677 bp of C57BL/6 mouse chromosome 11 containing the miR-142 precursors were amplified and cloned into an AscI site in-between the two loxP sites of the targeting vector. In addition, 2.2 kb (5’) and 3.5 kb (3’) of flanking arms of homology were inserted in BglII - NotI and HindIII – XhoI, respectively. The construct was then introduced into C57BL/6 embryonic stem cells (Artemis). Targeted ES cells were selected for Neomycin-resistance and correct genomic insertion of the floxed miR-142 fragment was evaluated using a Southern blotting strategy. A properly targeted embryonic stem cell clone was injected into C57BL/6 albino blastocysts and the obtained chimera crossed with C57BL/6 mice. The progeny was then bred to FLPe-deleter mice to eliminate the FRT-flanked Neomycin resistance gene (21), ultimately generating the miR-142^fl mouse strain. Finally, the LoxP-flanked miR-142 fragment was amplified from the genome of a miR-142^fl mouse and sequenced to validate the targeting of the locus.

Mb1-cre, R26-Stop^fl-hCD2 and deleter-cre mice have been previously described (22–24). The Mb1-cre strain was obtained from The Jackson Laboratory (stock: 020505). miR-142-deficient mice were generated by breeding miR-142^fl with deleter-cre animals and subsequently intercrossing the offspring carrying a deleted miR-142 allele. Mice were on the C57BL/6 background and kept under specific pathogen-free conditions. Animals were analyzed between 7 and 24 weeks.

Although no obvious segregation of the control genotypes could be noticed in the data, we grouped these animals for clarity as follows (+, f and Δ indicate wild-type, floxed and deleted alleles): Mb1-cre^–: miR-142^f/+, miR-142^f/f and miR-142^f/Δ. Mb1-cre⁺: Mb1-cre⁺ miR-142^f/+, Mb1-cre⁺ miR-142^+/Δ and Mb1-cre⁺. The group of mutant mice (Mb1-cre⁺ miR142^fl) comprises Mb1-cre⁺ miR-142^f/f and Mb1-cre⁺ miR-142^f/Δ. The presence of deleted alleles in mice stemmed from the sporadic activity of Mb1-cre in germ cells (25). Mice could additionally bear the R26-Stop^fl-hCD2 reporter gene.

Mice of both sexes were included in the present work. When analyzed separately, male (n=5-16) and female (n= 4-5) mutant mice tended to show similar alterations in pre- as well as mature B cellularity, frequency of lymph node B cells and S1P-mediated chemotaxis (data not shown). These observations suggest the absence of a strong gender bias induced by the inactivation of the miR-142 locus in B cells.

Pieces of spleens were embedded in Tissue-Tek OCT (Sakura Finetek) and frozen at -80°C. 7 μm-thick splenic sections were cut on a Microm HM 500 OM cryotome and laid on polylysine-coated slides (Thermo Scientific). Samples were fixed in acetone for 5 min, air dried, then washed twice in PBS (136.8 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄·7 H₂O, 1.76 mM KH₂PO₄; Carl Roth) and twice in 0.1% Tween 20 PBS. Slides were next incubated in a blocking solution of 5% skim milk (Sigma-Aldrich) in PBS for 30 min at room temperature, followed by two washes in Tween-PBS. Sections were subsequently stained with B220-PerCP-Cy5.5 (RA3-6B2; Biolegend) and CD169-Bv421 (3D6.112; Biolegend) in blocking buffer, first for 1-2h at room temperature and next at 4°C overnight in a humidifying chamber, then washed twice in Tween-PBS and once in PBS. Finally, cover slips were fixed to the slides using Fluoromount G (ThermoFischer). Data were acquired on a confocal microscope Cell Voyager CV1000 (Yokogawa) using the CV1000 software and processed with the Fiji software (26).

Perimeters of B220⁺ B cell populations contiguous to CD169⁺ macrophage rings, as well as the circumference of the latter rings, were manually outlined in confocal immunofluorescence images using the Fiji software to estimate white pulp and marginal zone areas (in pixels²). An index of circularity (4π x Area/Perimeter²) was also computed, with the maximum value of 1 representing a true circle. In addition, the surfaces occupied by B cells within the two defined areas were calculated using the Fiji software based on B220 fluorescence, selecting the signal intensity above background.

Single cell suspensions of spleen, bone marrow (1 hind leg), peripheral lymph nodes as well as peritoneal cavity were prepared in Dulbecco’s PBS 1x (Sigma-Aldrich) with 2% fetal bovine serum (ThermoFisher). In addition, spleen and bone marrow preparations were treated with Gey’s solution to remove red blood cells. Cells were stained with the subsequent antibodies conjugated to APC, biotin, Bv421, Bv510, Bv605, Bv785, FITC, PE, PE-Cy7 or PerCP-Cy5.5: B220 (RA3-6B2), CD1d (1B1), CD11b (M1/70), CD19 (6D5), CD21 (7E9), CD23 (B3B4), CD25 (PC61), CD3ε (145-2C11), CD5 (53-7.3), CD62L (REA828), CD93 (AA4.1), c-Kit (2B8), hCD2 (RPA-2.10, TS1/8), IgD (11-26c.2a), IgM (Fab fragment and RMM-1), Itgα4 (CD49d; R1-2), Itgβ1 (CD29; HMβ1-1), LFA-1 (CD11a/CD18; H155-78), Ly6G (1A8), S1PR1 (713412) and TCRβ (H57-597) obtained from Biolegend, Jackson ImmunoResearch, Miltenyi, R&D systems and ThermoFisher. When required, samples were first stained with biotinylated antibodies followed by a second step with mixes including streptavidin coupled to an aforementioned fluorochrome and other labelled antibodies. Data were acquired on FACSCanto II and LSRFortessa flow cytometers (BD Biosciences) and analyzed using the FlowJo software (BD Biosciences).

Splenic B cells were enriched by magnetic depletion, removing non-B cells using the following antibodies coupled to biotin: CD11b (M1/70), CD11c (N418), CD4 (GK1.5), CD5 (53-7.3), CD8α (53-6.7), Gr1 (RB6-8C5), NK1.1 (PK136) as well as Ter119 (TER-119) (Biolegend), and the MagniSort Streptavidin Negative Selection Beads according to the manufacturer’s instructions (ThermoFisher). Alternatively, splenic CD19⁺B220⁺CD93^–IgM⁺CD1d⁺ Fo B cells were FACS-isolated on a FACSAriaIII (BD Biosciences).

5x10⁶ splenocytes/ml were serum-starved for 2h at 37°C and 5% CO₂ in high glucose DMEM medium (Sigma-Aldrich) supplemented with 0.5% fatty acid-free BSA (Carl Roth), 2 mM L-glutamine (Sigma-Aldrich), 10 mM HEPES (Lonza), 100 U/ml penicillin and 100 μg/ml streptomycin (Sigma-Aldrich). Cell migration was assessed by placing 5 μm transwell inserts (Sarstedt) in 24-well plates. 600 μl medium containing or not 1 μg/ml mouse CCL21 (Peprotech), 30 or 100 ng/ml mouse CXCL12/SDF1α (Peprotech) and, 20 or 100 nM d18:1 sphingosine-1-phosphate (Sigma-Aldrich) (27–30), were supplied to the bottom chambers and 5x10⁵ splenocytes were added to the top chambers. In addition, 5x10⁵ cells were seeded in a well filled with 600 μl of medium without transwell insert (input). Plates were incubated for 3h at 37°C and 5% CO₂. Cells collected from the bottom chambers and the input wells were subsequently stained to identify CD23⁺ and CD23^lo/– CD19⁺B220⁺CD93^–IgM⁺CD1d⁺ Fo as well as CD19⁺B220⁺CD93^–IgM^hiCD1d^hi MZ B cells in flow cytometry. Events were recorded for 90s at medium speed on a LSRFortessa (BD Biosciences). B cells that migrated to the lower chambers were expressed as proportions of the inputs as follows: (events of B cell subset in the lower chamber/events of B cell subset in the input) x 100. A minimum threshold of 20 events/given population detected in the bottom chamber upon treatments was applied for inclusion in the analysis.

Splenocytes isolated from Mb1-cre⁺ control and mutant mice were labelled for 10 min at 37°C with Cell Proliferation Dye eFluor450 and eFluor670 (10 μM final), respectively, following the provided protocol (ThermoFisher). Cell suspensions were counted and equal numbers of labelled splenocytes (in Dulbecco’s PBS 1x) from pairs of control and mutant donor mice were mixed. Recipient animals (two for each donor mix) were injected intravenously with 200 μl (14 - 20x10⁶ cells) of the cell preparations using insulin syringes. After 2h, the spleen and peripheral lymph nodes of the recipient mice were collected for flow cytometry.

From these data, the B (% of CD19⁺ cells)/T (% of CD5⁺ cells) ratios within Mb1-cre⁺ and Mb1-cre⁺ miR142^fl donor cells in the spleens and peripheral lymph nodes of the recipients were determined and normalized to the B/T ratios in the respective donor mixes. In addition, normalized donor mutant/control ratios within IgM^+/hiCD21^+/hiCD93^– mature B cells were similarly calculated.

Recipient mice were of the following genotypes: wild type, miR-142^f/+ and miR-142^f/+ R26-Stop^fl-hCD2.

2x10⁶ magnetically-isolated splenic B cells were lysed in whole cell extraction buffer: 25 mM HEPES (Sigma-Aldrich), 0.3 M NaCl (Carl Roth), 1.5 mM MgCl₂ (Carl Roth), 0.2 mM EDTA pH 8.0 (Sigma-Aldrich), 0.5% Triton X-100 (Sigma-Aldrich), 10 mM NaF (Sigma-Aldrich), 10 mM Na-pyrophosphate (Sigma-Aldrich), 100 μM Na-o-vanadate (Sigma-Aldrich), 2 mM DTT (Sigma-Aldrich), supplemented with protease inhibitors (cOmplete Protease Inhibitor Cocktail, Roche). 30-50 μg of proteins were separated by SDS-PAGE (31) and transferred onto PVDF membranes (Merck). Membranes were subsequently probed with anti-FOXO1 (C29H4, Cell Signaling Technology) and anti-HSP90 (F-8, Santa Cruz Biotechnology) primary antibodies, goat anti-mouse as well as anti-rabbit IgG secondary antibodies conjugated to HRP (Bio-Rad), and signals visualized using ECL prime reagent (GE Healthcare). Data were recorded on Chemidoc Imaging System (Bio-Rad) and analyzed using the Image Lab software (Bio-Rad).

Determination of Sell, Klf2, Foxo1, Bcor, S1pr1, S1pr3 and Grk2 expression: RNA was prepared from 1-2x10⁶ magnetically-purified splenic B cells using the Quick-RNA Microprep kit (Zymo Research). cDNA synthesis was performed using random hexamers and the RevertAid First Strand cDNA synthesis Kit (ThermoFisher). Levels of mRNAs were assessed using AceQ SYBR qPCR Master mix (Vazyme) and the primer pairs indicated below on a QuantStudio 7 Flex instrument (ThermoFisher). Samples were run in duplicates and analyzed using the QuantStudio software. Transcript abundance in mutant B cells was normalized to Hprt mRNA levels as well as to their expression in control cells. Primers used to detect: Sell (5’-CTTACTGGGGCTCGAGGAAC and 5’-TCCAACAGTGAGTTCCATGGT), Klf2 (5’-CTGCGTACACACACAGGTGAGAA and 5’-AAGTGGCACTGAAAGGGTCTG), Foxo1 (5’-CGGCGGGCTGGAAGAATT and 5’-TTCTCCGGGGTGATTTTCCG), Bcor (5’-AGCCAAAGTCAGTCACCCTG and 5’-TGTCCTCTGGGGCTTCAAAG), S1pr1 (5’-CGGTGTAGACCCAGAGTCCT and 5’-AGCAGCAGATGAGAATGAAC) (32), S1pr3 (5’-AGTCTTAGCTGAGACACGGC and 5’-GCGCCAGGAACGTTCATTTC), Grk2 (5’-TCTGGAGGACCGAGGAGAAG and 5’-AGGCAGAAGTCCCGGAAAAG) and Hprt (5’-TCCTCCTCAGACCGCTTTT and 5’-CCTGGTTCATCATCGCTAATC).

Evaluation of miR-142-3p levels: RNA was isolated from FACS-sorted splenic CD19⁺B220⁺CD93^–IgM⁺CD1d⁺ Fo B cells using the Direct-zol RNA Microprep kit (Zymo Research). miR-142-3p and RNA U6 were specifically reverse transcribed out of 10 ng of total RNA using TaqMan MicroRNA assay (AssayID 000464 and 001973, respectively) and TaqMan MicroRNA reverse Transcription kit (ThermoFisher). Expression of these RNAs was assessed in duplicates using the primers from the TaqMan MicroRNA assay and TaqMan Fast Advanced Master Mix on QuantStudio 7 Flex instrument. Data were analyzed using the QuantStudio software. miR-142-3p abundance was normalized to RNA U6 levels.

The Prism software (Graphpad) was used to create graphs, calculate means as well as standard deviations, and determine statistical significance using one-way ANOVA with Tukey’s post hoc multiple comparisons test, two-tailed unpaired Student’s t test and one sample t test against a theoretical mean of one.

While the mature B compartment is altered in miR-142-null mice (17), the individual contributions of the miR-142 locus specifically to B cell physiology remain obscure. In addition, progenitor B cells lacking Dicer demonstrate an enrichment for miR-142-3p targets within upregulated genes (10), suggesting a role for this locus in B cell development. However, the various stages of the latter process have not been examined in the BM of miR-142-null mice. Thus, to gain insights into the cell intrinsic roles of the miR-142 locus in B cells, we generated a mouse allele (miR-142^fl) permitting the conditional elimination of the pre-miRNAs and mature miR-142-5p, miR-142-3p and miR-142b ( Figure 1A ), and crossed it to the Mb1-cre mouse strain to inactivate this locus specifically in the B lineage (23).

We first investigated the early developmental stages (pro-, pre- and immature B cells) in the BM of our mutant mice ( Figure 1B ). Pre-B cell numbers tended to be lower (1.75 – 1.9-fold) in Mb1-cre⁺ miR-142^fl compared to control mice ( Figure 1B ). Intriguingly, the pro- and immature B cell compartments seem unaffected by the absence of miR-142 miRNAs in B cells ( Figure 1B ), suggesting a specific role for these miRNAs in the accumulation of pre-B cells. Next, we examined the splenic B populations in Mb1-cre⁺ miR-142^fl mice. We predicted increased numbers of maturing transitional B cells as well as a strong accumulation of MZ B cells in these animals, based on the previous analysis of miR-142-null mice (17). Remarkably, similar numbers of transitional B cells were detected in Mb1-cre⁺ miR-142^fl animals and controls ( Figure 2A ). There was also no expansion of MZ B cells in the spleens of the conditional mutant mice. Instead, we could not recognize a distinct IgM^hiCD1d^hi MZ B population in the Mb1-cre⁺ miR-142^fl animals ( Figure 2A ). Accordingly, mutant MZ B cell numbers were strongly reduced (8- to 9-fold) ( Figure 2A ). We confirmed these results using an alternative identification strategy. The proportions of IgM^hiIgD^lo MZ B cells were similarly decreased in the spleens of Mb1-cre⁺ miR-142^fl mice ( Figure 2B ). Mutant MZ B cells do not seem to be misplaced. We did not observe them to be present in the blood ( Figure 2C ). Likewise, an IgM^hiCD21^hi MZ B population was not visible in the BM or LNs of Mb1-cre⁺ miR-142^fl mice, as illustrated in panels A and C of figure 3. We can, however, not exclude that mutant MZ B cells aberrantly move to a tissue not examined in our experiments. In contrast to MZ B cells, mature B cells with a Fo B phenotype (IgM⁺CD1d⁺ and IgM⁺IgD⁺) could be clearly identified in the mutant mice ( Figures 2A, B ). However, this population was about 1,7-fold smaller in the spleens of the mutant animals ( Figure 2A ). A sizable fraction of these splenic mutant IgM⁺CD1d⁺ (Fo) B cells displayed low CD23 expression compared to their control counterparts (36% vs 6.8-10%) ( Supplementary Figure 1A ). The levels of CD23 were also mildly reduced on miR-142-deficient compared to control CD23⁺ Fo B cells ( Supplementary Figure 1A ). In contrast, CD21 levels tended to be slightly elevated on the mutant B cells ( Supplementary Figure 1A ). Our observations are consistent with the reported altered expression of these two markers on splenic mature B cells from miR-142-null mice, which Kramer at al. interpreted as an indication of the increased presence of mutant MZ B cells (17). We decided to consider both CD23⁺ and CD23^lo/– mature B cells in Mb1-cre⁺ miR-142^fl mice as Fo B cells based on their IgM⁺IgD⁺CD1d⁺ phenotype. The vast majority of mutant and control Fo B cells expressed the Cre-inducible reporter gene hCD2, although the frequency was slightly lower for the miR-142-deficient population ( Figure 2D ). Furthermore, levels of miR-142-3p were substantially reduced (~170-fold) in splenic Fo B cells isolated from Mb1-cre⁺ miR-142^fl mice compared to control cells ( Figure 2E ). These findings indicate that the mature B compartment in Mb1-cre⁺ miR-142^fl mice is mainly composed of cells that had inactivated the miR-142 locus, i.e., had not escaped Cre-mediated recombination. We also assessed the generation of mature B1 cells. These cells were absent in the peritoneal cavity of Mb1-cre⁺ miR-142^fl mice ( Supplementary Figure 2 ), in agreement with the results of Kramer et al. in null animals (17).

Splenic MZ and Fo B cells are defined by their anatomical locations as well. Thus, we examined B cell positioning in the spleens of our mice by fluorescence microscopy ( Figure 2F ). The spleen architecture of Mb1-cre⁺ miR-142^fl animals appeared grossly similar to controls, showing ring-shaped formations of MZ macrophages and Fo B cell follicles in the white pulp ( Figure 2F ). A typical thin smooth layer of MZ B cells surrounding the ring of macrophages could be observed in the controls ( Figure 2F ). Instead, the distribution of mutant B cells in the marginal zone appeared irregular ( Figure 2F ). To quantify splenic B cell positioning, we estimated the areas covered by these cells in the white pulp and marginal zone ( Figure 2G ). In control animals, splenic B cells occupied expectedly more space in the white pulp than in the marginal zone ( Figure 2G ). On average, the area including control B cells in the MZ was equivalent to 25 – 37% of the surface covered by them in the white pulp ( Figure 2G ). Interestingly, B cells appeared more evenly distributed between these two locations in the spleen of Mb1-cre⁺ miR-142^fl mice ( Figure 2G ), with the area holding mutant B cells in the MZ corresponding to 86% of the region populated by the latter cells in the white pulp. In addition, B cells tended to occupy more space in the marginal zone of the mutant compared to control animals ( Figure 2G ). Finally, we determined an index of circularity for the border of the B cell population located in the MZ ( Figure 2G ). A value of 1 represents a perfect (smooth) circle. This index was noticeably lower in the mutant animals, in agreement with a more irregular distribution of B cells in the region of the marginal zone. Thus, these data are consistent with the flow cytometry analysis suggesting the absence of a bona fide MZ B population and the presence of Fo B cells in Mb1-cre⁺ miR-142^fl mice.

Since several of our observations on B cells in Mb1-cre⁺ miR-142^fl mice differ from published work in miR-142-null mice (17, 19), we evaluated whether the discrepancies were caused by the design of our mutant allele. Thus, we crossed our miR-142^fl mice with a deleter-cre strain to generate animals deficient for the miR-142 locus in the germline (22). Consistent with the literature, we observed an accumulation of splenic B cells in our miR-142^–/– mice ( Supplementary Figure 3A ). However, this higher cellularity appeared to result from the expansion of mature IgM⁺CD1d⁺ (Fo) B cells in our mice ( Supplementary Figure 3A ), instead of the anticipated MZ B cells (17). We could not detect a distinct IgM^hiCD1d^hi MZ B cell population in our miR-142-null animals ( Supplementary Figure 3A ), similar to the situation in Mb1-cre⁺ miR-142^fl mice ( Figure 2 ). Since no information is available, we also examined the early stages of B cell development in the BM of our miR-142 knockout mice. These mice demonstrated higher immature B cell numbers than controls (means of 2.1x10⁶ vs 1.2x10⁶), while pro- and pre-B cellularity appeared normal ( Supplementary Figure 3B ). Unexpectedly, we detected a clear pre-B cell population in the spleens of our miR-142^–/– animals ( Supplementary Figure 3B ). The results are compatible with enhanced production of B cells in miR-142-null mice, which could participate in the increased pool of splenic mature mutant B cells alongside the proposed pro-survival contribution of higher BAFFR levels (17). Of note, T cell numbers were reduced and neutrophil cellularity was increased in the spleens of our miR-142^–/– mice ( Supplementary Figures 3C, D ), consistent with previous work (16, 17, 19). Thus, the phenotype of our miR-142-deficient mice resembles the one reported in other null mice (16, 17, 19), excluding potential flaws in the design of the miR-142^fl allele and validating our findings in Mb1-cre⁺ miR-142^fl animals.

Overall, our results indicate cell autonomous roles for the miR-142 locus at several stages of B cell development. Thus, the inactivation of this locus selectively in B cells led to an undersized pre-B cell population and a lack of B1 cells. In addition, an aberrant Fo B population was detected in Mb1-cre⁺ miR-142^fl mice, with reduced cellularity and abnormal expression of CD23. Also, genuine MZ B cells could not be observed by flow cytometry or immunofluorescence microscopy in the mutant animals.

Since Fo B cells circulate between lymphoid tissues (1, 13), we examined the size of the mature population in peripheral LNs and BM upon conditional ablation of the miR-142 locus in B cells. Remarkably, the frequency of LN B cells was strongly decreased in Mb1-cre⁺ miR-142^fl compared to control mice (6,8- to 8,5-fold on average) ( Figure 3A ). Fo B cells, detected as IgM⁺CD21⁺ mature B cells, were the dominant B cell population in the LNs of all groups of mice ( Figure 3A ), implying a specific loss of LN Fo B cells in the mutant animals. In agreement, evaluations of CD19⁺ B cellularity in the inguinal LNs indicated that these tissues contained only 2.8x10⁴ B cells in Mb1-cre⁺ miR-142^fl mice compared to 8.5-11.8x10⁵ in controls ( Figure 3B ). Inguinal LN CD5⁺ T cellularity was not significantly impacted in the mutant mice ( Figure 3B ). Consistently, the B to T cell ratios in these tissues dropped substantially in the Mb1-cre⁺ miR-142^fl mice ( Figure 3B ). The limited impact on T cell numbers suggests that the paucity of B cells is not an indirect result of dysfunctional LNs in the mutant mice ( Figure 3B ). Similar to the spleen, a substantial fraction of LN Fo B cells was CD23^lo/– in Mb1-cre⁺ miR-142^fl mice ( Supplementary Figure 1B ). However, surface levels of CD23 and CD21 were not noticeably altered on LN mutant CD23⁺ and total Fo B cells respectively ( Supplementary Figure 1B ), differing from the observations made on their splenic counterparts ( Supplementary Figure 1A ). Mb1-cre⁺ miR-142^fl mice also demonstrated a reduction (3-fold) of mature Fo B cellularity in the BM ( Figure 3C ). Proportions of CD23^lo/– Fo B cells tended to be higher in the BM of the mutant mice as well, although the increase did not reach significance ( Supplementary Figure 1C ).

Thus, the overall numbers of Fo B cells appear more affected in the BM and especially LNs than in the spleens of Mb1-cre⁺ miR-142^fl mice. These findings point to a possible role for the miR-142 locus in the regulation of naïve Fo B cell migratory properties.

The miR-142 locus has been invoked in the modulation of mouse neutrophil and CD4⁺ T cell migration (33, 34). Focusing on the LNs, we hypothesize that altered responses to chemokines could contribute to the greatly reduced numbers of mutant B cells in these tissues.

Both CCR7 and CXCR4 promote the homing of naïve Fo B cells to LNs, with signals from CCR7 seemingly playing a more important role (35, 36). Accordingly, we determined the chemotactic responses of splenic IgM⁺CD1d⁺ Fo B cells towards CCL21 and CXCL12, performing in vitro transwell assays ( Supplementary Figure 4 ). Control IgM^hiCD1d^hi MZ B cells were included in these analyses for comparison since these cells exhibit migratory properties distinct from Fo B cells. In the absence of chemokines, the random access of mutant Fo B cells to the bottom chamber of the transwell was minimal and comparable to control Fo B cells ( Figures 4A, B ). miR-142-proficient and -deficient Fo B cells migrated similarly in the presence of CCL21 (1 μg/ml), although the chemotactic response of the mutant cells appeared slightly better ( Figure 4A ). Both types of Fo B cells demonstrated higher sensitivity to CCL21 signals than control MZ B cells ( Figure 4A ). These results are consistent with the reported chemotactic responses of wild-type splenic mature B populations to CCL19, a second CCR7 ligand (37, 38). In addition, we tested the possibility of distinct behaviors for mutant Fo B cells expressing high and low levels of CD23 in CCL21-mediated chemotaxis. CD23^lo/– miR-142-deficient Fo B cells migrated slightly less than their CD23⁺ counterparts in presence of the chemokine ( Figure 4A ). Yet, proportions of migrating mutant CD23^lo/– B cells remained within the range seen for control Fo B cells, excluding a compromised response to CCL21.

Mature B cell responses to CXCL12-mediated chemotaxis were also consistent with previous work ( Figure 4B ) (29). Both control Fo and MZ B cells migrated towards CXCL12 (100 ng/ml), with the former cells tending to be better responders ( Figure 4B ). The mutant Fo B cells seemed to migrate on average slightly more than their control counterparts ( Figure 4B ). Similar observations were made at a lower CXCL12 concentration (30 ng/ml) ( Figure 4B ). miR-142-deficient Fo B cells displayed some mild chemotactic response, while control cells barely migrated. Within the miR-142-deficient Fo B population, the CD23⁺ and CD23^lo/– cells did not obviously differ in their sensitivity to CXCL12 signals ( Figure 4B ), although the mutant CD23^lo/– B cells showed a modest tendency to migrate less well in presence of the chemokine.

Since the above in vitro migration assays suggest functional CCR7 and CXCR4 signaling, the expression of the LFA-1 (αLβ2) integrin could instead be affected upon inactivation of the miR-142 locus in B cells. A minor participation for α4-containing integrin, like VLA-4 (α4β1), interaction with VCAM1 in the firm adhesion of lymphocytes to HEVs has also been reported (39). We therefore monitored surface levels of LFA-1, Itgα4 and Itgβ1 on Fo B cells ( Figure 4C ). As expected (40), the abundance of LFA-1 and Itgβ1 was higher on MZ compared to Fo B cells, while Itgα4 levels were similar, in the spleens of control mice. The surface expression of these molecules was, however, not lower on the mutant Fo B cells compared to their control counterparts in either the spleen or LNs. Itgβ1 expression was even higher (1.5 – 2-fold) on the mutant cells, reaching levels observed for control MZ B cells ( Figure 4C ).

Overall, the responsiveness of splenic Fo B cells from Mb1-cre⁺ miR-142^fl mice to chemotactic signals from both CCL21 and CXCL12 in vitro appears comparable to their control counterparts. The levels of integrins mediating the attachment of B cells to the HEVs were also not decreased on the mutant cells. Thus, defective firm adhesion during homing seems unlikely the cause of the LN B cell paucity in the mutant animals.

Since the firm adhesion to HEVs of mutant Fo B cells does not appear problematic, the rolling/tethering step during their homing to LNs could be compromised instead. This latter process depends on the expression of CD62L by immune cells (5, 41, 42). Indeed, CD62L-null mice demonstrate a drastic reduction in peripheral LN cell content (43).

Proportions of Fo B cells positive for CD62L were on average reduced by 30% in the spleens of Mb1-cre⁺ miR-142^fl mice compared to controls (51,4% vs 80.04%) ( Figures 5A, B ). Similar observations were made in the LNs ( Figures 5A, B ). Of note, the fraction of CD62L⁺ mutant Fo B cells was higher in the LNs than in the spleens, consistent with a need for lymphocytes to express sufficient L-selectin to enter the former tissues. In comparison to controls, surface levels of L-selectin also tended to be lower (1.5- to 2-fold) on CD62L⁺ mutant Fo B cells, in particular in the LNs ( Supplementary Figure 5A ). Next, we evaluated whether cells with decreased surface L-selectin would be mainly found within CD23^lo/– mutant Fo B cells. Unexpectedly, splenic and LN proportions of CD62L⁺ cells were diminished in CD23^lo/– compared to CD23⁺ Fo B populations for both control and mutant mice ( Supplementary Figure 5B ). Nevertheless, frequencies of CD62L⁺ cells were consistently lower for the mutant B cells independently of their levels of CD23 ( Supplementary Figure 5B ). Thus, the reduction in L-selectin affects the whole mutant Fo B population and does not correlate with changes in CD23 expression.

We next attempted to identify molecular alterations that could lead to the impaired display of surface CD62L on mutant B cells. Levels of the CD62L transcript (Sell) in splenic B cells from Mb1-cre⁺ miR-142^fl mice were only about 32% of the expression observed in controls by real time PCR ( Figure 5C ). These results are compatible with inefficient transcription of L-selectin in the mutant Fo B cells. FOXO1 and KLF2 promote the transcription of Sell in mouse B cells (44–47). FOXO1 can also modulate to some extent Klf2 expression (45). Thus, we examined the levels of these two transcription factors in the B cells from our mutant mice. Klf2 and Foxo1 mRNA levels were moderately decreased by on average 30% and 40%, respectively, in splenic miR-142-deficient compared to proficient B cells ( Figure 5C ). The polycomb repressive complex 1.1 has been reported to participate in establishing a closed chromatin conformation at the Foxo1 locus in acute myeloid leukemia blasts (48). A component of the latter complex is BCOR, a predicted miR-142-5p target (Targetscan.org). Increased BCOR amount could thus contribute to limit Foxo1 expression in our mutant B cells due to enhanced formation of polycomb complexes. A major effect of eliminating miRNA-mediated repression is the improved stability of target transcripts (9). Yet, no clear increase in the abundance of Bcor was detected in splenic mutant B cells ( Figure 5C ), suggesting that this repressor may not be regulated by miR-142-5p in mature B cells. In agreement with the above results, the amount of FOXO1 was also reduced in splenic miR-142-deficient B cells compared to controls, by Western blotting ( Figure 5D and Supplementary Figure 6 ). The relative reduction in FOXO1 protein appeared however much stronger than the decrease seen at the transcript levels in the mutant B cells, implying a possible additional defective regulation of FOXO1 stability in the latter cells.

In summary, the miR-142 locus appears involved in establishing the typical surface levels of CD62L seen on Fo B cells. In particular, these miRNAs seem to favor the production of FOXO1, ultimately permitting an optimal transcription of Sell. These findings also support a contribution for impaired B cell tethering/rolling in HEVs to the reduced size of the mutant LN B population.

We decided to verify that the migration of miR-142-deficient B cells to the LNs is indeed compromised ( Figure 6 ). Splenocytes from control and mutant mice were isolated, labelled with proliferation dyes, mixed in equal parts, and injected into recipient animals. 2h post-transfer, the presence of labelled B and T cells in the spleens and LNs of the recipients was evaluated by flow cytometry. T cells were used as a reference since their physiology is not affected by our targeting strategy.

Proportions of labelled B and T cells in the spleens of recipients were comparable to their frequencies in the donor mixes ( Figure 6A ). Accordingly, the normalized B/T ratios within control and mutant donor cells averaged 1, indicating that these various lymphocyte populations similarly accumulated in the spleen. In the recipient lymph nodes, percentages of labelled T cells were higher while the fractions of labelled B cells were reduced compared to the donor mixes ( Figure 6A ), in agreement with the known propensity of the former cells to efficiently accumulate in LNs. Notably, the proportions of mutant B cells were clearly smaller than their control counterparts, leading to significantly lower (3.9-fold) B/T ratios in the case of mutant compared to control donor cells ( Figure 6A ). These observations reveal an impaired accumulation of miR-142-deficient B cells in the recipient LNs. They also remind one the decreased B/T ratios seen in the inguinal LNs of Mb1-cre⁺ miR-142^fl compared to control animals ( Figure 3B ). Likewise, the normalized ratios of labelled mutant to control donor cells within the IgM^+/hiCD21^+/hi mature B populations in the recipients were close to 1 in the context of the spleen whereas the values strongly dropped for the LNs ( Figure 6B ).

Thus, miR-142-deficient mature B cells cannot properly accumulate in LNs. Since the evaluations were done 2h after adoptive transfer, the reduced mutant population size appears primarily the result of defective homing likely due to impaired CD62L expression. In contrast, mutant B cell migration to the spleen seems unperturbed, consistent with L-selectin not being necessary for the entry of lymphocytes into this organ (43).

Finally, we checked whether the miR-142 locus could also play a role in the regulation of B cell egress from lymphoid tissues. Therefore, we determined the effects of S1P chemotactic signals on Fo B cells from Mb1-cre⁺ miR-142^fl mice in transwell assays in vitro. Migration of MZ B cells from the spleen of control mice was visibly induced by 100 nM S1P (13-31-fold increased vs no S1P) while the Fo B cells barely responded ( Figure 7A ), as expected from the literature (28, 37). miR-142-deficient Fo B cells migrated clearly more to 100 nM of S1P than their control counterparts and in proportions comparable to MZ B cells ( Figure 7A ). This trend was already observed at a lower concentration (20 nM) of the phospholipid ( Figure 7A ). The improved chemotactic response to S1P did not appear dependent on CD23 expression by the mutant Fo B cells, since both CD23⁺ and CD23^lo/– populations reacted similarly to the chemoattractant ( Figure 7A ).

Two receptors for S1P have well known roles in B cell migration. The predominant S1PR1 chiefly regulates the egress of mature B cells from lymphoid tissues in vivo, while the lesser expressed S1PR3 mediates S1P chemotaxis in vitro (6, 7, 28, 37, 49). The availability of a suitable commercial antibody allowed us to assess S1PR1 abundance on mature B cells by flow cytometry, although the detected signals were dim like in previous work (50, 51). Surface expression of S1PR1 appeared higher (1.7-fold) on mutant Fo B cells compared to controls in the spleen and LNs ( Figure 7B ). This expression did however not reach the levels seen on control MZ B cells ( Figure 7B ), the strongest expressors of S1PR1 among splenic B cells (51).

Absence of the kinase GRK2 in B cells results in higher surface levels of S1PR1 and limits their accumulation in peripheral LNs (50, 51). The kinase also interacts with S1PR3 in endothelial cells (52), suggesting a role for GRK2 in the internalization of the latter receptor as well. We thus examined Grk2 expression in control and mutant splenic B cells. Transcript levels of the kinase were only modestly reduced (25% on average) in B cells lacking miR-142 miRNAs ( Figure 7C ). It seems unlikely that such a subtle change in expression is the main driver of elevated S1PR1 surface levels on mutant Fo B cells. Therefore, we determined whether the abundance of S1pr1, as well as S1pr3, may be enhanced in miR-142-deficient cells. Remarkably, no clear accumulation of either receptor’s mRNAs could be observed in splenic mutant B cells compared to controls ( Figure 7C ). Thus, the higher surface levels of S1PR1 on mutant Fo B cells does not appear to stem from the increased transcription of its gene or, since S1pr1 is a predicted target of miR-142-5p (Targetscan.org, Pictar.mdc-berlin.de), from the relief of miRNA-mediated transcript destabilization. While the role of S1PR3, a possible target of miR-142-3p, in mutant B cell response to S1P is unclear, our data rule out increased mRNA levels as a possible contribution to the perturbed physiology of mature miR-142-deficient B cells.

Collectively, the above findings indicate that the miR-142 locus participates in weakening Fo B cell responsiveness to S1P signals and may thus slow down their exit from LNs in vivo. Although the molecular mechanism remains to be uncovered, miR-142 miRNAs could regulate B cell egress in vivo by limiting the amount of S1PR1 available at the cell surface.