Participants were 18-23 year old females from the Females Rising through Education Support and Health (FRESH) cohort, an ongoing longitudinal study based in Durban, KwaZulu-Natal, South Africa (10, 30). Participants at high risk of HIV-1 infection consent to twice-weekly finger-prick blood sampling for HIV-1 RNA testing to enable detection of HIV infection during acute HIV infection (AHI). Biological samples are collected pre and post-infection. At study inception, the study protocol adhered to the South African national HIV clinical management guidelines, which restricted ART initiation to CD4 counts at or below 350 cells/mm³. Accordingly, participants testing HIV-1 RNA positive during this period did not initiate therapy until they met eligibility criteria. Approval to initiate ART during AHI was obtained, allowing for immediate treatment upon detection of HIV-1 RNA. Participation was voluntary and all participants provided written informed consent in accordance with protocols approved by the Biomedical Research Ethics Committee of the University of KwaZulu-Natal and the Institutional Review Board for Massachusetts General Hospital. In this sub-study, twenty-seven individuals from the FRESH cohort were included according to HIV status and time of ART initiation as follows: HIV-1 negative (n=10), HIV positive who initiated ART at the hyperacute phase (n=10), and HIV positive who initiated ART during chronic infection (n=7). HIV positive individuals, either starting ART at hyperacute (Fiebig stage I-II) or chronic infection were assessed at 1- and 12- months post-infection. Fiebig stage I-II was characterised as HIV-1 RNA positive (NucliSENS EasyQ HIV-1 v2.0 kit, BioMérieux, Marcy-l’Étoile, France), p24 antigen negative or positive and HIV antibody rapid or ELISA assay negative (Elecsys HIV combi PT 4th Generation (Ag+Ab test) p24 Ag, Roche, Basel, Switzerland) and Western blot negative (GS HIV-1 Western blot kit, Bio-Rad Laboratories, Hercules, CA, USA). Individuals starting ART during chronic infection were treatment naïve at the 1- and 12-month post-infection time points. These individuals subsequently initiated ART and were further assessed at 12 months after starting ART (i.e., approximately 24-months post-infection). Samples for the participants initiating ART during chronic infection were limited at the 1- month (n=4) and 24-month (n=6) time points.

Blood samples were collected in ethylenediaminetetraacetic acid (EDTA) anticoagulated vacutainer tubes (Becton Dickinson (BD), Franklin Lakes, NJ, USA) for CD4+ T cell count measurement and viral load quantification. CD4 counts were measured using BD Trucount and analysed on a four-parameter FACSCalibur flow cytometer (BD). Viral loads were determined using the NucliSENS EasyQ HIV-1 v2.0 kit with a detection limit of 20 copies/ml (BioMérieux, Marcy-l’Étoile, France). Individuals initiating ART were prescribed a fixed-dose three drug oral combination pill of efavirenz, emtricitabine and tenofovir. For participants who started antiretroviral therapy in acute HIV infection, raltegravir was prescribed as a fourth drug until 90 days after viral suppression.

Monoclonal antibodies were obtained from BD Biosciences: fluorescein isothiocyanate (FITC) anti-human CD8 (RPA-T8), phycoerythrin (PE)-CF594 anti-human HLA-DR (G46-6), PE-Cyanine(Cy)7 anti-human CD86 (FUN-1), PE anti-human IFN-α[2b] (7N4-1) and BioLegend (San Diego, CA, USA): brilliant violet (BV) 650 anti-human CD3 (OKT3), BV510 anti-human CD19 (HIB19), BV510 anti-human CD56 (NCAM) (HCD56), allophycocyanin (APC) anti-human CD4 (SK3), BV785 anti-human CD16 (3G8), APC-Cy7 anti-human CD14 (HCD14), PE-Cy5 anti-human CD11c (3.9), BV421 anti-human CD123 (6H6), BV711 anti-human CD38 (HIT2), peridinin-chlorophyll-protein (PerCP)/Cy5.5 anti-human CD69 (FN50), BV605 anti-human TNF-α (MAb11). Anti-Mouse Ig, κ/Negative control particles were used for flow cytometry compensation (BD Biosciences). LIVE/DEAD Fixable Aqua dead cell stain kit and FIX & PERM cell permeabilization kit were sourced from Invitrogen (Carlsbad, CA, USA). Toll-like receptor (TLR) agonists for cell stimulation were as follows: TLR4 ligand lipopolysaccharide (LPS) was from Sigma-Aldrich (Saint Louis, MO, USA), TLR7/8 ligand imidazoquinoline compound (CL097) and TLR9 ligand class A CpG oligonucleotide (ODN 2216) were from InvivoGen (San Diego, CA, USA). The DOTAP liposomal transfection reagent was sourced from Roche. Brefeldin A (BFA) was obtained from Sigma-Aldrich. R10 medium was Roswell Park Memorial Institute (RPMI) 1640 supplemented with 10% gamma irradiated, heat-inactivated foetal bovine serum (Gibco, Waltham, MA, USA), 1% penicillin/streptomycin, 1% L-glutamine and 1% HEPES buffer (Lonza, Basel, Switzerland).

Blood samples were collected in acid citrate dextrose (ACD) vacutainer tubes (BD). Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll density gradient centrifugation and cryopreserved within 6 hours until use. Cryopreserved PBMCs were thawed, re-suspended in R10 medium and rested at 37°C and 5% CO₂ for at least 2 hours. For immunophenotyping, one million cells were stained with LIVE/DEAD Fixable Aqua dead cell stain and monoclonal antibodies for CD19 (to exclude B cells), CD56 (to exclude NK cells), CD3, CD4, CD8 (to identify T cells), CD11c, CD14, CD16, CD123 and HLA-DR (to identify monocytes, myeloid dendritic cells [mDCs] and plasmacytoid dendritic cells [pDCs]) and activation and co-stimulatory receptors CD38, CD69 and CD86, before fixation with Medium A and acquisition on a BD LSRFortessa. For in vitro functional assays: monocytes, mDCs and pDCs were assessed for cytokine production (TNF-α and IFN-α) by culturing one million PBMCs in either R10 medium alone, TLR4 ligand LPS, TLR7/8 ligand CL097, or TLR9 ligand CpG in the presence of DOTAP. BFA was added to all conditions at a final concentration of 10 μg/ml. Following overnight culture at 37°C and 5% CO₂, samples were stained with LIVE/DEAD Fixable Aqua dead cell stain, CD3, CD11c, CD14, CD16, CD19, CD56, CD123 and HLA-DR antibodies (as described above), fixed with Medium A, before permeabilization with Medium B and staining with monoclonal antibodies for TNF-α and IFN-α, and acquisition on a BD LSRFortessa.

Blood samples were collected in EDTA anticoagulated vacutainer tubes. Plasma component was separated by centrifugation within 6 hours followed by immediate storage at -80°C until use. Plasma concentrations for IL-6 were measured using Human IL-6 Quantikine ELISA Kit (R&D Systems, Minneapolis, Min, USA). sCD14 was measured using Human CD14 DuoSet ELISA (R&D Systems).

FACSDiva 7 (BD) was used for flow cytometry data acquisition. FlowJo software version 9 (FlowJo, LLC, Ashland, OR, USA) was used for flow cytometry data analysis. GraphPad Prism 5 (GraphPad software Inc., La Jolla, CA, USA) was used for statistical data analysis. Mann-Whitney U test was used for single group comparisons. Kruskal-Wallis test was used for multiple comparisons followed by Bonferroni’s correction test for multiple comparisons. Correlations were examined using Spearman rank analysis and p-values <0.05 were considered statistically significant. All statistically significant differences are marked with the p-value in the figures and non-significant differences are unmarked.

Early treated individuals achieved undetectable viremia (<20 RNA copies/ml) within 1-month post-infection, and suppression was maintained throughout the first year of infection, whilst those treated during chronic infection had a median viral load of 107,250 and 82,000 RNA copies/ml at 1- and 12-months respectively ( Table 1 and Figure 1A ). CD4+ T cell counts differed significantly across the 3 study groups and by time points analysed ( Table 1 and Figure 1B ). Median CD4+ T counts for individuals treated with ART during hyperacute infection were 848 and 907 cells/mm³ at 1- and 12-months respectively, which was not significantly different from the median count of 1,045 cells/mm³ for HIV uninfected individuals ( Figure 1B ). In contrast, individuals initiated on ART during chronic infection presented with CD4+ T cell decline, with counts significantly lower compared to the uninfected controls at both 1-month (p=0.004) and 12-months (p=0.0002) post-infection. Individuals initiating treatment during chronic infection also had significantly lower CD4+ T cell counts compared to hyperacute treated individuals at 1-month (p=0.014) and at 12-months (p=0.001) post-infection. After 12 months on ART (24-months post-infection), CD4 counts in ART chronically treated individuals remained significantly lower compared to HIV uninfected participants (p=0.001) and the ART hyperacute treated group at 12-months (p=0.008) ( Figure 1B ). We further evaluated the activation profiles of CD4+ and CD8+ T cells by assessing CD38 and HLA-DR co-expression ( Figures 1C, D and Supplementary Figure 1 ). ART hyperacute treated individuals displayed similar frequencies of activated CD4+ and CD8+ T cells compared to HIV negative individuals ( Figures 1C, D ). Higher CD4+ T cell activation was observed in the ART chronic phase treated group (before therapy initiation) at 1-month (p=0.076) and at 12-months (p=0.019) compared to HIV negative individuals. Chronic phase treated persons also displayed higher CD4+ T cell activation at 1-month (p=0.014) and 12-months (p=0.0004) compared to those initiating treatment during the hyperacute phase ( Figure 1C ). Following initiation of ART in the chronically treated group, CD4+ T activation reduced such that the frequency of activated cells was not statistically different to that in the HIV uninfected and ART hyperacute treated groups ( Figure 1C ). Similarly, chronically treated individuals displayed higher CD8+ T activation at 1-month (p=0.002) and at 12-months (p=0.001) compared to HIV negative participants ( Figure 1D ). Moreover, CD8+ T cell activation was higher in chronic phase treated individuals compared to ART hyperacute phase treated participants at 1-month (p=0.002) and at 12-months (p=0.0007) post-infection ( Figure 1D ). Following 12 months of therapy for the chronically treated individuals, activated CD8+ T cell frequencies declined and were significantly lower than in HIV negative and ART hyperacute treated individuals (p=0.042 for both comparisons) ( Figure 1D ). Overall, these results indicate that ART during the hyperacute phase of HIV infection prevented immune activation of the CD4+ and CD8+ T cell populations, and that initiation of ART during the chronic infection phase reduced immune activation to baseline pre-infection levels.

The impact of very early ART initiation on phenotypic and functional characteristics of professional APCs is not well characterised and yet this may have implications for the maintenance of optimal immunity following infection and for HIV cure or remission strategies. To address this, we performed ex vivo flow cytometry analysis of monocytes, mDCs and pDCs in individuals initiating therapy during Fiebig stage I-II in contrast to individuals initiating therapy in the chronic phase of infection and compared these groups to HIV negative individuals. We first assessed monocyte, mDC and pDC cell frequencies; and then further categorised monocytes as classical, intermediate or inflammatory based on CD14 and CD16 expression ( Supplementary Figure 2 ). Cell types were enumerated as a percentage of total cells. Monocyte frequencies were noticeably elevated in the ART hyperacute treated compared to the uninfected group, particularly in acute phase infection (p=0.005 at 1-month, p=0.029 at 12-months, Figure 2A ). Likewise, median monocyte frequencies were seemingly elevated in the ART chronically treated group at 1-month post-infection and were significantly higher at 12-months post-infection (p=0.033) compared to uninfected individuals ( Figure 2A ). mDC frequencies were not affected by HIV or ART intervention with no significant differences noted in median cell frequencies between the groups irrespective of HIV infection status or ART initiation timing ( Figure 2B ). pDCs were drastically elevated at 1-month post-infection (p=0.0008), with partial decline in frequencies by 12-months post-infection (p=0.01) for the hyperacute phase treated group ( Figure 2C ). In patients initiating treatment during chronic infection, pDC frequencies appeared elevated at 1-month post-infection and declined thereafter with no significant difference noted at 12-months post-infection when the individuals were treatment naïve and at 24-months post-infection (i.e. 12-months post-ART initiation). In fact, pDC frequencies were significantly higher at 12-months post-ART in the hyperacute treated group compared to 24-months for the chronic treated group (i.e. 12-months post-ART initiation) (p=0.04, Figure 2C ). We further investigated monocyte subsets – classical, intermediate and inflammatory monocytes were classified according to common nomenclature (31). Classical subset frequencies following early ART were maintained at levels that resembled uninfected individuals. In contrast, lower frequencies compared to uninfected controls were observed in the group initiating treatment in the chronic phase (p=0.024 at 1-month, p=0.055 at 12-months) and even after ART commencement at 24-months post-infection (p=0.042) ( Figure 2D ). Intermediate monocytes were significantly lower in treated groups regardless of ART timing (p=0.0007 at 1-month, p=0.009 at 12-months in ART hyperacute; p=0.009 at 24-months in ART chronic, Figure 2E ), whilst reduced inflammatory frequencies were linked to infection duration rather than ART status or timing (p=0.015 at 12-months in ART hyperacute; p=0.012 at 12-months, p=0.042 at 24-months in ART chronic, Figure 2F ). Interestingly, we observed a distinct CD14lowCD16- phenotype, which was considerably higher amongst HIV infected groups, irrespective of ART initiation or length of infection ( Figure 2G and Supplementary Figure 2 ). Overall, these data suggest that early initiation of antiretroviral therapy does not always prevent immunological changes in APC populations, with remarkable heterogeneity linked to HIV and ART initiation on different APC cell populations.

Next, we measured the ex vivo expression of inducible receptors CD69 and CD86 on APCs ( Supplementary Figure 3 ). Early treated individuals exhibited a non-significant dip in CD69 expression on monocytes, compared to HIV uninfected persons, with a recovery to pre-infection levels by 12-months post-infection ( Figure 3A ). In participants who initiated treatment in the chronic phase, CD69 expression on monocytes was decreased at 12-months post-infection compared to early treated individuals (p=0.014) but with recovery at 12-months post-ART initiation ( Figure 3A ). CD69 expression on mDCs was highly variable from participant to participant with no significant differences between the groups noted ( Figure 3B ). CD69 expression on pDCs in hyperacute treated individuals was similar to uninfected controls but levels were significantly lower at 12-months post-infection in participants starting ART in chronic HIV-1 infection when compared to hyperacute treated persons (p=0.008), with partial recovery after ART initiation ( Figure 3C ). Assessment of the co-stimulatory maker CD86 on APCs revealed significantly lower expression levels on monocytes and mDCs in all HIV-infected groups compared to HIV uninfected individuals ( Figures 3D, E ). Interestingly, CD86 expression on pDCs was only downregulated in the ART hyperacute group (p=0.0003 at 1-month, p=0.012 at 12-months, Figure 3F ), while ART chronic groups displayed no differences in pDC CD86 expression compared to uninfected individuals. Taken together, these results indicate differential impact of hyperacute ART on APCs such that whereas it may prevent the downregulation of CD69 on monocytes and pDCs, it does not prevent downmodulation of CD86 on monocytes and mDCs, and it may lead to downmodulation of CD86 on pDCs that does not appear to be HIV-1 infection driven.

We next assessed whether cellular activation of APCs, sometimes noted despite ART initiation during hyperacute infection, translated into systemic inflammation considering that the latter is commonly implicated in disease progression to AIDS. We therefore investigated the effect of early therapy on IL-6 and sCD14 circulatory levels. No significant differences in IL-6 levels were found in untreated infection compared to uninfected individuals. Surprisingly, early treated individuals exhibited significantly higher IL-6 levels at 12-months post-infection (p=0.036, Figure 4A ). Late treated individuals trended towards lower levels of IL-6 after one year on ART compared to the early treatment group (p=0.055, Figure 4A ). IL-6 positively associated with expression of activation marker CD69 on monocytes (rho=0.52, p=0.048; Figure 4B ) and mDCs (rho=0.46, p=0.081; Figure 4C ); whereas no correlation was found between IL-6 and pDC activation ( Figure 4D ). Median sCD14 was higher in all HIV-infected groups irrespective of ART timing compared to HIV negative individuals ( Figure 4E ). sCD14 plasma levels correlated with monocyte activation (rho=0.47, p=0.035; Figure 4F ) but not to mDC or pDC activation ( Figures 4G, H ). Overall, our results suggest that ART during HIV infection, irrespective of the timing of ART initiation, may be associated with specific systemic inflammatory markers, with some of these surprisingly more elevated in individuals with ART exposure during the hyperacute phase of infection.

We further ascertained functional capacity of APCs by assessing cytokine production (TNF-α and IFN-α) following stimulation of TLR4, 7/8 or 9 pathways ( Supplementary Figure 4 ). pDC cytokine expression capacity was impacted by the timing of ART initiation. TNF-α production by pDCs following TLR4 stimulation was reduced in the HIV infected persons not commencing immediate ART compared to uninfected controls (p=0.048 at 1-month, p=0.058 at 12-months) ( Figure 5A ). There were no changes in TNF-α production noted for the study groups following TLR7/8 stimulation of pDCs ( Figure 5B ), however, production of TNF-α following TLR9 stimulation was lower in HIV infected hyperacute treated group at 1-month (p=0.036) and in the untreated group at 1-month (p=0.054) and at 12-months (p=0.019) post-infection ( Figure 5C ). Production of IFN-α by pDCs following TLR4 stimulation was similar across the study groups ( Figure 5D ) but there was trend towards impaired IFN-α production following TLR7/8 stimulation for the HIV infected groups irrespective of ART initiation timing ( Figure 5E ). TLR9 stimulation of pDCs revealed comparable levels of IFN-α production in HIV uninfected compared to HIV infected hyperacute treated individuals but impaired production of the cytokine in late treated individuals compared to HIV uninfected and hyperacute treated persons (p=0.024 at 1-month, p=0.043 at 12-months, Figure 5F ). No significant differences were found in median cytokine release of monocytes in ART hyperacute or chronic treated groups in comparison to HIV negative individuals ( Supplementary Figures 5A–F ). Although trends of lower TLR9-induced TNF-α were exhibited by monocytes at 1-month post-infection (p=0.063 in ART hyperacute; p=0.077 in ART chronic, Supplementary Figure 5C ), cytokine production normalised by 12-months post-infection in both groups. IFN-α release trended lower following TLR7/8 stimulation in acute untreated infection compared to early ART individuals (p=0.076, Supplementary Figure 5E ), but this too stabilised by 12-months post-infection. Similarly, mDC cytokine production remained unaffected by HIV infection ( Supplementary Figure 6 ). There was altered TNF-α release following TLR7/8 stimulation in untreated infection (p=0.024 at 1-month, p=0.055 at 12-months; Supplementary Figure 6B ), but this function was restored following late ART initiation.