Essential requirement for polypyrimidine tract binding proteins 1 and 3 in the maturation and maintenance of mature B cells in mice

Abstract The maturation of immature B cells and the survival of mature B cells is stringently controlled to maintain a diverse repertoire of antibody specificities while avoiding self‐reactivity. At the molecular level this is regulated by signaling from membrane Ig and the BAFF‐receptor that sustain a pro‐survival program of gene expression. Whether and how posttranscriptional mechanisms contribute to B cell maturation and survival remains poorly understood. Here, we show that the polypyrimidine tract binding proteins (PTBP) PTBP1 and PTBP3 bind to a large and overlapping set of transcripts in B cells. Both PTBP1 and PTBP3 bind to introns and exons where they are predicted to regulate alternative splicing. Moreover, they also show high‐density of binding to 3’ untranslated regions suggesting they influence the transcriptome in diverse ways. We show that PTBP1 and PTBP3 are required in B cells beyond the immature cell stage to sustain transitional B cells and the B1, marginal zone and follicular B cell lineages. Therefore, PTBP1 and PTBP3 promote the maturation of quiescent B cells by regulating gene expression at the posttranscriptional level.


Introduction
The maturation of naïve B cells occurs through discrete stages. In the bone marrow, developing B cells express IgM at their cell surface and are defined as immature. These B cells migrate to the spleen where they mature through different transitional stages (T1 and T2) [1,2], express BAFF-receptor and establish a homeostatic survival-circuit dependent upon its ligand [3,4]. Mature Correspondence: Dr. Martin Turner e-mail: martin.turner@babraham.ac.uk naïve B cells express IgD in addition to IgM and may recirculate between lymphoid follicles or reside in the marginal zones of the spleen. The maturation and survival of mature B cells is dependent upon signaling by surface Ig, and gene regulation by DNA binding transcription factors [4]. The contribution of posttranscriptional regulation of mRNA processing, stability, or translation is less well characterized [5,6].
RNA binding proteins (RBP) acting posttranscriptionally determine the quantity and variants of transcripts produced from genes. They also regulate where, and in what amounts, mRNAs are translated. The role of these regulatory processes in B cell development is a new and exciting area. Polypyrimidine tract binding protein (PTBP) 1 is a widely expressed alternative splicing regulator. It has additional, less well-characterized, roles in regulating microRNA processing, mRNA abundance, alternative polyadenylation, and internal ribosome entry sitemediated translation [7]. Studies of B cells with Cd79a(Mb1) Cremediated conditional deletion of Ptbp1 showed it was dispensable for B cell development, but important for the generation of GC B cells and antigen-specific antibodies [8,9]. Absence of PTBP1 in B cells led to the expression of PTBP2 [8], which is expressed in the brain [7]. In mice with Cd79a Cre -mediated conditional deletion of both Ptbp1 and Ptbp2, B cell development is blocked at the pro-B cell stage [10]. This is associated with defective cell cycle regulation in pro-B cells and shows that PTBP2, which is silenced by PTBP1, compensates for the lack of PTBP1. The paralog PTBP3 is expressed specifically in cells of the hematopoietic system. In B cells, PTBP1 and PTBP3 are found in the nucleus whereas PTBP3 is also detected in the cytoplasm [11]. The prominent cytoplasmic localization of PTBP3 suggests that in addition to nuclear roles it may also affect mRNA stability, translation, or localization. Knowledge of PTBP3 in physiological systems is limited with a report that Ptbp3 -/mice have normal B cell development, but defective antibody responses [12]. However, when Ptbp3 was deleted conditionally in B cells with the Cd79a Cre allele we found normal B cell development, antibody amounts, and affinity maturation in response to immunization ( [10] and unpublished data). Here, we show that PTBP1 and PTBP3 act redundantly to promote the maturation and maintenance of B cells.

PTBP1 and PTBP3 bind redundantly to introns and 3'-UTRs
Since both PTBP1 and PTBP3 are highly homologous, particularly within their RNA binding domains, we compared their transcriptbinding profile by direct mapping binding sites in the B cell transcriptome using individual-nucleotide resolution UV cross-linking and immunoprecipitation (iCLIP) [13]. We used LPS-activated splenic B cells. A comparison of the genes expressed with at least one FPKM in LPS-activated B cells to ex vivo-isolated splenic B cells revealed a 91% overlap (Supporting Information Fig. S1A). These iCLIPs revealed the same polypyrimidine-tract binding motifs for PTBP1 and PTBP3 (Fig. 1A). In addition to the expected binding to introns, we found prominent binding of both proteins to 3' untranslated regions (3'-UTRs) (Fig. 1B). We found a higher density of PTBP3 binding to 3'-UTRs compared to PTBP1 (Fig. 1C). Conversely, there was a higher density of PTBP1 binding to introns compared to PTBP3 (Fig. 1C), which is consistent with the predominant nuclear localisation of PTBP1 [11].
Detailed analysis of individual binding sites, such as those in transcripts encoding IgM and IgD heavy chains (Ighm and Ighd) and the BAFF-receptor (Tnfrsf13c), showed that PTBP1 and PTBP3 bound largely to the same sites (Fig. 1D). We observed binding of both PTBPs to the IgM M2 exon where PTBP1 has been shown to suppress an enhancer of splicing that promoted exon inclusion [14]. To define reproducible binding sites, we chose a cut-off of at least three independent cDNA reads in the iCLIP (and FDR<0.05). With these settings, we found that PTBP1 and PTBP3 bound to transcripts of 9,035 genes with a large overlap of 64% (5,803 genes) (Fig. 1E, Supporting Information Table S1). This is an underestimate of genes bound by both PTBPs, since there will be genes bound by both PTBPs for which one PTBP did not pass our cut-off criteria. PTBP3 bound to the 5'-UTR and 3'-UTR of more genes than PTBP1 (Fig. 1E), reflecting the greater cytoplasmic localization of PTBP3 compared to PTBP1. Amongst the 12,657 polyadenylated genes detected in LPS-activated B cells with at least one FPKM, PTBP1 and PTBP3 bound to 7,644 (60%) (Supporting Information Fig. S1B), highlighting their potential to regulate numerous genes and pathways. Altogether, although we found some small differences, these RBPs showed a highly redundant pattern of binding consistent with them regulating the same RNAs in B cells.

PTBP1 and PTBP3 are necessary for B cell maturation
We found previously that Cd79a Cre -mediated deletion of Ptbp1 and Ptbp3 together did not affect the numbers of pre-and immature-B cells in the bone marrow in Cd79a Cre/+ Ptbp1 fl/fl Ptbp3 fl/fl (P1P3 CD79a dKO) mice [10]. In contrast, in the spleen, we observed a reduction in the frequency of B220+CD19+ cells ( Fig. 2A). Among CD93 + B cells in P1P3 CD79a dKO mice, there was an increase in the proportions of the first transitional T1 cells compared to littermate control mice ( Fig. 2A). The enumeration of cells showed an almost complete absence of mature B cells in P1P3 CD79a dKO mice (Fig. 2B). The numbers of cells in the T2 and T3 subsets were reduced by 9.7-and 17-fold, respectively, with the T1 subset the least affected being reduced by 3.5-fold. In the peritoneal lavage, there were few B cells and CD19 high B220 low B1 B cells were clearly depleted (Supporting Information Fig. S2A). Intracellular staining for PTBP1 and PTBP3 in B cells from P1P3 CD79a dKO mice showed efficient protein depletion but, unlike control cells, they express PTBP2 ( Fig. 2C; Supporting Information Fig. S2B).
We found that P1P3 CD79a dKO mice became severely unwell and had to be culled between 10-12 weeks of age. At necropsy, it was apparent that the kidneys were abnormal with signs of polycystic lesions (Supporting Information Fig. S2C). There have been reports of Cd79a Cre expression in the kidney [15] and conditional KO of Tsc1 with Cd79a Cre led to a polycystic kidney phenotype [16], suggesting that Ptbp1 and Ptbp3 deletion with the Cd79a Cre led to these lesions, but this will have to be confirmed. To establish if the B cell phenotype was intrinsic to the hematopoietic system we transferred bone marrow from Cd79a Cre/+ Ptbp1 +/+ Ptbp3 +/+ control or P1P3 CD79a dKO mice into lethally irradiated B6.SJL mice. There was no early-onset disease or kidney abnormalities found in these chimeras. Following reconstitution, as with intact mice, the numbers of all splenic transitional and mature B cell subsets in P1P3 CD79a dKO were substantially reduced compared to control chimeras ( Fig. 2D; Supporting Information Fig. S2D). The proportions of B cells and the ratio of B cells compared to T cells in the blood were reduced 2.4-and 5.7-fold, respectively, in P1P3 CD79a dKO compared to control chimeras (Supporting Information Fig. S2E and S2F). Histological examination of the spleens from P1P3 CD79a dKO chimeras showed the few B cells present localized around the T cell zone indicating unaffected migration of the PTBP1 and PTBP3-deficient B cells in the spleen (Fig. 2E). A defect in the export of PTBP1 and PTBP3 deficient immature B cells could contribute to the reduced number of T1 B cells. These data revealed the indispensable role of PTBP1 and PTBP3 in the differentiation or maintenance of mature B cells in the spleen and the peritoneum-a role that cannot be compensated for by PTBP2.

PTBP1 and PTBP3 are necessary for the maintenance of mature B cells
To study the mature B cell compartment further we used Cd23 CreTg [17], which mediates recombination in transitional B cells to generate Cd23 CreTg Ptbp1 fl/fl Ptbp3 fl/fl mice (P1P3 CD23 dKO) and analyzed splenic B cell subsets by flow cytometry (Fig. 2F). The numbers of T1, T2, and T3 B cells in P1P3 CD23 dKO mice were similar to those in Cd23 CreTg Ptbp1 +/+ Ptbp3 +/+ or Ptbp1 fl/fl Ptbp3 fl/fl mice (Fig. 2G). By contrast, the numbers of B cells with the follicular or marginal zone (MZ) phenotype in the spleen were substantially reduced (Fig. 2G). Identification of MZ B cells with staining of CD21 and CD23 also showed a reduction in MZ B cell numbers in the spleen and revealed lower staining of CD21 in PTBP1 and PTBP3-deficient cells compared to control B cells (Supporting Information Fig. S3A). In addition, the numbers of B cells from lymph nodes were reduced by three-fold in P1P3 CD23 dKO mice (Fig. 2H, Supporting Information Fig. S3B).
The presence of both PTBP1 and PTBP3 proteins in T1 and T2 cells from P1P3 CD23 dKO mice (Supporting Information Fig.  S3C) was consistent with the minimal impact on cell numbers at these stages. Follicular B cells in both the spleen and lymph node showed a large proportion of cells that retained expression of PTBP1 and PTBP3 (Supporting Information Fig. S3C and S3D) at the levels of control and Ptbp1 and Ptbp3 heterozygous CD23 KO mice (Supporting Information Fig. S3E). Yet their absolute numbers were reduced two to three-fold in the P1P3 CD23 dKO mice ( Fig. 2G and H). Moreover, cells with maintained PTBP1 and PTBP3 expression were abundant amongst MZ B cells (Supporting Information Fig. S3C).
We assessed IgM and IgD surface levels in PTBP1 and PTBP3deficient mature B cells from P1P3 CD23 dKO mice where we could assess consequences of recent PTBP1 and PTBP3 protein depletion in B cells. IgM and IgD staining in PTBP1 and PTBP3deficient mature B cells were unaffected (Supporting Information Fig. S3F). Also, BTK and AKT phosphorylation at Tyrosine 223 and Threonine 308, respectively, upon IgM crosslinking was unaffected in PTBP1 and PTBP3-deficient mature B cells (data not shown). PTBP1 and PTBP3-deficient mature B cells expressed PTBP2 (Supporting Information Fig. S3C) suggesting that PTBP2 compensated for some roles of PTBP1 and PTBP3. Nonetheless, PTBP2 in follicular and MZ B cells was unable to support these cells revealing a role for PTBP1 and PTBP3 in the maintenance of mature B cell subsets.

Reduced BAFF-receptor in the absence of PTBP1 and PTBP3
BAFF signaling is critical for the maturation and maintenance of B cells [3,4] and BAFF-receptor is induced by tonic signaling [18]. PTBP1 and PTBP3 bound to the 3'UTR of Tnfrsf13c (encoding BAFF-receptor) (Fig. 1D). BAFF-receptor staining was reduced in PTBP1 and PTBP3 deficient B cells from the P1P3 CD79a dKO chimeras and the P1P3 CD23 dKO mice (Fig. 3A-D). This reduction could be a direct effect of PTBP1 and PTBP3 in controlling Tnfrsf13c mRNA abundance or translation or also a result of impaired tonic signaling. We cultured B cells isolated from lymph nodes of P1P3 CD23 dKO mice and control littermate mice ex vivo with or without exogenous BAFF (Fig. 3E). The numbers of B cells from P1P3 CD23 dKO mice we recovered after several days in culture were unaffected even though B cells from P1P3 CD23 dKO in culture had lost PTBP1 and PTBP3 proteins (Fig. 3F). This showed that PTBP1 and PTBP3-deficient mature B cells respond to exogenous BAFF which promotes their survival ex vivo. Thus, we expect that in vivo PTBP1 and PTBP3 are necessary for controlling expression at the posttranscriptional level of many other genes necessary for B cell maturation and maintenance in addition to BAFF-receptor and the absence of B1 cells in P1P3 CD79 dKO mice reflects this.

Concluding remarks
The requirement of both PTBP1 and PTBP3 in transitional and mature B cells contrasts with early B cell development where PTBP2 compensated completely for the absence of these proteins [10]. In mouse brain, PTBP1 and PTBP2 bind redundantly largely to the same sites [19]. Therefore, we expect PTBP2 to bind to the same sites as PTBP1 and PTBP3 do in B cells. The inability of PTBP2 to sustain B cell maturation may reflect qualitative differences between the paralogs, as PTBP2 is not thought to enter the cytoplasm. However, it is also possible that all three paralogs have redundant roles and that PTBP2 abundance is insufficient to ensure B cell maturation and maintenance.
Given their widespread binding, we expect that PTBP1 and PTBP3 will control many genes necessary for the maintenance of mature B cell populations. In addition to Tnfrsf13c binding, PTBP1 and PTBP3 bound to exons, introns, and 3'UTRs Bcl2, Syk, Akt1, Cxcr4, and S1pr1 among many other genes (Supporting Information Table S1 and Fig. S1B). Further investigations are warranted to elucidate if and how PTBP1 and PTBP3 control expression of these targets necessary for the export of immature B cells from the bone marrow and survival, tonic signaling, and maintenance of mature B cells. We anticipate diverse and shared roles for PTBP1 and PTBP3 in B cells including the control of alternative splicing and also other functions via 3'-UTR-binding such as mRNA localization, stability, and translation.

Mice and bone marrow chimeras
Mice were maintained in the Babraham Institute Biological Support Unit. No primary pathogens or additional agents listed in the FELASA recommendations have been confirmed during health monitoring since 2009. Ambient temperature was ∼19-21°C and relative humidity 52%. Lighting was provided on a 12hr light: 12hr dark cycle including 15 min 'dawn' and 'dusk' periods of subdued lighting. After weaning, mice were transferred to individually ventilated cages with one to five mice per cage. Mice were fed CRM (P) VP diet (Special Diet Services) ad libitum and received seeds (e.g., sunflower, millet) at the time of cage-cleaning as part of their environmental enrichment. All mouse experimentation was approved by the Babraham Institute Animal Welfare and Ethical Review Body. Animal husbandry and experimentation complied with existing European Union and United Kingdom Home Office legislation. All mice were used experimentally between 8-and 25-weeks of age (for bone marrow chimeras) and were age-and sex-matched within experiments. No sex-associated differences were observed in the results obtained. Mice were on the C57BL/6 background and derived from crossing the following transgenic strains: Ptbp1 fl/fl (Ptbp1 tm1Msol ) [20], Ptbp3 fl/fl (Ptbp3 tm1Tnr ) [10], Cd79a Cre (Cd79 atm1(cre)Reth ) [15], and Cd23 CreTg (Cd23-cre Tg(Fcer2a-cre)5Mbu ) [17]. To generate chimeras, B6.SJL mice were lethally irradiated (2 × 0.5 Gy) and reconstituted with 3 × 10 6 bone marrow cells derived from control (Cd79a Cre/+ Ptbp1 +/+ Ptbp3 +/+ ) or P1P3 CD79a dKO mice.

iCLIP
The PTBP3 iCLIP reported in this study was carried out with B cells isolated from the spleen stimulated ex vivo for 48h with LPS (10 μg/ml, 127:B0, Sigma) as described in [8]. We used 10 μg of the anti-PTBP3 rabbit polyclonal antibody PTBP3-L2-3 [11] and 100 μl protein A magnetic Dynabeads (ThermoFisher, cat#:10002D). Details of the barcodes and oligos used can be accessed in GEO under: GSE168769. iCLIP libraries were mapped to the GRCm38 mouse genome M16 release version from Gencode with STAR [22]. cDNA reads were de-duplicated using random barcodes included in the library preparation and significant (FDR<0.05) xlink-binding sites were identified with iCount (https://icount.readthedocs.io/en/latest/cite.html). PTBP1 iCLIP was published previously [8,10]. FPKMs of ex vivo isolated and LPS-activated splenic B cells were calculated from mRNAseq data published under the GSE62129 accession number [23].