The miR-15a/16-1 and miR-15b/16-2 clusters regulate early B cell development by limiting IL-7 receptor expression

MicroRNAs are small non-coding RNAs that have emerged as post-transcriptional regulators involved in development and function of different types of immune cells, and aberrant miRNA expression has often been linked to cancer. One prominent miRNA family in the latter setting is the miR-15 family, consisting of the three clusters miR-15a/16-1, miR-15b/16-2 and miR-497/195, which is best known for its prominent tumor suppressive role in chronic lymphocytic leukemia (CLL). However, little is known about the physiological role of the miR-15 family. In this study, we provide a comprehensive in vivo analysis of the physiological functions of miR-15a/16-1 and miR-15b/16-2, both of which are highly expressed in immune cells, in early B cell development. In particular, we report a previously unrecognized physiological function of the miR-15 family in restraining progenitor B cell expansion, as loss of both clusters induces an increase of the pro-B as well as pre-B cell compartments. Mechanistically, we find that the miR-15 family mediates its function through repression of at least two different types of target genes: First, we confirm that the miR-15 family suppresses several prominent cell cycle regulators such as Ccne1, Ccnd3 and Cdc25a also in vivo, thereby limiting the proliferation of progenitor B cells. Second, this is complemented by direct repression of the Il7r gene, which encodes the alpha chain of the IL-7 receptor (IL7R), one of the most critical growth factor receptors for early B cell development. In consequence, deletion of the miR-15a/16-1 and miR-15b/16-2 clusters stabilizes Il7r transcripts, resulting in enhanced IL7R surface expression. Consistently, our data show an increased activation of PI3K/AKT, a key signaling pathway downstream of the IL7R, which likely drives the progenitor B cell expansion we describe here. Thus, by deregulating a target gene network of cell cycle and signaling mediators, loss of the miR-15 family establishes a pro-proliferative milieu that manifests in an enlarged progenitor B cell pool.


Introduction
B cells develop in the bone marrow, where they undergo two sequential steps of immunoglobulin gene segment rearrangements, finally giving rise to a unique B cell receptor (BCR). Heavy chain diversity (D) to joining (J) gene segment rearrangement is initiated between the pre-pro and pro-B cell stage and is followed by variable (V) to DJ gene segment recombination in late pro-B cells (1). Successful rearrangement results in expression of an immunoglobulin heavy chain that pairs with the surrogate light chain components VpreB and lambda5, thereby forming the pre-B cell receptor (pre-BCR) (2). Expression of a functional pre-BCR in pre-B cells induces several rounds of cell division before the cell cycle is stalled and light chain (VJ) recombination is initiated. The thereby formed light chain pairs with the heavy chain to form the BCR, marking the immature B cell stage (3). Immature B cells travel to the spleen where they undergo additional maturation steps through several transitional stages, eventually entering the mature B cell pool as marginal zone or follicular B cells (4,5).
As heavy and light chain gene segment recombination in early B cell development is based on the introduction of DNA double strand breaks, the stepwise progression through the abovementioned stages needs to be tightly regulated. This ensures the proper separation of proliferation and cell cycle arrest to allow somatic rearrangements, which has been demonstrated as critical to maintain genomic stability and to prevent malignant transformation and blood cancer (6)(7)(8). In early B cells, this biphasic sequence is mainly coordinated by the interplay of three receptors and their downstream signaling pathways: The IL-7 receptor (IL7R), which contributes to proand pre-B cell survival, is critical for proliferation and at the same time inhibits differentiation by downregulation of FOXO1 and repression of Rag1 and Rag2 genes (9); the abovementioned pre-BCR, as well mediating cell survival and proliferation, but also light chain recombination (10); and the chemokine receptor CXCR4, which instructs proper positioning of B cell progenitors in the bone marrow microenvironment and promotes cell cycle exit, pre-BCR repression and light chain rearrangement (11). However, even with these main players identified, it appears fair to say that the precise gene regulatory networks that govern early B cell development are not fully understood.
Since their discovery, microRNAs (miRNAs), small noncoding RNAs with a length of about 21 to 25 nucleotides, have emerged to establish an additional layer of gene regulation in almost all biological processes. They exert their function by sequence-specific binding to the 3' UTR of target mRNAs, resulting in inhibition of translation or mRNA decay and ultimately in post-transcriptional repression of their target genes (12). Given that more than 60% of all protein coding genes in mammals are predicted to be targeted by miRNAs (13), it is not surprising that also B cell development in the bone marrow is shaped by certain miRNAs such as the miR-17-92 cluster and miR-34a (14,15). However, with an estimated number of about 50 different miRNAs expressed at functionally relevant levels in early B cell development (16), it is likely that additional regulators will emerge upon thorough analysis, such as in in vivo loss-of-function mouse models. A possible candidate in this context is the miR-15 family, comprising the three bicistronic clusters miR-15a/16-1, miR-15b/ 16-2 and miR-497/195 (17). Of these clusters, only the first two are highly expressed throughout B cell development, whereas miR-497 and miR-195 are predominantly found in non-immune cells and consequently both have been described as dispensable for the immune system in a previous study (18). We therefore decided to focus on miR-15a/16-1 and miR-15b/ 16-2 for this particular study. The miR-15 family is best known for its tumor-suppressive activity in B cell chronic lymphocytic leukemia (CLL), in which the chromosomal region encoding the miR-15a/16-1 cluster is deleted in about 60% of all patients (19). This role has been recapitulated in mouse models, in which deletion of either the miR-15a/16-1 or the miR-15b/16-2 cluster also promoted a CLL-like disease (20,21). However, both mouse models have been reported to display normal B cell development, possibly indicating that the lack of obvious phenotypes in individual miR-15a/16-1 and miR-15b/16-2 knockout mice is the result of a compensatory effect by the remaining cluster. In fact, our own work using an in vitro knockdown approach that targets the whole miR-15 family suggested that the miR-15 family fine-tunes the pre-B to immature B cell transition (22).
To investigate this in detail and in a complex in vivo environment, we have therefore generated a conditional mouse model for the immune cell-specific deletion of both miR-15a/16-1 and miR-15b/16-2 clusters. Exploiting this model, we demonstrate that one of the main physiological functions of the miR-15 family is to restrict expansion of B cell progenitors in a dose-dependent manner. Mechanistically, our data show that the miR-15 family directly limits expression of the IL7R, and that loss of this repression results in increased IL7R protein levels as well as in enhanced activation of the PI3K/AKT signaling pathway. Thus, in synergy with the repression of cell-cycle regulatory genes that have been previously described to establish an anti-proliferative state in various systems, such as Cdc25a, Ccnd3 and Ccne1, the miR-15 family constricts growth factor receptor signaling and thereby contributes to the biphasic sequence with concomitant phases of proliferation and cell cycle arrest.

Animals
The conditional miR-15a/16-1 and miR-15b/16-2 alleles were generated by CRISPR/Cas9-facilitated homologous recombination in murine ES cells, as already described for targeting of the miR-497/195 cluster (18 (24)] as a marker and the targeting DNA template containing the miRNA cluster flanked by loxP sites (see Supplementary Table S1 for detailed sgRNA sequences). After 36 h, ES cells were sorted for GFP hi cells and plated at a low density on feeder cells. Individual ES cell clones were screened by PCR, sequenced, and then used for injection in C57BL/6NRj blastocysts. High percentage chimeras were bred with C57BL/6NRj females to confirm germline transmission and then further backcrossed to generate a congenic strain. Cremediated recombination in hematopoietic stem cells was induced by mating with C57BL/6N.Cg-Tg(Vav-Cre) mice (25) or C57BL/6N.Cg-Tg(Mb1-Cre) mice to mediate B-cell specific deletion (26).
Animals were kept pathogen-free according to FELASA recommendations (27) under controlled environmental conditions (temperature 22°C ± 1°C, relative humidity of 40-60%), a 12:12 h light/dark cycle, in a facility for laboratory rodents. Food (regular mouse diet; Ssniff, Germany) and water were provided ad libitum. Mice were maintained in small groups in individually ventilated cages lined with wood shavings as bedding and enriched with nesting material. If not stated otherwise, mice were analyzed at an age of 10-12 weeks. For all experiments, male and female mice were used in comparable numbers.

Preparation of single-cell suspension
Single-cell suspensions for flow cytometry were obtained by pulpifying spleens through a 70-µm filter. For bone marrow cell suspensions, femurs and tibiae were isolated, ground, and filtered through a 70-µm filter. Lysis of erythrocytes for spleen and blood samples was performed by incubating the cells for 3-5 min in 1 mL lysis buffer (155 mM NH 4 Cl, 10 mM KHCO 3 , 0.1 mM EDTA; pH 7.5). Cells were resuspended and washed in FACS buffer (PBS with 1% FCS). Cells were counted using a Neubauer counting chamber.
Flow cytometric analysis was performed on a FACS LSR Fortessa instrument (BD Biosciences), and sorting was performed on a FACS Aria III (BD Biosciences) cell sorter. Data were analyzed using FlowJo software (BD Life Science, USA).

MACS of bone marrow B cells for qPCR
Bone marrow cell suspensions were stained with 100 µl anti-B220-biotin antibody solution (BioLegend; 103204, 1:100) and incubated for 20 min at 4°C. Cells were washed in FACS buffer, the pellet was resuspended in 200 µl MACS beads (Invitrogen; MSNB-6002-74; dilution 1:10 in FACS buffer) and incubated for 5 min at RT. The cell/bead suspension was mixed with 3 ml FACS buffer and placed on a magnet for 10 min at RT. FACS buffer was discarded and the cell/bead suspension was washed once more by repeating the previous step. In the last step, the FACS buffer was discarded and the cell/bead suspension was resuspended in 500 µl Trizol (Invitrogen; 15596-018).

RNA sequencing
RNA sequencing was performed as a service by Novogene (Cambridge, UK). In short, total RNA was isolated using the RNeasy Micro Kit according to manufacturer's instructions (QIAGEN; 74004) and amplified and reverse transcribed into cDNA using the SMART-seq protocol (Takara), followed by a quality control and library preparation. Paired end sequencing (reads of 150 nt) was performed on a NovaSeq 6000 (Illumina). For data analysis, single end reads were trimmed using Trimmomatic (28), aligned using STAR aligner (29) and counted with FeatureCounts (30). Differential gene expression was analyzed using EdgeR (31,32).

Transfections and retroviral transductions
Retroviral supernatants were produced in HEK293T cells. Per sample, 400 ng of DNA for the plasmid of interest, 150 ng of HIT60 and 150 ng pVSVg were diluted in 20 ml IMDM medium (Sigma-Aldrich) and mixed with 20 ml IMDM medium containing 1.2 ml of 1 mg/mL polyethylenimine (PEI, Polysciences). Following a 20 min incubation step at room temperature, the DNA:PEI mix was pipetted into a 24-well tissue culture plate and 230 ml of HEK293T at a density of 6 × 10 5 cells per ml were plated on top. After 30 -40 hours viral supernatants were harvested, mixed with polybrene (16 mg/mL final concentration) and used for spin-infection of cells in 1.5 ml reaction tubes or 96-well plates (400 g at 37°C for 90 min).

3' UTR reporter assays
UTRs of interest were amplified by PCR (see Supplementary  Table S1 for details), purified by agarose gel isolation (Monarch DNA Gel Extraction Kit, NEB (T1020L)) and ligated into LMP-destGFP (22) 3′ of the destabilized GFP cDNA via XhoI and ClaI. The pre-B cell line WK3 was transduced with the respective retroviral supernatants and cells were selected for puromycin resistance by adding puromycin at a concentration of 1 mg/ml for 3-4 days. For the suppression assays, cells were further transduced with a previously described plasmid encoding dsRed as a marker together with the miR-15a/16-1 cluster (22) or an empty vector as a control. For the sponge-mediated derepression of endogenous miRNAs, a scrambled sponge or a sponge containing multiple binding sites for miR-16 were used (22). In experimental settings, cells were always cultured in the absence of puromycin.

Cell lines and cell culture
The pre-B cell line WK3 was derived by extended culture of total bone marrow of SLP-65-deficient mice in IL-7supplemented medium (22). Pre-B cells were cultured in IMDM (Sigma) containing 7.5% FCS (Biochrom Superior), 100 U/ml penicillin, 100 U/ml streptomycin (PAN), 50 mM 2mercaptoethanol and IL-7 (produced and secreted by transgenic J558L cells) in excess. In short, each batch of the IL-7-containg J558L supernatant was titrated to determine the lowest concentration to support maximal pre-B cell proliferation, and this amount was doubled for standard cell culture. HEK293T cells were cultured in the same medium, but without addition of IL-7.
For the culture of primary B cell progenitors, the bone marrow of miR-15a/b +/+ Vav-Cre or miR-15a/b fl/fl Vav-Cre mice was sorted for the c-KIT + B220 + CD19 -NK1.1population and cells were seeded in a 96-well plate in 100 µl IMDM medium (7.5% FCS (Biochrom Superior), 100 U/ml penicillin, 100 U/ml streptomycin (PAN), 50 mM 2-mercaptoethanol, 100 ng/ml SCF, 5 ng/ml IL-7 and 50 ng/ml FLT3L) at a density of 5 x 10 4 cells/ml. Upon a "recovery phase" of 4-5 days, progenitor cells typically started to proliferate and were subsequently kept in culture and split regularly. B cell progenitors were used for experiments while in this proliferative phase, and were discarded when the proliferation rate dropped significantly, typically after 3-4 weeks of culture. To induce differentiation of progenitors in vitro, cells were cultured in the absence of IL-7 for 48 hours.

EdU assay
To assess in vitro proliferation, cells were analyzed with the Click-iT Flow Cytometry Assay Kit (Invitrogen; C10419) according to the manufacturer's protocol, using a labeling time of 4 hours at an EdU concentration of 10 µg/ml.

Statistical analysis
Values in figures depict mean ± standard deviation (SD), and each mouse or experiment is represented as a dot. The two experimental groups were statistically analyzed by unpaired twotailed Student's t-tests using Prism 9 software (GraphPad). Pvalues < 0.05 were considered as statistically significant, and graphs were labeled according to following scheme: ***P < 0.005, **P < 0.005, and *P < 0.05.

Results
Generation of conditional miR-15a/16-1 and miR-15b/16-2 double knockout mice To study whether the miR-15 family is involved in early B cell development, we generated a conditional mouse model for the Cre-mediated deletion of both the miR-15a/16-1 and miR-15b/ 16-2 clusters, the main family members expressed within the hematopoietic system. The conditional alleles were generated using CRISPR/Cas9-assisted editing to flank the endogenous clusters -embedded within introns of non-coding (Dleu2 for miR-15a/16-1) and coding (Smc4 for miR-15b/16-2) genes, respectively -with loxP sites in embryonic stem (ES) cells ( Figures S1A, B). To delete the respective miRNA clusters in the hematopoietic system, mice derived from these ES cells were intercrossed and then initially bred with the Vav-Cre strain (25) to generate single miR-15a/16-1 fl/fl (hereafter miR-15a fl/fl Vav-Cre ) and miR-15b/16-2 fl/fl (miR-15b fl/fl Vav-Cre ) as well as miR-15a/ 16-1 fl/fl miR-15b/16-2 fl/fl (miR-15a/b fl/fl Vav-Cre ) double knockout (DKO) mice. To verify the loss of the respective miRNAs in all genotypes compared to the control, we analyzed their expression in sorted bone marrow B cells by quantitative PCR. As expected, expression of miR-15a was undetectable in miR-15a fl/fl Vav-Cre and miR-15a/b fl/fl Vav-Cre DKO mice, and likewise there was no miR-15b in miR-15b fl/fl Vav-Cre and DKO mice, respectively. Moreover, since the mature forms miR-16-1 and miR-16-2 are identical, a complete loss of miR-16 was only detectable in miR-15a/b fl/fl Vav-Cre DKO mice, whereas miR-15a fl/fl Vav-Cre and miR-15b fl/fl Vav-Cre mice showed an approximately 50% reduction compared to miR-15a/b +/+ Vav-Cre controls ( Figure 1A).

Loss of miR-15 interferes with NK cell development and slightly enhances myeloid cell numbers
To validate the deletion of the respective miR-15 family members also on the functional level, we furthermore compared our mouse model to previously published miR-15 loss-offunction phenotypes. A study by Sullivan et al. has described a partial block in NK cell development at the stage III to stage IV transition upon NK-cell specific deletion of the miR-15a/16-1 cluster (33). Our analysis clearly recapitulated this finding in miR-15a fl/fl Vav-Cre mice, and interestingly the developmental block was further aggravated in miR-15a/b fl/fl Vav-Cre DKO compared to the single knockout mice ( Figures 1B, S1C). However, while clearly contributing to the DKO phenotype, the miR-15b/16-2 cluster alone seemed to be dispensable for NK cell maturation, as miR-15b fl/fl Vav-Cre mice did not show any perturbations upon deletion of the cluster, pointing towards different roles of the individual miR-15 family clusters in this setting ( Figure 1B). In the myeloid compartment, a recent publication has revealed that concomitant loss of both miR-15a/16-1 as well as miR-15b/16-2, albeit not in a conditional model, promotes the development of acute myeloid leukemia (AML) at an age of approximately 5 months (34). According to our analysis, a dose-dependent myeloid expansion was already detectable in 10 to 12 weeks old mice, with the strongest impact in the DKOs and mainly affecting neutrophils ( Figure 1C, D).
Furthermore and in consequence of the myeloid phenotype, both with respect to relative frequencies and total cell numbers, spleen weight and splenic cellularity were significantly increased in miR-15a/b fl/fl Vav-Cre mice ( Figures 1E, S1B). Together, these data demonstrate a dose-dependent role of the miR-15a/16-1 and miR-15b/16-2 clusters in previously reported non-B cell hematopoietic lineages, and overall validate the function of our mouse model. This, of course, raised the question about the effects of their combined deletion in B cells as the lineage mostly affected by pathological loss of one of the miR-15 family clusters.
Loss of the miR-15a/16-1 and miR-15b/ 16-2 clusters impairs B cell development Hematopoietic deletion of miR-15a/16-1 and miR-15b/16-2 blocks NK cell development and promotes the accumulation of myeloid cells. (A) Quantitative PCR for the expression of mature miR-15a, miR-15b and miR-16 in B cells derived from miR-15a/b +/+ Vav-Cre , miR-15a fl/fl Vav-Cre , miR-15b fl/fl Vav-Cre and miR-15a/b fl/fl Vav-Cre mice. Each dot represents the data derived from one mouse. n.d. = not detected. (B) NK cells defined as NK1.1 + TCRbetawere classified as stage I to stage IV based on their expression of CD27 versus CD11b (schematic left panel and contour plots, numbers indicate the percentage of cells within the respective quadrant). The line and bar graphs show the analysis of all data sets for the depicted genotypes, and again each dot represents data from one mouse. (C) Schematic illustration of the gating strategy and representative primary data for myeloid cells and neutrophils in spleens derived from all genotypes. Numbers indicate the percentage of cells within the respective gate. (D) Bar graphs show the percentages of indicated populations within total cells or CD11b + myeloid cells for all analyzed mice. (E) Bar graphs indicate the total splenic cell numbers and spleen weight for the respective genotypes. Each dot represents the data derived from one mouse. Error bars depict the standard deviation of the mean. Sample groups were statistically compared as indicated by an unpaired two-tailed Student's t-test. *P < 0.05; **P < 0.005; ***P < 0.0005.
we observed a dose-dependent increase in the pro-B cell subset for all three knockout genotypes ( Figure 2B), with the effect being most severe in the DKO mice. Notably and as discussed later, we also observed a significantly higher surface expression of the pro-B cell marker c-KIT/CD117, in response to transcriptional upregulation of the respective gene, in mice lacking miR-15 expression (Figures 2A, S2B). In analogy to the pro-B cells, the pre-B cell fractions were increased in the single knockouts, but not significantly in the DKO, most likely due to the strong relative increase in pro-B cells ( Figure 2B

Loss of the miR-15 family results in pro-B cell expansion in vitro
To investigate the role of the miR-15 family in early B cell development and to dissect the molecular mechanism underlying the miR-15-mediated regulation in more detail, we made use of a stromal cell-free in vitro system that allows the cultivation of primary B cell progenitors (35). Here, B220 + c-KIT + CD19 -NK1.1cells with myeloid and lymphoid potential are sorted from the bone marrow, and in the presence of IL-7, FLT3L and SCF those cells differentiate into CD19 + cells that recapitulate all stages of early B cell development including pro-B (c-KIT + CD25 -), intermediate (c-KIT -CD25 -) and pre-B cells (c-KIT -CD25 + ; Figure 3A). Notably, under excessive growth factor conditions, cells derived from control mice as well as from miR-15a/b fl/fl Vav-Cre mice gave rise to comparable percentages of B cell progenitors in vitro, supporting our earlier finding that B cell lineage commitment per se is not strongly affected by loss of the miR-15 family ( Figure 3B). However, a significant advantage of this progenitor culture system is that it also allows an assessment of the differentiation potential of the cells. Upon IL-7 withdrawal as the key factor to instruct proliferation, the steady state distribution of progenitor cells normally shifts from a pro-B/ intermediate B towards a pre-B cell-centric pattern ( Figure 3C). Interestingly, loss of the miR-15 family interfered with this differentiation, as cells derived from miR-15a/b fl/fl Vav-Cre mice failed to generate pre-B cells as efficient as the control cells ( Figure 3C). As differentiation and proliferation appear to be mutually exclusive phases throughout early B cell development, we wondered whether the reduced differentiation capacity and the elevated progenitor B cell frequencies in vivo could be the consequence of enhanced expansion. To test this, actively proliferating miR-15a/b fl/fl Vav-Cre and control progenitor cells derived as described above were transduced with different fluorescent markers, mixed and monitored for the outgrowth of one or the other population. In this assay, miR-15a/b fl/fl Vav-Cre cells consistently outcompeted the control cells over time ( Figure 3D), pointing towards increased survival, proliferation or both. To further assess this, we quantified EdU incorporation and analyzed apoptosis by AnnexinV staining. While the percentage of apoptotic cells was similar for control and miR-15a/b fl/fl Vav-Cre cells, DNA synthesis as an indirect indicator for proliferation was significantly enhanced upon loss of the miR-15 family ( Figures 3E, F). Thus, our findings suggest that the miR-15a/16-1 and miR-15b/16-2 clusters limit progenitor B cell proliferation and thereby enable efficient differentiation under physiological conditions.

The IL-7 receptor is a direct target of the miR-15 family
Having identified a clear function of the miR-15 family in restricting progenitor B cell expansion, we wondered whether this phenomenon could be linked to the repression of particular target genes. With pro-B cells being affected the most in early B cell development, we performed RNA sequencing of sorted control and miR-15a/b fl/fl Vav-Cre cells from this particular stage. Anticipating that direct target genes become upregulated upon deletion of the miR-15 family, we focused on this fraction of the differentially expressed genes (346 genes in total; Figure 4A, grey quadrant). Using the well-established targetscan and miRDB algorithms (36)(37)(38), this list was further refined by filtering for genes harboring conserved miR-15 binding sites in their 3' UTRs. This led to the identification of 65 putative target genes including e.g. Ccne1 ( Figure 4B, Supplementary Table S2), which we have previously described as repressed by miR-15 in B cells, thus validating our strategy (22). Notably and in correspondence with other reports, a GO term analysis with these gene IDs revealed a strong enrichment for cell cycle processes (GO accession 0007049; data not shown). Given that progenitor B cells displayed enhanced proliferation upon miR-15 family loss, we therefore concentrated on genes known to be directly or indirectly involved in cell cycle regulation. To investigate whether this subset of interest encodes mRNAs that could in principle be direct targets of the miR-15 family, we expressed their individual 3' UTR regions together with GFP as a reporter and quantified reporter fluorescence upon co-expression of the miR-15a/16-1 cluster ( Figure 4C). This assay revealed miR-15-dependent repression for 11 out of 17 tested candidates, including genes that had been linked to the miR-15 family before, albeit not in context of B cells, as well as novel targets. In addition to Ccne1, reporter repression was also prominent for Cdc25a, both established drivers of the G1-to-S phase cell cycle transition, suggesting that the miR-15 family limits early B cell development at least in part by directly interfering with cell proliferation.
Beside those targets, our UTR assay also confirmed the direct repression of the Il7r gene encoding the alpha chain of the IL-7 receptor (IL7R; Figure 4C, blue bar). Notably, IL-7 receptor signaling is well established as a key regulator of early B cell survival, proliferation and differentiation, with both pro-and pre-B cells expressing high levels of the receptor on their surface (39-42). The murine Il7r 3' UTR contains two miR-15 binding sites that are largely conserved across species ( Figure 4D). To access the functionality of both sites in detail, we mutated the first, the second or both binding sites and performed 3' UTR reporter assays as described above. In the miR-15 overexpression setting ( Figure 4E, left panel), loss of either of the two binding sites already increased GFP reporter fluorescence, with a slightly stronger effect for binding site 1, and this derepression was further enhanced once both binding sites were disrupted. For the reciprocal loss-of-function approach, we made use of a miR-15 family-specific sponge construct expressing an mRNA with multiple repeats of a miR-16 binding site, and demonstrated that this sponge is indeed able to sequester endogenous miRNA/ RISC complexes in a sequence-specific manner ( Figure S4). In the context of the Il7r 3' UTR variants, sponge expression resulted in increased reporter fluorescence in the presence of both binding sites ( Figure 4E, right panel), reflecting that endogenous miR-15 family members bind to and repress the reporter mRNA under steady state conditions. However, this sponge-mediated derepression was reduced when one of the binding sites was disrupted, and was completely abrogated to control levels once both sites were mutated. Together, these experiments clearly indicate that both binding sites individually and additively contribute to miR-15-mediated Il7r mRNA repression, and importantly, that this repression manifests under physiological miR-15 expression levels. This raised the question about the functional consequence when this regulatory relationship is disrupted in B cell progenitors. The miR-15 family directly regulates IL-7 receptor expression and cell cycle related genes. (A, B) Volcano plot of RNA sequencing data comparing the transcriptome of pro-B cells sorted from miR-15a/b fl/fl Vav-Cre and miR-15a/b +/+ Vav-Cre mice. Significantly upregulated genes (grey quadrant) were filtered for predicted miR-15a, miR-15b and miR-16 binding sites in their 3′-UTRs using miRDB and targetscan, giving rise to 65 putative miR-15 target genes displayed as a heatmap. Each column corresponds to cells derived from one mouse. (C) 3' UTR assays of a subset of genes identified in (B). A vector lacking a 3′ UTR served as negative control, whereas constructs with a 3′ UTR containing two perfect miR-16 target sites or the 3' UTR of the already described miR-15 target gene Ccne1 were defined as positive controls. Error bars depict the standard deviation of the mean. The different groups were compared to the negative control by a one-way ANOVA followed by Bonferroni testing. (D) Schematic illustration of the location and the sequence of the two miR-15 binding sites within the Il7r transcript in different species. (E) 3' UTR assay with the Il7r sequence or mutants thereof as depicted upon overexpression or knockdown of the miR-15 family. The length of the bars indicates the ratio of GFP MFI comparing transduced and non-transduced cells as measured by flow cytometry. The constructs encoding mutated binding sites were statistically compared to the wild type 3' UTR construct by an unpaired two-tailed Student's t-test. Each dot represents the data derived from one independent experiment. *P < 0.05; **P < 0.005; ***P < 0.0005. n.s., not significant.

Loss of the miR-15 family in B cell progenitors enhances IL7R expression and signaling
Recapitulating our findings from the RNA sequencing, we first used quantitative RT-PCR to confirm that Il7r transcripts are indeed upregulated upon loss of miR-15 expression, as suggested by our UTR assays ( Figure 5A). To investigate whether the increased mRNA levels also translate onto the protein level, thus being functionally relevant, in vivo IL-7 receptor surface expression of different B cell developmental stages in the miR-15a/b fl/fl Vav-Cre mice was then analyzed by flow cytometry. Here, all early developmental stages known to be regulated by IL7R signaling, i.e. pro-B, intermediate B and pre-B cells as defined by c-KIT and CD25 surface expression, displayed significantly enhanced IL-7 receptor levels ( Figure 5B). Anticipating that this increase in receptor expression affects signaling downstream of the IL7R, we again exploited the in vitro culture system for primary B cell progenitors as a tool to enable easy cell manipulation and to reduce sample-to-sample variability. Of note, primary progenitors displayed the same increase in IL7R expression as the in vivo preparations, justifying our approach ( Figure 5C).
A key pathway downstream of the IL7R in early B cell development is the PI3K/AKT cascade, which has been demonstrated to mediate proliferation and survival of progenitor cells (42,43). Investigating activation of this pathway in detail, a Western Blot analysis as well as intracellular flow cytometry demonstrated that both total and phospho-AKT Ser473 levels were strongly upregulated in progenitors derived from miR-15 DKO mice compared to controls (Figures 5D, E). This raised the question whether increased activation of the PI3K/AKT axis would translate into any functional differences, such as enabling cells to proliferate better or longer under limiting IL-7 concentrations. To address this, control cells and primary miR-15a/b fl/fl Vav-Cre DKO progenitors were cultured with 5 ng/ml, 1 ng/ml and 0.1 ng/ ml IL-7 for 48 hrs and cell size as a proxy for proliferation was quantified by flow cytometry ( Figure 5F). Indeed, while the fraction of large, FSC hi cells was comparable when the growth factor was supplied in excess, limiting IL-7 concentrations significantly favored miR-15a/b fl/fl Vav-Cre DKO cells over the control, basically affecting all populations within the progenitor pool ( Figure 5G). We conclude that this pro-proliferative phenotype under limiting IL-7 concentrations likely contributes to the observed B cell progenitor expansion in vivo.

Discussion
In this study, we aimed to characterize the early B cell development of mice lacking expression of the miR-15 family, i.e. miR-15a/16-1 and miR-15b/16-2, anticipating that the absence of obvious phenotypes in individual knockout mice is mainly the result of a compensatory effect. This was motivated by our previous in vitro study, in which a functional knockdown of the miR-15 family in a pre-B cell line facilitated proliferation and counteracted pre-B to immature B cell differentiation. Mechanistically, this in vitro analysis revealed that the miR-15 family exerts its function via direct and indirect targeting of the cell cycle regulators cyclin E1 and cyclin D3, respectively (Lindner et al., 2017). In consequence, cells lacking miR-15 activity failed to exit the cell cycle under growth factor limiting conditions, which appeared critical to induce the transcriptional program required for proper differentiation.
Our experiments presented here, exploiting Cre deleter lines specific for the hematopoietic system (Vav-Cre) and for B cells only (Mb1-Cre), largely recapitulate this initial study and demonstrate that the miR-15 family also controls progenitor B cell proliferation and differentiation in a more complex in vivo environment. In particular, we show that B cell development in the absence of miR-15a, miR-15b and miR-16 is mainly affected at the pro-B cell stage, in which relative cell percentages were increased in a dosedependent manner, reaching levels of up to three-fold in the DKO. Our detailed analyses demonstrate that this is mainly due to enhanced proliferation, rather than resistance against apoptosis, resulting in a competitive advantage when compared to wild type progenitor B cells. Pre-B cell percentages and the pre-B to immature B cell ratio were elevated in deficient mice as well, albeit much more pronounced in the B cell specific deletion of the miR-15 clusters, suggesting that both stages are controlled by the miR-15 family under physiological conditions. We argue that the differences in the two Cre strains likely reflect the distinctive kinetics of gene deletion in both systems: Loss of the miR-15 family at an early developmental time-point, such as in the hematopoietic stem cell compartment by Vav-Cre, is fully established once cells enter the pro-B cell stage, and consequently all miR-15 target genes will already be derepressed. We hypothesize that this promotes proliferation and interferes with differentiation of cells at both the pro-and pre-B cell stages, but that the latter phenotype may be partially masked by the pronounced pro-B cell increase. In particular, we propose that a stronger, albeit incomplete block at the pro-B cell stage simply reduces the number of cells that proceed to the next developmental step. Here, fewer pre-B cells may have an advantage to compete for microenviromental factors that promote differentiation, such as CXCL-12, which may in turn weaken the block at the pre-B to immature B cell transition. Indeed, such a model is supported by the phenotype of the single knockouts, which show a milder pro-B cell increase, but already the same pre-B to immature B cell ratio as the double knockout. Mb1-Cre, in contrast, initiates deletion only in pro-B cells and thus at the developmental stage that is mostly affected by loss of the miR-15 clusters. It is therefore tempting to speculate that miR-15-mediated repression in the Mb1-Cre B cell progenitors decreases only gradually, possibly due to the residual presence of mature miR-15 family members even upon disruption of the alleles. Supporting such a model of a delayed knockout, the pro-B cell phenotype in the Mb1-Cre system is much smaller compared to the Vav-Cre-mediated deletion, however, the block at the pre-B to immature B cell transition becomes more pronounced, which recapitulates the phenotype we have previously described with pre-B cell lines in vitro (22).
A scenario in which both proliferative stages of early B cell development are controlled by the miR-15 family is not surprising, as pro-and pre-B cells rely on similar signals, mainly derived from the IL-7 receptor. Moreover, the differentiation of both pro-to pre-B and of pre-B to immature B cells is critically dependent on stalling of the cell cycle to enable RAG1/2-mediated VDJ recombination, making a similar regulatory mechanism likely. Supporting this, we found Ccne1 (encoding cyclin E1) and Ccnd3 (encoding cyclin D3) to be strongly upregulated upon miR-15 deletion also in pro-B cells in vivo, together with other cell-cycle regulators such as Cdc25a, all of which have also been described to be derepressed in pre-B or mature B cells lacking miR-15 function (20,22).
An additional key target gene of the miR-15 family in this context, however, appears to be the Il7r gene itself, which we found strongly upregulated on the transcriptional level in pro-B cells and whose enhanced surface expression was then confirmed on all stages of progenitor B cells in the bone marrow. Notably, our in vitro study already reported an increased IL7R expression in pre-B cells lacking the miR-15 family, but Il7r mRNA levels were not altered in this system (22). In consequence and due to our own selection criteria, we did not consider this gene as a direct target at that time, but have argued for an indirect effect. The discrepancy between these in vitro data and our in vivo findings may be due to different ratios of posttranscriptional verses translational inhibition of the miR-15 family on the Il7r gene, with an almost exclusive translational effect in the pre-B cell lines. That said, the experiments presented here clearly demonstrate a direct repression of the Il7r gene, which is in accordance with a T cell-related study (44), and identify two largely conserved binding sites that additively contribute to the posttranscriptional regulation of this gene.
On the functional level, the upregulation of the IL7R upon loss of the miR-15-mediated repression resulted in a clear activation of the PI3K/AKT pathway, indicating that receptor expression is rate-limiting for signal transduction at least under the tested experimental conditions. We therefore conclude, in particular in the light of the well-established pro-proliferative roles of PI3K signaling in B cell progenitors, that the increased IL7R expression drives the accumulation of pro-and pre-B cells in our system, and likely also interferes with proper differentiation as reported before (22). Notably, we have also observed an upregulation of c-KIT, a second growth factor receptor with a critical role in pro-B cells, upon deletion of the miR-15 clusters, and it is conceivable that this enhanced expression also contributes to the pro-proliferative phenotype. Indeed, several in vitro studies indicate a stimulatory role of signaling via c-KIT in pro-B cells (45)(46)(47), however, B cells lacking c-Kit gene expression have been shown to develop normally (48). While this points to a non-essential role of c-Kit in lymphopoiesis, it does not exclude a positive impact once upregulated, in particular in the context of c-Kit being considered as a proto-oncogene in several cancer entities (49). Thus, it remains to be determined whether the increased c-KIT protein levels are a bystander effect without any functional relevance, or whether it in fact aggravates the progenitor B cell phenotype induced by loss of miR-15 family members.
In addition to the derepression of cell cycle regulators, the IL7R and possibly c-KIT, other direct targets of the miR-15 family may be implicated in the pro-proliferative progenitor phenotype as well. Arl2 (ADP-ribosylation factor 2), for example, the gene with the strongest effect in the 3' UTR assay due to the presence of four conserved mir-15 binding sites, is involved in ADP/ATP transport in mitochondria and thus plays a role in the cellular energy metabolism. Confirming this gene as a target, loss-of miR-15b has been shown to upregulate Arl2 in cardiomyocytes, resulting in increased cellular ATP levels (50). Moreover, it has been reported that the miR-15-mediated regulation of proliferation and G0/G1 cell cycle arrest is at least to some extent established via Arl2 (51). It is thus tempting to speculate that the increased Arl2 expression upon loss of the miR-15 family also elevates ATP levels in B cell progenitors, resulting in enhanced cycling.
Beyond early B cell development and the progenitor B cell phenotype, we observe a slight reduction in mature B cells in Vav-Cre mice both in relative terms as well as absolute numbers, albeit not to a statistically significant extent with respect to the former. We attribute this to a B cell extrinsic effect that is likely secondary to the expansion of the myeloid compartment. In contrast, the mature B cell compartment is not affected upon Mb1-Cre-mediated deletion, which appears the more suitable system to address this point. This indicates that the loss of the miR-15 familywhile interfering with proper progenitor B cell differentiationcan be compensated and/or is tolerated on the systemic level, at least with regard to lymphocyte output. We speculate that the surprising absence of a reduction in mature B cells may be explained by two possible, not mutually elusive mechanisms. First, the phenotype may simply correct itself at the mature B cell stage because of the opposing nature of the progenitor B cell phenotype. From our data, we conclude that pro-B and pre-B cell differentiation are compromised upon loss of miR-15 expression, which is indicated by the altered pro-B to intermediate and pre-B to immature B cell ratios in the bone marrow. At the same time, however, miR-15-deficient cells proliferate faster and have a competitive advantage. Thus, it appears feasible that these two phenomena, enhanced proliferation on the one hand and a reduced rate of differentiation on the other hand, balance each other. Relative to the wild type situation, the differentiation rate may be lower, but this is possibly compensated by an increased number of cells that in theory can undergo differentiation, ultimately resulting in a largely unperturbed output of mature B cells. Alternatively, it is conceivable that the developmental block upon miR-15 deletion is so mild that the mature B cell pool is simply filled up to normal levels over a short period of time.
This, of course, raises the question of the physiological function of the miR-15 family. In our study, we have only assessed B cell development under steady-state conditions. Thus, it is possible that the impact of miR-15a/b/16-mediated regulation of pro-and pre-B cell proliferation and differentiation only becomes apparent when the system is challenged. Here, it will be particularly interesting to investigate whether loss of miR-15 promotes or accelerates development of ALL-like leukemia in suitable mouse models.
Moreover, loss of the two dominant hematopoietic miR-15 clusters alone may be sufficient to drive progenitor B cell leukemia, although we have not seen signs for this in the relatively small cohorts used for the immunophenotyping. Last, while we have excluded the third cluster of the miR-15 family for this analysis due to the absence of an immune cell-specific phenotype (18), we cannot exclude that miR-497 and miR-195 may partially compensate for the loss of the miR-15a/16-1 and miR-15b/16-2 clusters. It will therefore be important to generate triple knockout mice for all clusters, which may uncover additional aspects of the physiological role of the miR-15 family in B cells.
In the non-B cell population, our study furthermore confirms and reveals additional immune cell compartments that are particularly sensitive to a small decrease in miR-15 levels, as it occurs in either miR15a/16-1 or miR-15b/16-2 single knockout mice, as well as compartments that tolerate deletion of one cluster but are severely affected by the loss of the whole miR-15 family. With respect to the former, we have recapitulated the previous finding that NK cells are dependent on high miR-15 levels, since deletion of the miR-15a/16-1 cluster alone already resulted in a maturation block (Sullivan et al., 2015). Interestingly, loss of the miR-15b/16-2 cluster did not alter NK cell maturation in our hand, but a combined deletion of both clusters further aggravated the developmental block at the stage III to IV transition. We hypothesize that miR-15b/16-2 is simply expressed to a lower extent in this subpopulation, and that its loss fails to reduce miR-15 levels below the critical threshold that provokes a phenotype. The concomitant deletion of both clusters in such a setting, however, completely abolishes miR-15 family expression and consequently enhances the developmental delay.
The myeloid compartment, on the other hand, appears to be less sensitive to changes in miR-15 levels. Here, loss of either cluster is well tolerated, but a complete deletion of both clusters provokes an increase of myeloid cells, in particular neutrophils, in the bone marrow and spleen already in young mice. These findings are in accordance with recent reports describing the onset of acute myeloid leukemia (AML) at an age of 5 months upon combined whole body deletion of miR-15a/16-1 and miR-15b/16-2 (34). This has been linked to deregulated expression of Cyclin D1, Cyclin D2 and anti-apoptotic Bcl-2 (34), but it is likely that additional miR-15 target genes identified or confirmed in this study are involved as well. One interesting candidate in this context is Cdc25a, which has already been reported to be regulated by miR-16 in an AML subtype (FLT3-ITD AML) and serves as a critical factor for AML development (Sueur et al., 2020). Similarly, our analysis in B cell progenitors has identified Fam122a, encoding an inhibitor of the wellestablished tumor suppressor protein phosphatase 2A (PP2A), as a direct miR-15 target. Fam122a overexpression has been shown to enhance cell proliferation and colony formation (52), and a recent study even indicates that Fam122a is required for AML growth and survival and that it promotes AML development in vivo (53). For the future, it will therefore be interesting to investigate whether these genes are indeed causally linked to AML in the miR-15 DKO mouse model.
Taken together, our data confirm and build on previous findings regarding the role of the miR-15 family in NK cell maturation and myeloid expansion, and establish a novel function in early B cell development in vivo. In correspondence with its described tumor-suppressive function, physiological levels of miR-15 family members restrict or limit expansion of progenitor B cells at several stages, thereby ensuring proper transitions between phases of proliferation and differentiation. In analogy to most miRNAs that have been extensively studied so far, the miR-15 family appears to exert this role by modulating an extensive network of target genes, in this case including direct cell cycle regulators as well as prominent growth factor receptors and their downstream signaling cascades. However, more work is needed to elucidate which target genesboth direct and indirectorchestrate the precise sequence of proliferation and differentiation that is a prerequisite for proper B cell development.

Data availability statement
The datasets presented in this study can be found in the https://www.ncbi.nlm.nih.gov/geo/ online repository under accession number GSE201238.

Ethics statement
The animal study was reviewed and approved by animal welfare committees of the University of Veterinary Medicine Vienna and the Medical University of Innsbruck in accordance with good scientific practice guidelines and national legislation (license numbers: BMBWF-68.205/0023-II/3b/2014 and BMBWF-66.011/ 0021-V/3b/2019).

Author contributions
KH and SH designed the research, performed experiments, and analyzed data. LA and TR performed experiments. TS analyzed data. KH and SH wrote the first manuscript draft and prepared the figures. KH, AV, TR, and SH reviewed and edited the manuscript. AV, TR, and SH acquired funding. All authors contributed to manuscript revision, read, and approved the submitted version.

Funding
This work was supported through the Austrian Science Fund (FWF; P30196-B26) to SH.