Characterization of Plasmablasts in the Blood of HIV-Infected Viremic Individuals: Evidence for Nonspecific Immune Activation

Abstract

Terminal differentiation of B cells and hypergammaglobulinemia are hallmarks of B-cell hyperactivity in HIV disease. Plasmablasts are terminally differentiating B cells that circulate transiently in the blood following infection or vaccination; however, in HIV infection, they arise early and are maintained at abnormally high levels in viremic individuals. Here we show that only a small fraction of plasmablasts in the blood of viremic individuals is HIV specific. Assessment of plasmablast immunoglobulin isotype distribution revealed increased IgG⁺ plasmablasts in early and most prominently during chronic HIV viremia, contrasting with a predominantly IgA⁺ plasmablast profile in HIV-negative individuals or in aviremic HIV-infected individuals on treatment. Of note, IgG is the predominant immunoglobulin isotype of plasmablasts that arise transiently in the blood following parenteral immunization. Serum immunoglobulin levels were also elevated in HIV-infected viremic individuals, especially IgG, and correlated with levels of IgG⁺ plasmablasts. Several soluble factors associated with immune activation were also increased in the sera of HIV-infected individuals, especially in viremic individuals, and correlated with serum immunoglobulin levels, particularly IgG. Thus, our data suggest that while plasmablasts in the blood may contribute to the HIV-specific immune response, the majority of these cells are not HIV specific and arise early, likely from indirect immune-activating effects of HIV replication, and reflect over time the effects of chronic antigenic stimulation. Such B-cell dysregulation may help explain why the antibody response is inadequate in HIV-infected individuals, even during early infection.

INTRODUCTION

Immunologic abnormalities arise shortly after HIV infection and persist in the majority of individuals in the absence of antiretroviral therapy (ART). In the pathogenesis of HIV infection, CD4⁺ T cells are the primary targets of the virus (reviewed in reference 1), although other lymphocyte populations are affected in the absence of direct infection, including B cells. B-cell dysfunction in HIV infection and especially in viremic individuals is characterized by phenotypic and functional alterations in B-cell subpopulations resulting primarily in impaired humoral responses to vaccination and certain infections (2–14). One of the hallmarks of HIV infection in viremic individuals is hypergammaglobulinemia (9, 15–23). HIV viremia is also associated with increased terminal differentiation of B cells, identified phenotypically in the peripheral blood of HIV-viremic individuals, as well as functionally by increased frequencies of cells spontaneously secreting immunoglobulins (Igs) ex vivo (9, 16, 18, 19, 24, 25). Recently, we demonstrated a higher frequency of plasmablasts (PBs), defined as Ig-secreting B cells that are cycling (Ki-67⁺), in the blood of early compared to chronically infected HIV-viremic individuals (26), consistent with findings from another study showing rapid induction of polyclonal terminal B-cell differentiation shortly after infection (27). However, despite these observations, little is known regarding the origin of PBs and their association with hypergammaglobulinemia in HIV infection.

In healthy individuals at steady state (in the absence of immunization or infection), terminally differentiating B cells are present at very low levels in the peripheral blood, on the order of 1 to 3% of all circulating B cells (26, 28, 29). Although different terminologies have been used, the vast majority of terminally differentiating B cells in the blood are PBs. At steady state, IgA is the predominant Ig isotype of PBs circulating in the blood (29, 30), and these PBs are thought to reflect migration to and from mucosal sites and other secondary lymphoid tissues resulting from homeostatic events, immune surveillance, and low-level antigenic stimulation (29–31). After systemic immunization or infection, there is a rapid, yet transient, burst of PBs in the blood that likely reflect extrafollicular reactions in the case of a primary response (32) and stimulation of memory B cells during secondary responses (28, 33). Recent studies have shown that IgG becomes the predominant isotype of PBs that circulate transiently in the blood following a secondary response to T-cell-dependent immunogens such as those contained in tetanus and diphtheria vaccines (29), as well as in vaccination or natural infection with influenza virus (32, 34). Studies on the burst of PBs seen in peripheral blood that occurs during acute viral infections, including infections with influenza virus, dengue virus, and respiratory syncytial virus (RSV), have shown that a high fraction of these PBs is pathogen specific (34–37).

In the setting of HIV infection, relatively little is known about the nature of the PBs that circulate in the peripheral blood of infected individuals. In the present study, we used the enzyme-linked immunospot (ELISPOT) assay, which allows for the enumeration of HIV-specific antibody-secreting cells, to evaluate the extent to which the HIV-specific response contributed to the overexpression of PBs observed in early and chronic HIV infection. In an effort to better understand blood-derived PBs in HIV infection, we also concomitantly evaluated Ig isotype distribution on PBs as well as serum Igs and other soluble factors at different stages of disease. Our findings show that only a small fraction of PBs in the blood of infected individuals is HIV specific, and the majority of PBs are likely to be induced by systemic immune-activating effects of the virus.

MATERIALS AND METHODS

Study subjects.

Leukapheresis and blood draw products were obtained from study subjects after they provided informed consent in accordance with the Institutional Review Board of the National Institute of Allergy and Infectious Diseases, National Institutes of Health. Four groups of individuals were recruited: early infected HIV-viremic individuals, defined as having been infected within 6 months of donating baseline samples and not receiving ART; chronically infected HIV-viremic individuals, defined as having been infected for at least 6 months and not receiving ART; chronically infected HIV-aviremic individuals, defined as having initiated ART during the chronic phase of infection and maintaining a plasma viremia level of <50 copies HIV RNA per ml for at least 3 months; and HIV-negative individuals. HIV plasma viremia was measured by branched DNA assay (Bayer Diagnostics), with a lower limit of detection of 50 copies per ml. Characteristics for each group of study participants are shown in Table 1, and P values are indicated where there were statistical differences between groups. Of note, participants who presented with evidence of an ongoing opportunistic infection at the time of study were excluded.

Table 1.

Characteristics of study participants

Parameter	Value for group
	Early infected HIV viremic	Chronically infected HIV viremic	Chronically infected HIV aviremic	HIV negative
No. of participants	35	55	40	47
Agea	36 (20–58)	37 (21–59)	41 (20–61)	37 (22–63)
Male sex (%)	88.6	81.8	85	89.4
CD4+ T-cell count (cells/μl)a	475 (187–1,178)	393 (93–1,048)	482 (195–1,086)	757 (280–1,262)c
Differences in CD4+ T-cell count	P < 0.0001d	P < 0.0001d; P = 0.017e	P < 0.0001d
HIV RNA (copies/ml plasma)b	38,889 (107–1,093,280)	13,260 (445–399,052)	<50	NAg
Differences in HIV RNA	P = 0.001f

^aMedian (range).

^bGeometric mean (range).

^cAvailable for 28 individuals.

^dSignificant difference compared to HIV-negative group.

^eSignificant difference compared to chronically infected HIV-aviremic group.

^fSigniifcant difference compared to chronically infected HIV-viremic group.

^gNA, not applicable.

ELISPOT assay.

Peripheral blood mononuclear cells (PBMCs) were isolated from blood products by Ficoll-Hypaque density gradient centrifugation. Mature (CD10⁻) B cells were enriched to ∼95% purity from PBMCs by negative selection (StemCell Technologies), as previously described (11). ELISPOT assay was used to enumerate HIV-specific and total IgG-, IgA-, and IgM-secreting B cells as previously described (11, 26, 38). In brief, Immobilon-P polyvinylidene difluoride (PVDF) membrane plates (MAIPSWU10; Millipore) were coated with 5 μg/ml anti-Ig light-chain antibodies (Rockland Immunochemicals), followed by addition of cells, incubation for 5 h, and detection with biotinylated antibodies against each of the Ig classes or biotinylated YU2-gp140-F (39).

Phenotypic analyses.

PBMCs were isolated from blood by Ficoll-Hypaque density gradient centrifugation, and multicolor flow cytometric analyses were performed. The following conjugated monoclonal antibodies (MAbs) were used for cell staining: allophycocyanin (APC)-H7 anti-CD20 and APC anti-IgG (BD Biosciences), peridinin chlorophyll protein (PerCP)-Cy5.5 anti-CD19 and phycoerythrin (PE)-Cy7 anti-CD27 (eBioscience), biotin anti-IgM (Invitrogen Life Technologies), and fluorescein isothiocyanate (FITC) anti-IgA (Dako). The secondary Ab for anti-IgM was streptavidin-APC (BD Biosciences). For gp140 staining, B cells were isolated from PBMCs and assayed as previously described (40, 41), with the following modifications. Biotinylated trimeric YU2 gp140-F was conjugated to streptavidin-APC (Invitrogen) as described previously (41). The conjugated gp140 was used to identify the HIV-specific memory (CD20⁺ IgG⁺) B cells. Flow cytometry was performed on a FACSCanto flow cytometer (BD Biosciences), with data analyses on FlowJo Version 9.6 software (TreeStar).

Serum analyses.

Twenty-one cytokines, chemokines, and soluble receptors were measured by multiplex cytometric bead array (CBA) using a FACSArray instrument (BD Biosciences). The following factors were analyzed: soluble tumor necrosis factor receptor II (TNFRII), MIG, soluble ICAM, IP-10, alpha interferon (IFN-α), MIP-1α, soluble TNF receptor I (TNFRI), MCP-1, MIP-1β, granulocyte-macrophage colony-stimulating factor (GM-CSF), lymphotoxin alpha (LT-α), IFN-γ, interleukin 1β (IL-1β), IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-12p70, and TNF-α. The CBA analyses were conducted according to the manufacturer's instructions. Data were analyzed using FCAP Array software (version 1.01; BD Biosciences), which converts sample mean fluorescent intensity values into a concentration using the standard curve. Levels of serum Ig were measured by CBA according to the manufacturer's instructions (BD Biosciences), with the exception of IgG1, which was calculated by subtracting levels of IgG2 to IgG4 from IgG. Serum levels of human B-cell-activating factor belonging to the TNF family (BAFF) were measured by enzyme-linked immunosorbent assay (ELISA) (R&D Systems).

Statistical analyses.

Statistical analyses were performed using Prism software (version 6.0 for Mac). Four-group comparisons were performed with the Kruskal-Wallis test as previously described (42), followed by pairwise comparisons with the Mann-Whitney U test. Two-group comparisons were performed using the Mann-Whitney U test, and the Wilcoxon signed-rank test was used for comparisons within a group. The Spearman rank method was used to test for correlation. A P value of ≤0.05 was considered significant.

RESULTS

Patients.

Four groups of individuals were studied, with each group comprised of 35 to 55 individuals (Table 1). There were no significant differences in age or gender between the groups. As expected, there were significant differences in CD4⁺ T-cell counts between the HIV-negative and the three HIV-infected groups, as well as between the two chronically HIV-infected groups. The early viremic group also had a significantly higher HIV plasma viremia than did the chronic viremic group (Table 1). Both cross-sectional and longitudinal (before and after ART) analyses were performed.

Evaluation of frequencies of HIV-specific PBs in infected individuals.

Given previous observations that HIV viremia is associated with overexpression of PBs in the blood of infected individuals (9, 19, 21, 24, 43), especially those in early infection (26), we wished to determine the proportion of these PBs that were HIV specific. Accordingly, frequencies of B cells actively secreting IgG, IgA, or IgM, referred to as total Ig, were measured by ELISPOT assay using Ig class-specific biotinylated antibodies. Biotinylated HIV gp140 was used to detect secretion of HIV-specific antibodies, as previously reported (39). In early infected HIV-viremic individuals, the HIV-specific response was 1.3% of total antibody-secreting cells (ASCs), and significantly higher than the 0.5% for chronically infected HIV-viremic individuals (Fig. 1A). In addition, we evaluated the effect of reducing HIV viremia by ART on ASC frequencies in six individuals from each group. As shown in Fig. 1B, total Ig and HIV-specific ASCs, which ranged from 2,167 to 15,800 and 10 to 440, respectively, per 10⁶ B cells pre-ART, were significantly reduced in both groups post-ART. For total Ig ASCs, the reduction post-ART was, on average, 5.4-fold, whereas HIV-specific ASC frequencies were undetectable post-ART in 11 of the 12 individuals tested (Fig. 1B). Taken together, these data show that active HIV replication is associated with and possibly responsible for the presence of ASCs in the blood and that percentages of HIV-specific ASCs are highest in early viremia, although these frequencies are much lower than in other acute viral infections (34–37).

Fig 1.

Frequencies of total Ig and gp140-specific ASCs in the blood of HIV-infected individuals. Frequencies of ASCs were measured by ELISPOT assay on B cells isolated from the blood of early and chronically infected HIV-viremic individuals. (A) Percent gp140 ASC response, defined as percent gp140-specific of total Ig (IgG, IgA, and IgM) ASC frequency, was measured on peripheral blood B cells isolated from 6 early and 9 chronically infected HIV-viremic individuals. Horizontal bars represent medians. (B) Frequencies of total Ig and gp140-specific ASCs, as well as percent gp140-specific ASCs, were measured before and after ART for 6 HIV-viremic individuals each in early and chronic HIV infection. Duration on ART was 12 months, with a viral load below the limit of detection at the time of sample collection. (C) Flow cytometric profiles of HIV gp140 binding to B cells (CD20⁺ CD3⁻) (left graph) of a representative HIV-viremic individual are shown. The numbers in each box represent percent expression of IgG among gated B cells (middle graph) and percent gp140 binding among IgG⁺ B cells (right graph). (D and E) Correlation (D) and comparison (E) between percent gp140-specific ASCs and percent gp140 binding of IgG⁺ B cells measured by flow cytometry on B cells isolated from individuals in panel A.

The HIV gp140 used to evaluate HIV-specific ASC frequencies by ELISPOT assay can also be used to measure HIV-specific memory B-cell frequencies by flow cytometry, as described previously (39–41, 44) and shown in a representative profile (Fig. 1C). Of note, the flow cytometric approach excludes HIV-specific PBs because permeabilization of cells is required to detect most Igs being secreted by PBs and the gp140 probes used in the present study are not compatible with this approach. Costaining for IgG is also commonly included in the flow cytometry-based procedure because it helps identify true memory B cells by excluding cells that bind HIV envelope at low affinity through non-B-cell receptor (non-BCR) mechanisms (12, 45). When cells of HIV-infected individuals in the early and chronic viremic groups were used to perform both ELISPOT assay (to measure PBs) and flow cytometric analyses (to measure memory B cells), a strong direct correlation was observed between the frequencies of PBs and memory B cells that were specific for HIV (Fig. 1D). The highest frequencies of HIV-specific PBs and memory B cells were observed in the early infection group (Fig. 1A for PBs and P = 0.02 for memory B cells). In addition, the percentage of HIV-specific memory B cells, which ranged from 0.5 to 3.4% of IgG⁺ B cells, was higher than the corresponding ASC frequencies (Fig. 1E). It should be pointed out that while these two analyses are comparable, the memory response in reality is actually even higher than shown since only IgG-expressing cells are considered, whereas HIV-specific ASCs were not limited to cells secreting IgG. Nonetheless, these data demonstrate a strong correlation between the two cellular sources of antibodies in the humoral immunity to HIV infection. In addition, the similar range of frequencies between the HIV-specific PBs and memory B cells is in contrast to the much higher ratio of pathogen-specific PB at peak infection to memory B-cell frequencies seen in other viral infections (46, 47). These distinctions further underscore the notion that HIV-specific PB frequencies are comparatively low in HIV infection, consistent with previous findings (44), and strengthen the suggestion that the majority of PBs in the blood of HIV-viremic individuals arise from nonspecific effects of HIV replication.

Ig isotype distribution of PBs in the blood.

Given the relatively high frequency of non-HIV-specific (i.e., polyclonal) PBs in the blood of HIV-viremic individuals, we considered other PB properties that could help elucidate their origin. Accordingly, we determined the Ig isotype distribution of blood-derived PBs for the four groups described in Table 1. The PB immunophenotype was defined as CD19⁺ CD27⁺⁺ CD20⁻ CD3⁻ (where “++” indicates high-intensity expression) based on previous observations from our group and others (10, 48) and illustrated in a representative profile (Fig. 2A). In agreement with previous studies (29, 30), IgA was the predominant Ig isotype of PBs in the blood of HIV-negative and chronically infected HIV-aviremic individuals (Fig. 2B). In contrast, IgG was the more dominant Ig isotype expressed by PBs in chronically infected HIV-viremic individuals, whereas early infected HIV-viremic individuals had an intermediate profile (Fig. 2B). When the proportions of the three Ig isotypes were compared among the groups, the fraction of PBs expressing IgG was found to be significantly higher in chronically infected HIV-viremic than in HIV-negative, early infected HIV-viremic, and chronically infected HIV-aviremic individuals (Fig. 2C). This IgG PB fraction was also significantly higher in early infected HIV-viremic than in chronically infected HIV-aviremic individuals (Fig. 2C). Conversely, the fraction of PBs expressing IgA was significantly higher in HIV-negative and chronically infected HIV-aviremic individuals than in early and chronically infected HIV-viremic individuals, respectively (Fig. 2D). Finally, the fraction of PBs expressing IgM/D was significantly higher in early infected HIV-viremic than in HIV-negative and chronically infected HIV-viremic individuals (Fig. 2E). Furthermore, PB counts were directly correlated with HIV plasma viremia and modestly inversely correlated with CD4⁺ T-cell counts (Fig. 3), with the IgG PB isotype showing the strongest effects (data not shown).

Fig 2.

Frequencies of IgG⁺, IgA⁺, and IgM/D⁺ PBs in the blood of early infected HIV-viremic (EV; n = 24), chronically infected HIV-viremic (CV; n = 36), chronically infected HIV-aviremic (CAV; n = 30), and HIV-negative (HN; n = 25) individuals. (A) The percentage or fraction of PBs, identified as CD19⁺ CD20⁻ CD27⁺⁺ CD3⁻, expressing each Ig isotype was determined by flow cytometry. The profiles shown are of a representative chronically infected HIV-viremic individual. The percentages in the pie charts represent means for IgG, IgA, and IgM/D (B); the horizontal bars in graphs for IgG (C), IgA (D), and IgM/D (E) represent median values for each group.

Fig 3.

Correlations between peripheral blood PB counts and CD4⁺ T-cell count (left graph) and HIV plasma viremia (right graph). Data include samples from the four groups described in Table 1, with the exception that the HN group was not included in the right graph.

Longitudinal analyses were also performed on the two HIV-viremic groups before and after reduction of viremia by ART. Consistent with the cross-sectional findings, the fraction of PBs expressing IgG decreased significantly after ART in early infected (Fig. 4A) as well as chronically infected (Fig. 4B) HIV-viremic individuals. Conversely, the proportion of cells expressing IgA increased significantly in both groups after ART (Fig. 4). No difference was observed in the level of IgM/D-expressing cells pre- and post-ART in either group (Fig. 4). Taken together, the cross-sectional and longitudinal data indicate that IgG-expressing PBs become increasingly more prevalent during HIV viremia; furthermore, reduction of viremia by ART normalizes the Ig PB isotype distribution toward a predominantly IgA profile characteristic of uninfected individuals (29, 30).

Fig 4.

Longitudinal analysis of Ig isotype among PBs in the blood of HIV-infected individuals. PBs of eight individuals each from early (A) and chronically (B) HIV-infected groups were analyzed before and after initiation of ART for Ig isotype distribution. Individuals in the early group were receiving ART for an average of 12 months, and individuals in the chronic group were receiving ART for an average of 6 months.

Ig levels in the sera.

Increases in the frequency of PBs in the blood during viral infections may or may not be accompanied by hypergammaglobulinemia (37). In HIV infection, elevated PB and serum Ig levels have been reported (9, 15–23), although studies that have considered both parameters together and at different stages of HIV disease are lacking. Accordingly, we extended our observations on PBs in the blood to include quantitative and qualitative analyses of Igs in the sera of the four groups of individuals described in Table 1. As shown in Fig. 5A, serum IgG levels in the chronically infected HIV-viremic group were significantly higher than in all other groups. Furthermore, all HIV groups had serum IgG levels that were significantly higher than those of the HIV-negative individuals, and levels in the early infected HIV-viremic group were significantly higher than those of the chronically infected HIV-aviremic group (Fig. 5A). For IgA, significant differences were observed between the three HIV groups and the HIV-negative individuals (Fig. 5B). For IgM, levels in serum were significantly higher in early and chronically infected HIV-viremic individuals than in both HIV-negative and chronically infected HIV-aviremic individuals (Fig. 5C). Finally, analysis of IgG subclasses revealed that IgG1 was largely responsible for the differences observed in total IgG (Fig. 5D), with a lesser contribution from IgG3 (Fig. 5E), but no differences were observed between the groups for IgG2 and IgG4 (data not shown). These data demonstrate that hypergammaglobulinemia occurred at all stages of HIV infection and involved all Ig classes, although to different extents, with IgG being the most altered by HIV infection, especially in the chronically infected HIV-viremic individuals.

Fig 5.

Ig levels in the sera of early infected HIV-viremic (EV; n = 33), chronically infected HIV-viremic (CV; n = 33), chronically infected HIV-aviremic (CAV; n = 34), and HIV-negative (HN; n = 30) individuals. Horizontal bars represent median values for each group. Ig classes and subclasses reported include IgG (A), IgA (B), IgM (C), IgG1 (D), and IgG3 (E).

Serological analysis of cytokines, chemokines, and soluble receptors.

Given the indications that PBs in the blood of HIV-viremic individuals are likely to arise from indirect effects of viral replication, we extended our analyses to include many soluble serum factors that have been implicated in HIV-induced immune activation and especially B-cell hyperactivation and terminal differentiation (reviewed in references 12 and 49). A total of 21 cytokines, chemokines, and soluble factors were measured in the serum (see complete list in Materials and Methods); the five factors shown in Fig. 6 were those for which significant differences were observed between the four study groups described in Table 1. The patterns of significant difference were identical for TNFRII and ICAM: levels in serum were significantly elevated in the two viremic groups compared to the chronically infected HIV-aviremic and HIV-negative groups and also significantly elevated in the HIV-aviremic group compared to the HIV-negative group. Differences among groups for MIG and IL-10 were similar to those for TNFRII and ICAM except that there was no difference between HIV-aviremic and HIV-negative individuals. Finally, differences among groups for chemokine IP-10 were similar to those for MIG and IL-10 but also included a significantly higher level in the chronically infected than in the early infected HIV-viremic group, as shown in Fig. 5A for IgG. Thus, several soluble factors were elevated in the sera of HIV-infected individuals, especially HIV-viremic individuals, and with patterns that were similar to those observed for serum Ig.

Fig 6.

Levels of soluble factors TNFRII, MIG, ICAM, IL-10, and IP-10 in the sera of early infected HIV-viremic (EV; n = 33), chronically infected HIV-viremic (CV; n = 32), chronically infected HIV-aviremic (CAV; n = 34), and HIV-negative (HN; n = 29) individuals. Horizontal bars represent median values for each group.

Correlations between cellular phenotypes, HIV plasma viremia, and soluble serum factors.

Finally, we performed correlation analyses between the various cellular phenotypes and soluble serum factors discussed in the previous sections. As shown in Fig. 7A, for all groups combined, there were strong correlations between all serum Ig isotypes (IgG, IgA, and IgM) and the respective Ig isotype PB counts. Similar correlations were observed between serum Ig isotypes and corresponding Ig isotype PB percentages (data not shown). Regarding soluble factors in the serum that were significantly modulated in HIV infection (Fig. 6), strong correlations were observed between all five soluble factors and serum Igs, with the strongest correlation observed with IgG (Fig. 7B; data not shown for IgA and IgM). In addition, direct correlations were also observed between the soluble factors affected by HIV infection and levels of HIV plasma viremia, with strongest correlation observed with MIG (Fig. 7C). Taken together, these correlations indicate that PBs, serum Igs, and several soluble factors are interconnected and likely reflect the systemic activating effects of HIV viremia on the immune system.

Fig 7.

Correlations between serum Ig levels and peripheral blood PB counts for corresponding Ig isotype (A), levels of serum IgG and soluble factors (B), and HIV plasma viremia and soluble factors (C). Data in panels A and B include samples from the four groups described in Table 1. Data in panel C include samples from the three HIV groups described in Table 1.

DISCUSSION

Hypergammaglobulinemia and elevated frequencies of PBs in the blood have long been associated with HIV infection (18), and specifically with HIV viremia in more recent studies (9, 19, 21, 24, 43). However, the relationship between these two abnormalities and their underlying causes in HIV infection remain largely unknown. In the present study, we demonstrate that the majority of PBs in the blood of both early and chronically infected HIV-viremic individuals are not specific for the virus and likely arise from nonspecific immune-activating effects of HIV replication. Among PBs in the blood of HIV-viremic individuals, IgG was the more prevalent isotype secreted by these cells, in contrast to PBs of HIV-aviremic and HIV-negative individuals, in which IgA was the predominant Ig isotype. Furthermore, while all classes of Igs were elevated in the sera of both early and chronically infected HIV-viremic groups, IgG was most elevated in the chronic phase of viremia. Moreover, serum IgG correlated strongly with PB counts in the blood and soluble factors associated with immune activation. Taken together, these data demonstrate that hypergammaglobulinemia and elevated frequencies of PBs are associated with HIV viremia and suggest that these abnormalities are driven by systemic hyperactivating effects of the virus on the immune system.

Two features of PBs in the blood help identify the circumstances and mechanisms of their appearance: one is their antigen specificity, and the other is their Ig isotype. With regard to antigen specificity, accurate analyses rely on the capacity to measure a pathogen-specific response, which currently relies almost exclusively on ELISPOT-based analyses. Under certain circumstances, particularly with transient responses to initial infection, it is also critical to identify individuals during the acute phase of infection. Viral infections in this category include influenza virus, dengue virus, and RSV infections, in which PBs have been shown to undergo a rapid, transient, and mainly pathogen-specific burst in the blood of acutely infected individuals who then resolve their infection with a marked decline in the PBs in question (34–37). While acutely HIV-infected individuals are difficult to identify, our findings on early infected HIV-viremic individuals indicate that the proportion of PBs in the blood that are HIV specific is much lower than with the three viruses just mentioned. In early HIV infection, PBs comprise on average 13% of B cells that circulate in the blood (26), and among these PBs, less than 1.5% are specific for HIV (Fig. 1). In contrast, for dengue virus and RSV, PBs in the blood during acute infection account for as much as 30% of peripheral lymphocytes, and the majority of these are virus specific (37, 50). In acute influenza virus infection, the percentage of PBs in the blood is somewhat lower, reported at approximately 3% of B cells (36); however, the percentage of these PBs that are influenza virus specific is high, above 50% in most reported cases (36). One caveat to these analyses, especially in HIV infection, is that the antigen and assay used to measure virus-specific antibodies secreted by the PBs may capture only a fraction of the total virus-specific response. Furthermore, the antigen may not be compatible with the strain of virus, and the assay may lack sufficient sensitivity to detect such a response. For HIV, the majority of antibodies in early infection are directed against the envelope (51), and the strain used, YU2, is widely recognized by B cells of our study cohorts (data not shown). Furthermore, while the strong correlation found between HIV-specific PBs and memory B cells (Fig. 1D) is in itself novel and indicative that these two cellular sources of humoral immunity are linked, their relatively similar frequencies provide further evidence that the HIV-specific PB frequency is very low. In other viral infections, including infections with influenza and dengue viruses, the burst of virus-specific PBs that occurs during acute infection or even vaccination is orders of magnitude higher than HIV-specific memory B-cell frequencies that have been reported at any time during the course of HIV disease (34, 46, 47, 52).

It is unclear why HIV-specific PB frequencies are lower than the frequencies in other viral infections, but this finding may relate to the chronic and persistent nature of HIV viremia and its immune-activating effects. However, there are certain similarities between HIV and the other viruses that relate to kinetics of exposure and the Ig isotype of blood PBs. With regard to temporal relationships, the burst of PBs in the blood is surprisingly similar for most non-HIV viruses studied, occurring 6 to 7 days after infection or vaccination (34, 35, 37, 53, 54), although the duration that PBs remain present in the blood varies with the persistence of the pathogen (37, 46, 50). With regard to Ig isotype, PBs that circulate in the blood following parenteral immunizations are mainly IgG (29, 32, 34) and are likely transiting to and from draining lymph nodes and the spleen (55–57). In HIV infection, IgG PBs become increasingly prevalent in chronically viremic individuals, whereas the reduction of viremia by ART has an effect that is similar to that seen with resolved non-HIV viral infections, namely, a reduction of PBs in the blood and a normalization of the PB isotype from IgG toward IgA. Thus, the effect of chronic antigenic stimulation in persistently viremic HIV-infected individuals may represent a prolongation of the transient antigenic stimulation seen in the context of vaccination and likely those viral infections of short duration.

The relatively low percentage of HIV-specific PBs in the blood of HIV-viremic individuals argues for a strong nonspecific effect of HIV replication on B cells. HIV-induced immune hyperactivity has been suggested as a major cause of B-cell dysregulation, including aberrant terminal differentiation (12). In recent longitudinal studies where HIV-infected individuals were identified very soon after being infected, there was clear evidence that the rapid burst in HIV viremia is associated with a “storm” of proinflammatory cytokines, many of which are elevated transiently, followed by an increase in immunoregulatory cytokines (58). Among these cytokines are several known to induce B-cell differentiation, including proinflammatory cytokines IL-6 and TNF, as well as the immunoregulatory cytokine IL-10 (59). While not all these cytokines were significantly increased in our early HIV infection group, it is possible that some were missed because of the transient nature of the acute-phase cytokine storm and the fact that although our patients were in the “early” phase of HIV infection, they were likely initially infected weeks to months prior to study. Nonetheless, these factors associated with immune activation very likely contribute to the increased frequency of PBs observed in the blood of early HIV-infected individuals (19, 26, 60).

We identified several cytokines that were elevated in both early and chronically infected HIV-viremic individuals, including inflammatory soluble receptors TNFRII and ICAM and factors MIG and IP-10, as well as IL-10. Both soluble TNFRII and ICAM are strong indicators of immune activation and inflammation and have been shown to be increased in HIV-viremic individuals (61–65). Furthermore, MIG and IP-10 are among proinflammatory factors induced in the early phase of HIV viremia (58, 66). Of note, both IP-10 and IgG were elevated in early infected and particularly in chronically infected HIV-viremic individuals. The IP-10 gene is also among the genes induced by type I IFN in the setting of HIV infection (67), and we have shown that B cells of chronically infected HIV- viremic individuals upregulate several type I interferon-induced genes as well as genes associated with B-cell terminal differentiation (10). Finally, several of these soluble factors, including IP-10 (68), IL-10 (69), and ICAM (70), have been associated with B-cell activation and/or terminal differentiation. Thus, increases in several soluble factors may contribute to the IgG hypergammaglobulinemia and increased frequencies of IgG⁺ PBs in the blood of HIV-viremic individuals.

In conclusion, we provide new insight into the induction of B-cell terminal differentiation and hypergammaglobulinemia in the setting of HIV infection. Collectively, our observations suggest that both HIV-specific and nonspecific PBs arise early after HIV infection, the latter likely driven by inflammatory and B-cell differentiating factors that are sustained by the chronic antigenic stimulation reflecting uncontrolled HIV replication. Such dysregulated B-cell differentiation may contribute at least in part to the inefficiency of the humoral immune response, against both HIV and other pathogens, that afflicts patients with HIV disease.