B1 lymphocytes develop independently of Notch signaling during mouse embryonic development

ABSTRACT B1 lymphocytes are a small but unique component of the innate immune-like cells. However, their ontogenic origin is still a matter of debate. Although it is widely accepted that B1 cells originate early in fetal life, whether or not they arise from hematopoietic stem cells (HSCs) is still unclear. In order to shed light on the B1 cell origin, we set out to determine whether their lineage specification is dependent on Notch signaling, which is essential for the HSC generation and, therefore, all derivatives lineages. Using mouse embryonic stem cells (mESCs) to recapitulate murine embryonic development, we have studied the requirement for Notch signaling during the earliest B-cell lymphopoiesis and found that Rbpj-deficient mESCs are able to generate B1 cells. Their Notch independence was confirmed in ex vivo experiments using Rbpj-deficient embryos. In addition, we found that upregulation of Notch signaling induced the emergence of B2 lymphoid cells. Taken together, these findings indicate that control of Notch signaling dose is crucial for different B-cell lineage specification from endothelial cells and provides pivotal information for their in vitro generation from PSCs for therapeutic applications. This article has an associated ‘The people behind the papers’ interview.


INTRODUCTION
B1 cells are a subtype of B lymphocyte that colonize body cavities and mucosa. Unlike conventional B lymphocytes (B2 cells), B1 cells are able to mount a rapid response to antigenic stimuli, and they do not depend on T-cells to differentiate into effector antibodyproducing plasma cells (Martin et al., 2001). Studies in mice suggest that B1 cells generate many of the natural antibodies that are present at mucosal sites, the production of which does not depend on prior immunization (Smith and Baumgarth, 2019). In particular, B1 cells produce antibodies that are both antimicrobial, in particular against polysaccharide antigens of microorganisms, and homeostatic, i.e. reactive with auto-antigens (Baumgarth, 2017). Moreover, they display very peculiar features, such as phagocytic ability, antigen presentation and induction of proinflammatory T-cell differentiation (Aziz et al., 2015). For these reasons, they are often referred to as innate-like B-cells and play a pivotal role in innate immunity and tissue homeostasis.
B1 cells are generated predominantly during fetal life and persist throughout adulthood thanks to their self-renewal ability (Baumgarth, 2017;. Because fetal liver (FL) is the primary source for B1a cell production, as shown by transplantation assays (Hardy and Hayakawa, 1991), it has been assumed that FL hematopoietic stem cells (HSCs) are the main provider of postnatal B1a cell pool in the peritoneal cavity. Indeed, recent papers have shown FL HSC-derived B1a cell repopulation (Beaudin et al., 2016;Kristiansen et al., 2016); however, we and others have demonstrated that highly purified FL HSCs failed to repopulate B1a cells upon transplantation (Ghosn et al., 2016;Kobayashi et al., 2019). These contradictory reports led us to investigate the HSC dependency of B1 cell emergence, because understanding their exact origin is crucial for designing protocols for their in vitro generation from pluripotent stem cells (PSCs). Given that B1a cells are not replenished after bone marrow transplant, PSCs would serve as a unique source of B1 cells. In order to determine whether B1 cells have a truly HSC-independent origin, we studied the requirement of Notch signaling for their emergence. It is well accepted that Notch signaling is indispensable for the generation of HSC from hemogenic endothelial cells through the endothelial-to-hematopoietic transition (EHT) (Kumano et al., 2003;Robert-Moreno et al., 2005). However, the role of Notch signaling in HSC-independent hematopoietic programs, such as those generating tissue-resident immune cells (e.g. microglia, epidermal γδT-cells and B1a cells), remains unclear (Dzierzak and Bigas, 2018). Experiments in different animal models have shown that the specification of primitive as well as erythroid-myeloid progenitors (EMPs) in the yolk sac (YS) is largely unaffected in the absence of Notch signaling (Kumano et al., 2003;Hadland et al., 2004;Robert-Moreno et al., 2005;Bertrand et al., 2010), but its requirement for fetal B-lymphocytes is unknown. Using mouse embryonic stem cells (mESCs) to recapitulate murine embryonic development, we found that Notch-deficient (Rbpj −/− ) mESCs are able to generate B1 cells. Our results indicate that Notch signaling is dispensable for the emergence of fetal B1 cells and that its fine control induces B2 cell specification.
Next, we tested the ability of day 5.5 Flk1 + cells, which yield hematopoietic progenitors closely resembling EMPs (Clarke et al., 2015), to undergo EHT (Fig. 1A) and found that Rbpj −/− day 5.5 Flk1 + cells generated significantly fewer clonogenic progenitors (2.9-fold) ( Fig. 1B), in particular of the erythroid lineage (Fig. 1C). In line with previous reports (Clarke et al., 2015), the β-globin gene expression pattern of these erythroid colonies is similar to those of cultured YS-derived EMPs (McGrath et al., 2015), as they express a lower level of Hbb-y compared with EryP-CFC and contain mainly adult β1-globin transcripts (Fig. S2A,B). Flow cytometric analysis for the emergence of CD45 + hematopoietic cells during EHT cultures confirmed the hematopoietic impairment of Rbpj −/− day 5.5 Flk1 + cells by marked reduction of the CD45 + fraction compared with their Rbpj +/− counterpart. This reduction of hematopoietic output from Rbpj −/− day 5.5 Flk1 + cells was already apparent after 48 h of EHT cultures (Fig. 1D). Extending EHT cultures to 144 h resulted in even greater differences in hematopoietic output, measured as the proportion of CD45 + cells, CD11b + myeloid and Ter119 + erythroid cells, as well as clonogenic progenitors in EHT cultures ( Fig. S2C-E). This suggests the deficiency in hematopoietic potential observed in Rbpj −/− day 5.5 Flk1 + cells is unlikely due to variability in the differentiation kinetics of the two mESC lines but is rather an outcome of the absence of Notch signaling. Gene expression analysis via qPCR revealed that known Notch targets (Hes1 and Gata2), as well as pivotal hematopoietic transcription factors (Tal1 and Runx1c), are significantly downregulated in day 5.5 Flk1 + cells (Fig. 1E). These results, in line with those obtained with chimeric Notch1 −/− mice (Hadland et al., 2004), suggest Rbpj is not required for the generation of EMP hematopoiesis but instead alters the proliferation or survival of hematopoietic progenitors. Alternatively, day 5.5 Flk1 + cells, and probably ECs yielding EMPs in the mouse embryos, are heterogeneous and comprise both Notch-dependent and -independent precursors. As both Rbp +/− -and Rbpj −/− -derived Flk1 + cells are devoid of HSC potential (data not shown), collectively, these data indicate that the differentiation of Rbpj −/− mESCs represents a valuable tool for dissecting the Notch signaling requirement in different hematopoietic embryonic progenitors that are generated before HSC emergence.
We then investigated the lymphoid potential of Rbpj −/− ESCs. In the OP9-DL1 co-culture system (Yoshimoto et al., 2009), Rbpj +/− Flk1 + cells differentiated into CD4 + CD8 + double-positive (DP) T cells, whereas Rbpj −/− Flk1 + cells showed a maturation arrest at the CD4 − CD8 − CD25 − (DN1) stage and a lineage switch into CD19 + B cells (Fig. 1F,G), consistent with what was observed in conditional Rbpj −/− mice (Han et al., 2002). This is also compatible with a recent report showing that the first embryonic thymopoiesis-initiating progenitors were generated in the absence of Rbpj (Luis et al., 2016) but were unable to progress beyond the DN1 stage; however, this report did not assess the alternative B-cell fate. The earliest thymic Tprogenitors (ETPs) in DN1 have been reported to possess both B-cell and myeloid potential (Luc et al., 2012), and it is well accepted that lineage differentiation into T-or B-cells depends on Notch signaling (Han et al., 2002;Dallas et al., 2005). Therefore, our observation of alternative Rbpj −/− B-cell specification in T-cell cultures supports a model where early lymphoid precursors are produced and switch to B-cell fate in the absence of Notch signaling.
Next, we tested the effect of Notch signaling inhibition ex vivo in ECs derived from wild-type mouse embryos. FACS-sorted VE-cadherin + (VE-Cad-Cre + ) ECs from E9.5 YS and para-aortic splanchnopleura (P-Sp) were plated on OP9 stromal cells with or Fig. 3. B-cell production from ECs in the E9.5 YS and P-Sp was not affected by Notch signaling inhibition. (A) Sorting strategy of CD31 + VE-cad + ECs from E9.5 YS and P-Sp from wild-type C57BL6 mice is shown. (B,C) Representative FACS analysis (B) and percentages of CD45 + cells expressing CD11b or CD19 (C) in the supernatant of ECs co-cultured on OP9 stromal cells with or without GSI (n=3). (D) Representative FACS analysis of E9.5 YS/P-Sp coculture with OP9-DL1 with/without GSI (n=3 for each group). (E) Percentages of each blood lineage among mononuclear cells (MNCs) in the supernatant of E9.5 wild-type and VE-Cad-Cre: RbpjKO YS and P-Sp, co-culture with OP9 (n=3 for each group). Ter119/CD11b and CD19 expression were examined at day 6 and at day10 of OP9 co-culture, respectively. Student's unpaired t-test (*P<0.05). Data are mean±s.d.
without γ-secretase inhibitor (GSI), a small molecule that inhibits the cleavage of the Notch intracellular domain (NICD), thus preventing the activation of Notch-intracellular signaling (Souilhol et al., 2016) (Fig. 3A). CD11b + myeloid and CD19 + B220 + B cells were similarly produced regardless of GSI addition (Fig. 3B,C). In OP9-DL1 cultures supplemented with GSI, the T-cell development from VE-Cad-Cre + ECs was arrested at the DN1 stage, where B220 + B cells were detected, while DP cells were produced in DMSO-treated control culture (Fig. 3D). We also confirmed the Notch-independent emergence of B-cell potential using VE-Cad-Cre: Rbpj f/f embryos (Cre + RbpjKO). Rbpj deletion in ECs leads to midgestation lethality, thus preventing the assessment of hematopoietic development in vivo. For this, we analyzed ex vivo the hematopoietic potential of E9.5 YS and P-Sp cells. Both Cre + RbpjKO and control cells displayed similar erythromyeloid and B-cell production (Fig. 3E). These results indicate that, although they cannot give rise to HSCs, ECs that are unable to respond to Notch signaling harbor B-lymphoid potential. In addition, we tested the B-cell potential of the major HSC-independent hematopoietic progenitor population present at E9.5: the YS-derived EMP fraction. Sorted cKit + CD16/32 + CD41 + EMP cells from E9.5 wild-type YS (20-25 somite pairs) did not generate AA4.1 + CD19 + B cells (Fig. S2F), in line with the previous reports that only ECs have B-cell potential at E9-9.5 (Nishikawa et al., 1998;Yoshimoto et al., 2011;McGrath et al., 2015).
Collectively, these results indicate that the Notch-independent Bcell specification occurs in ECs.
Next, we examined the positive effect of Notch signaling on the ECs using doxycycline (Dox)-inducible NICD-overexpressing (iNICD) ESCs (Jang et al., 2015). iNICD Flk1 + cells were plated on OP9 with 0, 100 and 500 ng/ml Dox (Fig. 4A). In these culture conditions, Flk1 + cells from iNICD ESCs can produce B-lymphoid cells only in the absence of Dox. When cultured with 500 ng/ml Dox, Flk1 + cells can generate DP and DN CD25 + cells committed to the T-lymphoid lineage (Fig. 4B). Interestingly, both T and B cells were produced in the same culture when 100 ng/ml Dox was added. These results were in line with the previous report that Notch signaling dose determines B-or T-cell specification from BM hematopoietic stem/progenitor cells (Dallas et al., 2005). As we and others showed Notch signaling induced B2 potential in HSC precursors (Lu et al., 2016;Hadland et al., 2017;Kobayashi et al., 2019), we hypothesized that Notch signaling also regulates B1 and/ or B2 lineage determination from Flk1 + -derived progenitors. To test this, we used the modified B-progenitor CFC assay recently described (Montecino-Rodriguez et al., 2016;Kobayashi and Yoshimoto, 2020). Whereas B-progenitors generated in vitro from mESCs or ex vivo from mouse embryos do not express mature B1 or B2 cell markers (such as IgM, CD5 and CD23), the modified B-progenitor CFC assay distinguishes B1 and B2 progenitors based on their AA4.1, CD19 and B220 expression. Under these conditions, B1 progenitors are defined as AA4.1 + CD19 + B220 − , whereas B2 progenitors display an AA4.1 + CD19 − B220 + phenotype (Montecino-Rodriguez et al., 2016;Kobayashi et al., 2019). We cultured iNICD ESC-derived Flk1 + cells on OP9 with different doses of Dox. After 7 days of co-culture, the resulting CD11b − CD45 + hematopoietic progenitors were plated onto methylcellulose with OP9 cells. In the absence of Dox, colonies containing uniquely B1 progenitors were generated (Fig. 4C,D), whereas colonies containing both B1 and B2 progenitors were detected only in Dox-treated cultures. Of note, colonies composed exclusively of B2 progenitors were never observed. These results suggest that the activation of Notch signaling elicited B2 potential in Flk1 + -derived progenitors that otherwise would harbor only B1 potential. Thus, fine-tuning Notch signaling plays a role in the control B-cell lineage specification of mESC-derived Flk1 + cells.
Collectively, using Rbpj −/− ESCs and mouse models, we demonstrated that the development of the earliest lymphoid progenitors, in particular of B1 cells, can occur in the absence of Notch signaling. As HSC development is dependent upon Notch signaling, our results strongly support the hypothesis that at least a fraction of B1 cells arises from fetal precursors independently of HSCs. However, our data cannot rule out the possibility that other Notch-dependent hematopoietic progenitors emerging during embryonic development, including fetal HSCs, also contribute in parallel to the B-1 cell pool. These two scenarios are not mutually exclusive and, indeed, the generation of B1 cells is likely developmentally restricted to prenatal development with contribution from different waves of ontogenic progenitors (Beaudin et al., 2016;Ghosn et al., 2016;Kristiansen et al., 2016;Montecino-Rodriguez et al., 2016;Hadland and Yoshimoto, 2018;Kobayashi et al., 2019). Our results also indicate that Notch signaling is a potential regulator of the B1 versus B2 cell lineage specification in Flk1 + -derived progenitors. As conditional Rbpj deletion in adult HSCs does not alter B2 cell production (Han et al., 2002), the finetuning of Notch signaling in B1-and B2-cell specification may occur only in progenitors during embryonic development. Further investigation is required to clarify the lineage choice of B1 versus B2 cells in fetal precursors in vivo. In addition, it remains to be determined whether Notch activation in ECs specifies a B1 and B2 bipotent cell, a common lymphoid or multipotent progenitor, or elicits an HSC program.
Because B1a cells are mainly fetal derived and are not replenished by adult HSCs, the conditioning regimen for HSC transplantation may induce permanent B-1 cell deficiency in humans, which may increase the susceptibility of patients to some microbial infections (Moins-Teisserenc et al., 2013). Additionally, natural IgM antibodies produced by B1 cells have protective roles against atherosclerosis and other inflammatory diseases (Aziz et al., 2015). Therefore, there is an unmet clinical need for B1 re-establishing cell therapies. Our findings are crucial for the future development of strategies for the in vitro generation of B1a cells from PSCs. Although robust protocols for the in vitro derivation of B-cells from human PSCs are still lacking, our results suggest that a stage-specific Notch signaling manipulation should be taken into consideration for the in vitro generation of specific lymphoid subsets.
Mice, tissue collection, processing and transplantation C57BL/6 mice were purchased from Jackson Laboratory and used for breeding with transgenic mice. CD1 mice were purchased from Charles River Laboratories and used for isolating EMPs. VE-cadherin Cre (VE-Cad-Cre) mice (Chen et al., 2009) were obtained from Dr Nancy Speck (University of Pennsylvania, Philadelphia, USA). Rbpj-flox mice (Han et al., 2002) were obtained from Dr Tasuku Honjo (Kyoto University, Japan). VE-Cad-Cre mice were crossed with Rbpj-floxed mice, and double heterozygous mice were crossed with Rbpj-floxed mice to obtain VE-Cad-Cre: Rbpj KO embryos. Mice were timed mated, and YS and P-Sp tissues were dissected from embryonic day (E) 9.5 embryos. Embryonic tissues were staged according to somite pair counts. E9.5 YS and P-Sp were incubated in 0.125% collagenase type I (STEMCELL Technologies) at 37°C for 5-10 min and then dissociated by adding cell dissociation buffer (Thermo Fisher Scientific) with gentle manual pipetting. FL was harvested from E15.5 embryos and made into a single cell suspension by mechanical dissociation using 22G needles and syringes. Mononuclear cells (MNCs) were collected by gradient centrifugation using Histopaque 1083 (Sigma-Aldrich). For transplants, AA4.1 + CD19 + B220 + B progenitors derived from Rbpj +/− and Rbpj −/− mESCs cultured on OP9 were injected into the peritoneal cavity of sublethally irradiated (150 rad) NSG neonatal mice (day 2-5). E15.5 FL MNCs were injected as a positive control. Eight to 12 weeks after transplantation, the recipient peritoneal cells were analyzed for donor cell engraftment by flow cytometry. All procedures involving mice were carried out in compliance with the Animal Care and Use Committee at UTHealth or Ospedale San Raffaele (IACUC 841) and communicated to the Ministry of Health and local authorities according to Italian law.