These studies were approved and conducted in accredited facilities in accordance with The Home Office UK Animals (Scientific Procedures) Act 1986 (Home Office license number PPL 70/7102).

C57BL/6 (H-2b) mice were purchased from Harlan Laboratory. CD19^Cre mice were purchased from the Jackson Laboratory. The dominant negative (dn) RARα mice have been previously described (25). We generated the dnRAR^fl/flCD19^Cre/+ (heterozygous form for CD19^Cre), in a manner similar to previous reports (24). The CD19^Cre/+ control mice are haplosufficient and had a normal B cell compartment as previous described (24). To evaluate T-independent immune responses, mice aged 8–10 weeks were immunized with 50 µg of the hapten (Tri-4-hydroxy-3-nitrophenyl) acetyl (TNP)-Ficoll (Biosearch Technology). To study T-dependent humoral immune responses, 50 µg of TNP-KLH (Biosearch Technology) adsorbed in alum was injected i.p. in a volume of 200 µl. These studies were approved and conducted in accredited facilities in accordance with the United Kingdom Animals (Scientific Procedures) Act 1986 (Home Office license no. PPL 70/7102). All animals were cohoused and maintained in a specific pathogen-free facility at King’s College London.

To analyze B cells, single-cell suspensions of lymphocytes were prepared from spleens by mechanical disruption in DPBS followed by passing the cells through a 70 µm cell strainer. Cells were collected by centrifugation (5 min, 500 × g, 4°C) and red blood cells were lysed (2 min, 37°C) using RBC Lysis buffer (BioLegend). Total number of cells and cell viability were determined using a hemocytometer and trypan blue. For isolation of peritoneal cells, the peritoneal cavity was flushed with 5 ml warm (37°C) PBS, 2% BSA, 2 mM EDTA, 0.02% sodium azide, and 10 U/ml Heparin.

Antibodies against the following antigens were used: anti-mouse B220 (clone 6B2), CD38 (clone 90), IgM (clone 11–41), CD24 (clone M1/69), CD21/35 (clone 7E9), CD23 (clone B3B4), and IgD (clone 11-26c) from BD Biosciences. Anti-mouse CD5 (clone 53-7.3) and CD11b (clone M1/70) from Biolegend. Anti-mouse CD11a (integrin αL, clone M17/4), CD18 (integrin β2, clone M18/2), CD29 (integrin β1, clone HMb1-1), and CD49d (integrin α4, clone R1-2) from eBioscience. Murine cells were resuspended in DPBS + 3% FCS and stained for 30 min at 4°C with different anti-mouse antibodies in the presence of anti-mouse TruStain fcX (Biolegend). Control samples were labeled with matched isotype control antibodies. The samples were resuspended in Sytox Blue Dead Cell Stain (1:1,000 in DPBS; Invitrogen) before acquisition. Samples were acquired by flow cytometry (BD LSRFortessa cell analyser, Becton Dickinson) with FACSDiva software v6.2 (Becton and Dickinson) and data were analyzed using FlowJo (Tree Star) software.

B cells were purified from the spleens of dnRARαCD19^Cre or littermate control mice by negative selection using CD43 microbeads (Miltenyi Biotech) according to the manufacturer’s instructions. Purified splenic B cells were stained with anti-CD21/35, anti-CD24, anti-CD23, and anti-IgM as described previously (26). Sytox was added to exclude dead cells. Cells were sorted by BD FACSAria II (BD Biosciences). In some cases, MZ B cells were cultured with 1µM of LE135 (Tocris Biosciences), an antagonist for RAR receptors, for 24 h as previously described (27).

Total RNA was isolated from purified B cells using MiniKit RNeasy columns (QIAGEN) with a DNase-I treatment step. One microgram of DNA-free RNA was reverse transcribed to cDNA using Omniscript RT (QIAGEN). TaqMan gene expression assays containing FAM dye-labeled TaqMan MGB probe was used for mouse S1pr1 (Mm02619656_s1), RARα (Mm01296312_m1), RARβ (Mm01319677_m1), and RARγ (Mm00441091_m1) in multiplex with primer-limited assays for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) endogenous control containing VIC/MGB probes. Real-time quantification was performed using TaqMan gene expression master mix on a Bio-Rad CFX96 optical reaction module on a C1000 thermal cycler. Data were analyzed using CFX Manager Software (Bio-Rad).

Tissue was then snap frozen in OCT Tissue-Tek freezing medium (Sakura). Cryostat sections of 7 µm were air dried, fixed with acetone, and washed in DPBS. Blocking was performed with 20% normal horse serum (PAA Laboratories Inc.) for 15 min at room temperature. Sections were incubated with biotin-conjugated anti-mouse IgM (clone 11–41, BD Biosciences), followed by streptavidin Alexa Fluor 594 (Life Technologies) or FITC-conjugated anti-mouse monocyte and macrophage (MOMA; Abcam). Negative controls were no primary antibody or stained with isotype-matched antibody. Images were acquired on an Olympus BX51 microscope using Micro-Manager software (Vale Laboratory).

Migration assays were performed as previously described (28). FACS-sorted MZ B cells were treated with 1 µM of LE135, or left untreated, before analysis using a chemotaxis assay. Transwell assays were performed in duplicate, using 100 nM for S1P and 250 nM for SDF-1 as a chemoattractant. Two mice were pooled in each assay, in three independent experiments. It was not possible to perform ex vivo chemotaxis assays with cells from dnRARαCD19^Cre (unpublished data).

For ELISA analysis, plates were coated with 10 mg/ml TNP-BSA overnight in PBS, blocked with PBS + 5% FBS, washed, and then 1:2,000 diluted serum added at serial 1:2 dilutions. Anti-TNP IgG3 (provided by Thomas J. Waldschmidt) and anti-TNP IgM (BD Bioscience) was included on each plate as a reference between plates and between experiments and used to generate a standard curve. Antibody levels were detected with AP-conjugated IgG3 and IgM (Southern Biotechnology Associates) and developed with 1 mg/ml para-nitrophenylphosphate (Sigma-Aldrich) in 0.05 mM sodium carbonate buffer.

Single-cell preparations were obtained from the spleen and the number of TNP-IgM, TNP-IgG₃, and TNP-IgG₁ antibody-secreting cells (ASCs) were determined by ELISPOT as previously reported (29, 30).

Results are expressed as mean ± SD. Comparisons between groups were performed using T-test for two groups, or one-way ANOVA and a Dunnett or Bonferroni posttest for more than two groups. Analyses were performed using GraphPad Prism software.

To evaluate the role of RA signaling in T-independent humoral responses, we first evaluated the expression of the different RAR isoforms in splenic B cell subpopulations. Figure 1A shows the gating strategy used to electronically sort splenic B cell subsets, T1, T2, MZ, and FO splenic B cells, for analysis of RNA expression levels of RAR isoforms (all >95% purity). Our results showed that these subsets of splenic B cells exclusively express RARα, but not the β and γ isoforms (Figure 1B).

To study the functional role of RARα in splenic B cell function, we used mice engineered to express a dominant negative form of the RARα (RARα403) targeted to the ROSA26 locus downstream of a loxP-flanked “stop” cassette. Upon breeding with mice expressing Cre recombinase (in this case CD19^Cre), RA signaling is abrogated in the CD19⁺ B cell compartment (hereafter denoted as dnRARαCD19^Cre). We have previously used this approach and demonstrated that dnRARα is expressed exclusively in B cells from dnRARαCD19^Cre mice and is fully functional (24).

We then characterized whether RA signaling in B cells has an impact on the development of splenic B cell subsets in vivo. We observed an increase in the percentage (Figures 1C,D) and absolute number of T2 B cells in the dnRARαCD19^Cre mice compared with littermate controls (Figure 1E). In contrast, we observed a reduction in the percentage and absolute number of MZ B cells in the spleen of dnRARαCD19^Cre mice compared to control mice (Figures 1C–E). We then analyzed the splenic architecture by histology in dnRARαCD19^Cre mice. The data shown in Figure 1F demonstrates a reduction in MZ B cells in the spleen of dnRARαCD19^Cre mice compared to littermate controls. In addition, no difference was found in MHC-II, CD80 or CD86 expression in B cells from dnRARαCD19^Cre and littermate controls (data not shown). Together, these data suggest that RA signaling in B cells is necessary for the development of MZ B cells.

It has been previously reported that the integrins α_Lβ₂ and α₄β₁ facilitate long-term retention of B cells in the MZ (31). We assessed whether RA signaling in B cells affects the expression of these integrins in MZ B cells or T2 B cells. Our results showed no difference in the expression of these integrins in any of the studied B cell subsets (Figures 2A–D). Another molecule involved in the migration of B cells to the MZ is S1P₁ (28). We evaluated S1P₁ expression on T2 and MZ B cells from dnRARαCD19^Cre mice and observed that it was drastically increased in T2 B cells (Figure 2E), but reduced in MZ B cells from dnRARαCD19^Cre mice compared with littermate controls (Figure 2F). Together, these results suggest that RA signaling is necessary to regulate the expression of S1P₁ in T2 and MZ B cells.

In order to evaluate whether RA signaling in MZ B cells affects migration to S1P, we sought perform chemotaxis assays using MZ B cells from dnRARαCD19^Cre. However, due to low cell numbers, it was not possible to perform ex vivo chemotaxis assays with cells from dnRARαCD19^Cre (unpublished data). Instead, we compared wild-type MZ B cells treated with LE135, an antagonist for RAR signaling, to untreated wild-type MZ B cells (27). First, we evaluated whether LE135 affects S1P₁ expression in B cells. As observed in T2 and MZ B cells from dnRARαCD19^Cre mice (Figures 2E,F), LE135 increased the expression of S1P₁ in T2 B cells (Figure 2G) whereas S1P₁ expression in MZ B cells treated with LE135 was reduced (Figure 2H). In addition, LE135 treatment drastically reduced the ability of MZ B cell to respond to S1P (Figure 2I), without affecting the chemotaxis to SDF-1 (Figure 2J). Together, these results suggest that RA signaling in B cells plays an important role in the expression of S1P₁ and induces S1P-mediated chemotaxis in MZ B cells.

RA signaling in B cells is important for IgA plasma cell differentiation and to maintain IgA titers in the gut (24). In addition, it has been proposed that peritoneal B cells migrate from the peritoneum to the lamina propria of the small intestine to contribute to IgA antibody levels in the gut (32, 33). Thus, we studied whether RA signaling in B cells could also affect the generation of peritoneal B cells. We observed a 60% reduction in the proportion of B1a and B1b B cells in the peritoneum of dnRARαCD19^Cre mice compared to controls (Figures 3A–D), which suggests that RA signaling is necessary in B cells for the generation or migration of B1 B cells.

The main function of B1 and MZ B cells is the T cell-independent production of antibodies against blood-borne particulate antigens (34, 35). Since RA signaling in B cells affects the frequency of MZ and B1 B cells, we analyzed whether the absence of RA signaling affects the T-independent humoral response. To address this, we immunized dnRARαCD19^Cre and control mice with TNP-Ficoll and collected serum at different time points to quantify anti-TNP antibodies by ELISA. We observed a decrease in both anti-TNP IgM and IgG3 titers in the dnRARαCD19^Cre mice compared to controls (Figures 3E,F). This reduction in Ag-specific antibody levels correlated with a reduction in anti-TNP IgM and anti-TNP IgG₃ ASCs (Figures 3G,H). In contrast, no difference was observed in the number of anti-TNP-IgG₁ ASC between dnRARαCD19^Cre and littermate control mice immunized with TNP-KLH and alum to induce a T-dependent humoral response (Figure 3I). In summary, these results suggest that RA signaling in B cells is important for the generation of T-independent humoral responses.

To assess whether RA signaling affects the development of B cells in the bone marrow, we quantified the percentage and absolute numbers of pre-B cells (B220⁺IgM^negIgD^neg), immature B cells (B220⁺IgM^hiIgD^neg), and mature B cells (B220⁺IgM^intIgD⁺) in dnRARαCD19^Cre mice, as previously described (29). Our results showed an increase in the percentage of immature B cells in dnRARαCD19^Cre mice compared with controls (Figure 4B). However, there was no difference in the absolute number of immature B cells (Figure 4E). In addition, neither the proportion nor the number of pre/pro B cells and mature B cells were affected by the abrogation of RA signaling in B cells (Figures 4A,C,D,F). These results indicate that B cell intrinsic RA signaling does not have an impact in the early development of B cells in the bone marrow.