondrial mapping percentage as confounding source of variation. Of note, SCTransform automatically regresses out sequencing depth using a regularized negative binomial model. Next, we selected the shared 2000 highly variable features from each dataset identified by SCTransform and harmonized the samples to integrate cells that share common immune cell types features. In detail, we identified anchors and integrated datasets with SCT normalization and default parameters. Integration was based on canonical correlation analysis (CCA), before proceeding with downstream analysis, visualization and clustering of the integrated dataset. We performed dimensionality reduction by applying the Uniform Manifold Approximation and Projection (UMAP) algorithm and computed clusters based on the 20 top principal components with a resolution of 0.8 using the Louvain clustering. To obtain a finer-grained resolution of cell states, we extracted monocytes, B cells, CD4 and CD8 / NK cells based on canonical expression markers and re-run the integration workflow and downstream analyses and clustering for every groups separately (resolution of 0.5).

Cluster annotation and analysis

Differential expression analysis for clusters was performed to determine gene expression markers (compared to other cell states within the same cell type). We used the Wilcoxon Rank Sum test, limiting the testing to genes that showed (on average) at least 0.1 fold difference (log scale) between two groups for an adjusted p-value of 0.05. For cluster annotation, we combined Seurat’s FeaturePlot function and the mathSCore2 R package (https://github.com/elimereu/matchSCore2). All plots were created with ggplot2 R package (v3.3.3). Mann-Whitney test was used to evaluate the presence of significant differences between groups in the proportion of cell clusters. Gene Ontology (GO) was performed with the clusterProfiler R package (v3.14.3)⁷¹ and enrichGO function with default parameters: 10 as minimal size of genes annotated by ontology term for testing and a Benjamini Hochberg (BH) adjustment with a p-value cutoff of 0.05. In Data S1 we include the lists of genes with the most differential expression per cell cluster used for annotation; the lists of GO terms with the most differential expression per cell type in VNPs versus Progressors; and the lists of genes with the most differential expression per cell cluster in VNPs versus Progressors.

Exome sequencing

Reads were mapped to human build GRCh38 with BWA-MEM 0.7.17. Alignment files containing only properly paired, uniquely mapping reads without duplicates were processed using Picard 2.25 [http://broadinstitute.github.io/picard/] to add read groups and to remove duplicates. The Genome Analysis Tool Kit (GATK 4.1.8.0)⁷² was used for local realignment and base quality score recalibration. Variant calling was done using HaplotypeCaller from GATK. Functional annotations were added using SnpEff v.5.0 with the GRCh38.99 database. Variants were annotated with SnpEff v5.0⁷³ using population frequencies, conservation scores and deleteriousness predictions from dbNSFPv4.1a. Other sources of annotations, such as gnomAD, CADD and Clinvar were also used. Each variant had to be at least supported by 10 reads in one sample. Clinical interpretation of genetic variants, following ACMG/AMP 2015 guidelines, was added with InterVar⁷⁴. HLA typing was performed with xHLA⁷⁵. Fisher Test between Progressor and VNP groups was conducted with fisher.test from R stats package (v3.6.2) to identify variants with significantly different abundance and predicted high/moderate impact.

Plasma untargeted metabolomics

A final list of 190 metabolites was used in the analysis. ANOVA or Kruskal-Wallis tests were performed to identify those species with significantly different concentrations between groups. Association between relative concentration of metabolites and other parameters was assessed by Spearman correlation. Pathway enrichment analysis was performed with MetaboAnalyst v5.0. In Data S2 we include the relative abundance of all metabolites per individual.

Plasma marker determination

Kruskal-Wallis and Conover-Iman test were performed to evaluate the presence of statistical differences between groups. Non-parametric Peto left-censored test adjusted by the Holm method for multiple comparisons was used to evaluate IFNα production upon TLR7/9 stimulation.

Functional assays to evaluate IFN production

Kruskal-Wallis and Conover-Iman test were performed to evaluate the presence of statistical differences between groups. Non-parametric Peto left-censored test adjusted by the Holm method for multiple comparisons was used to evaluate IFNα production upon TLR7/9 stimulation to consider values below LOQ.

Graphic imaging

GraphPad Prism 9 was used to generate graphs and perform statistical analysis.

Supplementary Material

1

Data S1. Gene expression data used for cell cluster annotation and differential expression analysis by single-cell RNA sequencing, related to Figure 5.

2

Data S2. Quantification data of metabolites identified by mass spectrometry, relative to Figure 7.

3

Document S1. Figures S1-S4 and Table S3.

Table S1 and S2 are included as separate Excel files as they contain additional data related to individuals in Figure 1 too large to fit in a PDF.

4

Table S1. Demographic and clinical characteristics of the individuals in the study, related to Figure 1. Estimated seroconversion date was calculated as the midpoint between the last seronegative test, if available, and the diagnose date. Time off ART indicates the time between estimated seroconversion date, if available, or diagnose date and ART initiation. Viral load setpoint indicates the mean value of all plasma viral load determinations taken from 6 months after diagnose date. CD4+ T-cell decay rate is the ratio of CD4+ T-cell count slope in cells/μl/year and median CD4+ T-cell count; it indicates the % of CD4+ T cells respect to median CD4+ T-cell count that is lost each year of untreated infection. HLA-B haplotype, HLA-C haplotype, and CCR5 genotype were obtained by exome sequencing, as indicated in the text. Virus clade was determined together with viral tropism, as indicated in the text. Mann-Wilcoxon test was performed to evaluate statistically significant differences between VNPs and Progressors in: age at diagnose, time off-ART, viral load setpoint and CD4+ T-cell decay rate. **** indicates p<0.0001. Abbreviations. Age seroconv.: age at estimated seroconversion date; Age diagn.: age at diagnose; ART: antiretroviral treatment; y: years; VL: viral load; cop/ml: copies per milliliter; Prog: progressor; M: male; F: female; MSM: men who have sex with men; IVDU: intravenous drug user; HET: heterosexual transmission; SEX: homosexual or heterosexual transmission; Δ32het: CCR5Δ32 heterozygote; wt: homozygous for full-length CCR5.

5

Table S2. Clinical and biological characteristics of the sampling timepoints used in the study, related to Figure 1. Estimated seroconversion date was calculated as the midpoint between the last seronegative test, if available, and the diagnose date. Viral tropism was determined as indicated in the manuscript. Mann-Wilcoxon test was performed to evaluate statistically significant differences between VNPs pre-ART and Progressors pre-ART in: years from seroconversion or diagnose, viral load, CD4+ T-cell count, CD4/CD8 ratio. Mann-Wilcoxon test was performed to evaluate statistically significant differences between VNPs on-ART and Progressors on-ART in: years from seroconversion or diagnose, years from ART-mediated suppression, CD4+ T-cell count, CD4/CD8 ratio. * indicates p<0.05; ** indicates p<0.01; *** indicates p<0.001; **** indicates p<0.0001. Abbreviations. Seroconv.: seroconversion; y: years; ART: antiretroviral treatment; cop/ml: copies per milliliter; cells/μl: cells per microliter; HBV: Hepatitis B virus; HCV: Hepatitis C virus; scRNAseq: single-cell RNA sequencing; Prog: progressor; NA: not applicable.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal anti-CD3 PerCP (clone SK7)	BD Biosciences	Cat#345766; RRID: AB_2783791
Mouse monoclonal anti-CD4 APC-R700 (clone RPA-T4)	BD Biosciences	Cat#564975; RRID: AB_2744418
Mouse monoclonal anti-CD8 V500 (clone SK1)	BD Biosciences	Cat#561617;RRID: AB_10896281
Mouse monoclonal anti-TCRγδ APC Fire 750 (clone B1)	BioLegend	Cat#331228; RRID: AB_2650627
Mouse monoclonal anti-CD45RA AF647 (clone HI100)	BioLegend	Cat#304154; RRID: AB_2616995
Mouse monoclonal anti-CD95 PE (clone DX2)	BioLegend	Cat#305608; RRID: AB_314546
Mouse monoclonal anti-CCR7 PE/DAZZLE 594 (clone G043H7)	BioLegend	Cat#353236; RRID: AB_2563641
Mouse monoclonal anti-CD27 FITC (clone M-T271)	BD Biosciences	Cat#555440; RRID: AB_395833
Mouse monoclonal anti-CCR5 BV421 (clone 2D7)	BD Biosciences	Cat#562576; RRID: AB_2737661
Mouse monoclonal anti-HLADR PE/Cy7 (clone L243)	BioLegend	Cat#307616; RRID: AB_493588
Mouse monoclonal anti-CD38 BV605 (clone HIT2)	BioLegend	Cat#303532; RRID: AB_2562915
Mouse monoclonal anti-CD19 PerCP (clone 4G7)	BD Biosciences	Cat#345778; RRID: AB_2868806
Mouse monoclonal anti-CD14 FITC (clone M5E2)	BD Biosciences	Cat#555397; RRID: AB_395798
Mouse monoclonal anti-CD11c APC (clone B-ly6)	BD Biosciences	Cat#559877; RRID: AB_398680
Mouse monoclonal anti-HLADR BV786 (clone G46–6)	BD Biosciences	Cat#564041; RRID: AB_2738559
Mouse monoclonal anti-CD123 PE/Cy7 (clone 6H6)	eBiosciences	Cat#25–1239-42; RRID: AB_1257136
Mouse monoclonal anti-CD303 BV605 (clone V24–785)	BD Biosciences	Cat#748004; RRID: AB_2872465
Mouse monoclonal anti-CD169 PE (clone 7–239)	BioLegend	Cat#346004; RRID: AB_2189029
Mouse monoclonal anti-CD3 PE (clone HIT3a)	BioLegend	Cat#300308; RRID: AB_314044
Mouse monoclonal anti-CD4 BV605 (clone RPA-T4)	BD Biosciences	Cat#562658; RRID: AB_2744420
Mouse monoclonal anti-CD8 BV510 (clone SK1)	BioLegend	Cat#344732; RRID: AB_2564624
Mouse monoclonal anti-CD56 BV786 (clone NCAM16.2)	BD Biosciences	Cat#564058; RRID: AB_2738569
Biological samples
Cryopreserved PBMCs from HIV+ individuals	NA	NA
Cryopreserved EDTA-treated plasma samples from HIV+ individuals	NA	NA
Blood samples from HIV− individuals	Banc de Sang I Teixits	www.bancsang.net
Plasma samples from HIV− individuals	NA	NA
Plasma samples from HIV− individuals	NA	NA
Chemicals, peptides, and recombinant proteins
Flow Cytometry Staining Buffer	eBiosciences	Cat# 00-4222-26
ExoSAP-IT	Applied Biosystems	Cat#78200.200.UL
Lysis buffer: 10mM Tris-HCl (pH=9), 0.1% Triton X-100, 400 μg/ml Proteinase K.	Ambion (proteinase K)	Cat# AM2546 (proteinase K)
Heavy standard mix (amino acid standards, 2.5 mM)	Cambridge Isotope	Cat# MSK-A2-1.2
Methanol (LC-MS grade)	Acros Organics	Cat# 61513–0025
Imiquimod (R837)	Invivogen	Cat# tlrl-imqs
ODN 2216	Invivogen	Cat# tlrl-2216
LPS	Sigma Aldrich	Cat# L4391
Trypan blue	Thermo Fisher	Cat# 15250-061
MACS BSA Stock Solution	Miltenyi Biotec	Cat# 130-091-376
DTT 1M	Sigma Aldrich	Cat# 646563
SSC 20x	Thermo Fisher	Cat# AM9765
RNAse Inhibitor	Applied Biosystems	Cat# N8080119
Critical commercial assays
LIVE/DEAD Fixable Near-IR Dead Cell Stain kit	Invitrogen	Cat#L10119
FAM FLICA™ Caspase-3/7 Kit	Bio-Rad	Cat#ICT094
FLICA™ 660 Caspase-1 Kit	Bio-Rad	Cat#ICT9122
RNA/DNA Purification Micro Kit	Norgen Biotek	Cat#50300
QIAamp Viral RNA Mini kit	Qiagen	Cat#52904
SuperScript™ III One-Step RT-PCR System with Platinum™ Taq High Fidelity DNA Polymerase	Invitrogen	Cat#12574
Platinum® Taq DNA Polymerase High Fidelity	Invitrogen	Cat#11304
QIAamp DNA Blood Mini Kit	Qiagen	Cat#51104
Pierce™ Protein G Spin Plate	ThermoFisher	Cat#45204
GlycanAssure APTS Kit	ThermoFisher	Cat#A33952
Interferon-alpha (IFN-α) Assay	Quanterix	Cat#100860
Human IFN gamma ELISA Kit	Abcam	Cat#ab46025
Master mix PCR Taqman Universal	Applied Biosystems	Cat# 4364338
ProQuantum™ Human IFNα Immunoassay Kit	Invitrogen	Cat# A428975
Human Zonulin ELISA kit	MyBioSource	Cat# MBS167049
Glucatell Kit	CapeCod	Cat# GT003
LBP ELISA kit	R&D Systems	Cat# DY870–05
sCD14 ELISA kit	R&D Systems	Cat# DY383–05
sCD163 ELISA kit	R&D Systems	Cat# DY1607–05
Customized MSD U-PLEX multiplex assay	Meso Scale Diagnostic	Cat# K15067L-2
Next GEM Single Cell 3’ Reagent Kits v3	10X Genomics	Cat# 1000075
DNA High Sensitivity Bioanalyzer Kit	Agilent	Cat# 5067–4626
Kapa HyperExome probes	Roche Diagnostics	Cat# 09062564001
Deposited data
Raw data – Exome Sequencing	European Genome-phenome Archive (EGA)	Study ID: EGAS50000000079
Raw data – Single-cell RNA Sequencing	European Nucleotide Archive (ENA)	Project ID: PRJEB68223
Raw data – Metabolomics	Metabolomics Workbench	Project ID: PR001838. Study ID: ST002955.
Code for data analysis	GitHub / Zenodo	DOI (Zenodo): 10.5281/zenodo.12819667
Oligonucleotides
*See Table S3.
Software and algorithms
FlowJo v10.7.1	BD Life Biosciences	https://www.flowjo.com/
Quantasoft v1.6.6.0320	Bio-Rad	http://www.bio-rad.com/
Sequencher v5.3	Gene Codes Corporation	http://www.genecodes.com/
BWA-MEM 0.7.17	(Li 2013)	https://arxiv.org/abs/1303.3997v2
Cell Ranger v3.1.0	10X Genomics (PMID: 28091601)76	https://www.10xgenomics.com/support/software/cell-ranger
Genome Analysis Tool Kit (GATK) 4.1.8.0	McKenna et al. 2010 (PMID: 20644199)	https://gatk.broadinstitute.org/hc/en-us
Seurat v3.2.0	Stuart et al. 2019 (PMID: 31178118)	https://satijalab.org/seurat/
Geno2Pheno	Max-Planck Institute für Informatik	https://www.geno2pheno.org/
GraphPad Prism 9	Dotmatics	http://www.graphpad.com/
Picard v2.25	Broad Institute	http://broadinstitute.github.io/picard/
R	R Core Team	https://www.R-project.org/
SnpEff v5.0	Cingolani 2013 (PMID: 22728672)
xHLA	(Xie et al 2017) (PMID: 28674023)	https://github.com/humanlongevity/HLA
Compound Discoverer 3.3 SP1	Thermo Fisher Scientific	https://www.thermofisher.com/us/en/home/industrial/mass-spectrometry/liquid-chromatography-mass-spectrometry-lc-ms/lc-ms-software/multi-omics-data-analysis/compound-discoverer-software.html
Other
QX100 Droplet Digital PCR System	Bio-Rad	Cat# 186–3001
7500 Fast Real-Time PCR Instrument	ThermoFisher	---
Cryotubes for metabolite extraction	Nalgene	Cat#V4632–1000EA
SR-X100	Quanterix	---
Luminex200	ThermoFisher	Cat# APX10031
Chromium Controller Instrument	10X Genomics	Cat# 1000202
2100 Bioanalyzer Instrument	Agilent	Cat# G2939BA
TC20™ Automated Cell Counter	Bio-Rad	Cat# 1450102
Chromium i7 Multiplex Kit	10X Genomics	Cat# 120262
Chromium Single Cell B Chip Kit	10X Genomics	Cat# 1000073
NovaSeq 6000	Illumina	---

Context and Significance: purpose and implications in plain language.

Most people with HIV need lifelong antiretroviral treatment to keep the virus under control and prevent the development of immunodeficiency. However, in rare cases, no immune damage is observed in the absence of treatment and despite high levels of the virus. Researchers at IrsiCaixa have performed an integrated multiperspective analysis using cutting-edge technologies, revealing that most of these individuals have a genetic mutation that hinders viral infection of their cells. Additionally, other peculiarities of their immune system, such as a moderated chronic inflammatory response to the presence of the virus and preservation of equilibrium in their intestinal mucosa, play a complementary role. These results highlight the potential of modulating chronic inflammation to improve the health status of people with HIV.

Highlights.

• VNPs are a rare group of people with HIV who do not exhibit CD4+ T cell depletion

• Protection from HIV pathogenesis is driven by a complex combination of mechanisms

• Heterozygosity for the CCR5Δ32 coreceptor confers genetic predisposition

• Moderated gut disruption, T-cell activation and IFN responses jointly contribute