Murine Glucocorticoid Receptors Orchestrate B-cell Migration Selectively between Bone Marrow and Blood

Abstract

Glucocorticoids promote CXCR4 expression by T cells, monocytes, macrophages, and eosinophils, but it is not known if glucocorticoids regulate CXCR4 in B cells. Considering the important contributions of CXCR4 to B-cell development and function, we investigated the glucocorticoid/CXCR4 axis in mice. We demonstrate that glucocorticoids upregulate CXCR4 mRNA and protein in mouse B cells. Using a novel strain of mice lacking glucocorticoid receptors (GRs) specifically in B cells, we show that reduced CXCR4 expression associated with GR deficiency results in impaired homing of mature B cells to bone marrow, whereas migration to other lymphoid tissues is independent of B cell GRs. The exchange of mature B cells between blood and bone marrow is sensitive to small, physiologic changes in glucocorticoid activity, as evidenced by the lack of circadian rhythmicity in GR-deficient B cell counts normally associated with diurnal patterns of glucocorticoid secretion. B cell^GRKO mice mounted normal humoral responses to immunizations with T-dependent and T-independent (Type I) antigens, but antibody responses to a multivalent T-independent (Type II) antigen were impaired, a surprise finding considering the immunosuppressive properties commonly attributed to glucocorticoids. We propose that endogenous glucocorticoids regulate a dynamic mode of B cell migration specialized for rapid exchange between bone marrow and blood, perhaps as a means to optimize humoral immunity during diurnal periods of activity.

Introduction

Glucocorticoids are best-known for their immunosuppressive properties, and synthetic glucocorticoids represent a first-line of treatment for many inflammatory and autoimmune diseases. Glucocorticoids bind intracellular glucocorticoid receptors (GRs, encoded by Nr3c1) to repress pro-inflammatory transcription factors(1), activate anti-inflammatory genes(2-5), and induce lymphocyte apoptosis(6). However, the cell-type specific contributions of GR signaling to multicellular immune processes have been difficult to assess due to the ubiquitous nature of GR expression. Recent studies have used mice with genetic ablation of Gr in specific cell lineages to dissect direct glucocorticoid actions on T cells(7), macrophages(8), and dendritic cells(9) in vivo.

Among their various effects, glucocorticoids alter leukocyte trafficking. It is unclear if glucocorticoids regulate chemotactic cues generated by tissues or act directly on lymphocytes to modulate migration(10-13). Evidence for the latter includes observations that glucocorticoids upregulate CXCR4 expression by eosinophils(14), monocytes(15), macrophages(16), T cells(17), and hematopoietic progenitor cells(18). To our knowledge, it is not known if glucocorticoids regulate CXCR4 in B cells. CXCR4 plays key roles in the retention of B cell precursors in bone marrow(19), migration of mature B cells from Peyer’s patches(20), organization of germinal centers(21), and homing of plasma cells to lymphoid tissues(22). If glucocorticoids regulate CXCR4 in B cells, then these processes could be impacted by changes in glucocorticoid activity. We sought to determine if glucocorticoids regulate B cell expression of CXCR4 and, if so, to what extent glucocorticoid-induced changes in CXCR4 expression affect B cell function.

Materials and methods

Mice:

Intact and adrenalectomized C57BL/6 mice (Jackson Laboratories) were housed in specific-pathogen free conditions under 12 hour light/dark cycle at the National Institute of Environmental Health Sciences. Adrenalectomized mice were provided saline drinking water. Mb1-Cre mice(23) were crossed with mice harboring Nr3c1 alleles containing loxp sites flanking exons 3 and 4(24) to generate Mb1^Cre/wtNr3c1^fl/fl mice and Cre-negative littermates (Mb1^wt/wtNr3c1^fl/fl) on a C57BL/6 background. Mice were provided with sterilized chow and water ad libitum. Male mice aged 8-12 weeks were used in all studies unless otherwise stated. All experiments were approved and performed according to the guidelines of the Animal Care and Use Committee at the NIEHS.

Dexamethasone injections:

Dexamethasone (Steraloids) was injected i.p. at 2 mg/kg; control mice received an injection of vehicle. Mice received one i.p. injection daily for three days, and were sacrificed on day four.

Flow cytometry:

Bone marrow, blood, and spleen were prepared as single-cell suspensions and subjected to red blood cell lysis with ACK buffer. Cells were stained with fluorochrome-conjugated mAb for B220, IgM, IgD, CD138, CD43, CD3 or CD5, CXCR4, CD11c, and CD11b (eBioscience, BD Biosciences, and Biolegend). 7AAD staining identified dead cells. Data were acquired on a LSRII flow cytometer (BD Biosciences) and analyzed with FlowJo software (FlowJo).

To detect intracellular GR, cells were stained for surface antigens, and then fixed/permeabilized with eBioscience fixation/permeabilization reagents. Cells were stained with anti-GR (clone D8H2, Cell Signaling) or isotype-matched control mAb, and then Alexa fluor 647-conjugated anti-rabbit IgG (Southern Biotech). viSNE plots were generated in Cytobank using the Barnes-Hut implementation of the t-SNE algorithm(25).

Quantitative PCR:

B cells were enriched by negative selection (EasySep™ Mouse B cell Isolation Kit, StemCell Technologies). For Dex treatment studies, B cells were cultured in RPMI-1640/10% charcoal-stripped FBS/55 μM 2-ME/10 mM HEPES/penicillin-streptomycin containing 100 nM Dex or vehicle for 1, 3, and 6 hours. Total RNA was isolated using Qiagen RNeasy kits. RNA (50-100 ng) was reverse transcribed and amplified with iScript One-Step RT-PCR kit (Bio-Rad) using Taqman primer/probe sets (ThermoFisher Scientific). Real-time quantitative PCR was performed with a Bio-Rad CFX96 sequence-detection system. The fluorescent signal from each transcript was normalized to the housekeeping gene Ppib using the 2^−ΔCt method.

Transwell Migration Assay:

Transwell plates (24-wells) with 5 μm pore polycarbonate membrane inserts were seeded with 10⁵ splenic B cells. The lower chamber contained medium alone, 100 ng/ml recombinant mouse CXCL12 (Peprotech), 200 ng/ml recombinant mouse Macrophage Migration Inhibitory Factor (MIF) (Peprotech or Abcam), 500 ng/ml recombinant mouse CXCL13 (Biolegend), or 100 ng/ml recombinant mouse CCL19 (Peprotech). Plates were incubated at 37°C incubator with 5% CO₂ for 3 hours. Cells in the lower chamber were quantified by FACS, with a 30-second acquisition time for each sample.

Immunizations:

The succinic anhydride ester of (4-hydroxy-3-nitrophenyl)-acetyl (NP) was reacted with chicken γ-globulin (CGG; Sigma Aldrich) to generate NP₈-CGG. NP₈-CGG was added to Alhydrogel 2% (Accurate Chemical). Mice were injected i.p. with 10 μg NP₈-CGG per mouse. NP_0.15-LPS (Biosearch Technologies) was diluted in PBS and injected i.p.at a dose of 10 μg NP_0.15-LPS per mouse. NP₄₉-AECEM-Ficoll (Biosearch Technologies) was diluted in PBS and injected i.p.at a dose of 10 μg NP₄₉-AECEM-Ficoll per mouse. Mice were bled on days 0, 5, 8, 16, and 24 after immunization, and serum was collected for ELISA.

ELISA:

To measure IgM, IgG, and IgA in sera, wells of 384-well assay plates were coated with anti-Igκ and anti-Igλ (Southern Biotechnologies) in carbonate buffer. To quantify NP-specific antibody, 384-well assay plates were coated overnight with NIP₂₅-BSA. For quantification of anti-NP IgM, sera were subject to mild reduction with 2-ME to dissociate pentameric IgM into monomers(26). Serial dilutions of serum were made in PBS/0.5% BSA. Anti-NP IgG and IgM mAb standards (clones H33Lγ1 and B1-8, respectively) were run simultaneously as standards. HRP-conjugated anti-mouse IgM, IgG, or IgA (Southern Biotech) was added to detect isotype-specific immunoglobulin. TMB substrate (Biolegend) was added to wells. 2N H₂SO₄ was used to stop the enzymatic reaction. OD450 readings were recorded with background subtraction at OD630. Serum IgE was quantified using a commercial ELISA kit (Biolegend).

ELISpot:

For analysis of IgM, IgG, and IgA secreting cells, 96-well High Protein Binding plates (Millipore) were coated with anti-Igκ and anti-Igλ (Southern Biotech) in carbonate buffer. Bone marrow and spleen cell suspensions were added at 2.7, 0.9, 0.3, and 0.1x10⁵ cells/well in duplicate. Cells were incubated at 37°C/5% CO₂ for 3 hours. After removing cells, AP-conjugated anti-mouse IgG or IgM, or HRP-conjugated anti-mouse IgA were added to wells. SIGMAfast BCIP/NBT reagent (Sigma-Aldrich) and AEC substrate (BD Biosciences) were used to visualize AP-conjugated and HRP-conjugated detection reagents, respectively. Frequencies of antibody-secreting cells were calculated as the number of spots divided by the number of input cells.

Adoptive Transfer of B cells:

Splenic B cells were enriched by negative selection (Stem Cell Technologies); the purity of B220⁺ cells routinely exceeded 97%. Control B cells were labeled with 10 nM or 100 nM CellTrace™ CFSE reagent (ThermoFisher Scientific) according to the manufacturer’s protocol. GR-deficient B cells were labeled with 100 nM CFSE. Recipient mice received a 100 μl i.v. injection of B cells (2-3 million cells) containing a 1:1 mixture of 10 nM CFSE-labeled control B cells and 100 nM CFSE-labeled GR-deficient B cells, or a 1:1 mixture of 10 nM and 100 nM CFSE-labeled control B cells. After 18 hours, recipient animals were sacrificed, and single-cell suspensions from the blood, bone marrow, spleen, inguinal lymph nodes, and Peyer’s patches were analyzed by flow cytometry for the presence of CFSE^dim and CFSE^bright donor B cells.

Statistics:

Pair-wise comparisons were made with Wilcoxon Exact Tests. Comparisons for transwell migration studies were made using Sign Tests for matched pairs of WT and GR-KO cell count data for each assay condition. All statics were performed using JMP Pro version 15.0.0 (SAS).

Data Sharing Statement:

For original data, please contact cidlows1@niehs.nih.

Results

Glucocorticoids upregulate CXCR4 transcription in B cells

A link between GR signaling and Cxcr4 expression in B cells has not, to our knowledge, been reported. We treated C57BL/6 splenic B cells with the synthetic glucocorticoid dexamethasone (Dex) or vehicle ex vivo, then evaluated transcription of the chemokine receptors Cxcr4, Cxcr5, and Ccr7, alongside the glucocorticoid target gene Tsc22d3 (aka Gilz). After six hours of incubation, Gilz transcripts were significantly increased by Dex treatment, as expected (Figure 1A). We also observed more Cxcr4 transcripts in Dex-treated B cells compared to vehicle treatment (Figure 1A), whereas Cxcr5 and Ccr7 transcripts were unchanged (Figure 1A). A time-course study revealed that B cells upregulated Cxcr4 transcription within one hour of incubation with Dex (Figure 1B). Dex-induced increases in CXCR4 surface protein were evident after three hours of incubation (Figure 1C). As previously reported for myeloid and T cells(27), CXCR4 surface labeling gradually increased during the cultivation of both Dex-treated and control B cells, however Dex treatment provided an additive effect (Figure 1C). Notably, CXCR5 labeling diminished modestly with Dex treatment, suggestive of an inhibitory effect of glucocorticoids on this chemokine receptors, whereas CCR7 labeling was unaffected (Figure 1D). B cells were very sensitive to Dex effects on CXCR4 expression, as increased CXCR4 labeling occurred at 1 nM Dex and saturated at 10 nM and above (Figure 1E). The opposing effects of Dex on CXCR4 and CXCR5 expression suggested that glucocorticoids might alter B-cell trafficking patterns.

Figure 1. Glucocorticoids enhance CXCR4 expression by murine B cells.

(A) Splenic B cells from male C57BL/6 mice were cultured with vehicle (gray bars) or 100 nM Dex (open bars) for 6 hours, then analyzed by Q-PCR for Gilz, Cxcr4, Cxcr5, and Ccr7 transcription. Mean+SD number of gene transcripts normalized to the housekeeping gene Ppib are shown. Data were pooled from three independent experiments, N=8 biological replicates of each condition. (B) Cells were harvested after 1, 3, and 6 hours of incubation with vehicle (gray) or 100 nM Dex (open) for Cxcr4 expression analysis by Q-PCR; the black bar shows Cxcr4 transcripts of freshly isolated B cells. Mean+SD numbers of Cxcr4 transcripts normalized to PPIB are shown. Data represent two independent experiments; N=4 biological replicates of each condition. (C) The mean±SD median fluorescence intensities (MFI) of anti-CXCR4 labeling of vehicle (filled, dotted line histogram) vs. Dex (open, solid line histogram)-treated B cells at different time points of incubation are shown. Data are pooled from two independent experiments; N=4 biological replicates of each condition. (D) Mean±SD MFI of anti-CXCR5 and anti-CCR7 labeling of B cells after 6 hour incubation with vehicle (gray) or 100 nM Dex (open). Data represent 5 independent experiments; N=8-10 replicates per condition. (E) Anti-CXCR4 labeling (mean±SD MFI) of mouse B cells after six hours of incubation with various Dex concentrations. Data are pooled from two independent experiments; N=6 per condition. Experimental groups not connected by the same letter are significantly different (P<0.01). *, P<0.05; **, P<0.01.

To investigate glucocorticoid effects on B cells in vivo, we injected mice with Dex and enumerated B cell populations in various lymphoid tissues. Mouse lymphocytes are sensitive to glucocorticoid-induced death, so it was not surprising that Dex-treated mice had small spleens, and generalized reductions in B and T cell numbers in blood and tissues (Figure 2A and B). All subpopulations of B cells analyzed, including IgM⁻IgD⁻ pro/pre, IgM^lowIgD⁻ immature, IgM^hiIgD^low transitional B cells, and IgM^lowIgD^hi mature B cells were dramatically reduced in the spleen and blood (Figure 2A, B) (see Supplemental Figure 1 for detailed B cell immunophenotyping). However, the number of mature B cells (and T cells) in bone marrow was not changed by Dex treatment (Figure 2B). While it is possible that the bone marrow selectively protects mature B cells from glucocorticoid-induced death, another explanation is that Dex upregulates CXCR4 expression by B cells, thereby promoting their homing to or retention in bone marrow. In the second scenario, Dex-induced losses of mature B cells in bone marrow might be masked by Dex effects on B cell migration.

Figure 2. Effects of glucocorticoid perturbation on B cells in vivo.

Male C57BL/6 mice were injected i.p. with dexamethasone (Dex, 2 mg/kg) or vehicle once per day for three days and then B cell populations in the bone marrow, blood, and spleen were enumerated by flow cytometry. (A) Representative dot plots from flow cytometric analyses of vehicle- and Dex-treated mice are shown. CD11b⁺CD11c⁺7AAD⁺ cells were excluded prior to gating on B220⁺CD3⁻ cells (B cells) and B220⁻CD3⁺ (T cells). In bone marrow and spleen, B cells were further subdivided as pro/pre B cells (IgM⁻IgD⁻), immature B cells (IgM^lowIgD⁻), transitional B cells (IgM^hiIgD^low), and mature B cells (IgM^lowIgD^hi). In the spleen, the IgM^hiIgD^low compartment comprises both transitional B cells and marginal zone B cells (“T/MZ B cells”). (B) The numbers (mean±SD) of B cells and T cells in the bone marrow, blood, and spleen in mice treated with vehicle (black) or Dex (open) are shown. Data were pooled from three independent experiments; N=11 vehicle-treated mice and N=11 Dex-treated mice. (C) The numbers (mean±SD) of B cells and T cells in the bone marrow, blood, and spleen of intact mice (black) and adrenalectomized (ADX) mice (open) are shown. Data were pooled from three independent experiments; N=14 intact mice and N=12 ADX mice. *, P<0.05; **, P<0.01.

We next determined if adrenalectomized (ADX) mice, which are deficient for endogenous glucocorticoids, exhibit alterations in B-cell populations. In bone marrow of ADX mice, the pro/pre B compartment was enlarged, immature and transitional B cell populations were comparable to controls, but mature B-cell and T-cell numbers were lower than intact animals (Figure 2C). B- and T-cell numbers in the blood, however, were supranormal in ADX mice, and the spleens contained higher numbers of B cell precursors but normal numbers of mature B cells and T cells (Figure 2C). These findings suggested distinct effects of glucocorticoids at various stages of B cell development, yet it was unclear if changes represented glucocorticoid effects on survival vs. migration, or if glucocorticoids alter B cell populations intrinsically or extrinsically.

Characterization of B cell-specific GR knockout mice

To determine direct vs. indirect effects of glucocorticoids on B cells, we generated mice with B cell-specific GR deletion. Mb1-Cre mice(23) were crossed with Nr3c1 floxed mice(24) to generate B cell^GRKO mice (Mb1^Cre/wt Nr3c1^fl/fl) and Cre-negative littermates (Mb1^wt/wt Nr3c1^fl/fl mice). B cell^GRKO mice were born at Mendelian ratios and exhibited normal weight, appearance, and behavior. Histological analyses revealed no evidence of generalized pathology, and the lymphoid architecture in spleen and lymph nodes was normal (data not shown).

We used flow cytometry to verify B-cell specific ablation of Gr in B cell^GRKO mice. Whereas most cells in bone marrow, blood, and spleens of control mice expressed GRs, anti-GR labeling was specifically reduced in B cells of B cell^GRKO mice (Figure 3A). In B cell^GRKO mice, GR was expressed by less than 10% of bone marrow pro/pre B cells, and was undetectable in immature, transitional, and mature B cells (Figure 3B). Moreover, B cells from B cell^GRKO mice were protected from Dex-induced cell death in vitro, whereas T cells from B cell^GRKO mice were fully sensitive (Figure 3C), confirming functional and specific deletion of GR in B cells.

Figure 3. Characterization of B cell^GRKO mice.

Mb1-Cre mice were crossed with Nr3c1^fl/fl mice to generate animals lacking GR in B cells (B cell^GRKO mice). (A) GR expression in bone marrow, blood, and spleen cells from control and B cell^GRKO mice was determined via flow cytometry. After staining for surface antigens (B220, CD5, CD11b, and CD11c), cells were labelled intracellularly with anti-GR mAb. Multiparameter data were subjected to viSNE analysis based on B220, CD5, CD11b, and CD11c labeling, and discrete populations of B cells (B220+CD5−CD11b−CD11c−), T cells (B220−CD5+CD11b−CD11c−), myeloid cells (B220−CD5−CD11b+CD11c−), and DCs (B220+CD5−CD11b−CD11c−) were identified in each tissue. GR labeling is depicted as a heat map on each viSNE plot. (B) GR expression by B cell precursors from bone marrow of control and B cell^GRKO mice was investigated. B220⁺CD11c⁻CD11b⁻CD5⁻ B cells were further into pro/pre B cells (IgM⁻IgD⁻), immature B cells (IgM^lowIgD⁻), transitional B cells (IgM^hiIgD^low), and mature B cells (IgM^lowIgD⁺). Data are representative of three independent experiments; N=3 mice of each genotype. (C) Susceptibility of GR-deficient B cells to Dex-induced cell death. Splenocytes from control (filled) and B cell^GRKO (open) mice were cultured overnight with graded doses of Dex, and then viable B cells (B220⁺, left graph) and T cells (CD3⁺, right graph) were enumerated by flow cytometry. For each dose of Dex, the number of cells in each compartment was normalized to that in wells lacking Dex. The mean±SD relative number of cells from control (filled) and B cell^GRKO (open) mice are shown. Data were pooled from three independent experiment, with three biological replicates per experiment. *, P<0.05; **, P<0.01.

Decreased CXCR4 expression in GR deficient B cells impacts chemotactic responses to CXCL12

To determine if GR deficiency impacted Cxcr4 expression in B cells, we isolated splenic B cells from B cell^GRKO and control mice for Q-PCR analysis. As expected, Gr and Gilz transcripts were reduced in GR-deficient B cells, whereas transcription of the mineralocorticoid receptor (Mr, encoded by Nr3c2) was unaltered (Figure 4A). Cxcr4 transcripts in GR-deficient B cells were reduced ~50% compared to B cells from control mice (Figure 4A), consistent with a role for endogenous glucocorticoids in basal Cxcr4 expression. Cxcr5 transcripts were modestly, but not significantly, elevated in GR-deficient B cells (P=0.054), whereas Ccr7 transcripts were comparable in control and GRKO B cells (Figure 4A). Antibody labeling revealed a similar pattern of chemokine receptor expression; GR-deficient B cells had less CXCR4 on the cell surface than control B cells, whereas CXCR5 and CCR7 labeling were comparable between control and GR-deficient B cells (Figure 4B). These results indicate that B-cell CXCR4 is regulated in vivo by GR signaling at physiologic concentrations of glucocorticoids.

Figure 4. Decreased CXCR4 expression in GR-deficient B cells.

(A) The mean+SD number of mRNA transcripts for Gr, Mr, Gilz, Cxcr4, Cxcr5, and Ccr7 (normalized to Ppib) in control (gray bars) and GR-deficient (open) splenic B cells. Data represent five independent Q-PCR experiments, with two mice of each genotype per experiment (N=10); except for Mr, where N=4. (B) The mean±SD MFIs for APC anti-CXCR4, anti-CXCR5, and anti-CCR7 labeling of control (gray bars) and GR-deficient (open) splenic B cells. Data represent five independent flow cytometry experiments, with two mice of each genotype per experiment (N=7-10). (C) The mean±SD MFIs for anti-CXCR4 labeling of bone marrow (BM), blood, and spleen cells from control and B cell^GRKO mice, as defined in Figure 2b. Data reflect three independent experiments; N=8 control mice and 6 B cell^GRKO mice. (D) Effect of GR deficiency on B cell migration in ex vivo transwell assays towards CXCL12, CXCL12+AMD3100 (a CXCR4 antagonist), MIF, CXCL13, and CCL19. Each pair of connected points represents the mean number of control and GR-deficient B cells counted in the lower chamber of transwells from an independent study (N=2-4 experimental replicates per independent study). *, P<0.05; **, P<0.01, ***, P<0.001.

We analyzed CXCR4 expression by B cell populations in the bone marrow, blood, and spleens of B cell^GRKO and control mice by flow cytometry. In both control and B cell^GRKO mice, anti-CXCR4 labeling decreased with B cell maturation in bone marrow, but starting at the immature stage of differentiation, GR-deficient B cells exhibited less CXCR4 labeling than their control counterparts (Figure 4C). Similarly, blood and splenic B cells of B cell^GRKO mice expressed lower levels of CXCR4 than in control mice (Figure 4C), whereas T cells labeled comparably (Figure 4C). GR signaling plays an intrinsic role in regulating CXCR4 expression starting at the immature stage of B cell differentiation.

Next, we determined if GR deficiency impacts B-cell migration towards the CXCR4 ligands CXCL12(28, 29) and macrophage migration inhibitory factor (MIF)(30). Splenic B cells from GR-deficient mice exhibited a modest but reproducible reduction in migration towards CXCL12 in transwell assays (P<0.01) (Figure 4D). Migration of control and GR-deficient B cells towards CXCL12 was fully blocked by the CXCR4 antagonist AMD3100 (Figure 4D), confirming that CXCL12 mediated B cell migration through CXCR4. However, we did not observe a chemotactic response of B cells, either control or GR deficient, toward MIF (Figure 4D) (Recombinant MIF from two commercial sources at various concentrations were tested). In contrast, GR-deficient B cells were fully competent to migrate towards CXCL13 and CCL19 (Figure 4D), chemokines bound by CXCR5 and CCR7, respectively, that organize B cells in secondary lymphoid tissues(31). GR deficiency does not impart a general defect in B cell migration but specifically impairs B cell responsiveness to CXCL12.

Intrinsic GR deficiency alters B cell numbers in bone marrow and blood, but not in secondary lymphoid tissues

To determine if GR deficiency affected B cell populations in tissues, we enumerated B cell populations in bone marrow, blood, and spleens of B cell^GRKO mice and control littermates. B cell^GRKO mice exhibited normal numbers of pro/pre, immature, and transitional B cells in bone marrow, but the number of mature B cells was reduced compared to controls (Figure 5A), similar to ADX mice. B cell^GRKO mouse blood contained 1.7-fold more B cells than controls (Figure 5A), primarily due to excess mature B cells, although immature and transitional B cell numbers were also elevated (Figure 5B). The spleens of B cell^GRKO mice were normal in size and in numbers of developing and mature B cells (Figure 5A). The lymph nodes of B cell^GRKO were also normal size and contained similar numbers of B cells as control mice (data not shown). Splenic T-cell numbers were comparable across tissues in B cell^GRKO and control littermates (Figure 5A). B cell^GRKO exhibited some of the features observed in ADX mice, including decreased numbers of mature B cells in bone marrow and increased numbers of B cells in blood, but B cell precursor numbers in spleen were not elevated in B cell^GRKO mice as in ADX mice, suggesting that some aspects of the ADX phenotype reflect B-cell extrinsic effects of glucocorticoid deficiency.

Figure 5. Redistribution of B cells from bone marrow to blood in B cell^GRKO mice.

B and T cell populations in male 8-12 week-old control and B cell^GRKO mice were enumerated by flow cytometry. (A) Representative dot plots from flow cytometric analyses of control (top row) and B cell^GRKO (bottom row) mice are shown. CD11b⁺CD11c⁺7AAD⁺ cells were excluded prior to gating on B220⁺CD5⁻ cells (B cells) and B220⁻CD5⁺ (T cells). In bone marrow and spleen, B cells were further subdivided as in Figure 2b. The mean±SD numbers of B cell subsets and T cells in the bone marrow (femur and tibia), blood, and spleen of control mice (black) and B cell^GRKO mice (open) are shown. Data were pooled from three independent experiments; N=8 control mice and 6 B cell^GRKO mice. (B) Blood B cells were subdivided by IgM and IgD expression, as in bone marrow and spleen. (C) The numbers (mean±SD) of cells in B cell subsets and T cells in bone marrow (femur and tibia), blood, and spleen were enumerated in aged (13-15 month) male control (black) and B cell^GRKO (open) mice. Data were pooled from two independent experiments; N=7 control mice and 5 B cell^GRKO mice. *, P<0.05; **, P<0.01.

To investigate effects of GR deficiency on B cells in a non-lymphoid tissue, we lavaged the peritoneal cavities of control and B cell^GRKO mice, where specialized B1a and B1b B cells are common alongside canonical B2 B cells. As expected, B1a and B1b B cells from B cell^GRKO mice lacked GR expression and, like B2 B cells, they exhibited significantly less surface CXCR4 than controls (Supplemental Figure 2). However, we found no evidence that GR deficiency affected B1a, B1b, or B2 populations in the peritoneal cavity, as comparable numbers of each cell type were found in B cell^GRKO mice and control mice (Supplemental Figure 2).

We expanded our studies of B cell^GRKO animals to female mice and aged male mice. Female B cell^GRKO mice (8-12 weeks old) exhibited a similar B-cell phenotype as males: decreased numbers of mature B cells in bone marrow, increased numbers of blood B cells, and normal counts of mature B cells in spleens (Supplemental Figure 3). Notably, in female B cell^GRKO mice, we observed a 1.7-fold increase in the number of immature B cells in the spleen compared to controls, a difference which was less apparent in male mice. In 12-15-month-old male mice, mature B cell numbers in bone marrow approached that of control littermates (P>0.05), but B-cell numbers in blood were still elevated (Figure 5C). Splenic B-cell numbers in aged B cell^GRKO mice were similar to controls (Figure 5C). The consistent phenotype of B cell^GRKO mice across age and sex was an increase in blood B cell numbers.

Intrinsic GR signaling regulates diurnal exchange of B cells between blood and bone marrow

Diurnal patterns of glucocorticoid secretion drive circadian rhythms of lymphocyte trafficking, with blood glucocorticoid concentrations correlating inversely with blood lymphocyte counts(17, 32). To determine if intrinsic GRs contribute to diurnal trafficking patterns of B cells, we enumerated blood B cells in control and B cell^GRKO mice at 9:00 a.m. and at 5:00 p.m (Zeitgeber times ZT2 and ZT10, respectively), when corticosterone concentrations were low and high, respectively (Figure 6A). In control mice, blood B cell counts were significantly higher in morning compared to evening (Figure 6A), consistent with previous reports(32). In B cell^GRKO mice, however, circulating B cell counts were unchanged from morning to evening, despite a normal rise in serum corticosterone concentration (Figure 6A). B cell GRs orchestrate circadian patterns of exchange between blood and tissues.

Figure 6. GR deficiency in B cells disrupts trafficking between blood and bone marrow.

(A) Effect of GR-deficiency on B-cell circadian trafficking patterns. A cohort of male control and B cell^GRKO mice were bled two hours after the start of the light cycle (9:00h; Zeitgeber time ZT2) and another cohort bled two hours before the start of the dark cycle (17:00h; ZT10). Serum corticosterone concentrations were measured by ELISA; the left plot shows mean±SD corticosterone concentrations at each time point. The mean±SD number of blood B cells (B220⁺CD11b⁻CD11c⁻CD5⁻) at each time point are shown at right. Data represent pooled data from three independent experiments, with N=7-12 mice for each genotype at each time point. (B) Effect of GR deficiency on B cell localization to different lymphoid tissues. A 1:1 mixture of CFSE^dim control B cells and CFSE^bright control B cells (left column) or CFSE^dim control B cells and CFSE^bright GR-deficient B cells (right column) were adoptively transferred into control recipient mice. Histograms show representative data of CFSE^dim and CFSE^bright B cells in the cell mixture used for adoptive transfers (“Input”), and the proportions of CFSE^dim and CFSE^bright B cells in recipient tissues 18 hours after transfer (CFSE negative cells have been excluded). (C) A Recovery Index for control (closed) or GR deficient (open) B cells was calculated for each tissue of each recipient mouse. Each point represents the recovery index from one recipient animal. Data reflect three independent experiments, with N=6 recipients receiving control:control B cell mixtures, and N=7 recipients receiving control:GRKO B cell mixtures. *, P<0.05; **, P<0.01, n.s., not significant.

To determine if GR deficiency affects B cell homing in a tissue-specific manner, we employed competitive adoptive transfers. Splenic B cells from control and B cell^GRKO mice were differentially labeled with CFSE to allow discrimination of CFSE^dim and CFSE^bright cells by flow cytometry. CFSE^dim control B cells and CFSE^bright GR-deficient B cells were mixed at a 1:1 ratio and adoptively transferred into control mice. A second cohort of mice received a 1:1 mixture of B cells in which both CFSE^dim and CFSE^bright cells were GR-sufficient. After 18 hours, we determined the relative proportions of CFSE^dim and CFSE^bright cells in blood, bone marrow, and various secondary lymphoid tissues of recipient mice. We calculated a “Recovery Index” as:

$$\text{Recovery Index}=\frac{[{\text{Freq. of CFSE}}^{\text{bright}}{\text{cells/(Freq. of CFSE}}^{\text{dim}}+{\text{CFSE}}^{\text{bright}}\text{cells) in tissue]}}{[{\text{Freq. of CFSE}}^{\text{bright}}\text{cells in adoptive transfer}]}$$

A Recovery Index greater than 1 represented an enrichment of CFSE^bright cells in tissue compared to the original cell mixture whereas a value less than 1 indicated a reduction in CFSE^bright cells compared to the input. We observed a decreased Recovery Index of GR-deficient B cells in the bone marrow but not in the blood, spleen, lymph nodes, or Peyer’s patches (Figure 6B, C). The Recovery Index of GR-sufficient CFSE^bright B cells, however, was comparable to the input mixture (Figure 6B, C). We conclude that GR signaling in B cells enhances homing specifically to bone marrow.

Humoral responses in B cell^GRKO mice

We hypothesized that GR deficiency in B cells would affect humoral immunity, either through dysregulated B cell migration or other modes of GR activity. However, serum concentrations of IgM, IgG, IgA, and IgE were comparable in control and B cell^GRKO male mice (Figure 7A), aged male mice (Figure 7B), and female mice (Supplemental Figure 3). We confirmed that GR was ablated in plasma cells of B cell^GRKO mice (Supplemental Figure 4) but, consistent with serological data, the numbers of plasma cells/-blasts in the bone marrow and spleens of B cell^GRKO mice were similar to those in littermate controls, as assayed by flow cytometry (Figure 7C) and ELISpot (Figure 7D). GR deficiency in the B cell lineage did not impact the basal generation of immunoglobulin or antibody secreting cells.

Figure 7. Humoral immunity in B cell^GRKO mice.

(A) Serum concentrations of IgM, IgG, IgA, and IgE from control (closed circles) and B cell^GRKO (open circles) mouse are shown. Data reflect three independent ELISA experiments; N=8 control mice and 7 B cell^GRKO mice (N=7 control mice and 6 B cell^GRKO mice for IgE). (B) Serum IgM, IgG, IgA, and IgE concentrations from aged (12-15-month old) naïve male mice are shown. Data represent two independent ELISA experiments; N=6 control mice and 4 B cell^GRKO mice. (C) Representative dot plots of CD138^hi FSC^int plasma cells/-blasts in bone marrow (BM) and spleen from control (top row) and B cell^GRKO (bottom row) mice are shown. The mean±SD number of plasma cells/-blasts are shown at right. Data were pooled from three independent experiments; N=7 control mice and 6 B cell^GRKO mice. (D) The mean±SD numbers of IgM, IgG, and IgA-secreting cells in bone marrow (femur+tibia) and spleen are shown. Data were pooled from two independent ELISpot experiments; N=6 control (filled) and 6 B cell^GRKO (open) mice. (E) T-dependent humoral responses to NP₈-CGG/alum immunization of control and B cell^GRKO mice. The mean±SD concentrations of NP-specific IgM and IgG from control (closed circles) and B cell^GRKO (open circles) at each time point are shown. Data shown are results from one of two independent experiments. N=4 for each genotype. (F) T-independent (Type I) humoral responses to NP-LPS by control and B cell^GRKO mice. The mean±SD concentrations of NP-specific IgM and IgG from control (closed circles) and B cell^GRKO (open circles) are shown. N=7 for each genotype. (G) T-independent (Type II) humoral responses of control and B cell^GRKO mice. The mean±SD concentrations of NP-specific IgM and IgG from control (closed circles) and B cell^GRKO (open circles) on days 0, 5, 8, 16, and 24 are shown. Data were pooled from three independent experiments; N=14 control mice and 9 B cell^GRKO mice. *, P<0.05; **, P<0.01.

Lastly, we quantified antibody responses in B cell^GRKO and control mice to immunizations with various antigens. Following immunization with the T-dependent antigen NP-CGG, B cell^GRKO mice generated antigen-specific IgM and IgG responses comparable to controls (Figure 7E). Similarly, B cell^GRKO and control mice mounted equivalent IgM and IgG responses to NP-LPS, a T-independent (Type 1) antigen (Figure 7F). In response to the T-independent (Type 2) antigen NP-Ficoll, B cell^GRKO mice mounted an NP-specific IgM response that was similar to control littermates, but generated significantly less NP-specific IgG than controls (Figure 7G). The diminished IgG response of B cell^GRKO mice to NP-Ficoll was surprising considering the immunosuppressive properties typically attributed to glucocorticoids, and indicates a positive role for GR signaling in certain aspects of humoral immunity.

Discussion

In this study, we determined that glucocorticoids promote B cell expression of Cxcr4 (Figure 1), and that alterations in GR signaling affect the distribution of B cells in blood and bone marrow (Fig. 2 and 5). Using a B-cell specific GR knockout mouse, we address long-standing questions regarding glucocorticoid effects on lymphocyte migration and arrive at the following conclusions: (1) glucocorticoids act directly on B cells to promote homing specifically to bone marrow in association with increased CXCR4 expression; and (2) GR-dependent regulation of B cell trafficking between blood and bone marrow occurs at physiologic concentrations of glucocorticoids associated with diurnal patterns of adrenal activity.

In humans and animals, glucocorticoid-induced changes in immune cell numbers have typically been attributed to well-characterized lympholytic properties. Our study, however, highlights that glucocorticoid effects on B cell populations in vivo are complex and likely reflect regulation of both survival and migration. This point is perhaps best illustrated in Dex-treated mice, in which B cell populations in blood and tissues were dramatically reduced, yet the number of mature B cells in bone marrow remained stable, giving the appearance that this particular population was “protected” from glucocorticoid-induced death (Figure 2B). Based on our findings that glucocorticoids upregulate B-cell CXCR4 and enhance homing to and/or retention in bone marrow, we postulate that mature B cells are universally susceptible to glucocorticoid-induced death regardless of tissue location, but apoptotic losses in bone marrow are hidden by the glucocorticoid-induced influx of mature B cells from blood. Indeed, in the three mouse models we employed – a pharmacologic model (Dex treatment, Figure 2A, B), a surgical model (adrenalectomy, Figure 2C), and a genetic model (B cell-specific GR KO mice, Figure 5) – the common observation was that increased glucocorticoid/GR activity supported mature B cell populations in bone marrow at the expense of blood, whereas decreased glucocorticoid/GR activity had the reverse effect. We posit that migration effects of glucocorticoids likely dominate over survival effects in vivo, as increases in Cxcr4 transcripts were evident at lower concentrations of Dex (1 nM, Figure 1E) than death effects (10 nM, Figure 3C). Moreover, our observations of altered B cell populations in bone marrow and blood of B cell^GRKO mice (Figures 5 and 6A), and defective migration of GR-deficient B cells in adoptive transfer studies (Figure 6B,C) occurred under physiologic conditions of adrenal activity, not stress-induced or pharmacologic states where glucocorticoid-induced death is evident. We also suspect that survival factors, such as BAFF, provide some protection for B cells from glucocorticoid-induced death in vivo, further separating the in vivo concentrations at which glucocorticoids alter migration vs. induce death.

Our finding that glucocorticoids regulate CXCR4 expression in B cells to affect their migration to bone marrow explains the recent findings of Courtiers et al., who observed in mice that ischemic stroke provokes the same changes in B cell populations that we observed with Dex treatment – decreased numbers of developing B cells but an abundance of mature B cells in bone marrow(33). Importantly, this phenomenon occurred in association with a surge in inflammatory cytokines and serum corticosterone, and was prevented by GR deletion in B cells. We suspect that glucocorticoid-induced changes in B-cell migration may explain, at least in part, the blood B lymphopenia associated with infection, inflammation, and severe physical stress.

We observed similarities between B cell^GRKO mice and B cell-specific CXCR4 knockout mice, although the latter strain exhibits a more dramatic phenotype. In B cell-CXCR4 KO mice, the bone marrow is nearly devoid of mature B cells, and substantial populations of B cell precursors occur in the blood and spleen, highlighting the role of CXCR4 in retaining both developing and mature B cells in bone marrow(19). Similarly, in B cell^GRKO mice, the mature B cell population in bone marrow was reduced, and the numbers of immature and transitional B cells in blood were elevated (Figure 5B), suggesting that glucocorticoid-dependent regulation of CXCR4 supports the retention of precursors and mature B cells in bone marrow. Since CXCR4 expression is highest in early stages of B cell maturation [Figure 4B and ref (19)], we hypothesize that GR-independent expression of CXCR4 is sufficient to retain most B cell precursors in bone marrow, but reduced CXCR4 expression associated with GR deficiency in mature B cells impacts this population more severely, resulting in the redistribution of a substantial portion of this population to the blood.

While glucocorticoids have been described to alter lymphocyte trafficking, it has been unclear if glucocorticoids act on tissues/endothelium to alter production of chemoattractants or if GR signaling in lymphocytes modifies responsiveness to migratory cues(10-13). Studies in humans showed that diurnal increases in circulating cortisol are accompanied by increased T cell expression of CXCR4 and decreased T cell counts in blood, a correlation ablated by pharmaceutical antagonism of endogenous glucocorticoids(17). In mice, we similarly observed that blood B cell counts correlated inversely with serum glucocorticoid concentrations throughout the day, and this relationship was uncoupled in B cell^GRKO mice, which exhibited stable concentrations of B cells in blood from morning to evening (Figure 6A). We provide genetic evidence that endogenous glucocorticoids act directly on B cells to regulate diurnal traffic patterns.

Our study of B cell^GRKO mice provides insight into the direct vs. indirect effects of glucocorticoids that were indiscernible in surgical and pharmaceutical models of glucocorticoid perturbation (Figure 2). Whereas ADX mice had increased numbers of B cell precursors in the spleen (Figure 2C), the phenotype of B cell^GRKO mice was subtler, with a B lymphocytosis that was primarily restricted to the blood (Figure 5). In addition to glucocorticoids, the adrenal glands produce mineralocorticoids and catecholamines, and their absence in ADX mice may affect lymphocyte accumulation and/or turnover in peripheral lymphoid tissues. On the other hand, ADX also disrupts the hypothalamic-pituitary-adrenal axis; dysregulated production of hypothalamic and/or pituitary hormones may contribute to GR-independent features of the ADX phenotype(34).

Tsc22d3 (or Gilz), a well-characterized glucocorticoid target gene, regulates B-cell activation and survival(35, 36). GILZ KO mice exhibit B lymphocytosis(35) and increased concentrations of serum autoantibody(36). Other than increased numbers of circulating B cells, we found no evidence of generalized B lymphocytosis or pathology in B cell^GRKO mice. GR-deficient B cells exhibited a 50% reduction in Gilz transcripts (Figure 4A), indicating that residual, GR-independent Gilz expression prevents the pathology associated with Gilz ablation.

A fundamental tenet of immunology holds that recirculation of B cells and T cells through lymphoid tissues promotes interactions between antigens and rare antigen-specific lymphocytes. Our adoptive transfer studies indicate that intrinsic GRs play a minimal role in the recirculation of B cells through secondary lymphoid tissues, but may drive a distinct mode of B cell migration specifically through bone marrow (Figure 6B, C). Since bone marrow is not subject to lymphatic drainage(37), we hypothesize that migration through bone marrow and secondary lymphoid tissues represent independent modes of recirculation, and that GR-mediated regulation of CXCR4 provides a mechanism for rapid exchanges of B cells specifically between blood and bone marrow. Whereas B-cell activation is normally associated with secondary lymphoid tissues, recent studies suggest that bone marrow can support antigen-specific humoral responses. Recirculating B cells localize to perivascular spaces in bone marrow populated with T cells and dendritic cells(38), and mature B cells in bone marrow are capable of responding to blood-borne, multivalent antigens(39, 40). We speculate that the shortage of mature B cells in the bone marrow of B cell^GRKO mice may explain the diminished IgG response to the multivalent antigen NP-Ficoll (Figure 7G); indeed, antibody responses to NP-Ficoll – but not the T-dependent antigen NP-KLH - are also diminished in B cell-CXCR4 KO mice, which lack mature B cells in bone marrow(19). Nonetheless, glucocorticoids likely regulate other aspects of B cell biology, and determining the mechanism behind reduced humoral responses in B cell^GRKO mice warrants further investigation.

Based on findings that bone marrow contributes to humoral immunity(39), and that glucocorticoids regulate diurnal exchange of mature B cells between blood and bone marrow, we propose the following hypothesis. As an animal transitions from resting to active periods of each day (morning for humans, evening for rodents), adrenal production of glucocorticoids increases, promoting circulating B cells to migrate to bone marrow. During the active period of the day, in which the animal is more likely to encounter pathogens and/or suffer injury in the process of foraging, avoiding predation, etc., GR-dependent maintenance of mature B cells in bone marrow ensures optimal humoral defense against T-independent antigens. As glucocorticoid activity subsides with the transition to the resting period of the day, reduced CXCR4 expression results in release of mature B cells from bone marrow pools into blood, allowing for migration into secondary lymphoid tissues.

Key Points.

• Glucocorticoid binding to Glucocorticoid Receptors on B cells upregulates CXCR4.

• Glucocorticoids control diurnal exchange of B cells between blood and bone marrow.

• B cell^GRKO mice exhibit impaired IgG response to T-independent(Type II) antigen.