Germinal center hypoxia potentiates immunoglobulin class switch recombination

Abstract

Germinal centers (GCs) are anatomic sites where B cells undergo secondary diversification to produce high affinity, class switched antibodies. We hypothesized that proliferating B cells in GCs create a hypoxic microenvironment that governs their further differentiation. Using molecular markers, we found GCs to be predominantly hypoxic. Compared to normoxia (21% O₂), hypoxic culture conditions (1% O₂) in vitro accelerated class switching and plasma cell formation and enhanced expression of GL-7 on B and CD4⁺ T cells. Reversal of GC hypoxia in vivo by breathing 60% O₂ during immunization resulted in reduced frequencies of GC B cells, T follicular helper (T_FH) cells and plasmacytes, as well as lower expression of ICOS on T_FH. Importantly, this reversal of GC hypoxia decreased antigen-specific serum IgG1 and reduced the frequency of IgG1⁺ B cells within the antigen specific GC. Taken together, these observations reveal a critical role for hypoxia in GC B cell differentiation.

Introduction

Development of effective vaccination strategies requires a detailed understanding of the mechanisms that govern adaptive immunity. The germinal center (GC) is the anatomical site in which antigen-activated B- and T lymphocytes interact, initiating immunoglobulin class switch recombination (CSR), somatic hypermutation, and the affinity maturation associated with effective antibody responses (1). However, surprisingly little is known about the physiological mechanisms within the GC microenvironment that regulate these processes.

It has become evident that local oxygen tension and the physiological response to hypoxia play roles in regulating inflammation (2) through synergistic and independent mechanisms including, but not limited to, hypoxia inducible factors (e.g. HIF-1α, HIF-2α and HIF-3α), NF-κB, mammalian target of rapamycin kinase (mTOR), and the unfolded protein response (UPR) (2-4). Moreover, the cellular response to tissue hypoxia often results in metabolites in the extracellular space, which have diverse signaling capacities that can facilitate both pro and anti-inflammatory responses (2, 5). Indications that hypoxia may be important in B cell physiology originally came from experiments that revealed HIF-1α is required for B cell development and prevention of autoimmunity (6). Furthermore, HIF-1α has also been detected in human tonsillar GCs (7).

We hypothesized that rapidly proliferating B cells within GCs develop a hypoxic microenvironment that promotes CSR. We show that GCs contain hypoxic regions linked to accelerated class switching, plasma cell development and antibody secretion. Following vaccination, administration of clinically relevant (8) respiratory hyperoxia (60% O₂) via supplemental oxygen dramatically suppressed the GC response and subsequent antibody production, revealing a previously unappreciated functional role of hypoxia within the GC microenvironment.

Materials and Methods

Animal studies

10-12wk female C57BL/6J mice were from Jackson Laboratories (Bar Harbor ME). Animal work was in accordance with IACUC at Northeastern University. Mice were immunized i.p. with Alum hydroxide (5mg/mouse) / NP(6)-OVA (1μg/mouse). 10% Alum sulfate (Sigma) in PBS was mixed 1:1 with diluted stock of NP(6)-OVA or NP(11)-CGG (Biosearch Technologies / produced in-house) and precipitated by 1M KOH, washed 3x, and injected in 200μl/PBS. For hyperoxia experiments, mice cages were placed in custom made chambers hooked up to an oxygen concentrator (AirSep). Statistics were calculated using two tailed students t-test in Microsoft excel.

Flow cytometry

Cells were FC blocked (10m/4°C) and stained (20m/4°C), washed 2x and fixed (FoxP3 kit, eBioscience). Hypoxyprobe compound was injected into mice i.v.(100mg/kg), circulated for 90 minutes, and detected with Hypoxyprobe mAB (4.3.11.3 (FITC)) was added for 1h/4°C after fixation. FACS buffer was 1xPBS/5%FCS. mABs (eBioscience, Becton Dickenson, Biolegend) were: IgG1 (A85-1) B220 (RA3-6B2), CD4 (RM4-5)(GK1.5), CD43 (S7), CD138 (281-2), CD16/32 (2.4G2), GL-7 (GL7), CD38(90), CD86 (GL-1), CD40 (1C10), CD73 (Ty/11.8), IgM(II/41), CXCR5 (SPRCL5), PD-1 (29F.1A12), ICOS (7E.17G9)(C398.4A), BCL-6(GI191E), FAS(Jo2), Zombie Green, FoxP3 (FJK16S), NP(36)-PE (Biosearch Technologies). Acquisition was on a Cytek DxP8 FACSCalibur, and analyzed on Flowjo X (Treestar).

In Vitro Assays

Napco 7000 incubators (37°C/5% CO₂) were used. Freshly prepared media consisted of IMDM, 25mM HEPES, 10% FCS (defined high grade (GE Healthcare)), glutamine, 55mM 2-ME, and 100U/ml Pen/Strep. Media was equilibrated in incubators for at least 6h before experiments. Cells were cultured in a 24 well plate (Costar #3474) in 1mL. Splenocytes were Ficoll separated (GE Healthcare) and stained with CD43 FITC (CSR Assay), CD8/GL-7 FITC (CD4 Assay), or CD43/GL-7 FITC (B cell assay). Labeled cells were depleted using αFITC Microbeads (Miltenyi Biotec) and AutoMACS. For CSR assay αCD40 mAb (clone 1C10, 2.5μg/ml, Biolegend) and rmIL-4 (10ng/ml, R&D Systems) were used and seeded at 0.1× 10⁶ cells/ml. For CD4 T cell stimulation cells were stimulated with αCD3 mAb (clone 2c11, 1μg/ml BD Biosciences) and seeded at 0.5× 10⁶ cells/well. For B cell assay cells were stimulated with αCD40 mAb (1C10, 2.5μg/ml Biolegend) and αIgM ab (6μg/ml, Jackson Immunoresearch), and seeded at 0.5× 10⁶ cells/ml.

Visualization of Vessels and Histology

Immunized mice were injected i.v. with 150μg Dylight 594 conjugated Tomato Lectin (Vector Labs). After 10-30m, the mice were euthanized and perfused with 30ml of 1% PFA followed by 30ml of PBS. Spleens frozen in OCT compound were cryo-sectioned (5μm), air dried and fixed in 1:1 mix of acetone/methanol at -20°C/10m. Slides were warmed to RT, Pap Pen applied and re-hydrated (0.1% Tween 20, 0.5% BSA in 1xPBS) for 20m, Fc blocked for 20m, primary stained for 3h, washed 3x and stained with αFITC Oregon Green (Invitrogen) for 1h, washed 3x and mounted. Images were acquired on a Nikon e80i microscope or a Zeiss LSM 710. ImageJ FIJI was used for analysis. Mask files were created for GCs (GL7⁺), follicles (B220⁺), and vasculature (tomato lectin⁺). Vasculature mask was expanded radially to 40μm and percent GC or follicle area covered was determined.

ELISA and ELISPOTS

For ELISA, 96 well plates (Costar 2595) were coated with 2μg/ml NIP 25- BSA in carbonate buffer for 1h/RT, washed 3x (0.5% BSA, 0.1%Tween 20 in PBS), blocked for 1h (0.5%BSA in PBS), washed 3x, and samples added for 1h. Plates were washed 3x and HRP-conjugated detection Abs for IgG1 and IgM (Bethyl Labs) were added for 1h, washed 3x and TMB substrate was added (BD Biosciences). Reaction was stopped using 2N H₂SO₄ and read on a Varioskan (Thermo Electron) at 450nm. Mid points of dilution curves were used and values calculated relative to standards (B-18 (IgM) and H33lγ1 (IgG1)). For ELISPOTS, Immobilon-P plates (Millipore) were coated overnight at 4°C with 2μg/ml NIP-25 BSA in carbonate buffer (pH 9.5). Plates were washed 5x (0.5%BSA/0.1% Tween 20/PBS), blocked in same buffer for 1h. After wash (5x) complete RPMI media was used to wash (2x) and cells were serially diluted, added and sat for 3h in incubator. Plate was washed with DI water and left in wash buffer overnight at 4°C. AP-conjugated antibodies to IgG1 or IgM (Southern Biotech) were added. Plates were washed (5x) with was buffer and with DI water (1x) and developed using BCIP/NBT reagent (Sigma Aldrich).

RT-PCR for IgG1 circle transcripts

RNA isolation was done using RNeasy mini kit (Qiagen) and Superscript III (Invitrogen) were used. For RT-PCR (done on Applied Biosciences 7300) initial step was 95C 9m followed by 40 cycles of PCR (94C 30s, 58C 1min) by using RT2 SYBR green PCR master mix (Life Technologies). Primers (Eurofins) were IgM reverse (CμR, 5′-AATGGTGCTGGGCAGGAAGT-3′) and IgG1 forward (Iγ1F, 5′-GGCCCTTCCAGATCTTTG AG-3′) (15). L32 (Qiagen) was used to normalize expression.

Results

The germinal center is hypoxic and poorly vascularized

To ascertain whether the GC microenvironment is hypoxic, we immunized female C57BL/6J mice with the hapten-protein antigen, (nitrophenyl) acetyl-ovalbumin (NP-OVA) in aluminum hydroxide (alum) adjuvant. Identification of areas of tissue hypoxia or cells in hypoxic areas was determined by immunohistology using a well-documented molecular marker of tissue hypoxia in vivo, Hypoxyprobe (pimonidazole) (9). Injected intravenously, Hypoxyprobe creates thiol adducts with proteins under hypoxic conditions (<1.3% O₂), which are then specifically recognized by a monoclonal antibody. To assess GC hypoxia in situ, we studied Hypoxyprobe labeling of splenic GCs (Fig.1A). These histologic studies indicated that GCs are enriched for hypoxic areas (Fig.1A).

Figure 1.

The germinal center (GC) microenvironment is hypoxic and poorly vascularized. A) Tissue histology of hypoxic marker Hypoxyprobe of splenic GCs. Magnification is 20x. B) Flow cytometric analysis of splenic GCs for Hypoxyprobe staining. Red histogram is GC gate, grey histogram is non GC B cell gate. Gated on scatter, B220⁺ CD4^-. C) Quantification of GC B cells depicted in B. D) Tissue histology of splenic GCs of mice perfused with fluorescent tomato lectin to stain vasculature. E) Schematic of masking strategy for determining percent GC and follicle area within 40μm of lectin staining. F) Quantification of GC area that is within 40μm of perfused vessels. Each dot represents one GC or B cell follicle. Each bar represents an individual mouse. Bars placed at mean. Data is of splenic GCs 8 days following i.p. immunization with either NP-OVA/Alum or NP-CGG/Alum. Representative of at least two independent experiment. Four to ten mice per group.

We subsequently confirmed Hypoxyprobe labeling of GC B cells by multi-color flow cytometric analysis of GC B cells from spleens (Fig. 1B, C) (Supplemental Fig. 1D), Peyer's Patches (Supplemental Figs. 1A-B) and mesenteric lymph nodes (Supplemental Fig. 1C) of immunized and naïve mice. To a lesser extent, GC associated T_FH cells (CD4⁺CXCR5⁺Bcl-6⁺GL-7⁺) showed increased Hypoxyprobe staining compared to respective controls (Supplemental Figs. 1E-F), indicating hypoxia. In support of this demonstration of GC B- and T-cell hypoxia, we injected fluorescently-labeled tomato lectin that binds to blood vessel walls and then used fluorescence microscopy to measure the relative distances between GC B cells and the splenic blood vasculature (Figs. 1D-E). Due to the fact that earlier studies have shown that pO₂ can drop to hypoxic levels 30-40 μm from blood vessels (10), this distance was used to evaluate the likelihood of a hypoxic environment at GC sites. We found that 8 days after NP-OVA immunization, the substantial majority of GC area lies ≥40 μm from the nearest blood vessel (Fig. 1F). Although the majority of GCs were located ≥40 μm from the nearest blood vessel, occasional GCs were adjacent to or even surrounded tomato lectin stained vessels (data not shown). However, in the three dimensional space of lymphoid tissue, GCs are preferentially sited in hypoxic regions (Fig. 1A-F).

Hypoxic culture conditions promote Gl-7 expression

To assess the functional role of GC hypoxia, we used in vitro experiments with defined gas mixtures. Briefly, resting B cells (CD43^-GL7^-) were isolated by magnetic depletion and stimulated in cultures containing anti-CD40 and anti-IgM antibodies in normoxic (21%O₂) or hypoxic (1% O₂) incubators. We found that the GL-7 activation marker was upregulated on a greater fraction of hypoxic B cells and to higher levels than normoxic controls (Fig. 2A-C). Although GL-7 expression can be upregulated on B- and T cells in vitro (11), it is widely used to identify GC B cells in vivo [1]. The GL-7 carbohydrate epitope may be influenced by oxygen levels directly, as it depends on the repression of CMP-Neu5Ac hydroxylase (12). Since GL-7 was recently shown to mark a subset of GC associated T_FH (13), we assessed the effect of hypoxia on GL-7 up-regulation on CD4⁺ T cells in vitro. Upon stimulation with anti-CD3 antibody, we observed that hypoxic culture conditions increased the frequency of GL-7 expressing CD4⁺ T cells and also increased the expression level of GL-7 in CD4⁺ T cells when compared to normoxic controls (Supplemental Figs. 2A-C).

Figure 2.

Hypoxic culture conditions increase GL-7 Expression on B cells. A) Representative flow cytometry plots of purified resting B cells stimulated with αIgM and αCD40 in hypoxic or normoxic culture conditions. Gated on lymphocyte, singlet, live, B220⁺, CD4^-. B) Quantification of GL-7⁺ cells in A C) Relative expression of GL-7 on positive cells in A. Bars placed at geometric mean of quadruplicate wells from representative experiment. Representative of at least two independent experiments. *=p<0.01

Hypoxic culture conditions promote CSR and plasmacyte differentiation

To identify any additional consequences of hypoxia and increased the expression of GL-7 by B cells activated in hypoxic cultures, we determined the frequency of cells exhibiting IgM⁺→IgG1⁺ class switch recombination after stimulating resting B cells with anti-CD40 antibody and IL-4 (14). We then placed matched cohorts under normoxic (21%O₂) or hypoxic (1% O₂) conditions for four days to determine the numbers of class switched IgG1⁺ B cells by flow cytometry. Interestingly, hypoxic culture conditions accelerated IgM⁺→IgG1⁺ class switch kinetics on day 3, while the frequency of IgG1⁺ B cells in hypoxic chambers failed to increase on day 4 (Figs. 3A-B). Consistent with the accelerated presence of IgG1⁺ B cells in hypoxic cultures on day 3 was a decrease in IgM⁺ B cells (Supplemental Fig. 3B). Furthermore, by CFSE labeling B cells at the start of culture, we assessed if IgG1⁺ B cells in hypoxic chambers proliferated comparably to normoxic controls. We found that the early increase in IgG1⁺ B cells in hypoxic cultures does not represent biased proliferation (Supplemental Figs. 3D), but rather increased rates of class switch recombination (IgM to IgG1) determined by the increase in IgM excision circles (Fig. 3C), which are hallmarks of CSR (15). Interestingly, the rate of B cell division in hypoxic chambers was accelerated on day 3 but slowed significantly and specifically in the IgG1⁺ compartment on day 4 (Supplemental Figs. 3C-D). Further analysis revealed that on day 4, the IgG1⁺ B cell compartment in hypoxia had significantly increased frequencies of apoptotic cells as determined by active caspase 3 staining, while apoptosis in the plasma cell compartment was unaffected (Supplemental Fig. 3E).This differential distribution of frequencies of apoptotic cells may account for the fact that we did not observe major differences in viable cells in hypoxic cultures (Supplemental Fig. 3A).

Figure 3.

Hypoxia accelerates Ig class switch recombination and plasma cell formation in vitro. A-B) Flow cytometric analysis and kinetics of B cell stimulated in normoxic or hypoxic incubators to undergo class switch recombination with αCD40 and IL-4. Gated on lymphocyte, live, singlet, CD138^-.C) RT-PCR analysis of looped-out circular IgG1 transcripts of B cells stimulated as in A. D, E) Representative flow cytometric plots and analysis and kinetics of plasma cell formation of cells stimulated as in A. Cells gated on lymphocyte, live, singlet. F,G) IgM and IgG1 antibody secreting cell (ASC) formation on day 4 of hypoxic or normoxic cultures stimulated as in A. *=p<0.05, **p<0.01. Representative of two to ten independent experiments. Samples were run in quadruplicate.

Plasma cell differentiation, as determined by the appearance of CD138⁺B220^- cells in hypoxic and normoxic cultures, was dramatically increased under conditions of low oxygen tension (Figs. 3D-E)(Supplemental Fig. 3F), suggesting that hypoxia accelerates both class switch recombination and differentiation to plasmacytes. This interpretation is supported by the observation that on day 4, hypoxic cultures had an increased frequency of IgG1⁺ antibody secreting cells (ASC) but not IgM⁺ ASCs as determined by ELISPOT (Figs. 3F-G). Taken together, the accelerated plasma cell differentiation and ∼50% increase in total IgG1⁺ ASC in hypoxic cultures may partially account for reduced expansion of IgG1⁺ B cells in these cultures on day 4 (Fig. 3B).

Reversal of tissue hypoxia by treatment with respiratory hyperoxia during immunization suppresses CSR and the GC reaction

If hypoxia promotes B cell differentiation during GC responses, we predicted that reversing tissue hypoxia during immunization would suppress the GC reaction. To test this prediction, we systemically reduced tissue hypoxia, as done clinically, by placing mice in chambers containing 60% O₂ to induce respiratory hyperoxia (8). Female C57BL/6J mice immunized with NP-OVA or NP-chicken γ-globulin (NP-CGG) in alum and held under conditions of respiratory hyperoxia exhibited dramatic suppressions of total and NP-specific GC responses 8 days after immunization (Fig. 4A and Supplemental Fig. 4 A-C). Similarly, but to a lesser extent, T_FH and GC associated T_FH, defined as CD4⁺B220^-CXCR5⁺PD1⁺GL7^- and CD4⁺B220^-CXCR5⁺PD1⁺GL7⁺, respectively, were reduced in frequency by hyperoxia (Fig. 4B and Supplemental Fig. 4E). Moreover, administration of respiratory hyperoxia in mice resulted in lower ICOS expression by both T_FH and GC associated T_FH (Fig. 4C and Supplemental Fig. 4D).

Figure 4.

Respiratory hyperoxia suppresses the germinal center (GC) reaction during immunization. A) GC frequencies of mice 8 days following immunization breathing either normoxic (21%) or hyperoxic air (60%) from day 0 to 8 following immunization. Gated on lymphocyte, singlet, live, CD4^-, B220⁺. B) Flow cytometric plots and frequencies of T follicular helper (T_FH) cells of immunized mice 8 days following immunization and treated as in A. Gated on lymphocytes, singlet, live, B220^-, FoxP3^-, CD4⁺. C) Effect of breathing 60% O₂ on relative ICOS expression of T_FH. D) Representative IgM and IgG1 antibody secreting cell frequencies determined by ELISPOT. E) Effect of breathing 60% O₂ on plasma cell frequencies, gated on lymphocytes, singlet, live, B220^low. F) Effect of breathing 60% O₂ on serum antigen specific IgM. G) Effect of breathing 60% O₂ on serum IgG1. H-I) NP-binding GC IgM and IgG1 compartments and effect of breathing 60% O₂. Gated on lymphocytes, singlet, live, CD4^-, B220⁺, GL7⁺, CD38^-. J) Histological images of IgG1 within GCs on day 8 of mice treated as in A. Bars placed at geometric mean. Representative plots of two independent experiments where mice were immunized with either NP-OVA/Alum or NP-CGG/Alum and analyzed 8 days following immunization. Each dot represents one mouse. Representative of at least two independent experiments. Five to ten mice per group.

Functional impairment of the GC response by respiratory hyperoxia was also manifested by fewer IgM⁺ and IgG1⁺ (nitroiodophenyl)acetyl-bovine serum albumin (NIP-BSA) specific ASCs (Figure 4D). To ascertain if plasma cells contributed to the reduced frequencies of ASCs in hyperoxic treated mice, we identified plasma cells using flow cytometry (CD138⁺B220^-) and found that hyperoxia significantly suppressed plasma cell formation (Fig. 4E). This was in excellent correlation with our in vitro finding that hypoxic cultures accelerated plasmacytic differentiation (Figs. 3D-E). Functionally, serum concentrations of NIP-BSA-specific IgM and IgG1 antibody were significantly reduced in mice treated with respiratory hyperoxia (Figs. 4F-G). Moreover, among NP-specific GC B cells, hyperoxia significantly reduced the frequency of class switched IgG1⁺ GC B cells whereas the proportion of IgM⁺ GC B cells was unchanged (Figs. 4H-I). We subsequently confirmed the reduction in IgG1⁺ GC B cells by respiratory hyperoxia through histology (Fig. 4J). Taken together, the reduction in IgG1⁺ NP-binding GC B cells and diminished serum NIP-specific IgG1 (Figs. 4G-I) suggests that respiratory hyperoxia likely has a more robust effect on Ig class switching than on initial, extrafollicular interactions between activated T and B cells.

Discussion

In summary, we have described a previously unappreciated role for hypoxia in GCs that acts on both B- and T cells and promotes class switch recombination as well as plasmacyte differentiation. These data show that administration of clinically relevant (8) 60% O₂ dramatically suppresses the GC reaction and production of IgG1⁺ antibody (Figs. 4A-J).

Our observations of the GC microenvironment offer new avenues of investigation into the role of hypoxia and hypoxia-induced immune-regulatory pathways during vaccination. It would be interesting to determine if any HIF proteins are responsible for the effects of hypoxia in promoting CSR and the GC response. Future studies will be needed to asses if hypoxia plays a role in directly regulating activation induced cytidine deaminase (AID) at the transcriptional and/or post transcriptional level, as AID is known to be regulated through phosphorylation (16, 17) at serine 38 and tyrosine 140 which affects CSR (18, 19).

Given our findings that class switched B cells cultured in hypoxic conditions appear to be more likely to undergo apoptosis, it would be interesting to determine if hypoxia within the GC promotes apoptosis and helps to “set the stage” within a GC for a competitive microenvironment to facilitate clonal competition and affinity maturation. This hypothesis is consistent with observations that both transgenic mice overexpressing the anti-apoptotic protein bcl-_XL and mice that specifically lacking caspase 8 within GC B cells both exhibit delayed affinity maturation (20, 21).

Our study also raises the question of whether hypoxia plays a role in regulation of somatic hypermutation (SHM) of variable region genes of GC B cells. We speculate that the hypoxic microenvironment within the GC may be an evolutionarily conserved regulatory mechanism to both promote as well as constrain potentially detrimental DNA-damaging genetic events such as CSR and SHM to the anatomical site of the GC. While there are documented examples of CSR and SHM outside of the GC (22-24), these examples come from either multi-valent T independent antigens such as NP-Ficoll which induce very strong BCR crosslinking or in disease states such as in lupus prone mice. While such examples exist, the GC is generally considered to be the “professional” site of CSR and SHM, and perhaps the hypoxic microenvironment within the GC helps promote this. In line with this thinking is the curious observation that while human GCs can be somewhat larger than murine GCs (25), many are the same size (26, 27)(and personal communication with G. Kelsoe), lie toward the center of the follicle (28), have been shown to express HIF-1α and are poorly vascularized (7). The development of hypoxia within the GC also raises the question if this hypoxic microenvironment is sufficient to drive neovascularization of secondary follicles over the course of a GC reaction.

It is important to note that reversal of tissue hypoxia appears to affect multiple aspects of the GC reaction: GC B cell frequency, CSR, T_FH and plasma cell induction. While insights from our in vitro CSR assay shows that hypoxia can have a direct effect on purified B cells in accelerating CSR and plasmacytic differentiation, it will be a goal of future studies to identify which cell types are most directly affected by oxygen tension in vivo (e.g. T_FH or GC B cells). From our studies of reversing in vivo hypoxia utilizing supplemental oxygen, we observed greater reductions in GC B cell frequencies than in comparable reductions in T_FH. These data coupled with our observation that hypoxia can accelerate overall proliferation of B cells in hypoxia in vitro, brings up the possibility that our observation of reduced T_FH frequencies may be secondary to the reduction in GC B cells, as GC B cells have been shown to support the maintenance of T_FH (29).

However, it certainly is a possibility that GC T_FH may be functionally affected in 60% O₂ which contributes to reduction in CSR to IgG1 as IL-4 secretion has been shown to dramatically increase in T cells cultured in hypoxic conditions (31), and IL-4 is dramatically up-regulated in GC T_FH when compared to non GC T_FH (13). These interpretations raise a unique point that T and B cells appear to react quite differently in hypoxia, in that secondary diversification such as CSR and early proliferation and as well as apoptosis are promoted in B cells under hypoxic conditions, while proliferation of CD4 T cells is suppressed in hypoxic culture conditions (30) but cytokine production (such as IL-4 and IFN gamma) is increased (31). Extrapolating and coupling these observations to the function of T_FH and B cells in the GC makes sense in the fact that GC B cells are intensely proliferating, undergoing secondary diversification, and highly apoptotic, while T_FH are not heavily proliferative in the GC but able to provide B cell help through multiple mechanisms and thus limiting for clonal selection (32). It is certainly possible that hypoxia within the GC microenvironment may be a significant environmental requirement of the GC so that B cells and T_FH develop their uniquely opposite but complimentary functions. Future studies will be needed to investigate these possibilities.

Our study suggests the possibility of unintended, negative consequences of supplemental oxygenation following vaccination or during ongoing humoral responses to infection. On the other hand, our study raises the possibility that supplemental oxygen could be utilized to suppress GCs which produce pathogenic antibodies such as in systemic lupus erythematosus (33). We view this study as opening the door to further investigations of the functional role of hypoxia within the GC microenvironment.

Abbreviations

ASC

antibody secreting cell

CSR

class switch recombination

GC

Germinal Center

HIF

hypoxia inducible factor

mTOR

mammalian target of rapamycin

NIP-BSA

(nitroiodophenyl) acetyl bovine serum albumin

NP-OVA

(nitrophenyl) acetyl ovalbumin

NP-CGG

(nitrophenyl) acetyl chicken gamma globulin

TFH

T follicular helper

UPR

unfolded protein response