|Home | About | Journals | Submit | Contact Us | Français|
The role of antibody and B cells in preventing infection is established. In contrast, the role of B cell responses in containing chronic infections remains poorly understood. IgG2a (IgG1 in humans) can prevent acute infections and T-bet promotes IgG2a isotype switching. However, whether IgG2a and B cell-expressed T-bet influence the host-pathogen balance during persisting infections is unclear. Here we demonstrate that B cell specific loss of T-bet prevents control of persisting viral infection. T-bet in B cells not only controlled IgG2a production, but also mucosal localization, proliferation, glycosylation, and a broad transcriptional program. T-bet controlled a broad antiviral program in addition to IgG2a since T-bet in B cells was important even in the presence of virus-specific IgG2a. Our data supports a model in which T-bet is a universal controller of antiviral immunity across multiple immune lineages.
Pathogens such as Human Immunodeficiency Virus (HIV), Hepatitis B Virus (HBV), HCV, mycobacterium Tuberculosis and malaria establish persistent infections despite ongoing T and B cell responses (1). These responses are often dysfunctional or exhausted, and unable to eradicate the pathogen. However, ongoing immune responses during chronic infections apply critical pressure on the pathogen to establish a host-pathogen equilibrium and, in some cases, contain the pathogen and limit or delay disease. Despite a clear role for T cells, it has become increasingly clear that humoral immunity is important during persistent viral infections(2). Indeed, despite the potent ability of HIV to evade antibodies through escape mutations, naturally occurring broadly neutralizing antibodies can develop in some individuals and these antibodies may help contain chronic infection(2). However, precisely how ongoing B cell and antibody responses are involved in the outcome of chronic viral infections remains poorly understood.
To investigate B cell responses during ongoing persistent viral infection, we used LCMV clone 13 infection (cl13). LCMV cl13 causes viremia that is partially contained over time, with viral replication limited to some tissues after 2-3 months. In contrast, acute infection with LCMV Armstrong (Arm) is rapidly cleared from all organs by d8-10 p.i. While Arm infection generates functional B and T cells, cl13 is associated with lymphocyte exhaustion(1, 3). Despite this exhaustion, T and B cells collaborate to hold cl13 replication in check long-term(1, 3). B cells are required for this viral containment since B cell KO mice or mice with an unrelated BCR transgene fail to control chronic infection(4, 5). However, the characteristics of antibody and B cells associated with viral containment during chronic infection remain poorly understood.
All animals were housed at, and used in accordance with protocols approved by the University of Pennsylvania Institutional Animal Care and Use Committee. Tbx21−flox, T-bet-GFP, CD19−cre, and pURF-Tg (i.e. IgG2a−/−) mice have been described (6-8). Tbx21flox/flox and Tbx21flox/wt mice were crossed to CD19−cre mice. CD19−creTbx21wt/wt mice were used as WT controls. Mice were infected with 2×105 PFU of Arm i.p. for acute infections or 4×106 PFU of cl13 i.v. for chronic infections. Viral titers were measured as described (6).
Live/Dead® Fixable Dead Cell Stain (Invitrogen) was used to identify live cells. Surface and intracellular staining and stimulation and cytokine staining was performed as described (6). Antibodies for flow cytometry were from eBioscience, Biolegend, BD Pharmingen, and Life Technologies. MHC tetramers were generated and used as described(6). Data were collected on a BD LSRII (BD Biosciences) and analysed with FlowJo (Tree Star, Ashland, OR). Cell sorting was done with a BD Aria II (BD Biosciences).
ELISAs were done using lysates of cl13 infected BHK cells. Serum ELISAs were done with serial dilutions starting at 1:100, while fecal ELISAs were done with serial dilutions of PBS:feces solution starting neat. Isotype detection antibodies were performed for IgM, IgG1, IgG2a, IgG2b, IgG3, and IgA using antibodies from Southern Biotechnologies. Relative absorption is reported as Log10 as values have been normalized to account for dilution factors.
B cells for microarray were FACS sorted into Trizol, and processed with Qiagen RNeasy kits. RNA QC was performed by spectrophotometry and gel analysis. B cell purity was >93% for each sample. Heat maps and gene lists were generated with the class neighbors function in GenePattern (FDR < 0.05). Gene set enrichment analysis (GSEA) was done via the Broad Institute (http://www.broadinstitute.org/gsea/index.jsp) using published gene sets for T-bet−/− CD8+ T cells(6). Microarray data has been deposited (GSE81189; http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE81189).
Student’s t-test (paired and unpaired), Mann-Whitney tests, and Log-rank tests were performed using Prism software (Graphpad, La Jolla, CA).
IgG2a (IgG1 in humans) is thought to be an effective antiviral isotype based on induction by acute infections and passive transfer where pathogen acquisition and/or acute infection is prevented (9, 10). However, it has remained unclear whether a causal relationship exists between IgG2a and containing ongoing chronic viral infections. The transcription factor T-bet has been implicated in IgG2a expression by B cells in non-infectious settings (11, 12). We, therefore, examined the role of T-bet in B cells and IgG2a in chronic LCMV infection. After cl13 or Arm infection a population of T-bet+ B cells was induced by d5 post infection (p.i.). This population persisted after cl13 infection through d30 p.i., while the number of T-bet+ B cells declined after Arm between d14 and d30 p.i. (Fig 1A). These T-bet+ B cells initially expressed markers of pre-GC B cells on d5 p.i. and later acquired GC and memory B cell phenotypes (data not shown), consistent with previous data on T-bet expression in B cells after protein immunization(12).
To examine the functional role of B cell expressed T-bet during chronic viral infection, we bred Tbx21flox/flox (called T-betflox/flox hereafter) mice to Cd19cre/+mice, to achieve B cell specific deletion of T-bet. The immune compartments of naïve T-betwt/wtCD19cre(WT), T-betflox/wtCD19cre (cHet), and T-betflox/floxCD19cre (cKO) mice had normal numbers of mature B cells, conventional T cells, and Treg cells (data not shown). A robust T-bet+IgMlo B cell response was observed in WT mice after cl13 infection, but this response was almost completely absent in cKO and substantially reduced in cHet mice (Fig 1B). To test whether the absence of T-bet+ B cells had an impact on the outcome of chronic LCMV infection we infected WT, cHet, and cKO mice with cl13. Serum viral titers rose to 104-105 pfu/ml by d7 p.i. in all groups. While WT mice contained viremia below detection by ~d45, cKO mice remained viremic until at least 8 weeks with viral titers remaining above 104 pfu/ml. cHet mice displayed an intermediate phenotype with delayed control of viremia compared to WT mice (Fig 1C). A similar effect on viral replication in tissues was also observed (Fig 1C). Development of neutralizing antibody titers are delayed after cl13 infection, non-neutralizing antibodies may be important(9, 13, 14) (data not shown) defective FcR function impairs viral control during cl13 infection(15, 16). Thus, non-neutralizing functions of T-bet dependent antibody or B cell responses may play a role in chronic LCMV infection. In contrast to the importance of B cell-expressed T-bet in controlling cl13 infection, the resolution of acute Arm infection was unaffected by the absence of T-bet in B cells (data not shown).
High titers of LCMV-specific IgG2a were present by 3 weeks p.i. in WT mice. In contrast, virus-specific IgG2a was nearly absent in cKO mice (Fig 1D). Early anti-viral IgM has been implicated in control of peak viremia during chronic LCMV infection(17). However, LCMV specific IgM at d7 p.i. was not different in T-bet WT and cKO mice (Fig 1E). In addition, the amount of LCMV-specific serum IgG1, IgG2b, and IgG3 was not affected in the absence of T-bet in B cells, while serum IgA was reduced (Fig 1F). This antibody production defect was restricted to IgG2a, as total amount of serum IgG was unaffected by the absence of T-bet (Fig 1F). Interestingly, deletion of only one copy of T-bet in cHet mice had little impact on IgG2a (Fig 1D). This lack of impact on IgG2a levels in cHets is in contrast to the delay in viral control in these mice (Fig 1C) and suggests that the effect of T-bet in B cells may be broader than only IgG2a class switching.
To directly test the importance of antibody in the T-bet KO phenotype, we performed two additional experiments. First, we examined viral control in mice with a specific deletion in IgG2a(8). Consistent with the data above, IgG2a KO mice were unable to contain cl13 infection, supporting a key role for IgG2a (Fig 1G). Next, we tested whether serum from WT mice containing LCMV-specific antibody including IgG2a could rescue the phenotype of T-bet cKO mice. Serum transfer only partially restored viral control in cKO mice (Fig 1H). These observations are consistent with our data that cHet mice have delayed viral control despite normal IgG2a levels, and suggest that T-bet controls aspects of B cell biology beyond IgG2a production.
Before the divergence in viral load (d7-d21 p.i), virus-specific CD8+ and CD4+ T cell responses were similar between WT, cHet, and cKO mice (Supplemental Fig 1). Moreover, T-bet+CD4+ T cells and the T-bethiPD1int and T-betintPD1hi subsets of CD8+ T cells(6) developed normally until later in infection when major differences in viral load prevented comparison(data not shown). Normal anti-viral CD8 and CD4 T cell responses were observed after Arm infection in cHet and cKO compared to WT mice (Supplemental Fig 1). Although a deficiency in IgG2a+ B cells and production of serum IgG2a was also observed following Arm infection (Fig 2A,B), this had no detectable impact on clearance of acute infection. Thus, B cell-expressed T-bet plays a critical role in the control of chronic, but not acute LCMV infection. Together, these data provoke the hypothesis that anti-viral IgG2a is necessary, but not sufficient, to contain persistent viral infection and that IgG2a class switching is one of multiple mechanisms that rely on B cell-expressed T-bet.
We next examined B cell differentiation in WT, cHet, and cKO mice after cl13 infection. GC B cells were generated after infection in cHet and cKO mice, as were memory B cells with no obvious alterations in GC or memory B cell numbers or frequency at time points before viral loads diverged (Fig 3A). Despite an increase in GC B cells at d64 p.i. in cHet and cKO mice when viral load was higher in these mice (Fig 3B), the frequencies of IgG2a+ cells were substantially reduced in a gene dose-dependent manner, while IgG1+ memory B cells were present in normal frequencies (Fig 3C). At d8 p.i. with cl13 there was no difference in the frequency or total number of LCMV-specific CD4+ or CD8+ T cells in the spleens of WT or cKO mice (Supplemental Fig 1 and data not shown). Furthermore, T cell production of IFNγ, IL2, and TNFα was largely unchanged between WT and T-bet cKO mice (Supplemental Fig 1 and data not shown). Finally, TFH cells were formed normally in the absence of T-bet in B cells (Supplemental Fig 1). These data suggest that the effects seen in cKO mice were not due to alterations in the T cell compartment. As initial GC reactions and B cell differentiation occurred normally in T-bet cKO mice after infection, we investigated other aspects of B cell biology that might contribute to impaired viral immunity.
Our data in cHet mice and serum transfer experiments suggested that the role of B cell-expressed T-bet may extend beyond IgG2a (Fig 1). To further interrogate the role of T-bet in anti-viral B cell responses, we examined genome-wide transcriptional profiles of naïve B cells, T-bet+, and T-bet− memory phenotype B cells at d10 p.i. with cl13 prior to the divergence of viral loads (Fig 1C) using T-bet-GFP reporter mice(7). Examination of key lymphocyte lineage genes confirmed the isolation of B cells and absence of other lineage markers in purified T-bet+ and T-bet− memory-phenotype B cells (Supplemental Fig 1). T-bet+ and T-bet− memory-phenotype B cells differentially expressed genes involved in cell migration (e.g.Cxcr3), differentiation and costimulation (e.g. Cd80, Ctla4, Fas), antibody diversification and mutation (Aicda), proliferation, gamma chain cytokine receptors (Il2ra, Il7r) and the glycosyltransferases B4Galt1, Fut8, and Fut11 (Fig 4A). In particular these enzymes may be important given their role in antibody glycosylation and function(18). These data suggest that, in B cells, T-bet can control multiple key immune functions, including anatomical localization and expression of glycosylation enzymes, properties that likely influence the effectiveness of B cell and antibody responses in vivo.
Gene set enrichment analysis revealed enrichment of the gene expression signature of T-bet+ B cells in WT compared to T-bet−/− CD8+ T cells responding to Arm infection (Fig 4B), suggesting that at least some of the transcriptional program regulated by T-bet is conserved between these two lymphocyte lineages. The genes that drove this enrichment included transcription factors, cell surface receptors, adhesion molecules, and cell cycle regulators (Supplemental Table 1). T-bet controls antiviral gene programs in CD4 and CD8 T cells, NK cells and group 1 ILCs(19). The identification of a T-bet-dependent antiviral program in B cells now suggests that T-bet controls a conserved lymphocyte viral immunity module, though there are likely lineage specific targets of T-bet in each case.
In agreement with the transcriptional profiles, T-bet+ memory phenotype B cells expressed high CXCR3, CD80, and Fas protein compared to T-bet− memory phenotype B cells (Fig 4A, C). Furthermore, the antibody clones GL7 and 1B11 (CD43), that bind specific glycosylation changes, stained T-bet+ B cells more brightly than naïve and T-bet− B cells, suggesting that glycosylation enzymes such as B4Galt1, Fut8, and Fut11 are indeed differentially regulated by T-bet (Fig 4A, C).
To better understand the potential role of T-bet-dependent genes in cell migration, we examined anatomical location of activated B cells following infection. T-bet+ memory phenotype B cells were highly enriched in the small intestinal mucosa, a peripheral site of inflammation and infection during cl13 infection(20), consistent with the up-regulation of CXCR3 in T-bet+ B cells (Fig 4A, C, D). Intraepithelial lymphocytes were especially enriched in T-bet+ memory phenotype B cells, with greater than 65% of B cells in this compartment exhibiting a T-bet+IgM− phenotype (Fig 4D). Consistent with a role for T-bet in regulating the quality and type of antibody producing cells in the intestinal mucosa, we observed an increase in LCMV-specific IgA in the feces of cKO mice compared to WT mice (Supplemental Fig 1). We hypothesized that B cell numbers in the gut would be reduced in T-bet cKO mice. Surprisingly, the total number of memory B cells at d10 p.i. was equivalent in the intestinal mucosa between WT and cKO mice (Fig 4E top). We, therefore stained these B cells for T-bet. As shown above, the frequency of T-bet+ memory B cells was greatly reduced in the spleens of cKO mice. However, this difference was partially mitigated in the peyer’s patches, and was completely abrogated in the LPL and IEL of cKO mice. Nearly all B cells present at these mucosal sites were T-bet+ escapees in the cKO mice (Fig 4E bottom). These data suggested selective pressure for T-bet in migration to the intestine. These data suggest that B cell intrinsic T-bet may regulate an important balance between systemic and mucosal antibody responses.
Together, our data suggest that T-bet+ B cells may be an important component of ongoing immune resposnes during chronic viral infections even when neutralizing Ab is not a major factor. These data point to possible therapeutic opportunities to treat and/or prevent chronic infections through modulation of T-bet in B cells, further emphasizing the importance of understanding the in vivo mechanisms that drive T-bet+ B cell responses. It will be interesting to test whether specific adjuvants or vaccine modalities enhance B cell expressed T-bet and whether such approaches improve the isotype profile of antibody responses, change the functional quality of these antibodies and/or increase the rate of somatic hypermutation. Further, these studies support a model where T-bet is a central regulator of antiviral immunity in all lymphocyte lineages. Future studies interrogating the transcriptional mechanisms of T-bet control of antiviral B cell responses should shed light on the extent of the core, conserved T-bet antiviral module.