The expression pattern of the transcription factor Arid3a is believed to be largely restricted to the B cell lineage within the hematopoietic system (8). We first determined the pattern of Arid3a expression by examining the ImmGen database (9) and could confirm that Arid3a is restricted to B cells with the exception of prominent expression also within granulocytes. Surprisingly, Arid3a was not strongly expressed in hematopoietic stem cells or progenitors, but instead showed upregulation within B cell progenitors resident in the bone marrow (Figure 1A). Specifically, Arid3a expression was upregulated through successive maturation steps from the pro-B to immature B cell stages. Interestingly, B1 cells did show only moderately increased expression of Arid3a.

We next wished to interrogate the in vivo function of Arid3a by genetic loss-of-function analysis. We generated a floxed allele of Arid3a by first removing the Lacz/Neo cassette of Arid3a^{tm1a(KOMP)Wtsi} by FLPE-mediated recombination to create an allele that contains Loxp sites flanking exon 4 (Figure S1A in Supplementary Material). This encodes for amino acids 237–261 that are crucial for Arid3a DNA binding (4), and additionally loss of exon 4 will result in an out of frame allele and a nonsense protein. We then crossed this floxed allele with the B cell-specific Cre line Cd79a^Cre (Mb1^Cre), which initiates deletion at the transition from the common lymphoid progenitor to the pro-B cell stage of development (10). We therefore created littermate cohorts of Arid3a^fl/fl (herein referred to as control mice), Arid3a^fl/+Cd79a^Cre/+ (herein referred to as heterozygous mice), and Arid3a^fl/flCd79a^Cre/+ (herein referred to as homozygous mice) and analyzed the B cell compartment in the bone marrow of these mice at 4–6 weeks of age. We found that loss of Arid3a influenced the relative abundance of the B1 (defined B220^loCD19⁺) populations in the bone marrow, with approximate 1.7-fold decreases in B1 cells in heterozygous versus control mice (p = 0.0047) and 1.8-fold decreases in homozygous versus control mice (p = 0.0006). In contrast, the B2 population (defined B220⁺CD19⁺) was expanded 1.7-fold in heterozygous versus control (p < 0.0001) and 2-fold increased in the homozygous versus control cohorts (p = 0.0006) (Figure 1B). We next determined which stage of B cell development was responsible for this increase in B2 cells. The first committed pro-B stage of B cell development (defined B220⁺CD19⁺IgM⁻IgD⁻c-Kit⁺CD25⁻) was increased 2.5-fold in homozygous mice upon loss of Arid3a (p = 0.038). Later stages of B cell development were also expanded. The pre-B population (defined B220⁺CD19⁺IgM⁻IgD⁻c-Kit⁻CD25⁺) including the large pre-B and small and pre-B subpopulations increased approximately 1.9-fold in both heterozygous and homozygous mice (control versus heterozygous p = 0.0001: control versus homozygous p < 0.0001). Immature-B (defined B220⁺CD19⁺IgM⁺IgD⁻) were increased 2-fold in the homozygous versus control mice (p = 0.038), whereas recirculating B cells (defined B220⁺CD19⁺IgM⁻IgD⁺) demonstrated dosage-dependent increases in cell number upon loss of Arid3a ranging from 2.1-fold (control versus heterozygous p = 0.0043) to 4.8-fold (control versus homozygous p = 0.0002) (Figure 1C; Figure S1B in Supplementary Material), resulting in an increase in bone marrow cellularity (Figure S1C in Supplementary Material).

Cd79a^Cre is well established as an efficient deleter of floxed alleles in B cells. We further confirmed this by sorting pro-B cells by fluorescence-activated cell sorting and assessing deletion of Arid3a at the mRNA level. Cells from heterozygous mice contained both wild-type and exon-4 deleted Arid3a, whereas cells sorted from homozygous mice only contained the deleted allele (Figure 1D). We further sequenced this deleted allele and confirmed that an out of frame mRNA is produced upon loss of exon 4 resulting in a protein with multiple premature stop codons (Figure S1D in Supplementary Material). In summary, loss of Arid3a leads to reductions in bone marrow B1 cell numbers and conversely a more noticeable expansion of all stages of B2 cell development.

The expansion of B cell numbers in the bone marrow could result in further abnormalities in peripheral B cell numbers. We analyzed control, heterozygous, and homozygous mice at 10–12 weeks of age and quantified the absolute cell numbers of various B cell subsets in the spleen. We observed a twofold expansion (p = 0.0001) in B2 cell numbers in homozygous versus control mice, but no difference in B1 cell numbers. This increase in B2 cells was dosage dependent as loss of one Arid3a allele also resulted in increases in absolute numbers of B cells (p = 0.0032) (Figure 2A; Figure S2A in Supplementary Material), with subsequent increases in spleen cellularity and CD3⁺ cell numbers (Figures S2B,C in Supplementary Material). We next wished to determine which subset was responsible for the increase in B2 cell number. Transitional 1 B cells (B220⁺CD19⁺IgM⁺CD23⁻) were increased in both heterozygous (p = 0.0024) and homozygous (p = 0.0027) versus control mice. Transitional 2 B cells (B220⁺CD19⁺IgM⁺CD23⁺) were increased up to 2.4-fold (p = 0.0036) in homozygous mice versus controls (Figure 2B). An alternative gating strategy incorporating CD93 also revealed increases in immature splenic populations (Figure S2D in Supplementary Material). Similarly, the number of follicular B cells (defined B220⁺CD21⁺CD21^loCD23⁺) was also increased in cell numbers and again appeared to be dosage sensitive to loss of Arid3a (control versus heterozygous p = 0.0022: control versus homozygous p < 0.0001). Marginal zone B cells (defined B220⁺CD21⁺CD21^hiCD23^lo) were also increased in homozygous versus control mice (p = 0.0095) (Figure 2C). Interestingly, surface IgM levels were decreased on follicular and marginal zone B cells in homozygous mice (Figure 2D). In summary, peripheral B cell numbers are expanded upon loss of Arid3a in similar fashion to B cell progenitors in the bone marrow.

Arid3a has recently been positioned as a key regulator of the B1 versus B2 lineage (5, 11). The reduction of B1 cells in the bone marrow indicated that this phenotype is also present upon conditional loss of Arid3a in the adult mouse. We therefore analyzed a key reservoir of B1 cells in the mouse, namely the peritoneal cavity. We observed a strong decrease in the relative abundance of B1a (CD19⁺B220^loCD5a⁺) cells in homozygous mice, being 5.2-fold (p = 0.0306) reduced relative to controls. B2 cells and B1b (CD19⁺B220^loCD5a⁻) cells were not significantly reduced in numbers (Figure 3A). These B1a cells were further confirmed to have a cell surface phenotype of IgM^hi, IgD^lo, and CD11b⁺ (Figure 3B). An alternative gating strategy to define B1 cells (CD19⁺CD11b⁺) produced similar results (Figures S3A,B in Supplementary Material). Finally, we confirmed that peritoneal lymphocytes were deleted for Arid3a by sorting B1 and B2 cells and determining Arid3a deletion. Similar to the results from bone marrow pro-B cells, homozygous mice displayed efficient deletion of exon 4 of Arid3a (Figure 3C). We could therefore confirm that Arid3a is required for normal peritoneal B1a cell development.

The humoral immune response of mouse B cells has been reported to be influenced by the transcription factor Arid3a. We measured total plasma immunoglobulin levels in naïve mice and found significant decreases in IgM (p = 0.0003), IgA (p = 0.039), and IgG (p = 0.0001) in homozygous mice compared to controls (Figure 4A), similar to that reported in mice expressing dominant-negative Arid3a (7) and the very rare survivor mice from the germline deleted Arid3a strain (6). Within the IgG fraction, levels of IgG1, IgG2b, and IgG3 were all significantly reduced (Figure 4B). We did not observe any significant differences in heterozygous mice relative to control mice.

A consistent feature of mice with perturbed Arid3a function is reduced antibody titers against the hapten phosphocholine, which in C57/Bl6 mice is partially dependent on V(D)J rearrangements that have incorporated V_HS107.1.42 (12, 13). We assayed titers of IgM against phosphocholine and observed significant decreases in naïve homozygous mice versus that of controls (p = 0.0006) (Figure 5A). We next immunized mice with the antigen phosphocholine linked to Keyhole limpet hemocyanin (PC-KLH) emulsified in Alum to study germinal centre formation and specific T cell-dependent immune responses. We observed a reduction in anti-phosphocholine IgM responses 14 days after immunization (p = 0.0056) (Figure 5B). We next tested if this reduced anti-PC response was due to an inability to form germinal centers. Activated B cells (CD19⁺B220⁺IgD^loCD95⁺GL7⁺) could be readily detected in the spleen following immunization in all genotypes and were also present in normal numbers in the Peyer’s patches and mesenteric lymph nodes regardless of the Arid3a genotype (Figure 5C).

The Arid3a^{tm1a(KOMP)Wtsi} allele was imported from the KOMP resource and in vitro fertilized with sperm from the B6.Cg-Tg(ACTFLPe)9205Dym/J strain to remove the Frt flanked selection and reporter cassette to create Arid3a^fl/+ mice. Mice were then further crossed with the Cd79a^Cre/+ strain to create control and homozygous mice. All experiments used mice at 4–6 or 10–12 weeks of age and were approved by the Stockholm Regional Board for Animal Ethics, permit numbers N46/14, N164/14, and N111/11. Mice were regularly controlled by the Swedish Veterinary Authorities.

Polymerase chain reaction was performed after DNA extraction from ear biopsies of mice. The Arid3a^fl/fl and Arid3a^fl/+ DNA fragments were amplified with the primers: 5′-CTCCTTCCTCTTGTCTCCTGTGTGG-3′ (ttR) and 5′-TGCCTGTTGGAAGGATGATCTGG-3′ (F); and with the primers 5′-GAGATGGCGCAACGCAATTAATG-3′ (loxF) and 5′-GATAGCTCAGGAGCTTCCTCACACC-3′ (R). The Cd79a^Cre/+ fragment was amplified with the following primers: 5′-CCTTGCGAGGTCAGGGAGCC-3′ (WT reverse), 5′-GTCCTGGCATCTGTCAGAG-3′ (Cre reverse) and 5′-GGCTCTGACCCATCTGTCTC-3′ (common forward). Primers ttR and F amplified a 354-bp wild-type fragment and a 470-bp postFlp fragment. Primers loxF and R amplified a 253-bp floxed fragment. Primers WT reverse and common forward gave a 477-bp wild-type fragment, and primers Cre reverse and common forward gave a 500-bp Cd79a^Cre/specific product. PCR products were subsequently separated by electrophoresis through agarose gels and visualized by GelRed nucleic acid gel stain (Biotium).

Bone marrow from Femur and Tibia of the two hind legs of 4–6 weeks old mice, and spleen and peritoneal cavity from 10 to 12 weeks old mice was isolated. Tissues were then fractioned to single-cell suspensions and filtered through 100-µm cell strainers (Sigma-Aldrich) with phosphate-buffered saline (PBS) containing 0.5% fetal bovine serum (FBS). Erythrocytes were thereafter lysed with Ammonium-Chloride-Potassium Lysing Buffer. Single-cell suspensions of bone marrow, spleen, and peritoneal cavity were pre-incubated at a density of 100 × 10⁶ cells/ml with Fc Block (BD Biosciences), and cells were subsequently analyzed by a CyAn ADP (Beckman Coulter) flow cytometer with the following antibodies B220 (APC-Cy7 clone: RA3-6B2), anti-CD19 (Alexa Fluor 647 clone: eBio1D3), anti-CD25 (PE-Cy7 clone: PC61.5), anti-c-kit (PE clone: 2B8), anti-CD5 (PE clone: 53-7.3), anti-GL7 (APC clone: GL-7), anti-CD95 (PE-Cy7 clone: Jo2), anti-IgD (PerCP clone: 11-26c.2a), anti-IgM (Pacific Blue clone: Il/41), anti-CD23 (PE-Cy7 clone: B3B4), anti-CD21 (FITC clone: 7G6), and anti-CD11b (PerCPclone: M1/70). Immune cell populations were defined as follows: B-1 (CD19⁺B220^lo) or (CD19⁺CD11b⁺), B-1a (CD19⁺B220loCD5⁺) or (CD19⁺CD11b⁺CD5⁺), B-1b (CD19⁺B220^loCD5⁻) or (CD19⁺CD11b⁺CD5⁻), B-2 (CD19⁺B220⁺), pro-B (CD19⁺B220⁺IgM⁻IgD⁻c-Kit⁺CD25⁻), pre-B (CD19⁺B220⁺IgM⁻IgD⁻c-Kit⁻CD25⁺), immature-B cells (CD19⁺B220⁺IgM⁺IgD⁻) recirculating-B cells (CD19⁺B220⁺IgM⁻IgD⁺), Transitional-1 (CD19⁺B220⁺IgM⁺CD23⁻) or (CD19⁺B220⁺CD93⁺IgM⁺CD23⁻), Transitional-2 (CD19⁺B220⁺IgM⁺CD23⁺) or (CD19⁺B220⁺CD93⁺IgM⁺CD23⁺), marginal zone B cells (CD19⁺B220⁺CD21^hiCD23^lo), follicular B cells (CD19⁺B220⁺CD21^intCD23^hi) germinal center B cells (CD19⁺B220⁺IgD⁻GL7⁺CD95⁺). Absolute cell numbers were calculated by first counting live cells using a BioRad TC-20 cell counter based on trypan blue exclusion. Cells within a defined gated quadrant were divided by the cells contained in a large gate encompassing all live cells and excluding debris as determined by forward and side scatter properties from the flow cytometry. This ratio was then multiplied by the cell count to give the absolute cell number.

Pro-B cells from bone marrow and B1 and B2 peritoneal cavity cells of Arid3a^fl/+Cd79a^Cre/+ and Arid3a^fl/flCd79a^Cre/+ mice were sorted into PBS containing 0.5% FBS before RNA extraction with the RNeasy Mini Kit (Qiagen). cDNA was synthesized with High capacity RNA-to-cDNA Kit (Applied Biosystems) followed by PCR with primers with the following Arid3a oligonucleotides: 5′-TGGACCTTTGAGGAGCAGTT-3′ (Primer 1) and 5′-GATGGAGGTAGGCAGGTTGA-3′ (Primer 2). A 255-bp PCR product was amplified from the floxed Arid3a allele and a 182-bp product from the deleted Arid3a allele. The Hypoxanthine guanine phosphoribosyl transferase (HPRT) gene was used as a control with primers GGGGGCTATAAGTTCTTTGC and TCCAACACTTCGAGAGGTCC generating a 312-bp PCR product. PCR products were then separated by electrophoresis through agarose gels and visualized by GelRed nucleic acid gel stain.

For immunization with PC-KLH (Biosearch Technologies Inc.), 100 µg of PC-KLH was precipitated on Alum and subsequently injected intraperitoneally in mice. Spleens were analyzed by flow cytometry 14 days after immunization. Specific IgM responses against PC-KLH was determined by ELISA using plates coated with PC-BSA (Biosearch Technologies), alkaline phosphatase coupled goat anti-mouse IgM (Southern Biotechnology Associates), and alkaline phosphatase yellow Liquid Substrate (Sigma-Aldrich). Purified rat anti-mouse IgG (H + L) (Sigma-Aldrich) was used as capture antibody to measure total plasma IgM, IgA, and IgG levels. AP-labeled goat anti-mouse IgM, IgA and IgG, IgG1, IgG2b, and IgG3 (Sigma-Aldrich) was used for detection.

Data were analyzed using GraphPad Prism. p-Values of <0.05 from Student’s t-tests were reported.