Helios + Regulatory T cell frequencies are correlated with control of viral replication and recovery of absolute CD4 T cells counts in early HIV-1 infection

A total of 37 individuals, 14 with early HIV infection (< than 3 months), 8 with HIV chronic infection (> 9 months) and 15 uninfected controls were enrolled in this study. All subjects were part of a large cohort study (RV363) established for assessment of HIV incidence and associated risk factors, and willingness to participate in future HIV vaccine trials. The HIV incidence study enrolled participants in Maputo, Mozambique, at Centro de Investigação e Treino em Saúde da Polana Caniço (CISPOC) between December 2013 and November 2014. A total of 500 uninfected individuals between 18 and 35 years, at high risk for HIV infection, were enrolled for assessment of HIV incidence every three months, over a period of two years. A smaller group of HIV-infected individuals, as part of the chronic infection group, were also included in the study, as a community masking purposes. Pregnant women were excluded from the study. Peripheral blood samples were collected from HIV early infected participants, those whom the rapid test shift from negative to positive in a period of 3 months (median) and masking group individuals for assessment of virologic and immunologic parameters. For the Tregs sub-study, peripheral blood mononuclear cells (PBMC) were available for 14 out of 19 individuals who seroconverted during the HIV incidence study. In addition, PBMC from 3 seroconverted individuals and 5 chronically infected individuals, were evaluated 9 months after the first PBMC collection and at least 2 years after infection, respectively. Randomly selected individuals who did not seroconvert and were negative for HIV, hepatitis B virus (HBV), hepatitis C virus (HCV) and syphilis were considered as control subjects (uninfected). Immunophenotyping of PBMCs was done on 22 HIV infected individuals and 15 uninfected controls.

The HIV-1 diagnosis was performed on fresh venous whole blood samples, following the Mozambican national algorithm for HIV testing, which consists of two sequential rapid immunochromatographic tests for detection of anti-HIV-1/2 antibodies. Screening was first performed using the Alere Determine™ HIV-1/2 (Alere Medical Co., Japan) rapid test. Non-reactive specimens were classified as negative. Reactive specimens were confirmed by a second rapid test, the Uni-Gold™ HIV (Trinity Biotech PLC, Ireland). Indeterminate results, reactive for Determine but non-reactive for UniGold, were resolved by a fourth-generation ELISA, Genscreen Ultra HIV Ag-Ab (BioRad, France). The antibody reactivity pattern for HIV-1 (p31, gp160, p24 and gp41) proteins was analyzed in all seroconverted individuals, using the Geenius HIV 1/2 Confirmatory Assay (BioRad, France) at seroconversion visit and at time of PBMC collections (Fig. 1a and b). CD4 counts were performed on whole blood cells as previously described [24]. Acquisition was performed on the four-color flow cytometer FACS CALIBUR (BD, USA). Plasma HIV-1 viral loads were determined using the COBAS AmpliPrep/COBAS TaqMan HIV-1 Test, v2 (Roche, USA).

Peripheral blood mononuclear cells (PBMC) were isolated from freshly obtained heparin anti-coagulated blood, using Ficoll-Paque Plus (GE Healthcare, Sweden) and Leucosep tubes (Greiner Bio-One, German), and stored in liquid nitrogen, in a freezing media (10% dimethyl sulfoxide (DMSO) + 90% fetal calf serum (FCS)). After thawing in a water bath at 37°C, PBMC were washed twice in complete RPMI medium supplemented with 20% of FCS followed by 10% of FCS (R20 and R10, respectively) and viable cells were counted using the Nucleocounter NC-100 (Chemometec, Denmark). Prior to cell staining, PBMC were allowed to recover overnight in R10 medium at 37 °C in an atmosphere of 7.5% CO₂. Subsequently, PBMC were counted and washed in phosphate buffered saline (PBS). Cell viability was assessed by adding 50 μl of the viability dye (fixable viability stain (FVS) 510 (BD, USA)) to one million PBMC. Finally, cells were washed again and stained for cell surface markers. Three eight-color panels were prepared per subject, each containing 500,000 cells. One panel was used to stain cell surface markers only and two other panels were used for combined staining of surface markers and transcription factors, FoxP3 and Helios, all following the instructions from the manufacturer. The FoxP3 buffer set from BD was used to fix and permeabilize PBMC for FoxP3 and Helios staining following manufacturer recommendations.

The following eight-color panels were used: (1) CD3^FITC/ CD45RA^PerCP/ HLADR^PE-CY7/ CD38^APC/ CD8^APC-H7/ CD4^V450/ FVS510 for assessment of levels of immune activation of CD4 and CD8 T cells; (2) CD195^FITC/ FoxP3^PE/ CD49d^PerCP/ CD25^PE-CY7/ B7^APC/ CD3^APC-H7/ CD4^V450/ FVS510, and (3) CD31^FITC/ FoxP3^PE/ CD45RA^PerCP/ CD25^PE-CY7/ Helios^APC/ CD3^APC-H7/ CD4^V450/ FVS510 were used for co-expression analysis of lineage or activation markers and expression of HIV-1 co-receptors on classic Tregs. After staining, cells were fixed with 1X BD CellFix (BD, USA) and analyzed within 24 h. The description of each monoclonal antibody is summarized in Additional file 1. Extensive titration experiments were performed on monoclonal antibodies prior to testing of participant samples.

A minimum of 100,000 events were acquired on FACSCanto II (Becton Dickinson (BD), USA) with Diva software version 8 (Becton Dickinson, USA). The BD Cytometer Setup & Tracking (CST) beads and BD Comp beads were used to ensure consistency of results over time and for compensation, respectively. The post-acquisition analyses, including compensation, were performed using the FlowJo software, version 10 (FlowJo LLC, USA). The fluorescence minus one (FMO) controls were used for definition of positivity applied to markers with continuous or dim levels of expression during post acquisition analysis. Representative plots, showing the gating strategy used for definition of T cells, including the classic Tregs and their phenotypes, are shown in Additional file 2 and Additional file 3.

Statistical analyses were performed using the software GraphPad Prism version 6.0 h (San Diego, CA). The Mann-Whitney test was used to test the heterogeneity among different groups. The analysis of correlations between two variables was performed by the Spearman Rank correlation. Difference or correlation with p-values less than 0.05 were considered statistically significant.

It is well known that HIV infection and disease progression is associated with systemic immune activation. T-cell activation starts at the acute phase of HIV infection and persists during the chronic phase, even after initiation of highly effective antiretroviral therapy [25]. Both CCR5 and CCR3 participate in T cell activation and migration during the inflammatory process, and are expressed predominantly on memory Th1 cells [26].

CD8 T cell activation was high in HIV early-infected group as compared to the HIV negative control group (p < 0.0001), and correlated positively with viral load (r = 0.73; p = 0.008) and inversely with the frequency of antibodies reactive to two or more HIV-1 proteins, namely p31, gp160, p24 and, gp4 (r = −0.59; p = 0.03) (Fig. 2a–c). We found an increased proportion of CD8 T cell (CD3⁺CD8⁺) expressing CXCR3 (p = 0.05) and CCR5 (p = 0.002), as compared to controls (Fig. 2d and e). However, the proportion of CD8 T cells expressing CCR5 and CXCR3 was not associated with either viral load (r = 0.01; p = 0.9 and r = 0.17; p = 0.57, respectively) or frequency of antibody responses against HIV-1 proteins (r = 0.20; p = 0.48 and r = −0.18; p = 0.71, respectively).

The activation profile of CD4 T cell (CD3⁺CD4⁺), measured by analyzing the co-expression of HLA-DR and CD38, was significantly higher in the HIV early infected group as compared to controls (p = 0.02). The proportions of both CD4⁺CD38⁺HLADR⁺ and CD4⁺HLADR^High tended to correlate inversely with the frequency reactive bands of HIV proteins (r = −0.45; p = 0.12 and r = −0.66; p = 0.01, respectively) and with the absolute CD4 T cell counts (r = −0.51; p = 0.07 and r = −0.4; p = 0.17, respectively). However, no significant correlation was found between the CD4 T cell activation and viral load (=0.34; p = 0.26). We also found a positive correlation between the percentage of recent CD4 thymic emigrants, defined as CD3⁺CD4⁺CD31⁺CD45RA⁺, and the frequency of reactivity to HIV-1 proteins (r = 0.6; p = 0.04), but percentages did not differ between HIV early infected patients and controls (p = 0.45).

With regard the expression of chemokine receptors on CD4 T cells, we did not find significant differences of CD4⁺CCR5⁺ (p = 0.24) and CD4⁺CXCR3⁺ T cells (p = 0.97) expression levels between the two groups. In addition, the proportion of CD4⁺CCR5⁺ (p = 0.24) and CD4⁺CXCR3⁺ T cells did not correlate significantly with viral load (r = −0.16; p = 0.59 and r = −0.16; p = 0.59, respectively) or with the frequency of antibody response against HIV proteins (r = 0.15; p = 0.60 and r = 0.15; p = 0.60, respectively).

The absolute counts and percentages of classic Tregs was analyzed. We observed a decrease of absolute numbers of classic CD4⁺CD25^HighFoxP3⁺ Tregs in HIV early infected patients compared with control group (p = 0.02) but not with those with chronic infection. No difference was observed in proportion when compared with control group (p = 0.60) (Fig. 3a and b). However, the proportion of classic Tregs, in the HIV early infected group correlated positively with viral load (r = 0.8; p = 0.002), the proportion of activated CD8⁺CD38⁺HLADR⁺ (r = 0.68; p = 0.03) cells, and absolute numbers of CD4 T cells (r = −0.9; p = 0.0004) (Fig. 3c-e). Despite the absence of significant correlation between the frequency of Tregs and the frequency of bands reactive to HIV proteins (r = −0.53; p = 0.1), at time of PBMC collection, we found a significant correlation with the number of bands at seroconversion visit (r = −0.67; p = 0.03) (Fig. 3f and g).

The absolute decrease of classic Tregs did not correlate with viral load (r = 0.21; p = 0.53), frequency of antibodies reactive to HIV-1 proteins (r = −0.26; p = 0.43) nor total number of CD4 T cells (r = −0.17; p = 0.61).

Chemokine receptor expression on Tregs can predict their function and their potential to accumulate in non-lymphoid tissues [27]. It has been shown that CXCR3⁺ and CCR5⁺ Tregs are recruited to sites with ongoing Th1 inflammation [28, 29]. Furthermore, increased expression levels of CCR5 and integrin β7 are associated with higher susceptibility to HIV infection and poor immunological response [30‐32].

We observed an increased proportion of CCR5 (p = 0.01) and CXCR3 (p = 0.03) on classic Tregs in patients with a recent HIV infection as compared to controls (Fig. 4). However, the increased expression of CCR5 and CXCR3 on classic Tregs did not correlate with viral load, nor with reactivity to HIV-1 proteins or with T cell activation.

The proportion and absolute number of classic CD4⁺CD25^HighFoxP3⁺ Tregs expressing β7 integrin did not differ between HIV patients and controls (p = 0.45 and p = 0.22 respectively). Their expression did not correlate with viral load, nor with frequency of antibodies reactive to HIV-1 proteins, T cell activation or with absolute numbers of CD4 T cells.

It has been shown that co-expression of FoxP3 and Helios identifies Tregs that do not produce inflammatory cytokines [33]. In addition, it has also been shown that deletion of Helios contributes to progressive systemic immune activation [34]. We analyzed the proportion of classic CD4⁺CD25^HighFoxP3⁺ Tregs expressing Helios (Fig. 5a) and found an inverse correlation between Tregs expressing Helios and viral load in early infection (r = −0.63; p = 0.04) (Fig. 5c). This correlation was preserved, when including chronically infected subjects with detectable viral loads (r = −0.62; p = 0.01) (Fig. 5d). In addition, the proportion of Tregs expressing Helios showed a trend to correlate positively with the recovery of CD4 T cells absolute counts, in early infected group (r = 0.6; p = 0.056). Furthermore, when adding data from progressive and chronically infected individuals with detectable viral loads, the correlation between Helios expressing Tregs and absolute CD4 T cell counts became significant (r = 0.56; p = 0.03) (Fig. 5e). We also found tendencies to correlations, between proportions of Tregs expressing Helios in early infected patients with the proportions of activated CD8 T cells (r = −0.48; p = 0.15) and frequency of antibodies reactive to HIV-1 proteins (r = 0.47; p = 0.17). The absolute numbers of classic Tregs expressing Helios did not differ significantly between early infected patients with either controls (p = 0.24), despite the trend to a decrease of proportions in early infected patients (p = 0.06).