Early Antiretroviral Therapy Preserves Functional Follicular Helper T and HIV-Specific B Cells in the Gut Mucosa of HIV-1–Infected Individuals

This article is featured in In This Issue, p.3315

Natural immunity to many viral diseases involves either circulating neutralizing Abs produced by long‐lived plasma cells in the bone marrow or the production of neutralizing Abs by memory B cells reactivated by the infecting pathogen, frequently many years after the first exposure. For HIV, however, the natural immune response appears ineffective (1). Among HIV-1–infected individuals, ∼20% develop high titers of cross-reactive neutralizing Abs to various regions of the HIV-1 envelope protein. In a few of these patients, known as elite neutralizers (∼1% of HIV-1–positive individuals), the cross-reactive Abs include broadly neutralizing Abs (bNAbs) capable of neutralizing most of the known HIV-1 strains (2). Unusual characteristics of bNAbs include high frequencies of V(D)J mutations, significantly extended third CDRs in the heavy-chain V region (CDRH3), and polyreactivity and/or autoreactivity with human lipids and proteins (3).

The affinity maturation process leading to the generation and selection of bNAb-expressing B cells remains poorly understood but must occur in germinal centers (GCs). Data from animal models demonstrate a critical role for follicular helper T (T_FH) cells in the induction of GCs needed for the development of a high-affinity, pathogen-specific Ab response (4). The T_FH cells are targeted by HIV-1 very early postinfection and constitute a major cellular compartment for HIV-1 replication and viral particle production in the lymph nodes of viremic individuals (5). Despite their high susceptibility to HIV-1 infection, HIV-1–infected patients have been shown in many studies to have abnormal T_FH cell accumulation compared with uninfected individuals (6). Interestingly, T_FH cell frequencies correlate positively with plasma viremia levels, and T_FH cell accumulation diminishes with combination antiretroviral therapy (cART) (6). Circulating T_FH (c-T_FH) cells were recently identified as a memory compartment of tissue-resident T_FH cells and were shown to share with these an ability to produce IL-21 and provide helper signals to B cells (7). Therefore, T_FH cell function must be preserved to achieve efficient HIV-specific B cell responses. T_FH cells isolated from lymph nodes of HIV-1–infected individuals do not provide adequate B cell help in vitro (8). One of the complex mechanisms involved in T_FH cell dysfunction concerns the regulatory protein programmed death 1 (PD1). PD1 blockade has been shown to reinvigorate exhausted T cells (9). Incidentally, the PD1 ligand (PD-L1) has been described as highly expressed at the surface of B cells and dendritic cells in HIV-1–infected individuals (10, 11).

Previous studies have shown significant blood B cell abnormalities in HIV-1–infected patients, including an imbalance among peripheral mature B cell subsets, with overexpression of tissue-like and activated memory (AM) B cell subsets (12). HIV-associated exhaustion of tissue-like memory (TLM) B cells has been described based on a range of features and on similarities with T cell exhaustion (13). These features include increased expression of multiple inhibitory receptors and weak proliferative and effector responses to various stimuli. Chronic immune activation appears to play a critical role in phenotypic and functional B cell exhaustion. Conversely, resting memory (RM) B cells, which induce efficient secondary humoral responses, are depleted in the blood during the chronic stage of HIV-1 infection. When initiated at the chronic stage, cART fails to restore normal counts of blood memory B cells (14). In contrast, starting cART at the early stage of HIV-1 infection was associated with better restoration of RM B cells, in terms of both phenotype and function, as measured by the memory B cell response to a recall Ag (15). Low RM B cell counts may contribute to poor vaccine responses and weakened serological memory in HIV-1–infected individuals (16).

We previously reported that cART initiation during the early phase of HIV-1 infection, or early combination antiretroviral therapy (e-ART), ensured preservation of the mucosal gut lymphoid follicles (17). The tertiary lymphoid structures (TLSs) that develop during chronic inflammation can activate the molecular machinery needed to sustain in situ Ab diversification, isotype switching, B cell differentiation, and oligoclonal expansion, in keeping with their ability to function as active ectopic GCs. These observations raise the question of whether the timing of cART initiation may affect the development of the anti–HIV-1 humoral response in the gut. We designed a study to investigate this possibility.

The objective of this study was to compare e-ART to late combination retroviral therapy (l-ART)—started during the chronic stage of the infection—in terms of frequency, function, and specificity of mucosal T_FH cells and B cells in the gut mucosa of HIV-1–infected individuals. We compared PBMCs and rectal biopsy specimens from patients identified retrospectively after several years of e-ART or l-ART. Frequencies of functional T_FH cells and RM Env gp140–specific B cells in the gut mucosa were higher in the e-ART group. This finding supports a heretofore unsuspected role for the gut in generating Abs against HIV-1.

Paired PBMCs and rectal biopsy specimens were collected from 22 HIV-1–infected individuals who had been taking effective cART for several years. This treatment was started within 4 mo after the diagnosis of primary HIV-1 infection in nine patients (e-ART group) and later on (i.e., during the chronic stage of HIV-1 infection [Fiebig stage VI]) (18) in 13 patients (l-ART group). The diagnosis of primary HIV-1 infection was defined as a negative or weakly positive ELISA with no more than four bands by Western blot and positive viremia and/or positive HIV-1 ELISA following a negative ELISA within the preceding 3 mo. Gut biopsy specimens from six HIV-1–seronegative individuals were included as controls. All rectal biopsy specimens (∼2 μm³ each) were collected from the same site (10–15 cm from the anal margin) to avoid potential bias because of regional variations among participants. Table I reports the main features of the HIV-1–infected patients.

All study participants provided written informed consent to participate in the study. This study was approved by our local ethics committee (Tours, France) (Comité de Protection des Personnes de Tours, 17th of December 2014, no.: 2011-R26 [2011-CHRO-2011-02])

The recombinant HIV-1 Env YU-2 gp120 protein (gp120) and unlabeled HIV-1 and biotinylated YU-2 gp140 proteins (gp140 and gp140-biotin, respectively) were produced and purified as previously described (19, 20). Purified recombinant HIV-1 mature naive (MN) gp41 was provided by the National Institutes of Health AIDS Reagent Program.

Rectal biopsy specimens were collected by rectoscopy at the regional hospital center in Orléans, France. Intraepithelial lymphocytes and lamina propria lymphocytes were obtained as previously described (17). Cells were used without further processing for immunophenotyping, ELISPOT, and/or cell culture.

PBMCs (5 × 10⁵) were incubated for 6 d at 37°C in a final volume of 300 μl complete RPMI 1640 medium (Life Technologies) supplemented with 10% human AB serum in 96- well deep-well plates (Greiner MasterBlock; Sigma-Aldrich), with or without stimulation by pools of 150 HIV-1 15-mer Gag or Env peptides (1 μg/ml; JPT Peptide Technologies, Berlin, Germany). Staphylococcus aureus enterotoxin B (SEB) superantigen (50 ng/ml) served as a positive control. Supernatants were collected after 6 d of culturing, aliquoted, and stored at −80°C until use (21). IL-21 and IFN-γ produced in the supernatant by stimulated or unstimulated PBMCs were quantified using Luminex kits (ProcartaPlex; Affymetrix; eBioscience; Thermo Fisher Scientific, San Diego, CA) according to the manufacturer’s instructions. All samples were acquired on a Bio-Plex 200 instrument (Bio-Rad Laboratories, Marnes-la-Coquette, France).

The phenotypes of isolated mucosal B cells were assessed using FACS staining with the following Abs: anti–CD3-BV605 (SK7; BD Biosciences, Le Pont de Claix, France), anti–CD19-PECF594 (HIB19; BD Biosciences), anti–CD10-PECy7 (HI10a; BioLegend, Ozyme, Saint-Quentin en Yveline, France), anti–CD21-BV711 (B-Ly4; BD Biosciences), anti–CD27-APC (L128; BD Biosciences), anti–CD38–PerCP-Cy5.5 (HIT2; BioLegend, Ozyme), anti–IgG-BV421 (G18-145; BD Biosciences), and anti–IgA-FITC (IS11-8E10; Miltenyi Biotec, Paris, France). Mucosal T_FH cells were stained with Abs to CD3-BV605 (SK7; BD Biosciences), CD4-PECF594 (RP4-T4; BD Biosciences), CXCR5–Alexa 488 (RF8B2; BD Biosciences), and PD1-BV421 (EH12.EH7; BioLegend, Ozyme). For intracellular staining, cells were fixed and permeabilized using the Foxp3 staining buffer set (eBioscience, Thermo Fisher Scientific), washed, and incubated with anti–BCL6-PE (K112-91; BD Biosciences). For all cell stainings, dead cells were excluded from the gating by using the LIVE/DEAD fixable dead cell stain kit (Molecular Probes, Invitrogen, Saint-Aubin, France). Cytometry acquisition was performed on an LSR II flow cytometer (BD Biosciences), and the data were analyzed using FlowJo software (version 7.6.5; Tree Star, Ashland, OR).

DNA was extracted from frozen biopsy specimens using the Qiasymphony automated extraction device (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. The B cell repertoire was evaluated by detection of heavy-chain Ig (IgH) gene rearrangements according to BIOMED-2 guidelines (22). Briefly, three sets of human Ig H chain–V region (VH) primers corresponding to the three VH framework regions (FR1, FR2, and FR3) were used. Each set of primers consisted of six or seven oligonucleotides capable of annealing to their corresponding VH segments (VH1–VH7). These VH primer sets were used in conjunction with a single HEX-labeled human Ig H chain–joining region consensus primer. After PCR, CDR3-derived products were loaded on a 3130 Genetic Analyzer (Applied Biosystems, Foster City, CA), and fragment sizes were analyzed by GeneScan (Thermo Fisher Scientific).

The total cell suspension isolated from rectal biopsy specimens was incubated for 6 d in complete RPMI 1640 medium (Life Technologies) supplemented with 10% FCS in 96-well plates (Nunc MaxiSorp; Thermo Fisher Scientific, Roskilde, Denmark), alone or with immobilized LEAF purified agonist anti-CD40 Ab (5 μg/ml; BioLegend, Ozyme), recombinant human IL-4 (50 ng/ml; Cell Signaling, Ozyme), and IL-21 (50 ng/ml; Cell Signaling Technology, Ozyme). Supernatants from 6-d-old cultures were collected and stored at −20°C.

Polystyrene 96-well ELISA plates (Nunc MaxiSorp) were coated with anti-human IgG (2.5 μg/ml; Jackson ImmunoResearch, Interchim, Montluçon, France) and anti-human IgA (5 μg/ml, HB200 (23)) in PBS overnight at 4°C. Plates were blocked by 2 h of incubation with PBS containing 1% BSA (Sigma-Aldrich). After washings, the plates were incubated for 2 h with supernatants from cultures of differentiated B cells and 3-fold serial dilutions in PBS–1% BSA. The plates were then washed and incubated for 1 h with HRP-conjugated anti-human IgG, IgA, and IgM Abs (Jackson ImmunoResearch, Interchim). Purified 10-1074 monoclonal IgG (24) and IgA1 (23) Abs (12 μg/ml⁻¹ starting concentration) were used as standards. To test HIV-1 gp140 reactivity, purified recombinant YU-2 gp140 trimers were coated (5 μg/ml) on polystyrene 96-well ELISA plates (Nunc MaxiSorp) overnight at 4°C in PBS. The plates were then blocked as described above and incubated for 2 h with IgGs secreted in B cell culture supernatants, adjusted to a concentration of 2 μg/ml in PBS–1% BSA. After washings, the plates were incubated for 1 h with HRP-conjugated anti-human IgG Abs (Jackson ImmunoResearch, Interchim), then revealed using tetramethylbenzidine substrate (Life Technologies). Anti–HIV-1 gp140 monoclonal IgG Abs 2F5 and 2G12 (National Institutes of Health AIDS Reagent Program) were used as positive controls.

Deparaffinized tissue sections were stained with mouse anti-human Pax-5 (DAK-Pax5; DakoCytomation, Glostrup, Denmark) and rabbit anti-human PD-L1 (E1L3N; Cell Signaling Technology, Ozyme) Abs, then with the anti-rabbit and anti-mouse IgG avidin–biotin complex system (ABC Kit Universal, VECTASTAIN; Vector Laboratories, Les Ulis, France). Cell staining was performed using the DAB Substrate Kit for peroxidase (Vector Laboratories). All slides were counterstained with hematoxylin. Immunohistochemical images were acquired on a ZEISS Axioplan 2 (Göttingen, Germany) microscope equipped with a ×20 (0.45 NA) objective using a Zeiss MRc digital camera (Göttingen, Germany) and AxioVision microscope software (ZEISS).

Total RNAs were isolated from rectal biopsy specimens using the RNeasy Micro Kit (Qiagen) according to the manufacturer’s protocol, then retrotranscribed into cDNA molecules using the Affinity Script QPCR cDNA Synthesis Kit (Agilent, Santa Clara, CA). Quantitative PCRs were performed using the Brilliant II SYBR GREEN QPCR Master Mix kit (Agilent) on the Mx3005 QPCR machine (Agilent). OAZ-1 mRNA, whose expression was found to be stable across the three groups of participants, was used as a control for sample normalization. The relative levels of each gene were calculated using the 2^−∆∆CT method.

The following primers were used (forward and reverse, 5′–3′): OAZ-1, 5′-ACTTATTCTACTCCGATGATCGAGAATCCTCGTCTTGTC-3′ (Invitrogen); IL-6, 5′-CTCAGCCCTGAGAAAGGAGATTCTGCCAGTGCCTCTTTGC-3′ (Eurofins Genomics, Les Ulis, France); IL-27p28, RefSeq accession no. NM_145659.3 (Qiagen); EBI-3, RefSeq accession no. NM_005755.2 (Qiagen); AICDA, RefSeq accession no. NM_020661 (Qiagen); IL-12A, 5′-AATGTTCCCATGCCTTCACCCAATCTCTTCAGAAGTGCAAGGG-3′ (Eurofins Genomics).

Groups were compared using either the two-sided Mann–Whitney U test or the Kruskal–Wallis test. The Spearman rank test was applied to assess bivariate correlations, and linear regression analysis was performed to produce an accompanying best-fit line. All statistical analyses were performed using GraphPad Prism (version 6.0; GraphPad Software, La Jolla, CA).

The phenotypes of cells freshly isolated from rectal biopsy specimens were compared in HIV-1–infected patients given e-ART (n = 7) or l-ART (n = 8) and in HIV-negative controls (n = 6). Table I reports the characteristics of the participants. Absolute counts of total CD19⁺ B cells were significantly higher in the l-ART groups than in the control groups, with no difference between the e-ART and l-ART groups (Fig. 1A). No differences were observed in terms of CD19⁺ B cell frequency between HIV-1–infected groups and controls. By assessing differences in surface CD27 and CD21 expression in CD19⁺ CD38^low CD10⁻ B cells, we identified MN (CD27⁻ CD21⁺), RM (CD27⁺ CD21⁺), TLM (CD27⁻ CD21⁻), and AM (CD27⁺ CD21⁻) B cells (Fig. 1B). As shown in Fig. 1C, patients in both the cART groups exhibited lower frequencies of AM B cells (1.3 ± 0.5% and 1.6 ± 0.4%, respectively) compared with the controls (4.4 ± 1.5%) (p < 0.001 for both comparisons). Other B cell subsets in e-ART patients were comparable to those from controls but differed significantly from those in l-ART patients, who had a lower frequency of RM B cells (39.6 ± 10.8% versus 71.8 ± 15%, p < 0.001) and higher frequencies of MN and TLM B cells (54.5 ± 11.6% and 2.6 ± 0.9% versus 23.3 ± 14.1% and 0.9 ± 0.6%, p < 0.001 and p < 0.01, respectively). Blood B cell phenotype was not substantially different between the e-ART and l-ART groups (Supplemental Fig. 1).

The frequencies of terminally differentiated B cells (Ab-secreting cells [ASCs]), defined as CD19⁺ CD27^hi CD38⁺ CD10⁻ and known to be abundant in the gut mucosa (25), were comparable in the e-ART and l-ART groups (38.5 ± 15.9% versus 30.7 ± 13.3%, respectively) (Fig. 2A) but were significantly lower than in the control group (71.7 ± 12.3%, p < 0.001 for both comparisons). The total amount of Igs spontaneously released by freshly isolated mucosal ASCs cultured for 6 d was not different between the e-ART and l-ART groups (data not shown). In contrast, significant differences were noted regarding the Ig isotype profile. Thus, IgA release by ASCs from e-ART patients was greater compared with ASCs from l-ART patients (9.4 ± 13.4 μg/ml versus 2.8 ± 1.4 μg/ml, p < 0.05, respectively) and similar to that seen with ASCs from controls (Fig. 2B). In contrast, the total amount of IgGs released by ASCs was significantly greater in the l-ART group than in the e-ART group (6.4 ± 5.8 μg/ml versus 2.5 ± 1.5 μg/ml, p < 0.05) (Fig. 2B). The IgA/IgG ratio indicated skewing from IgAs to IgGs in the l-ART group compared with the e-ART group (0.68 ± 0.72 arbitrary units [AU] versus 9.4 ± 15.24 AU, p < 0.001) (data not shown). The IgA/IgG ratio differed significantly between the l-ART and control groups (0.68 ± 0.72 AU versus 2.89 ± 1.93 AU, p < 0.05) but was comparable between the e-ART and control groups (9.4 ± 15.24 AU versus 2.89 ± 1.93 AU, p = 0.289).

To evaluate whether the time of cART initiation might influence the mucosal B cell repertoire in HIV-1–infected individuals, we studied B cell clonality in the e-ART and l-ART groups (26). All patients in both groups displayed a normally distributed polyclonal profile (Fig. 2C) similar to that typically observed in healthy humans (27). Interestingly, some of the treated patients exhibited abnormal, pre-eminent single peaks in their Ig spectratype (Fig. 2D), suggesting clonal B cell expansions in ectopic mucosal lymphoid structures, such as those described in reactive lymphoproliferation (28, 29). Interestingly, these peaks were more common in the e-ART group than in the l-ART group, although the difference was not statistically significant (25 versus 8.3%, p = 0.34) (Fig. 2E).

The development of memory B cells within GC follicles depends heavily on the presence of T_FH cells (30). The frequency of mucosal T_FH cells, defined as CD3⁺CD4⁺PD1^hiCXCR5⁺Bcl6⁺ cells (Fig. 3A), was significantly higher in the e-ART group than in the l-ART group (9.5 ± 5.1% versus 1.6 ± 1.5% of CD3⁺CD4⁺ cells, p < 0.05); the values in the e-ART and l-ART groups were significantly higher than in the controls (0.3 ± 0.3%, p < 0.0001 and p < 0.05, respectively) (Fig. 3B). As shown in Fig. 3C, the frequency of T_FH cells correlated significantly with the frequency of RM B cells (r = 0.7542, p < 0.01). As illustrated in Fig. 3D, an analysis of differential CXCR3 and CCR6 expression (31) allowed us to define three main c-T_FH cell subsets within blood CCR7⁻CXCR5⁺CD4⁺T cells: CXCR3⁺CCR6⁻ (c-T_FH1), CXCR3⁻CCR6⁻ (c-T_FH2), and CXCR3⁻CCR6⁺ (c-T_FH17). No differences in frequency or phenotype of c-T_FH cells were observed between the e-ART and l-ART patients (Fig. 3E).

The above-reported results and the role for PD-L1^hi B cells in regulating T_FH cell expansion and function (8, 10) led us to investigate whether PD-L1 expression in mucosal follicles differed between the e-ART and l-ART groups. Single-cell expression of Pax5 and PD-L1 was sought by immunohistochemistry of rectal biopsy specimens from e-ART (n = 6) and l-ART (n = 6) patients (Fig. 4A). Based on Pax5 staining, B cell follicle architecture differed between the two groups. All e-ART patients displayed well-defined secondary follicles, whereas most l-ART patients had some degree of B lymphoid area disorganization, with diffuse B cell distribution in four of the six biopsy specimens (patients g, h, j, and m). PD-L1 expression was clearly detectable in a single e-ART patient (patient e) and was not located in the B cell area (Fig. 4A). In contrast, PD-L1 expression was high in the follicles of five of the six biopsy specimens from l-ART patients and was located within the B cell area in three of these five biopsy specimens (patients g, k, and h) (Fig. 4B).

Previous studies have established the importance of soluble factors such as IL-6 and IL-27 for the development and maintenance of T_FH cells in mice and humans (32, 33). We used real-time quantitative PCR technology to quantify the transcripts for IL-6 and the two IL-27 subunits (IL-27p28 and EBI3) in the rectal biopsy specimens from patients in both HIV-1–positive groups (Supplemental Fig. 2). All three mRNAs were expressed at significantly higher levels in the e-ART group than the l-ART group (IL-6 mRNA: 4.39 ± 6.14 AU versus 0.76 ± 0.57 AU, p < 0.05; IL-27p28 mRNA: 0.12 ± 0.22 AU versus 0.04 ± 0.03 AU, p < 0.05; and EBI3 mRNA: 0.21 ± 0.09 AU versus 0.08 ± 0.04 AU, p < 0.05). The expression of control mRNAs encoding the IL-12A subunit, which also dimerize with EBI3 to form IL-35, was not different between the two HIV-positive groups (0.51 ± 0.06 AU versus 0.49 ± 0.15 AU; p = 0.4127). Cross-talk between B cells and T_FH cells was investigated by real-time quantitative PCR quantification of activation-induced cytidine deaminase transcripts. Activation-induced cytidine deaminase mRNA expression tended to be higher in mucosal GCs from e-ART patients compared with l-ART patients without reaching significance (Supplemental Fig. 2, 0.07 ± 0.11 AU versus 0.006 ± 0.007 AU, p = 0.111).

Next, we investigated whether HIV-1–specific B cell responses were affected by the timing of cART initiation. We used flow cytometry to evaluate the frequency and phenotype of YU-2 gp140-reactive B cells. Fig. 5A shows representative dot plots of CD19⁺ gp140-reactive cells from the patients and controls. Importantly, the frequency of mucosal gp140-reactive CD19⁺ cells was significantly higher in the e-ART group than in the l-ART group (0.26 ± 0.09% versus 0.07 ± 0.05%, p < 0.05 (Fig. 5B), and the phenotype of these cells differed between the two groups (Fig. 5C), with a predominance of RM cells in the e-ART group and a mixture of MN and RM cells in the l-ART group. Moreover, the frequency of mucosal gp140-reactive RM B cells was significantly higher in the e-ART group compared with the l-ART group (0.22 ± 0.14% versus 0.05 ± 0.08%, p < 0.05) (Fig. 5C). In line with these results, the frequency of gp140-reactive B cells expressing membrane-bound IgG was significantly higher in the e-ART group than in the l-ART group (0.28 ± 0.18% versus 0.06 ± 0.04%, p < 0.05) (Fig. 5D). Finally, the frequency of total mucosal gp140-reactive B cells correlated significantly with that of T_FH cells in the HIV-infected patients (r = 0.7821, p < 0.001) (Fig. 5E). We used an ELISA against trimeric YU-2 gp140 to test supernatants of freshly isolated mucosal B cells from both groups of HIV-1–infected patients after 6 d of stimulation. In the e-ART group, compared with unstimulated mucosal B cells, stimulated cells released larger amounts of gp140-reactive IgGs (0 AU versus 0.21 ± 0.27 AU, p < 0.05 (Fig. 5F). Stimulation did not have this effect on cells from the l-ART group (Fig. 5F).

To further investigate the association between the time of cART initiation and the role for T_FH cells in the development of HIV-1–specific B cell responses, we evaluated IL-21 production by PBMC. Indeed, IL-21 is primarily produced by CD4⁺ T cells and is particularly critical to generation of Ag-specific IgG Abs and expansion of class-switched B cells and plasma cells in vivo. Blood IL-21–secreting CD4⁺ T cells share phenotypic and transcriptional similarities with lymphoid T_FH cells in HIV-1–infected individuals (34). Given the very limited number of total cells that can be retrieved from rectal biopsy specimens, we used blood samples to quantify HIV-1–specific IL-21⁺ and HIV-1–specific IFN-γ⁺ T cells in both HIV-1–positive groups. PBMCs stimulated with SEB released significantly more IL-21 in the e-ART group than in the l-ART group (43.9 ± 35.9 pg/ml versus 10.9 ± 9.4 pg/ml, p < 0.01) (Fig. 6A, left panel). Similarly, SEB-stimulated PBMCs released more IFN-γ in the e-ART group than in the l-ART group (17 866.5 ± 9 512.6 pg/ml versus 9 186.4 ± 11 244.4 pg/ml, p < 0.05) (Fig. 6A, right panel). In addition, IL-21 release by PBMCs stimulated with pools of HIV-1 Env and Gag peptides was significantly more marked in the e-ART group (5.3 ± 3.5 pg/ml versus 2.7 ± 2.9 pg/ml in the l-ART group, p < 0.05) (Fig. 6B). In contrast, no statistically significant differences were observed between the e-ART and l-ART groups for the secretion of IFN-γ by PBMCs stimulated with HIV-1 Env and Gag peptides (Fig. 6B). In line with these results, the frequency of gp140-reactive B cells in blood was higher in the e-ART group (0.21 ± 0.12% versus 0.1 ± 0.07%, p < 0.05) (Supplemental Fig. 3).

Previous studies have shown that persistent infection with viruses such as HIV-1 lead to severe abnormalities in the dynamics of B cell distribution and function in the blood and lymphoid organs, which very likely interfere with the establishment of an optimal antiviral humoral response (35). Far less is known about whether these alterations also occur in the gut mucosa and whether the timing of cART initiation influences B cell phenotype and function. Our previous gene profiling study distinguished two groups of patients based on pathway signatures of gut mucosal lymphoid structures and dendritic cell function, which perfectly matched the timing of cART initiation (17). In this article, we extend those data by demonstrating, at the cellular level, that e-ART initiation preserves gut TLSs, which may function as active ectopic GCs characterized by high frequencies of functional T_FH cells and gp140-reactive memory B cells. These GCs may play a critical role in the development of the Ab response.

In the gut, e-ART was associated with partial correction of the abnormal expansion of AM and TLM B cells and with preservation of RM B cells, conferring on e-ART patients a phenotype comparable to that of uninfected controls. In contrast, patients treated only at the chronic stage, despite experiencing long-term control of HIV replication, exhibited a profile suggesting impaired B cell maturation, with a significant reduction in RM B cells. Finally, both groups of HIV-1–infected patients had similarly lower frequencies of ASCs compared with the uninfected control group. In contrast to findings at the mucosal levels, we did not find significant differences in the blood of early- and late-treated patients. These results differ from results reported by others (15, 36) and might be explained by a longer duration of ART treatment (more than 10 y on with an average of 4–21 y) in our cohort of l-ART patients. However, according to our results, a partial restoration of a normal homeostasis of B cell populations with a decrease of AM and TLM B cells has been reported in chronically ART-treated patients (37, 38). Altogether, our results underscore the interest to study changes in B cell populations in various compartments, revealing in this study that if e-ART may limit the major B cell subset alterations in the gut, they are only partially restored even in the long term.

Although ASC frequencies in the gut were comparable in the e-ART and l-ART groups, the proportion of IgG-secreting cells was higher and the proportion of IgA-secreting cells commensurately lower in the l-ART group. The abnormal predominance of IgG in l-ART patients may reflect mucosal inflammation, which may contribute to impair gut mucosal homeostasis, as observed in inflammatory bowel disease (39). Moreover, the IgA secretion deficiency may cause changes in the composition of the intestinal microbiota (40) that may further activate the inflammatory processes seen in the gut of patients with chronic HIV-1 infection despite effective cART (41). The skewing of IgA-secreting cells toward IgG-secreting cells is probably linked to microbial translocation and noninfectious complications associated with systemic inflammation. We therefore looked for abnormalities in global mucosal B cell repertoires in the e-ART and l-ART groups, using the spectratyping method. Surprisingly, all HIV-1–infected patients displayed polyclonal profiles, although single expansions were noted, more often in the e-ART group than in the l-ART group. However, the global repertoire analyses by Ig spectratyping were performed on mucosal Ig-expressing and Ig-secreting cells in the gut, including a high proportion of ASCs resulting from T cell–independent differentiation of mucosal B cells. Given the massive T cell depletion associated with HIV-1 infection (42), complete disorganization of ectopic mucosal GCs (17, 43), and crucial role for these GCs in memory B cell development within TLSs (44), any disturbances in the B cell repertoire would mainly concern the GC B cells (6) rather than the T cell–independent ASCs.

In GCs, T_FH cells are strongly involved in the development of memory B cells. In this study, we found that T_FH cells were expanded in the gut of HIV-1–infected patients compared with controls. Importantly, the frequency of gut T_FH cells correlated with the frequency of RM B cells in the gut. These results are in line with previous reports showing T_FH cell expansion in lymph node, spleen, and gut tissues of rhesus macaques infected with SIV (45, 46) and in mucosal tissues from humanized DRAG mouse models of HIV-1 infection (47). T_FH cells decreased substantially with cART (6). Surprisingly, gut T_FH cells remained significantly higher in the e-ART group than in the l-ART group, whereas no differences were observed for c-T_FH frequency between the two groups. Our results underline the critical impact of tissue compartmentalization on T_FH cell and B cell dynamics during HIV infection. In SIV-infected rhesus macaques, T_FH cell dynamics differ from one compartment to another (peripheral blood versus lymph nodes or spleen) (48). Indeed, microenvironment is essential for the differentiation and the maintenance of T_FH cells. In the gut, the microbiota induces the differentiation of CD4⁺ T cells into T_FH cells, thereby promoting the secretion of microbial-specific IgAs, which are important for controlling the microflora and maintaining gut homeostasis (49). Thus, T_FH cell expansion in HIV-1–infected patients may be seen as a mechanism that compensates for the massive Th17 depletion, thereby helping to maintain gut homeostasis. This hypothesis is supported by studies in RORγt-deficient mice, in which large numbers of TLSs are required to contain the microbiota (50).

T_FH cell dynamics vary according to the severity of the disease. Slow-progressor rhesus macaques display an increased frequency of T_FH cells in lymph nodes, whereas their numbers drastically decreased in fast-progressor rhesus macaques (51, 52). Evidence supports the pivotal role of persistent viral Ag within the GC in driving T_FH cell expansion, and the disruption of GC organization coincides with the loss of T_FH cells and the onset of AIDS in terminal stages of HIV infection (51). CXCL13 has been described to be a plasma biomarker of GC activity in HIV-infected humans (53). In our cohort of 56 l-ART patients and 17 e-ART patients, CXCL13 tended to be higher in the sera of e-ART patients compared with l-ART patients without reaching significance (data not shown). We have previously reported a loss of follicular dendritic cells networks and TLSs in the gut of l-ART patients. Thus, this may impact T_FH cell maintenance in l-ART patients.

The difference of T_FH cell frequencies between e-ART and l-ART patients may also reflect distinct immune response of T_FH cells depending on the nature of help signals, consisting of both cytokines and cell surface molecules. We therefore investigated the signaling factors that contribute to T_FH cell expansion. Studies in SIV infection models (46) and in mice (33) established a key role for IL-6 and IL-27 signaling in T_FH cell function and GC responses. In our study, IL-6 and IL-27 transcript levels in the gut were higher with e-ART than with l-ART. In addition to signaling mediators, B cell dysregulation may also be involved in the reduced frequency and impaired function of T_FH cells in HIV-1 infection (8). PD-L1 expression on B cells and PD-1 receptor engagement on T_FH cells decrease IL-21 secretion and cell proliferation (8, 10). We found that the proportion of TLS B cells expressing PD-L1 was greater in the l-ART group than in the e-ART group. These results suggest that e-ART patients had functional T_FH cells capable of contributing to the development of Ag-specific B cell responses in gut GCs.

Recent work highlighted the importance of maintaining functional GCs for the development of HIV-1 bNAbs (54). We therefore hypothesized that functional HIV-1–specific T_FH cells enhanced HIV-1–specific B cell responses in the gut. In keeping with this hypothesis, the frequencies of T_FH cells and gp140-reactive memory B cells in gut TLSs were higher in the e-ART group than in the l-ART group. Interestingly, the frequency of gut T_FH cells correlated with the frequency of gut gp140-reactive memory B cells in cART-treated HIV-1–infected patients. In line with these results, it has been recently shown that HIV Env-specific CXCR5⁺ CD4⁺ T cells that secrete IL-21 are strongly associated with B cell memory phenotypes and function (55). The results suggested that circulating total and HIV-1–specific IL-21–producing T cells were more abundant with e-ART than with l-ART. In contrast, counts of circulating IFN-γ–secreting HIV-1–specific T cells were not significantly different between the two groups. It is tempting to speculate that the HIV-1 specificity of gut T_FH cells may be extrapolated from the amount of IL-21 released by T cells in response to stimulation with a pool of HIV-1 peptides. Thus, TLSs may act as active ectopic GCs and may play a critical role in the development of the affinity-matured HIV-1–specific Ab response.

The first HIV-1–reactive Abs become detectable ∼13 d after HIV-1 transmission (56) and are mainly directed against the Env gp41 in both blood and the terminal ileum. Most of these gp41 Abs are polyreactive, affinity-matured IgGs that target self-antigens and microbial Ags (57). Using an in vitro B cell–to–ASC differentiation assay, we confirmed that the frequency of gp140-reactive memory B cells was higher with e-ART than with l-ART, as shown by the larger amount of anti–gp140-reactive IgGs detectable by ELISA in the e-ART group. The anti-gp140 IgGs targeted the gp41 portion of the HIV-1 Env protein (data not shown). Mucosal B cell clones can re-enter a GC, where they undergo further somatic hypermutation to produce high-affinity IgA that is adapted to the changing composition of the microbiota. It is tempting to speculate that gp140-reactive memory B cells may be a good target for a therapeutic vaccine. Indeed, a recent study of the prevaccination B cell repertoire identified a preexisting pool of microbiome–gp41 cross-reactive B cells that was stimulated by the vaccine (58). Extensive molecular characterization of the gp140-reactive B cells would be important to explore the potential beneficial effects of e-ART on the development of a potent HIV-1–specific humoral immune response in the gut.

The beneficial impact of e-ART on the circulating B cell populations is now well documented (15). In this article, we demonstrated that e-ART may also lessen the alterations in mucosal B cell subsets. The protection afforded by a potent mucosal humoral response is particularly important in the gut, where the intestinal barrier is continuously attacked by the microbiota. GC preservation may contribute to diversification of the mucosal B cell repertoire, thereby helping to control the billions of microorganisms found in the gut lumen (59). Thus, e-ART may contribute to reducing the appearance of non–HIV-1 AIDS-related gastrointestinal syndromes. Considerable effort is being put into creating a vaccine-based strategy for developing HIV-1 bNAbs in infected patients. A common feature of bNAbs is a higher level of somatic hypermutations compared with that seen in typical immune responses (60, 61), which are generated after multiple cell passages through GCs containing target Ags. We demonstrated that e-ART helped to preserve intestinal GC functions and was associated with a higher frequency of HIV-1 Env gp140-specific B cells in the gut compared with l-ART. This finding suggests that eliciting potent anti–HIV-1 Abs at mucosal sites may require e-ART to maintain an optimal mucosal GC response by preserving T_FH cell function and, therefore, maturing GC B cells.

The authors have no financial conflicts of interest.