Dynein light chain regulates adaptive and innate B cell development by distinctive genetic mechanisms

Mechanistic differences in the development and function of adaptive, high-affinity antibody-producing B-2 cells and innate-like, “natural” antibody-producing B-1a cells remain poorly understood. Here we show that the multi-functional dynein light chain (DYNLL1/LC8) plays important roles in the establishment of B-1a cells in the peritoneal cavity and in the ongoing development of B-2 lymphoid cells in the bone marrow of mice. Epistasis analyses indicate that Dynll1 regulates B-1a and early B-2 cell development in a single, linear pathway with its direct transcriptional activator ASCIZ (ATMIN/ZNF822), and that the two genes also have complementary functions during late B-2 cell development. The B-2 cell defects caused by loss of DYNLL1 were associated with lower levels of the anti-apoptotic protein BCL-2, and could be supressed by deletion of pro-apoptotic BIM which is negatively regulated by both DYNLL1 and BCL-2. Defects in B cell development caused by loss of DYNLL1 could also be partially suppressed by a pre-arranged SWHEL Igm-B cell receptor transgene. In contrast to the rescue of B-2 cell numbers, the B-1a cell deficiency in Dynll1-deleted mice could not be suppressed by the loss of Bim, and was further compounded by the SWHEL transgene. Conversely, oncogenic MYC expression, which is synthetic lethal with Dynll1 deletion in B-2 cells, did not further reduce B-1a cell numbers in Dynll1-defcient mice. Finally, we found that the ASCIZ-DYNLL1 axis was also required for the early-juvenile development of aggressive MYC-driven and p53-deficient B cell lymphomas. These results identify ASCIZ and DYNLL1 as the core of a transcriptional circuit that differentially regulates the development of the B-1a and B-2 B lymphoid cell lineages and plays a critical role in lymphomagenesis.

Introduction B lymphoid cells are essential for the humoral immune response by producing a diverse range of antigen-specific antibodies. Antibody-mediated immunity is provided by two distinctive B cell lineages that diverge early in life. The better-known conventional, or B-2, B cells provide adaptive immunity by producing high-affinity pathogen-specific antibodies, typically in a T cell-dependent manner. The pool of B-2 cells is constantly replenished from hematopoietic stem and progenitor cells in the bone marrow [1]. B-1a cells provide innate-like immunity by producing "natural" low-affinity, broad-specificity antibodies against a wide range of overlapping antigens, including self-antigens. In contrast to B-2 cells, the pool of B1-a cells is largely established at birth and these cells are able to self-renew as mature cells in the periphery [2,3].
During their differentiation in the bone marrow, B-2 lineage precursors go through a series of well-defined developmental stages [1,4]. Critical events are the rearrangement of the immunoglobulin (Ig) heavy chain-encoding Igh locus through V(D)J recombination at the pro-B cell stage (Hardy fraction B); association of this unique IgH with a generic surrogate Ig light chain (λ5 and v-preB) to form the pre-B cell receptor (pre-BCR) in pre-B cells (fraction C); pre-BCR-dependent clonal expansion by~5 cell division cycles during the large pre-B cell stage (fraction C') [5]; VJ recombination of the Igλ/κ light chain (IgL) loci during the small pre-B cell stage (fraction D); and association of IgH and IgL chains to form an IgM complex/B cell receptor (BCR) at the immature B cell stage (fraction E). The immature B cell stage represents a critical quality control phase during which cells with high-affinity self-reactive BCRs are eliminated through BIM-dependent apoptosis [6]. Immature B cells that pass this quality control stage exit the bone marrow, and undergo further maturation in peripheral lymphoid tissues, such as the spleen and lymph nodes. There, upon stimulation by cognate antigens and T cell help, activated B cells can again enter the cell cycle, alternating with further diversification of the V(D)J-rearranged Ig loci through somatic hyper-mutation of Ig variable regions and class-switch recombination of Igh constant regions.
Whereas adult B lymphopoiesis in the bone marrow almost exclusively produces B-2 cells, B-1a cells are derived from fetal stem cells, and B-1a precursors dominate early B cell development in the fetal liver and early post-natal spleen [2,3,7]. Thus, in mice, B-1a lineage cells constitute the major B cell compartment until~3 weeks of age [8,9]. In adults, B-1a cells are less numerous than B-2 cells and most reside in the peritoneal cavity, where they also can undergo further AID-mediated antibody diversification by somatic hyper-mutation and Ig class-switch recombination, albeit in a stochastic, age-dependent manner that appears to be independent NHMRC program grant (1016701), Cancer Council Victoria grant (1052309), Leukemia and Lymphoma Society (LLS) SCOR grant (7001-13) to AS; NHMRC program grant (1054925) and NHMRC Principal Research Fellowship (1060675) to DMT; and NHMRC Independent Research Institutes Infrastructure Support, and Victorian State Government Operational Infrastructure Support grants. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. of exogenous antigens [10]. B-2 and B-1a cells differ in their transcriptional programs as well as growth factor dependence [3,11], and they have markedly different Ig repertoires [2]. In particular, B-1a pools seem to be biased towards expressing V(D)J-rearranged IgH chains that associate only poorly with the surrogate light chain [2]. Although the understanding of the functional differences between B-1a and B-2 cells is continuously increasing, the developmental mechanisms that underlie these differences still remain poorly understood.
The Zn 2+ -finger protein ASCIZ (also known as ATMIN [12,13]), which functions as a highly specific transcription factor for the multifunctional dynein light chain, DYNLL1 (also known as LC8) [14][15][16], plays critical roles in B-2 cell development [13,17] and B cell lymphomagenesis [18]. DYNLL1 is a common subunit of the cytoplasmic, intra-flagellar and axonemal Dynein motor complexes [19][20][21][22], but also binds numerous Dynein-independent targets [23,24], including the apoptosis initiating BH3-only protein BIM [25]. We have previously shown that ectopic expression of DYNLL1 could rescue the severe defect in B cell development caused by the absence of ASCIZ, confirming that defective regulation of DYNLL1 plays a key role in the defects in B lymphopoiesis caused by Asciz deficiency.
Based on its established role in the B cell defects observed in ASCIZ-deficient mice, we sought to directly investigate the role of DYNLL1 during B cell development. We show here that conditional deletion of Dynll1 largely phenocopies the B-2 cell developmental defects seen in Asciz-deleted mice, although the two genes function in part by different mechanisms. We also show that the ASCIZ-DYNLL1 axis is particularly critical for the development of B-1a cells. Surprisingly, the B-1a and B-2 cell deficiencies in Dynll1-deleted mice were differentially affected by genetic modifiers, indicating that DYNLL1 regulates the development of the two B cell lineages via distinct genetic mechanisms.

DYNLL1 is required for B-2 cell homeostasis in mice
To investigate the role of DYNLL1 in B cell development, we deleted a conditional floxed Dynll1 allele (Dynll1 fl ) using the Mb1-Cre knock-in allele, which is efficiently expressed from the late pre-pro-B cell stage onwards [26]. Peripheral blood cell analyses at 4 weeks-of-age revealed a severe depletion of circulating B cells in Mb1-Cre ki/+ Dynll1 fl/fl -deleted mice compared to Mb1-Cre ki/+ or Dynll1 fl/fl controls ( Fig 1A). Similar lymphopenia was observed in the spleens of 8-week-old Mb1-Cre ki/+ Dynll1 fl/fl -deleted mice, with an >8-fold reduction of mature B cells numbers (Fig 1B-1D). Although transitional B cells, which give rise to both follicular and marginal zone B cells, were reduced in the spleens of Dynll1-deleted mice (Fig 1C), only the follicular B cell numbers were affected by the absence of DYNLL1 (S1 Fig), presumably because marginal zone B cells have a much longer half-life than follicular B cells [27]. Genotyping and Western blot analyses of MACS-purified splenic B cells from Mb1-Cre ki/+ Dynll1 fl/fl mice confirmed the efficient deletion of the targeted Dynll1 alleles, and complete loss of DYNLL1 protein (Fig 1E and 1F). These results indicate that DYNLL1 is required for normal B cell production in mice.

DYNLL1 is required for the proliferation and survival of developing B-2 linage cells in the bone marrow
To determine how loss of DYNLL1 leads to reduced peripheral B-2 cell numbers, we monitored B cell development in the bone marrow of 8-week-old mice using surface marker staining according to Hardy [5] (Fig 2). There were no differences during the first two stages (fractions B and C) following Mb1-Cre mediated recombination of the floxed Dynll1 alleles, but loss of DYNLL1 led to a >2-fold reduction of the cycling pre-B cell pool (fraction C') ( Fig  2B and 2C). This difference was maintained during the small (post-cycling) pre-B cell stage (fraction D), and was further exacerbated to an 8-to-10-fold reduction of immature B cells (fraction E) and recirculating B cells (fraction F) in Dynll1-deleted mice compared to Mb1-Cre controls ( Fig 2C and S2 Fig). These data reveal that DYNLL1 has critical functions during two distinct developmental stages: the cycling pre-B cell stage and the immature B cell stage.
To explore how loss of DYNLL1 lead to reduced cell numbers at these two stages, we performed cell cycle analyses of fraction C' cells (which represents the only developmental stage during which B-2 cell precursors proliferate [5]) and viability analyses of fraction E cells. FACS-sorted pro-B and pre-B cells (CD19 + IgM -) were stained for DNA content analysis, and gated on the larger cycling pre-B cells (Fig 2D). This analysis revealed a >2.5-fold increased proportion of pre-B cells in G 0 /G 1 phase in Dynll1-deleted mice compared to the control animals, and a corresponding reduction of S and G2/M phase cells (Fig 2D and 2E). Moreover, even though Dynll1-deleted mice contained~10-fold fewer immature B cells, the proportion of apoptotic cells within this fraction was increased by >2-fold compared to the Mb1-Cre control animals (Fig 2F and 2G). There were no significant differences in the cell surface levels of the pre-BCR or BCR in early pre-B cells or immature B cells, respectively (S3 Fig). This indicates that the developmental defects at these stages are not due to altered pre-BCR or BCR expression. These results indicate that loss of DYNLL1 impairs pre-B cell proliferation to a large extent by delaying them in G1 phase, and that the additional cell loss at the immature B cell stage is due to increased apoptotic cell death.
Defective B cell development in Dynll1-deleted mice can be suppressed by loss of pro-apoptotic BIM but not by loss of p53 To evaluate possible reasons for the G1 phase delay in pre-B cells and the increased apoptosis of immature B cells caused by loss of DYNLL1, we performed additional gene deletion studies. Increased proportions of cells in G1 phase are often the result of a p53-mediated checkpoint arrest, for example in response to endogenous or exogenously induced DNA damage, or mitotic chromosome separation defects during the previous cell cycle [28]. To test this hypothesis, we additionally deleted the Tp53 gene in these mice. However, concomitant conditional deletion of Tp53 did not increase the numbers of cycling fraction C' cells (and subsequent fraction D and E cells) in Mb1-Cre Dynll1-deleted mice ( Fig 3A). This demonstrates that the delayed cell cycling was not a consequence of a p53-mediated cell cycle arrest.
The pro-apoptotic BH3-only protein BIM plays key roles in the survival of B lymphoid cells during multiple stages of their development, for example in response to limiting concentrations of cytokines [29,30] and in the negative selection against self-reactive BCRs [6]. To test if BIM, which is known to be negatively regulated by DYNLL1 through direct binding, was involved in the B cell development defects of Dynll1-deleted mice, we also deleted Bim in these cells. Remarkably, co-deletion of Bim completely restored the number of fraction C' cells, as well as the subsequent fraction D and E cells, in the absence of DYNLL1 back to the amounts found in Mb1-Cre control mice ( Fig 3B)[note that the increased numbers of recirculating B cells (fraction F) cells are a consequence of abnormally increased peripheral B cell numbers, a characteristic of Bim knockout mice [6,17,31]].
Collectively, these results indicate that the numerical aberrations at different stages of B cell development of Dynll1-deficient mice, are predominantly driven by BIM-dependent apoptosis but independent of p53.

Loss of DYNLL1 leads to reduced expression of BCL-2 in B lymphoid cells
To determine how deletion of Bim may rescue B cell development in the absence of DYNLL1, we measured the expression of BIM and other BCL-2 family proteins in Dynll1-deleted mice  (Fig 4). The total levels of BIM protein, and its subcellular distribution, were not altered in Dynll1-deleted B cells compared to control cells (Fig 4A and 4B). Unexpectedly, we found that expression of the BIM antagonist BCL-2 was consistently and significantly reduced by~2-fold in both mature and developing Dynll1-deleted B lineage cells compared to cells from control animals ( Fig 4B, 4C and 4E). This finding was independently confirmed in B cells that lack the Dynll1-transcription factor ASCIZ, and consequently contain only very low amounts of DYNLL1 ( Fig 4D). Finally, reverse transcription PCR analyses indicated that the expression of Bcl2 was reduced at the mRNA level ( Fig 4F).
These data indicate that DYNLL1 is required for efficient Bcl2 expression and to restrain BIM-mediated apoptosis in developing B cells. The relative cell losses at the pre-B and immature B cell stages in Dynll1-deleted mice were reminiscent of what we have previously reported for mice with a B cell-specific conditional deletion of ASCIZ, the critical transcriptional regulator of Dynll1 [17]. Thus, to determine the extent to which these two genes function in the same pathway, we examined additional cohorts of Mb1-Cre Dynll1-deleted and Mb1-Cre Asciz-deleted, as well as Mb1-Cre Asciz/Dynll1-double-deleted mice for side-by-side comparisons of B-2 cell development. Interestingly, this analysis showed that loss of either Dynll1 or Asciz caused quantitatively almost identical effects on B cell development from fraction C' onwards ( Fig 5A). Moreover, compared to the single mutants, the reduction in the cycling fraction C' cells (which is carried over to the subsequent small pre-B stage, fraction D) was not further increased in Asciz/Dynll1 double mutants compared to the single mutant animals. This demonstrates that ASCIZ regulates the proliferation of cycling pre-B cells entirely through DYNLL1.
In contrast to their epistatic interaction at the cycling pre-B cell stage, we found that concomitant deletion of both genes led to a further 5-to-10-fold depletion of immature B cells B-2 lymphopenia in Dynll1-deleted mice can be rescued partially by transgenic expression of a rearranged B cell receptor The more severe phenotype of the Asciz/Dynll1 double mutant animals (which completely lack DYNLL1) compared to the Asciz single mutants (which still retain~5-10% of normal levels of DYNLL1 [14,17,32]) indicated that even very low residual amounts of DYNLL1 can sustain some of its functions either in immature B cells, or during the transition from the small pre-B to the immature B cell stage. As the level of apoptosis of immature B cell was comparable between Dynll1-deleted ( Fig 2G) and Asciz-deleted mice [17], we hypothesized that the low remaining DYNLL1 levels in Asciz mutants may have a specific role in the generation, rather than the survival, of immature B cells. As the generation of immature B cells (which are defined by the expression of a complete BCR) depends on the successful recombination of the Igλ/κ loci during the small pre-B cell stage (fraction D), we tested whether complementation with a pre-arranged BCR transgene could restore the numbers of immature B cells in Mb1-Cre Dynll1-deleted mice.
Remarkably, the anti-hen egg lysozyme-specific SW HEL Igh knock-in and Igl transgene [33] restored the numbers of immature B cells in Mb1-Cre Dynll1-deleted mice to a level comparable to those seen in Mb1-Cre control mice (Fig 6A and 6B), although they did not quite reach the abnormally elevated levels of immature B cells in Mb1-Cre SW HEL control mice (presumably because the chicken-specific SW HEL BCR does not activate auto-reactive apoptotic programs in mice). It should be noted that the SW HEL transgene did not increase the numbers of immature B cells in Mb1-Cre Asciz-deleted mice [17]. Thus, the specific rescue effect of the pre-arranged BCR for Mb1-Cre Dynll1-deleted mice supports the notion that DYNLL1 may play an ASCIZ-independent role in some aspect of Igl gene rearrangement or BCR surface expression.
Loss of DYNLL1 or ASCIZ leads to severe peripheral B-1a lymphopenia Next, we quantified B-1a cells in 4-week-old Dynll1-deleted, Asciz-deleted and Asciz/Dynll1double deleted mice. At this age, B-1a cell pools are established in the peritoneal cavity and thereafter expand through self-renewal [2,3,7]. Remarkably, in all three mutant strains, B-1a cell numbers in the peritoneal cavity were reduced by >100-fold compared to age-matched controls (Fig 7A and 7B). This demonstrates that the ASCIZ-DYNLL1 axis is crucial for the development of the initial B-1a cell pool. Peritoneal B-1a cells are derived from precursors in the fetal liver that transition through the spleen during the early post-natal period. In fetal livers at day E17.5, B-1a progenitors (Lin -CD93 + B220 -/low CD19 + ) [34] were reduced by~2.5-fold in Dynll1-deleted embryos compared to control littermates (Fig 7C and 7D), while early B-2 progenitors were present at normal levels (Fig 7C and 7D). In the fetal livers of Dynll1-deleted embryos there was also an~6-fold reduction in the more prominent Lin -CD93 + B220 int CD19 + fraction, which most likely represents more differentiated B-1a precursors (Fig 7C and 7D). Interestingly, at 4-5 days of age, the spleens of Dynll1-deleted mice contained >20-fold fewer transitional B cells, which at this stage primarily represent B-1 lineage cells [8], than littermate controls (Fig 7E and 7F). Overall, these results indicate that the defects resulting in the impaired production of B-1a lineage cells in the absence of DYNLL1 start at the progenitor stage in the fetal liver and become more severe around the neonatal transitional stage, when these cells first become surface IgM-positive. As all three mutant mouse strains had a quantitatively identical defect in establishing the initial B-1a cell pools at 4 weeks of age, we assume that Asciz-deleted and Asciz/Dynll1-double deleted mice would have comparable B-1a progenitor and transitional cell deficiencies to Dynll1-deleted embryos and newborns.
It should be noted that between 4 and 8 weeks of age, peritoneal B-1a cell numbers increased considerably in Dynll1 and Asciz single mutants, presumably through accelerated self-renewal, but interestingly, they did not increase in the Asciz/Dynll1-double mutants ( Fig  7B and S5 Fig). These results reveal that ASCIZ and DYNLL1 function in a single linear pathway that is essential for the initial establishment of the B-1a cell pool, but they may have some partially independent functions during the subsequent self-replicating expansion of B-1a cells.
Differential rescue of the B-2 but not B-1a lymphopenia of Dynll1-deleted mice by the SW HEL B cell receptor transgene or loss of Bim As shown above, the B-2 lymphopenia in Dynll1-deleted mice could be ameliorated by expression of the pre-arranged SW HEL BCR knock-in/transgene or by deletion of pro-apoptotic Bim. We therefore also investigated how these genetic modifiers affect the B-1a cell lineage in these mutant mice.
Surprisingly, in contrast to its pronounced rescue effect on B-2 cell numbers (Fig 3B), deletion of Bim had no protective effect on peritoneal B-1a cells in DYNLL1-deficient mice ( Fig  8A and S6 Fig). The numbers of B-1a cells in Dynll1-deleted mice were also not increased, but rather further reduced by expression of the SW HEL transgene ( Fig 8B).

Differential synthetic lethal interaction between loss of Dynll1 and oncogenic MYC expression in B-2 but not B-1a cells
We recently reported that loss of DYNLL1 (or ASCIZ) leads to a dramatic synthetic lethal interaction with oncogenic MYC expression driven by the Eμ-Myc transgene in B-2 lineage cells [18]. We therefore also monitored the effect of the Eμ-Myc transgene on B-1a cells in Mb1-Cre Dynll1-deleted mice. In pre-malignant 8-week-old Eμ-Myc mice, B-1a cell numbers in the peritoneal cavity were similar to control animals ( Fig 9A and S7 Fig). Notably, B-1a cells in Dynll1-deleted Eμ-Myc mice were not further reduced relative to the low levels seen in DYNLL1-deficient mice (Fig 9A). Similar results were also observed in Asciz-deleted Eμ-Myc mice compared to Asciz-deleted controls (Fig 9A). In fact, in both Asciz or Dynll1 mutant mice, there was a trend towards a modest rescue of the B-1a lymphopenia by Eμ-Myc, possibly reflecting increased MYC-driven self-renewal of the B-1a cell pool. Nevertheless, together with our previous findings, these results show that oncogenic MYC expression in combination with loss of DYNLL1 (or ASCIZ) has a specific synthetic lethal effect on the B-2 but not the B-1a cell lineage.
The ASCIZ-DYNLL1 axis contributes to the early onset of B-1a-like lymphomas in p53-deficient Eμ-Myc mice Heterozygosity of the tumour suppressor Tp53 in Eμ-Myc mice leads to the extremely early development of aggressive B cell lymphomas that cause severe morbidity requiring euthanasia by 40 days of age [35]. As conventional B-2 lymphopoiesis only becomes prevalent shortly before this age, the early onset of these lymphomas in Eμ-Myc Tp53 +/mice suggests that they are most likely derived from the B-1a cell lineage. To test if the ASCIZ-DYNLL1 axis contributes to the development of early postnatal lymphomas, we generated compound mutant Mb1-Cre Tp53 fl Asciz fl Eμ-Myc mice. Similar to germline Tp53 +/heterozygosity, conditional B-lymphoid-specific loss of one allele of p53 caused early-onset MYC-driven lymphomas with a median survival of 37 days (Fig 9B), compared to a median survival of~120 days for the Mb1-Cre Eμ-Myc control mice on this genetic background [18]. Median survival was further reduced to 26 days in homozygous Tp53-deleted Eμ-Myc mice (Fig 9B). Importantly, conditional deletion of Asciz significantly extended the median survival of homozygous Tp53deleted Eμ-Myc mice by 11 days and by 21 days in heterozygous Tp53-deleted Eμ-Myc mice. Thus, these data indicate that the ASCIZ-DYNLL1 axis is required for the rapid development and severity of these early onset B cell lymphomas in mice.

Discussion
The data reported here demonstrate that the multi-functional dynein light chain DYNLL1 is essential for the normal development of both the adaptive and innate-like B-2 and B-1a cell lineages. Overall, the phenotypes for B cell-specific conditional Dynll1-deleted mice are remarkably similar to the phenotypes reported for Asciz-deleted mice. This congruence directly confirms the importance of DYNLL1 for B cell development in general, and also highlights DYNLL1's role as the key effector of ASCIZ in B cells, as was previously invoked by the rescue of Asciz-deficiency phenotypes by retroviral complementation with DYNLL1 [17].
Our finding that ASCIZ and DYNLL1 act in a single linear pathway during the development of pre-B-2 cells (Fig 5) and B-1a cells (Fig 7A and 7B) represents a rare example where the biological functions of a transcription factor can be explained by the reduced expression of a single target gene. In the context of B cell development, this specificity is remarkable as the progression from one stage to the next is typically regulated by networks of transcription factors that activate a wide range of functionally important effectors [36,37]. However, the synergistic interaction between loss of Asciz and loss of Dynll1 at the immature B cell stage implies that these proteins must also have some non-overlapping, developmental stage-specific functions. Moreover, this synergistic effect also shows that-on an Asciz-deficient backgroundvery low residual levels of DYNLL1 can still support some molecular functions that are lost in the complete absence of the protein in Asciz/Dynll1 double mutant animals.
An important mechanistic finding was that the defects in B-2 lineage development caused by the absence of DYNLL1 could be rescued by deletion of Bim, but not deletion of tp53, and that they were associated with significantly reduced levels of BCL-2. The p53-independence of the G1-phase delay seen in Dynll1-deleted cycling pre-B cells indicates that this delay is not due to a primary cell cycle defect or unrepaired DNA lesions. Cell cycle entry at the fraction C' stage of B cell development depends on mitogenic signals from stromal cell derived cytokines, such as IL-7, and tonic ligand-independent activity of the pre-BCR [38,39]. Interestingly, impaired IL-7 signalling is known to diminish cell proliferation at the pre-B cell stage and also causes apoptosis at this stage of B cell development and also at the immature B cell stage [29,30]. IL-7 is required for high-level BCL-2 expression [40], and B cell development defects in IL7-or IL-7R deficient mice can be suppressed by the loss of BIM [29,30]. Based on these similarities between the phenotypes of these animals and those of the Mb1-Cre Dynll1-deleted mice (Figs 3 and 4), a plausible explanation for our observations is that DYNLL1 may be playing an important role in the IL-7/IL-7R signaling pathway in developing B lymphoid cells. Likewise, as BCR activation in mature B cells, which utilizes similar signal transductions pathways to the pre-BCR, has recently been shown to be enhanced by cytoplasmic dynein-  Fig. (B) Tumour-free survival of mice with the indicated genotypes. Mice were euthanised when judged to be critically ill by animal house staff who were blinded to their genotypes. (C) Distribution of B lymphoid cells in tumours of 6 randomly chosen individual mice per group. Pre-B, B220 + IgM -; immature, B220 + CD43 -IgM + IgD -; B1, B220 + CD43 + IgM + IgD -CD19 + CD23 -CD11b + CD5 +/-. https://doi.org/10.1371/journal.pgen.1007010.g009 DYNLL1 regulates B-1a and B-2 cell development 1-mediated receptor clustering [41], it would be conceivable that DYNLL1 could stimulate pre-BCR signaling through a similar dynein-mediated process. Finally, another relevant DYNLL1 effector in this setting could the human-immunodeficiency-related guanine-nucleotide-exchange factor RasGRP1 [42], which is known to be particularly critical for B-1a cell development in mice [43].
A notable difference between the phenotypes of Mb1-Cre Dynll1-deleted and Mb1-Cre Asciz-deleted mice was that the pre-arranged SW HEL BCR transgene could partially rescue immature B cell numbers in the absence of DYNLL1, in contrast to having no effect on ASCIZ-deficient cells [17]. A possible explanation for this difference could be that low levels of DYNLL1 may be necessary and sufficient for V(D)J recombination of the Ig gene loci. As our epistasis analysis indicates that Asciz and Dynll1 function in a linear pathway during early pro-B/pre-B cell development, including during the stage when the Igh locus gets rearranged, but have partially divergent functions during later developmental stages (Fig 5A), this suggests that the effect of DYNLL1 would be specific to the recombination of the Igλ/κ light chain loci. In this context it is interesting to note that DYNLL1 preferentially binds to TQT sequence motifs, which are also potential phosphorylation site motifs for the family of ATM-related kinases [44], including the DNA-dependent protein kinase that is critical for V(D)J recombination [45] and the repair scaffold protein 53BP1 [46].
Within the B-2 cell lineage, it may seem somewhat paradoxical that loss of DYNLL1 only affected the number of follicular B cells but not the number of marginal zone B cells (S1 Fig), even though both are generated by the same B cell development pathway in the bone marrow. This finding is reminiscent of other mouse models with severely reduced follicular B cell numbers but normal marginal zone B cell numbers, for example mice lacking the BAF chromatin remodelling complex subunit BAF155 [47], or the phosphatidylserine translocase ATP11C [48,49]. As indicated above, marginal zone B cells have a much longer lifespan than follicular B cells [27], and they are also present at a much lower number (S1 Fig). So, it is plausible that the marginal zone compartment reaches a steady state faster than the more ephemeral follicular compartment.
A major finding of this study is that loss of DYNLL1 or ASCIZ had even more severe quantitative effects on the development of the innate-like B-1a cell lineage (~100-fold reductions in initial B-1a cell pools; Fig 7B) than on B-2 cells (~10-fold reductions of mature splenic B cells; Fig 1D). In both B-2 (Fig 2) and B-1a cell (Fig 7) development, loss of DYNLL1 leads to significant cell losses at the pre-B stages, which become more dramatic during the immature/transitional stages when B-2 and B-1a lineage cells first become surface-IgM positive. It is thus notable that the loss of Bim and expression of the SW HEL BCR transgene, which can overcome developmental defects caused by loss of DYNLL1 during this stage in the B-2 cell lineage, did not have a comparable rescue effect for B-1a cells. This differential effect of the SW HEL transgene, which even worsened the B-1a lymphopenia in Dynll1-deleted mice (Fig 8), is best explained by the concept that B-1a and B-2 cells have distinctive Ig repertoires. Selection of B-1a lineage-biased IgH chains seems to involve their relatively poor association with the surrogate light chains, and weak antigen affinity of their ultimate BCRs [2]. In contrast, the SW HEL BCR is specific to B-2 lymphocytes, and even B-1a cells of V H 10 heavy chain heterozygous SW HEL mice do not recognize the HEL antigen [33]. This explains why the homozygous V H 10 heavy chain knock-in (which restricts its allelic exclusion during Igh VDJ recombination in pro-B cells) in our SW HEL mice, even without the V κ 10 light chain transgene, led to a notable reduction of B-1a cells, and a corresponding compounding effect on the B-1a lymphopenia in Dynll1 mutant animals (Fig 8B). It would be interesting to test in the future if a transgenic B-1a cell-specific BCR transgene-for example, the phosphatidylcholine-specific V H 12/V κ6 BCRcould suppress the B-1a cell developmental defect of Dynll1-deleted mice, similar to the partial rescue of their B-2 cell defect by the SW HEL transgene.
The pro-apoptotic BH3-only protein BIM has a well-documented critical function in B-2 development to eliminate immature B cells expressing auto-reactive BCRs [50]. In contrast, B-1a cells appear to be selected, at least in part, towards weak auto-reactive BCR specificities [2]. So, it makes sense that BCR ligation by self-antigens does not activate apoptotic cell death programs in this lineage. This is also consistent with the notion that survival of B-1a cells, in marked contrast to B-2 cells, does not depend on the BAFF-R/NFκB pathway, one of the key effector functions of which is to upregulate anti-apoptotic BCL-2 family proteins, such as BCL-2 and MCL-1, which both inhibit BIM-driven apoptosis [51].
B-1a cells have attracted renewed attention based on their widely appreciated roles in innate immune signaling [52,53] and their role in the origin of subsets of both acute lymphoblastic and chronic lymphocytic leukemias [54,55]. It has become clear that functional and developmental differences between B-1a and B-2 cells are driven by divergent transcriptional programs, growth factor sensitivities, and BCR repertoires. Based on the differential roles of Dynll1 in B-2 versus B-1a cells, our findings here add Bim and Myc to the list of genetic factors with highly distinctive lineage-specific functions in B cell development.

Ethics statement
Animal experiments were performed according to the Australian Code for the Care and Use of Animals for Scientific Purposes, 8th Edition (2013), and approved by the St. Vincent's Hospital Melbourne Animal Ethics Committee, approval numbers 047/12, 019/13, and 002/17.
Mice were analysed at the ages indicated in the figure legends. Ethical endpoints of survival analyses of tumor-prone mice were determined by trained animal technician who were blinded to the genotypes. Mice with different genotypes were co-housed in the same microisolators.

Flow cytometry
Spleen, fetal liver, bone marrow and tumour cell suspensions were prepared as described [18]. Bone marrow cellularities included tibias and femurs of two legs. Peritoneal B-1a cells were isolated by lavage using a syringe connected to 26G and then 23G needles with PBS containing 2% FBS. Cell counts were determined using an automated cell counter (KX-21N, Sysmex). Stained cells were analyzed using an LSRFortessa (BD) using FACSDiva (BD) and FlowJo (Tree Star) software. Our gating strategy for analysis of bone marrow cells is shown in S3A  Fig of reference [18]. Mature splenic B cells were isolated using the MACS mouse B cell isolation kit (Miltenyi Biotec). FACS sorting of bone marrow B cells was performed using a FAC-SAria (BD). IgM-depleted bone marrow pre-B cells were isolated using MACS Streptavidin MicroBeads (Miltenyi Biotec). Leukocytes were then incubated overnight in DMEM with 10% FBS at 37˚C to enhance pre-BCR expression before staining and FACS analysis.
Total RNA was isolated using Isolate II RNA Micro Kit (Bioline) and reverse transcribed using a High Capacity cDNA Reverse Transcription kit (Applied Biosystems) following the manufacturer's protocol. Real time PCR was performed using AmpliTaq Gold DNA Polymerase (Applied Biosystems) and a LightCycler 480 (Roche). TaqMan primers for Bcl2 and Hmbs were purchased from ThermoFisher Scientific. Ct values were normalized to Hmbs as the endogenous control and expressed as deltaCt (dCT).

Statistical analysis
Statistical analysis was performed using GraphPad Prism software (San Diego). P values were calculated by two-tailed unpaired Student's t test, or Mantel-Cox test for survival analyses. Grouped data were analyzed using the multiple T test function (one upaired t test per row, each row analysed individually) using the two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli. Error bars represent the mean ± SEM. Numbers of mice per analysis, and numbers of independent experiments are indicated in the figures or legends.