B Cell–Intrinsic Expression of the HuR RNA-Binding Protein Is Required for the T Cell–Dependent Immune Response In Vivo

The development and function of cells such as B lymphocytes require finely tuned and dynamic changes in gene expression. These changes are instigated by cell-intrinsic and cell-extrinsic factors and controlled by a combination of transcriptional, posttranscriptional, and posttranslational mechanisms. Posttranscriptional mechanisms allow cells to rapidly alter protein expression by modulating the stability and/or translation of specific mRNAs, or by modifying mRNA-processing events, such as alternative polyadenylation and splicing. RNA-binding proteins (RBPs) are major posttranscriptional regulators of gene expression. RBPs bind their target sequences in 3′ untranslated regions or internal elements of mRNAs to positively or negatively regulate mRNA processing, stability, and/or translation in cellular context-dependent manners (1, 2). RBP function is regulated via changes in RBP abundance, subcellular localization, posttranslational modification, and interactions with other mRNA-binding factors, permitting quick and specific changes in gene expression in response to developmental cues or other stimuli (1). Although RBPs are appreciated to play a role in lymphocyte biology and function, their role in regulating B cell development and function remains poorly understood (3).

B lymphocytes are comprised of two main populations, B-1 and B-2 cells, that develop through distinct programs that each link cell-intrinsic and cell-extrinsic signals with cellular survival, proliferation, and continued differentiation. The larger B-2 cell population arises in bone marrow (BM) from common lymphoid progenitors starting shortly after birth and extending throughout life (4, 5). The smaller B-1 population develops mainly from fetal liver precursors during fetal and early neonatal development, although these B-1 B cells may continue to develop at a low level in adult BM (4, 6). Developmental stage-specific assembly and expression of IgH and Igκ or Igλ L chain (IgL) genes result in expression of BCRs on immature B cells (5). Depending on their Ag specificity, these BCRs signal gene expression changes to induce apoptosis or mediate differentiation of immature transitional B cells (7, 8). Transitional B-2 cells migrate to the spleen and differentiate into mature naive quiescent marginal zone or follicular (Fo) B cells that express BCRs and traffic throughout lymphatic tissues (4, 7). In contrast, transitional B-1 cells migrate to serous cavities where they become mature B-1 cells that continually proliferate and secrete Abs (4, 6).

B lymphocytes mediate protective humoral immunity through their ability to express cell surface BCRs and secrete Abs that bind Ags. B-2 cells recognize and respond to Ags through T cell–dependent or –independent mechanisms (4, 9). During a T cell–dependent germinal center (GC) immune response, Fo B cells are activated by encounter with Ag, usually on the surface of a professional APC, such as a dendritic cell (10). These B-2 cells induce their expression of surface molecules and cytokines that further activate Ag-primed CD4⁺ T cells and promote their differentiation into T follicular helper (Tfh) cells (11–13). Tfh cells in turn further activate B cells and direct them to become GC B cells (10, 12, 14–17). GC B cells rapidly proliferate, conduct IgH isotype switching and Ig somatic mutation, and alter gene expression to differentiate into short-lived high-affinity Ab-secreting plasma cells or long-lived memory cells (14). The GC reaction is critical for adaptive responses and affinity maturation of Abs against an enormous number and variety of Ags, as well as for more rapid secondary responses against previously encountered Ags (14). B-1 cells predominantly recognize and respond to Ags through T cell–independent mechanisms that occur outside of GCs and largely do not involve Ag-driven IgH isotype switching or Ig mutation, although T-dependent responses by B1 B cells do occur (4, 6). B-1 cells express BCRs that bind common pathogen epitopes and also spontaneously secrete Abs to protect against commensal and other opportunistic bacteria (4, 6). Throughout life, B-1 and B-2 cells function together to protect host organisms from universally encountered common foreign organisms and random unanticipated infections.

The development and function of B cells require exquisite regulation of gene expression to coordinate cellular survival, proliferation, and differentiation. The ubiquitously expressed protein HuR (also called Elavl1) controls posttranscriptional expression of many genes that mediate these cellular processes (18–22). Mice with germline or postnatal global deletion of HuR are not viable (23, 24), whereas global HuR deletion in adult mice causes increased apoptosis and loss of immature, but not mature B cells (24). Because global HuR deletion leads to death of mice within 14 d (24), the role of HuR in B cell function is not known. To determine the roles of HuR in B cell development and function, we established and analyzed mice with B lineage–specific deletion of HuR initiating in pro-B cells. These HuRΔ/Δ mice have reduced numbers of immature BM and mature splenic B cells, with only the former rescued by p53 inactivation, indicating that HuR supports B lineage cells through developmental stage-specific mechanisms or cellular processes. We discovered that HuR is required for normal numbers of splenic B-2 cells and peritoneal B-1 cells; however, HuR is not necessary for B cell development per se, enabling us to study roles of HuR in B cell function. Upon in vitro stimulation of splenic B cells, HuR is dispensable for B cell survival, isotype switching, and induction of GC B cell markers, and HuRΔ/Δ B cells exhibit only mild defects in proliferation and Ig secretion. In contrast, HuRΔ/Δ mice have dramatically low serum titers of all Ab isotypes and severely decreased GC B cells, GC structures, Tfh cells, and high-affinity Abs after immunization with a T cell–dependent Ag. These data indicate HuR expression in B lineage cells is essential for aspects of B cell activation unique to the in vivo environment. Consistent with this notion, we find in vitro stimulated HuRΔ/Δ B cells exhibit modestly reduced surface expression of costimulatory molecules whose expression is similarly decreased in humans with common variable immunodeficiency. Because in vivo immune responses require activated B cells to undergo more nuanced and sophisticated processes than can be recapitulated in vitro, HuRΔ/Δ mice provide a model to better understand B cell–intrinsic contributions to T cell–dependent immune responses in vivo.

All mice were on a 129S1/SvImJ and C57BL/6 mixed background and housed, bred, and used under pathogen-free conditions at the Children's Hospital of Philadelphia. HuR^flox/flox mice were provided by T. Hla (Weill Cornell Medical College, New York, NY) (24) and bred with Mb1-Cre mice (25), p53^flox/flox mice (26), or V_H147 (anti-GPI) IgH transgenic mice (27). For most breedings, Mb1-Cre males were bred to Cre-negative females because Mb1-Cre is expressed in the female germline (28). However, Mb1-Cre⁺HuR^flox/flox females were bred with HuR^flox/flox males to generate HuR^flox/− mice with one germline HuR-deleted allele. Animal husbandry and experiments were performed in accordance with national guidelines and regulations and were approved by the Children’s Hospital of Philadelphia Institutional Animal Care and Use Committee.

Peritoneal lymphocytes were obtained by peritoneal lavage with PBS, or cells were isolated from spleen or BM, as previously indicated (28, 29). Equal numbers of cells were incubated with live dead viability dye (Life Technologies) and then stained with Abs against surface Ags in PBS with 3% FBS. Following washing, cells were either analyzed directly or treated with Cytofix/Cytoperm buffer (BD Biosciences) and then stained with Abs against intracellular Ags. The Abs used are listed in Supplemental Table I. Samples were run on a FACSCalibur or LSR Fortessa cytometer (BD Biosciences) and analyzed with Flowjo software (Tree Star). Sorting was performed on a MoFlo Astrios (Beckman-Coulter). Annexin V assays were performed according to manufacturer instructions (BD Biosciences), except that annexin V Ab was used at a 1:100 dilution. BrdU incorporation assays were performed by incubating cells in medium containing 10 μM BrdU and staining as instructed (BD Biosciences).

Ninety-six–well polystyrene assay plate (Corning) medium binding was coated with goat-mouse Ig(H + L), 4-hydroxy-3-nitrophenylacetyl (NP)4-BSA, or NP33-BSA (Biosearch Technologies). After blocking with 2% BSA, serum and unlabeled isotype standard dilutions were applied. The NP standard was provided by G. Kelsoe (Duke University, Durham, NC). Detection of Ab isotype was achieved with appropriate goat anti-mouse–conjugated Ab (see Supplemental Table 1). Tetramethylbenzidine substrate (OptiEIA; BD Biosciences) was used to develop according to manufacturer’s instructions, and 2 M sulphuric acid was used to stop the reaction. Signal was read at 450 nm on a Molecular Devices Emax.

Splenic B cells were isolated using EasySep negative selection B cell isolation kits (Stem Cell Technologies), or Fo B cells were isolated by positive selection using biotinylated anti-mouse CD23 (B3B4; BD Biosciences) in conjunction with streptavidin microbeads (Miltenyi Biotec) on a LS column (Miltenyi Biotec). Isolated cells were labeled with CFSE (Life Technologies), as described (30, 31). Equal numbers of cells were stimulated for indicated time periods with 25 μg/ml LPS (0111:B4; Sigma-Aldrich) and 80 ng/μl mouse rIL-4 (R&D Systems), or 10 μg/ml anti-mouse CD40 (HM40-3; BioLegend) and 10 μg/ml F(ab′)₂ fragment goat anti-mouse IgM (Jackson ImmunoResearch Laboratories) with or without 50 ng/ml IL-21 (Shenandoah Biotechnology). Where not specified, cells were stimulated in RPMI 1640 supplemented with 10% heat-inactivated FBS, antibiotics, 50 μM 2-ME, 2 mM l-glutamine, 10 mM HEPES, 1 mM sodium pyruvate, and nonessential amino acids.

NP-OVA is the hapten NP conjugated to the OVA carrier protein. NP-OVA (Biosearch Technologies) resuspended in PBS was added to a solution of 10% aluminum potassium sulfate and precipitated by dropwise addition of potassium hydroxide. All solutions were sterilized, and precipitate was washed thoroughly with sterile PBS before injection of 50 μg NP-OVA in alum into the peritoneal cavity of 8-wk-old HuRΔ/Δ or HuRf/f mice. Injected mice were euthanized and analyzed at 9 or 14 d postimmunization.

Cells were resuspended in a Tween 20 containing lysis buffer and sonicated at intervals of 30 s on and 30 s off for 5 min at 4°C. Cells were incubated for 5 min on ice and then spun to remove insoluble material. A total of 30 μg lysate prepared under reducing conditions was loaded into each well of a NuPage 10% Bis-Tris gel (Life Technologies). Electrophoresed proteins were transferred to polyvinylidene difluoride, and membranes were blocked with Odyssey blocking buffer (Li-Cor) and incubated with anti-HuR Ab (3A2; Santa Cruz) or anti–β-actin (polyclonal; Sigma-Aldrich) for 1 h at room temperature or overnight at 4°C. After washing, blots were incubated with IRDye800 secondary Abs (LiCor) for 1 h at room temperature. Following washing, blots were scanned on an Odyssey infrared scanner (Li-Cor).

Genomic DNA was isolated as described (32). Total RNA was isolated using TRIzol reagent (Life Technologies) and DNase treated according to manufacturer directions (Promega), primed with random nonamer (New England Biolabs), and reverse transcribed with Moloney murine leukemia virus (NEB). Quantitative PCR and quantitative RT-PCR were performed with SYBR Green mastermix (Applied Biosystems) and run on an Applied Biosystems 7500 Fast machine. The primers for quantitative PCR and quantitative RT-PCR are found in Supplemental Table 1.

RNA immunoprecipitations (RNA-IPs) for HuR were performed as previously described (33). Per IP, 100 μl protein G Dynabeads (Life Technologies) were incubated with 15 μg anti-HuR (3A2; Santa Cruz) or 15 μg normal mouse IgG (Santa Cruz). B cells stimulated for 72 h with LPS plus IL-4 (6–9 × 10⁷ cells per IP pair) were lysed in polysome lysis buffer containing 0.5% Nonidet P-40 supplemented with protease inhibitor (Roche) and RNase inhibitor (NEB). Half of each sample was added to IgG or HuR-coated Dynabeads and incubated with shaking for 2 h at 4°C. After washing of beads, RNA was extracted using TRIzol as for total RNA.

Cells were lysed in urea buffer (34), protein concentration was measured via micro bicinchoninic acid assay (Thermo), and peptides were prepared. A total of 15 × 10⁶ stimulated B cells at the start of stimulation yielded ∼0.25 mg total protein. The proteins were digested in solution with trypsin (protein:trypsin = 50:1) overnight, and the tryptic peptides were desalted via SepPak (50 mg SepPak; Waters). Tryptic digests were lyophilized and dissolved in solvent A (2% acetonitrile, 5 mM ammonium formate, pH 10). A total of 48 μg peptides was injected to H-class HPLC (Waters) and were separated in C18 column at 300 μL/min, 39-min gradient (ZORBAX 300 Extend-C18, 2.1 × 100 mm, 3.5-micron 300A, P.N. 761775-902). The separated peptides were collected in 96-well plates for 39 fractions, and 12 combined fractions were lyophilized for liquid chromatography/mass spectrometry/mass spectrometry (LC/MS/MS) analysis. The peptides were dissolved in 0.1% trifluoroacetic acid/water and analyzed by LC/MS/MS on a hybrid LTQ Orbitrap Elite mass spectrometer (Thermofisher Scientific) coupled with a nanoLC Ultra (Eksigent). Peptides were separated by reverse-phase HPLC on a nanocapillary column, 75 μm ID × 15 cm Reprosil-pur 3 μm (Maisch) in a Nanoflex chip system (Eksigent). Mobile phase A consisted of 0.1% formic acid (Thermofisher Scientific) and mobile phase B of 0.1% formic acid/80% acetonitrile. Peptides were eluted into the mass spectrometer at 300 nl/min with each reverse-phase LC run comprising a 90-min gradient from 10 to 25% B in 65 min, 25–40% B in 25 min, followed by column re-equilibration. The mass spectrometer was set to repetitively scan m/z from 300 to 1800 (R = 240,000 for LTQ-Orbitrap Elite), followed by data-dependent MS/MS scans on the 20 most abundant ions, with a minimum signal of 1500, dynamic exclusion with a repeat count of 1, repeat duration of 30 s, exclusion size of 500 and duration of 60 s, isolation width of 2.0, normalized collision energy of 33, and waveform injection and dynamic exclusion enabled. Fourier transform mass spectrometry full-scan automatic gain control target value was 1e6, whereas MSn AGC was 1e4, respectively. FTMS full-scan maximum fill time was 500 ms, whereas ion trap MSn fill time was 50 ms; microscans were set at 1. Fourier transform preview mode, charge state screening, and monoisotopic precursor selection were all enabled with rejection of unassigned and 1⁺ charge states.

Whole proteomes were analyzed together in MaxQuant version 1.5.1.2, using the Uniprot complete mouse reference proteome, including isoforms (updated April 9, 2015) and common laboratory contaminants with a minimum peptide length of 7 aa and 1% false discovery rate; requantify and match between runs were turned off. Label-free quantification was used to identify mass spectra (MS) counts for relative abundance.

Except where otherwise indicated, p values were generated by two-tailed unpaired Student t test using Prism (GraphPad Software). Error bars for all figures indicate the SE.

To identify roles of the HuR RBP in B lymphocyte development and function, we generated and analyzed Mb1Cre:HuR^flox/flox (HuRΔ/Δ) mice with B lineage–specific deletion of the HuR gene initiating in pro-B cells. We performed cell counting and flow cytometry analyses for B cell developmental stage-specific markers on BM and spleen cells of 4- to 6-wk-old HuRΔ/Δ mice and littermate HuR^flox/flox (HuRf/f or wild-type [WT]) mice. Compared with control mice, HuRΔ/Δ mice have 1.4-fold and 1.9-fold reduction in the numbers of B lymphocytes in their BM and spleen, respectively (Fig. 1A). Although HuRΔ/Δ mice have higher than normal numbers of BM pro-B cells, they have reduced numbers of BM pre-B cells and BM B cells that represent immature B cells and recirculating mature B cells (Fig. 1B). We also evaluated cell surface expression of CD43, BP-1, and CD24 to divide BM B-2 cells into subpopulations referred to as Hardy fractions (35). Although HuRΔ/Δ mice have normal numbers of cells in each of the pro-B cell subpopulations (Hardy fractions A, B, C, and C′), they have fewer cells in each of the pre-B cell and mature B cell subpopulations (D, E, and F) (Fig. 1C). The spleens of HuRΔ/Δ mice contain reduced numbers of T1 transitional, T2 transitional, marginal zone, and Fo B cells (Fig. 1D). HuRΔ/Δ mice have normal ratios of Igκ⁺ to Igλ⁺ B cells in their BM and spleen (data not shown). We detected comparable substantial deletion of floxed HuR exons in pro-B, pre-B, and mature B-2 cells of HuRΔ/Δ mice, and confirmed substantial loss of HuR protein in HuRΔ/Δ mature B cells (Supplemental Fig. 1A, 1B). Although Cre expression can negatively affect development independent of gene deletion (36, 37), we found no differences in numbers of mature B cells or B cells at each developmental stage in Mb1-Cre mice as compared with WT controls (Supplemental Fig. 1C). These data indicate that, whereas HuR is not necessary for BM B-2 cell differentiation per se, HuR is required to generate normal numbers of B lineage cells at each developmental stage beyond the pro-B cell stage.

The decreased numbers of B-2 lineage cells at all developmental stages beyond the pro-B cell stage could reflect important roles of HuR in promoting IgH gene assembly, proliferation, and/or survival during B cell development. To investigate potential roles of HuR in these processes, we first generated and analyzed HuRΔ/Δ mice expressing the V_H147 IgH transgene (IgHTg) that blocks IgH gene assembly and enables formation of pre-BCRs to signal proliferation and survival in the absence of IgH gene recombination (27). The numbers of BM pre-B cells and B cells in the BM and spleens were all lower in 4- to 6-wk-old IgHTg:HuRΔ/Δ mice relative to age-matched control IgHTg:HuRf/f mice (Fig. 2A). Because these differences were comparable to those between HuRΔ/Δ and HuRf/f mice (Fig. 1A), our data indicate that IgH gene assembly is grossly normal in HuRΔ/Δ mice. We next compared cell cycle profiles of pro-B cells and the most proliferating immature B cell subpopulation (fraction C) between HuRΔ/Δ and HuRf/f mice. By combining cell surface staining of B cell developmental markers with DNA content analysis, we detected similar percentages of pro-B cells and fraction C cells in S phase between HuRΔ/Δ and control HuRf/f mice (Fig. 2B). Whereas these data suggest normal proliferation of immature HuRΔ/Δ B-2 cells, they cannot identify potential changes in which cells spend proportionally more time in each cell cycle phase. Thus, we also quantified immature and mature splenic B-2 cell populations in HuRΔ/Δ and HuRf/f mice at 12–14 wk of age when the mouse B cell compartment is replete (38), reasoning that the number of splenic B-2 cells in HuRΔ/Δ mice should reach the number of cells in WT mice at this age if HuRΔ/Δ B cells merely transit the cell cycle more slowly. We found that B cell numbers in the BM and spleens of these older HuRΔ/Δ mice are not restored to normal (Fig. 2C), indicating that the reduced number of splenic HuRΔ/Δ B cells is not caused by decreased proliferation of immature HuRΔ/Δ B-2 cells. Next, we assayed apoptosis of pro-B and pre-B cells in HuRΔ/Δ and HuRf/f mice by performing flow cytometry for B cell developmental stage-specific surface markers and the annexin V protein that is expressed on apoptotic and dead cells. We detected a near significant increase in the fraction of annexin V⁺ pre-B cells in HuRΔ/Δ mice as compared with HuRf/f mice (Fig. 2D), consistent with a role for HuR in promoting survival of immature B cells. Because HuR has been proposed to counter proapoptotic p53 signaling in hemopoietic stem cells and early lymphocyte progenitors (24), we generated and analyzed Mb1Cre:p53^flox/floxHuR^flox/flox (p53Δ/Δ:HuRΔ/Δ) mice and control Mb1Cre:p53^flox/flox (p53Δ/Δ) mice. We detected no significant differences in the numbers of BM pro-B cells and pre-B cells between p53Δ/ΔHuRΔ/Δ and p53Δ/Δ mice (Fig. 2E), indicating that HuR promotes survival of pre-B cells by countering p53-dependent proapoptotic signals. Notably, deletion of p53 in HuRΔ/Δ B lineage cells did not restore to normal the numbers of mature splenic B-2 cells (Fig. 2E), despite inactivation of p53 rescuing immature B-2 cell numbers. Collectively, these data indicate that HuR supports B-2 lineage cells via developmental stage-specific mechanisms or cellular processes that antagonize p53-dependent elimination of immature, but not mature B-2 cells.

Despite the reduced numbers of immature and mature B cells in HuRΔ/Δ mice, we found that HuR is not necessary for the development of mature splenic B cells. Thus, we investigated roles of HuR in B cell function beginning with tractable in vitro approaches to monitor survival, proliferation, IgH isotype switching, and gene expression changes during B cell activation. We isolated total splenic B cells from HuRΔ/Δ and HuRf/f mice and incubated equal numbers of cells with LPS and IL-4, which mimics T cell–independent activation of B cells. Cell counting showed that HuRΔ/Δ cells expanded less than HuRf/f cells after 72 h of stimulation (Fig. 3A). Although Cre expression can antagonize cell growth (39, 40), Mb1Cre⁺ B cells expand normally following addition of LPS and IL-4 (Supplemental Fig. 2A), suggesting that in vitro stimulated HuRΔ/Δ B cells exhibit increased apoptosis and/or reduced proliferation. To determine whether LPS- and IL-4–stimulated splenic HuRΔ/Δ B cells exhibit increased apoptosis, we labeled HuRΔ/Δ and HuRf/f cells with an amine-reactive viability dye to identify dead cells. We detected similar frequencies of dead cells in HuRΔ/Δ and HuRf/f B cell cultures after 72-h stimulation and a lower frequency of dead HuRΔ/Δ cells after 96-h stimulation (Fig. 3B). These data indicate that HuRΔ/Δ B cells do not exhibit increased apoptosis, but instead survive better than HuRf/f B cells after in vitro stimulation with LPS and IL-4. To determine whether splenic HuRΔ/Δ B cells exhibit reduced proliferation, we labeled HuRΔ/Δ and HuRf/f B cells with the fluorescent dye CFSE, which is diluted 50% by each round of cell division, before 72-h incubation with LPS and IL-4. We detected fewer cell divisions in HuRΔ/Δ cultures (Fig. 3C), indicating that LPS- and IL-4–stimulated HuRΔ/Δ B cells exhibit reduced proliferation. As an independent means to monitor proliferation, we measured incorporation of BrdU into replicating DNA combined with DNA content staining to quantify cells in each cell cycle phase. We found that fewer HuRΔ/Δ cells entered S phase during the 45-min BrdU pulse (Fig. 3D), providing further evidence that HuRΔ/Δ B cells stimulated by LPS and IL-4 exhibit reduced proliferation. To monitor IgH isotype switching, we measured surface expression of IgG1, because LPS and IL-4 promote IgH class switch recombination (CSR) predominantly to this isotype. After 72 h of stimulation, HuRΔ/Δ cultures harbor IgG1⁺ B cells at the same frequency as HuRf/f cells (Fig. 3E). Consistent with normal signaling upstream of CSR, we detected normal levels of noncoding germline transcripts for IgG1 and IgE in stimulated HuRΔ/Δ cells (Supplemental Fig. 2B) (41). The data from our analyses of HuRΔ/Δ and HuRf/f B cells following in vitro stimulation with LPS and IL-4 indicate that HuR is necessary for optimal proliferation of B cells, yet dispensable for survival and isotype switching to IgG1.

In addition to expressing cell surface BCRs, the ability to secrete Abs is crucial for B cell function. Thus, we used ELISAs to quantify IgM and IgG1 Abs secreted into the supernatants of HuRΔ/Δ and HuRf/f cells stimulated in vitro with LPS and IL-4. Despite equivalent frequencies of IgM⁺ and IgG1⁺ B cells in HuRΔ/Δ and HuRf/f cultures, we detected modestly lower levels of IgM and IgG1 in supernatants from HuRΔ/Δ stimulations (Fig. 4A). Accounting for differences in cell numbers arising from impaired proliferation of HuRΔ/Δ B cells, these in vitro stimulated HuRΔ/Δ B cells secrete normal amounts of IgG1, but 50% less IgM than HuRf/f B cells (Fig. 4A). Membrane-bound and secreted IgH chains are generated from mRNAs that differ in 3′ translated sequences, with only membrane-bound mRNA forms encoding a transmembrane domain (Fig. 4B) (42, 43). Alternative polyadenylation controls the relative abundance of membrane-bound versus secreted transcripts (42, 43). Because HuR can regulate pre-mRNA processing (44), we investigated whether HuRΔ/Δ B cells normally generate mRNAs encoding secreted Abs. We used distinct primer sets to amplify mRNAs encoding the membrane-bound or secreted isoforms of IgM after 72 h of stimulation. We found that the levels of secreted IgM mRNAs were reduced, whereas levels of membrane-bound IgM mRNAs were unchanged in HuRΔ/Δ cells as compared with HuRf/f cells (Fig. 4C), reflecting the relative amounts of secreted and membrane-bound IgM in the cultures. These data show HuR promotes generation of alternatively polyadenylated IgH mRNAs and secretion of Abs following in vitro stimulation by LPS and IL-4.

We next assayed in vitro activation of HuRΔ/Δ B cells by conditions that mimic a T cell–dependent B cell response. For this purpose, we isolated total splenic B cells from HuRΔ/Δ and HuRf/f mice and incubated equal numbers of cells with anti-IgM and anti-CD40 with or without IL-21. To assess whether splenic HuRΔ/Δ B cells also exhibit reduced proliferation when stimulated by these conditions, we labeled cells with CFSE prior to incubation for 60 h with anti-IgM, anti-CD40, and/or IL-21. We observed an equivalent number of cell divisions in HuRΔ/Δ and HuRf/f cultures (Fig. 4D), indicating that HuRΔ/Δ B cells exhibit normal proliferation following in vitro stimulation under conditions that mimic T cell–dependent activation. We also noted that cell viability after stimulation was not decreased by HuR deletion (Fig. 4D). To determine whether HuR affects the induction of GC markers on HuRΔ/Δ B cells in vitro, we measured expression of the BCL6, CD95, and TACI proteins in unstimulated and stimulated cells. After stimulation for 60 h, we found no significant difference in expression of each of these GC B cell markers between HuRΔ/Δ and HuRf/f cultures (Fig. 4E, Supplemental Fig. 2F), revealing that HuRΔ/Δ B cells normally express GC markers following in vitro stimulation by conditions that mimic T cell–dependent activation. These data demonstrate that HuR expression in B-lineage cells is dispensable for the ability of naive B cells to become activated and GC-like in vitro.

To determine whether HuR has a role in humoral immunity in vivo, we first analyzed standing serum Ab levels in nonimmunized 8- to 10-wk-old HuRΔ/Δ and WT mice. We observed that serum levels of all Ig isotypes were reduced in HuRΔ/Δ mice (Fig. 4F), indicating that HuR expression in B lineage cells is necessary for optimal Ab production in vivo. Because B-1 B cells produce the majority of standing IgM and IgA serum titers, we assessed the effects of HuR deletion on these B lineage cells. We detected 75% fewer peritoneal B-1 B cells in HuRΔ/Δ mice (Fig. 4G), revealing that HuR expression in B lineage cells is required to support normal numbers of B-1 cells. We also quantified GC cells, plasma cells (PC), and switched memory (Sw-mem) B cells, which arise from T cell–dependent B cell responses and together generate the majority of standing IgG and IgE serum titers. We detected reduced numbers of GC (50-fold lower) and PC (2-fold lower) cells, and a trending reduction in Sw-mem (2.5-fold lower) B cells in HuRΔ/Δ mice (Fig. 5A), indicating that expression of HuR in B lineage cells is required to support normal numbers of Ab-secreting B-2 cells. Collectively, these data demonstrate that B-lineage intrinsic functions of HuR are necessary for normal numbers of Ab-secreting B-1 and B-2 cells and normal titers of standing Abs of all isotypes.

We next evaluated the role of HuR in T cell–dependent humoral immunity in vivo by testing the ability of HuRΔ/Δ mice to mount a B cell response to immunization with a T cell–dependent Ag. For this purpose, we injected 8-wk-old HuRΔ/Δ or WT mice with NP-OVA precipitated in alum. After 9 or 14 d, we quantified numbers of naive B cells, PCs, total GC B cells, and NP-specific GC B cells in immunized mice and nonimmunized littermate controls. Because some Abs generated from NP-OVA injection will recognize epitopes within OVA rather than NP, we measured total GCs as well as NP-specific GCs. The numbers of naive B cells in HuRΔ/Δ mice remained 50% lower than normal after immunization (Fig. 5B). WT mice mounted a robust humoral response, generating half a million or more total and NP-specific GC B cells by 9 and 14 d postimmunization (Fig. 5B). In contrast, HuRΔ/Δ mice were substantially impaired in their ability to mount a humoral immune response, generating only thousands of total and NP-specific GC B cells over the same period following immunization (Fig. 5B). Reflecting fewer GC B cells, the production of plasma cells was reduced 4.5-fold in HuRΔ/Δ mice relative to WT mice (Fig. 5B). We also performed ELISAs to quantify low- and high-affinity NP-specific Abs in sera of HuRΔ/Δ and WT mice at 9 or 14 d postimmunization. We detected 6-fold reduced levels of low-affinity IgG NP-specific Abs in HuRΔ/Δ mice at each time point assayed (Fig. 5C). Consistent with the requirement of GCs for robust affinity maturation, we found a 20-fold increase in high-affinity IgG anti-NP Abs in immunized WT mice between days 9 and 14, but only a 4-fold increase in HuRΔ/Δ mice over the same time (Fig. 5C). Furthermore, HuRΔ/Δ mice had reduced levels of IgM anti-NP Abs (Fig. 5C), consistent with the idea that low IgG Ab titers are not the result of a CSR defect in HuRΔ/Δ B cells. Notably, deletion of p53 in HuRΔ/Δ B lineage cells did not rescue GC numbers after immunization beyond those observed in mice with deletion of only p53 in B lineage cells (Fig. 5D). The lower than normal numbers of GC B cells after immunization of mice with deletion of only p53 in B lineage cells is consistent with the reported effect of p53 deletion on IgG2a switching (45). Although we cannot rule out the possibility that HuRΔ/Δ B cells undergo p53-independent death during an immune response, these in vivo data are consistent with our in vitro findings that HuR deficiency does not impair viability of activated B cells.

The compromised ability of HuRΔ/Δ mice to produce GC B cells and high-affinity Abs in vivo after immunization contrasts with the normal ability of HuRΔ/Δ B cells to become GC-like and make Abs in vitro. Because the formation of GC structures in lymphoid tissues is critical for differentiation of GC B cells and production of high-affinity Abs in vivo (14, 46), we conducted histology to evaluate GCs in the spleens of HuRΔ/Δ and WT mice. As measured by H&E staining, gross splenic architecture was normal in HuRΔ/Δ mice before and after immunization with NP-OVA precipitated in alum (Fig. 6A), consistent with the modest effects of HuR deletion on naive B cell pools. As measured by H&E staining and immunohistochemical staining for peanut agglutinin, which marks GC B cells (47), GCs were detectable in the spleens of immunized WT mice, but not of immunized HuRΔ/Δ mice or unimmunized mice of either genotype (Fig. 6A and data not shown). Because the light zone of the GC consists of rapidly dividing B cells (48), we also performed immunohistochemical staining for Ki67, which marks proliferating cells (49). Clusters of Ki67⁺ cells were detectable in the spleens of immunized WT mice, but not of immunized HuRΔ/Δ mice or unimmunized mice of either genotype (Fig. 6A and data not shown). These data indicate HuR expression in B cells is required to support the GC structures needed for humoral immunity in vivo.

The formation of GCs requires activated B cells to interact with activated CD4⁺ T cells at the periphery of lymphoid follicles, promoting migration of both cell types into follicles where they differentiate into GC B cells and GC Tfh cells, respectively (11, 50). Because the failure of activated B cells to functionally interact with activated CD4⁺ T cells prevents GC Tfh formation and ablates the in vivo GC response (11, 51), we conducted flow cytometry to quantify total (CXCR5⁺PD-1⁺) and mature (GC) Tfh cells (CXCR5^highPD-1^high) in the spleens of immunized HuRΔ/Δ and WT mice. We found spleens of immunized HuRΔ/Δ mice had 1.6-fold and 17-fold lower numbers of total and GC Tfh cells, respectively, than spleens of WT mice (Fig. 6B). These data indicate HuR expression in B cells is required for the presence of GC Tfh cells after immunization of mice with a T cell–dependent Ag.

Our GC B and GC Tfh cell data indicate HuR expression in B cells is critical for T cell–dependent immune processes requiring B–T cell interactions within the Fo milieu. We sought to identify potential mechanisms by which HuR expression in B cells supports normal GC structures and populations of GC B and GC Tfh cells. Because we cannot detect HuRΔ/Δ GC B cells in vivo, we studied in vitro stimulated HuRΔ/Δ and control B cells to find differential expression of proteins that might account for the failure of HuRΔ/Δ mice to mount a normal GC response. Using unbiased label-free whole proteome quantification, we identified 6132 proteins in both cells, but limited our inspection to 4678 proteins for which we obtained at least 8 total MS and had nonzero MS count values for 3 of 4 biological replicates (Supplemental Table II). We observed a greater number of proteins decreased in HuRΔ/Δ B cells than increased, consistent with the canonical role of HuR as promoting target gene expression (19, 22). HuRΔ/Δ B cells showed reduced expression of multiple proteins involved in chromatin modification and organization, suggesting roles of HuR-dependent chromatin regulation in B cell activation. HuRΔ/Δ B cells also had reduced expression of proteins involved in cytoskeletal remodeling, which could lead to impaired Ag presentation, cell migration, and/or cell-to-cell adhesion; however, our initial follow-up studies did not show a defect in any of these processes with HuRΔ/Δ B cells stimulated in vitro.

Because this proteomics approach lacks sensitivity to detect small magnitude changes in protein expression, we used a candidate-based approach to study expression of B cell factors known to regulate B–T cell interactions. Activated B cells interact with activated T cells via contacts between surface costimulatory receptors and ligands (15, 51, 52). Using flow cytometry to quantify surface expression of costimulatory molecules known to control B cell activation and/or B–T cell interactions, we found HuRΔ/Δ B cells showed reduced expression of the CD81, CD70, and CD86 proteins (Fig. 6C) (29, 52–58). We then used RNA immunoprecipitation to show HuR binds to CD81, CD70, and CD86 transcripts in B cells activated in vitro (Fig. 6D), consistent with the notion that HuR directly regulates expression of these costimulatory molecules in activated B cells. Impaired expression of CD70, CD81, or CD86 has been observed on B cells of humans with common variable immune deficiency (CVID) (59–61). The subtle reduction in CD70 and CD86 expression on HuRΔ/Δ B cells mimics the modest lower levels of these molecules on B cells from CVID patients (60, 61). Although these clinical data from CVID patients suggest lower costimulatory factor expression may be sufficient to impair the GC response of HuRΔ/Δ mice, it is equally plausible that HuR controls expression of additional proteins, including those identified in our proteomics screen, that together with these costimulatory factors promote the GC response in vivo.

Posttranscriptional regulation of gene expression contributes to the dynamic and precise regulation of intracellular processes and communication among cells. In this study, we set out to determine roles of the ubiquitously expressed HuR RNA-binding protein in B cell development and function by analyzing mice with B lineage–specific HuR deletion. Deletion of HuR initiating in pro-B cells results in lower numbers of B cells at each developmental stage beyond the pro-B cell stage, including the GC and PC B cell populations that arise from Fo B cells that encounter and recognize Ag. Mature HuRΔ/Δ B cells display functional defects in vitro when stimulated under conditions that mimic a T cell–independent response, but not by conditions that mimic a T cell–dependent GC immune response. In contrast with normal ability of HuRΔ/Δ B cells to become GC-like in vitro, HuRΔ/Δ B cells exhibit profound functional deficiencies in vivo, generating low levels of Abs and failing to participate in a T cell–dependent humoral immune response. As we were submitting our work, another group published that a different strain of mice with B lineage–specific HuR deletion is defective in the GC reaction and Ab generation in response to T cell–dependent and –independent Ags (62). They concluded HuR loss results in defective mitochondrial metabolism that leads to large amounts of reactive oxygen species (ROS) and resultant death of activated B cells in vivo (62). Their conclusion was based on data showing in vitro stimulated HuR-deficient B cells exhibit increased ROS and death and impaired proliferation and isotype switching, which were all restored to normal upon addition of sodium pyruvate or other ROS scavengers to the culture media (62). In contrast, HuR-deficient B cells from our mouse strain exhibit normal apoptosis and isotype switching during in vitro stimulation regardless of whether the media has sodium pyruvate (Supplemental Fig. 2C, 2D). Moreover, the modest in vitro proliferation defect of our HuR-deficient B cells was equivalent in the presence or absence of sodium pyruvate (Supplemental Fig. 2C). Although the reason for identical in vivo but different in vitro results between these two studies is unclear, possibilities include differences in experimental conditions and distinct roles for HuR on a 129SvEv versus a C57BL/6 background. Because our in vitro stimulated HuRΔ/Δ B cells survive, isotype switch, and induce expression of GC B cell markers normally under conditions that mimic a T cell–dependent response, altered B cell metabolism most likely is not the major cause of the impaired GC reaction and T cell–dependent immune response of our HuRΔ/Δ mice. We cannot rule out that the ∼50% lower than normal numbers of HuR-deficient mature B cells in either mouse strain might contribute to the more substantial reduction in GC B cells; however, we consider this possibility unlikely. Finally, although our data indicate HuR expression in B cells is critical for normal B and T cell interactions during a T cell–dependent immune response, they cannot distinguish among roles of HuR in promoting initiation of the GC B cell fate, maintaining survival of GC B cells, or directing functional communication between GC B cells and Tfh cells. Future analyses of both mouse strains lacking HuR in B cells will be needed to elucidate the B cell–intrinsic mechanisms by which HuR is essential for supporting normal humoral immunity in vivo.

Our analyses of HuRΔ/Δ and HuRΔ/Δ:p53Δ/Δ mice indicate that HuR supports B lineage cells through developmental stage-specific mechanisms or cellular processes that antagonize p53-dependent elimination of immature, but not mature B cells. A role for HuR in promoting survival of immature B-2 lineage cells had been suggested from analysis of adult mice with global HuR deletion (24). In contrast to our findings that B lineage–specific HuR deletion results in lower numbers of immature and mature B cells, T lineage–specific inactivation of HuR results in higher numbers of immature thymic T cells, but decreased numbers of mature splenic T cells (63). These findings indicate that HuR has lineage-specific and developmental stage-specific roles in supporting normal numbers of B and T cells. The reduced numbers of naive splenic HuRΔ/Δ B-2 cells in older mice and in the situation in which p53 inactivation restores immature B-2 cell numbers suggest that HuR has B lineage–intrinsic functions in controlling homeostasis of naive B-2 cells. In this context, the decreased numbers of peritoneal HuRΔ/Δ B-1 cells may reflect a role for HuR in promoting homeostasis of naive B-1 cells. The BLyS family cytokine BAFF secreted from fibroblastic reticular cells and radiation-resistant stromal cells promotes B-2 cell homeostasis by signaling through BAFF-R, TACI, and BCMA receptors expressed on B cells (4, 64–66). We detected normal expression of TACI on stimulated and unstimulated cells of both genotypes; however, HuR might regulate expression of BAFF-R or BCMA, or their downstream signaling factors, such as the mTORC2 complex (67). Although evidence suggests that the BLyS family cytokine APRIL secreted from peritoneal macrophages promotes B-1 cell homeostasis by signaling through heparin sulfate proteoglycans expressed on B cells (68), the cytokines and their receptors on B cells that regulate homeostasis of B-1 cells are less well understood than for B-2 cells. Identifying HuR mRNA targets in splenic B-2 and peritoneal B-1 cells may yield novel molecular insight into the molecular factors and mechanisms that govern B cell homeostasis.

Our analysis of HuRΔ/Δ mice demonstrates expression of HuR in B lineage cells is critical for normal Ab levels in nonimmunized mice. The reduced serum titers of all Ab isotypes in HuRΔ/Δ mice are consistent with the reduced numbers of peritoneal B-1 cells and GC, PC, and Sw-mem B-2 cells. However, the impaired ability of HuRΔ/Δ B cells to produce mRNAs that encode the IgH chains of secreted Abs suggests that decreased Ab secretion most likely contributes to the low level of circulating Abs. HuR may regulate alternative polyadenylation of IgH mRNAs by binding IgH pre-mRNAs and directly regulating pre-mRNA processing at the level of splicing or polyadenylation, which are in competition at the IgH locus (43). This role of HuR in mature B cells would be consistent with HuR regulating the splicing of many transcripts in other cells (22, 44). Alternatively, HuR might regulate the expression of a mRNA processing factor that acts on IgH mRNAs. We could not detect HuR bound to IgH pre-mRNAs in normal mature B-2 cells, whereas in HuRΔ/Δ B-2 cells we did not observe decreased expression of the ELL2, PTB, hnRNPF, or active XBP1 proteins (data not shown), which have been implicated in the switch between membrane and secreted forms of IgH mRNAs (42, 43, 69–72). Because the trans-acting factors and molecular mechanisms that regulate the switch from membrane to secreted forms of IgH mRNAs remain to be elucidated, the identification of mRNAs to which HuR binds in mature B cells could provide new insights into how posttranscriptional changes in gene expression promote Ab secretion.

Our analyses of HuRΔ/Δ B cells during in vitro stimulation and in vivo following immunization indicate B cell–intrinsic functions of HuR are required for GC structures, GC B cells, high-affinity Abs, and GC Tfh cells each to be detectable at appreciable levels. Because HuRΔ/Δ mice lack HuR specifically in B lineage cells, any defects in Tfh cells are attributable to failure of B cells to appropriately coordinate gene expression programs during the immune response. In vivo GC responses require intimate codependent interactions between B and T cells (16, 17, 51). B cells traffic in and out of the follicle with defined kinetics to encounter Ag and subsequently engage Ag-primed CD4⁺ T cells (10, 12). Activated B cells must physically interact with Ag-primed CD4⁺ T cells to induce pre-Tfh cells to migrate into the follicle and differentiate into mature Tfh cells that provide survival and differentiation signals essential for generating GC B cells and high-affinity Abs (11, 15, 50, 51). B cells drive contact with CD4⁺ T cells through B cell surface MHC II molecules loaded with processed Ag fragments and through costimulatory proteins expressed on B cells (12, 51, 73). In addition to physical interactions with CD4⁺ T cells, B cells enhance Tfh differentiation by secreting IL-6 and other cytokines (13, 74). We found that HuRΔ/Δ B cells exhibit normal expression of chemokine receptors that control B cell trafficking, MHCII proteins, and IL-6 mRNA (Supplemental Fig. 2E and not shown); however, we did not confirm chemokine signaling and Ag processing are normal in HuRΔ/Δ B cells. In contrast, we found that activated HuRΔ/Δ B cells had reduced expression of the CD81, CD70, and CD86 costimulatory molecules (29, 52–58). HuR binds to CD81, CD70, and CD86 transcripts in stimulated WT cells, consistent with the notion that HuR directly regulates expression of these molecules. Because CD86 binding to CD4⁺ T cells also activates B cell intracellular signals to promote Ab secretion (75), this reduced CD86 expression might impair humoral immunity through distinct mechanisms. Impaired expression of CD81, CD70, or CD86 has been observed on B cells of humans with CVID (59–61). Notably, the subtle differences in CD86 and CD70 expression on HuRΔ/Δ B cells mimic the modest reduction of these costimulatory molecules on B cells from CVID patients (60, 61). Although, to our knowledge, genetic mutations or polymorphisms of the HuR gene have not been associated with CVID, identifying HuR target mRNAs in naive B-2 cells and Ag-activated GC B cells could lead to greater understanding and perhaps improved therapies for humans with impaired T cell–dependent immune responses.

C.H.B. is a consultant for Regeneron Pharmaceuticals. The other authors have no conflicts of interest.