A transient developmental hematopoietic stem cell gives rise to innate-like B and T cells

GFP+ HSCs exist during fetal development

We recently established a lineage tracing model that tracks cells with a history of expression of the tyrosine kinase receptor Flk2 (Flt3). (Figure 1A). Importantly, we have shown that all adult HSCs in this “FlkSwitch” mouse model express Tomato (Tom) and that only Tom+ cells within the adult bone marrow (BM) possess LTMR potential (Boyer et al., 2012; Boyer et al., 2011; Buza-Vidas et al., 2011) (Figures S1A and S1B). Because the switch from Tom to GFP expression is due to the irreversible excision of the Tom gene, GFP+ cells cannot give rise to Tom+ cells (Figure S1C). Thus, adult HSCs cannot be derived from cells with a history of Flk2 expression. The FlkSwitch model thereby provides a unique tool with which to interrogate the developmental relationships between fetal and adult HSCs.

As we previously demonstrated that all adult HSCs in the FlkSwitch model are Tom+ (Figure 1J, middle panel, and S1B) (Boyer et al., 2012; Boyer et al., 2011), all developmental precursors of adult HSCs must be derived via a Flk2-negative lineage. Surprisingly, investigation of reporter expression in phenotypic stem and progenitor cell compartments of FlkSwitch embryos revealed the existence of both Tom+ and GFP+ cells (Figure 1). Neither Flk2 surface protein nor GFP expression was observed prior to embryonic day (E) 10 (Figures 1C, S1D, and S1I). By E10.5, coincident with emergence of definitive hematopoiesis and Flk2 surface expression, GFP+ cells were also found., Both Tom+ and GFP+ cells were consistently observed within the phenotypic compartments enriched for primitive stem/progenitor activity in the yolk sac (YS), aorta-gonad-mesonephros region (AGM), placenta (PL), and fetal liver (FL) between E10.5 and E14.5 during fetal development (Figure 1B,and 1D-H), as well as in the neonate BM at postnatal day 14 (P14) (Figure 1I, middle panel), regardless of phenotypic markers used (Figure S1E-H). Cells appearing both Tom+ and GFP+ had undergone excision of the Tom gene and had thus recently switched from Tom to GFP expression (Figure S2A). As shown previously (Boyer et al., 2012; Boyer et al., 2011; Buza-Vidas et al., 2011), GFP+ cells were not present in the phenotypic (Figure 1J) or functional (Figure S1B) adult HSC compartment, whereas the majority of adult Flk2+ multipotent progenitors (MPPs) expressed GFP (Figure 1J). The presence of both Tom+ and GFP+ cells within the phenotypic stem cell compartments across fetal development opened the possibility of coexistence of two HSC populations during fetal and neonatal hematopoiesis.

GFP+ FL cells are capable of serial, multilineage reconstitution

To determine which fetal population contained functional HSCs, we transplanted Tom+ or GFP+ cells sorted from the cKit+Lin-Sca1+ (KLS) compartment of E14.5 FL of FlkSwitch embryos into sublethally (750 rad) irradiated adult recipients (Figure 2A). We focused on the E14.5 FL stage, as developmental HSCs are relatively abundant and well characterized at this time point. We transplanted more GFP+ cells (1000) as compared to Tom+ cells (500) based on a greater proportion of CD150+ cells in the Tom+ KLS fraction (Figure S2C) (Kim et al., 2006). Surprisingly, and in direct contrast to transplantation of the equivalent populations from the BM of adult FlkSwitch mice (Figure S1B and (Boyer et al., 2012; Boyer et al., 2011; Buza-Vidas et al., 2011), both Tom+ and GFP+ FL populations were capable of long-term multilineage reconstitution (LTMR) (Figure 2B-D and Table S1). Average reconstitution from both Tom+ and GFP+ cells was remarkably robust (6% or greater) and sustained without evidence of decline over time. GFP+ FL KLS cells similarly reconstituted lethally irradiated adult recipients in competition with co-transplanted adult cells (Figure S4A,B). Although Tom+ FL KLS cells gave rise to both Tom+ and GFP+ cells upon transplantation, as previously described (Boyer et al., 2011; Boyer et al., 2012), GFP+ FL KLS cells never gave rise to Tom+ cells (Figure S2B). The absence of Tom+ cells in mice transplanted with GFP+ cells confirmed that the switch to GFP expression within the HSC compartment is irreversible, and that Tom+ cells did not contaminate transplanted GFP+ cells. Although limiting dilution analysis indicated that GFP+ HSCs were present at lower frequency than Tom+ HSCs (Figure 2E), the efficiency of Flk2-Cre mediated recombination is incomplete during fetal development (Figure S3A), and therefore the frequency of GFP+ HSCs is therefore somewhat underestimated by the limit dilution assays of Figure 2E.

To further test the heterogeneity of the FL HSC populations, we subfractionated E14.5 FL KLS with additional cell surface markers. GFP+ and Tom+ FL HSCs expressed similar levels of CD34 (Figure S1L), CD11b (Mac1) (Figure S1M) and CD41 (Figure S1N), but the GFP+ KLS fraction expressed lower levels of CD150 as compared to the Tom+ fraction (Figure S2C). Transplantation experiments revealed that FL HSCs were enriched in the CD150+ fraction for both Tom+ and GFP+ FL KLS cells (Figure S2D-G), and that the CD150- KLS compartment contains little or no LTMR activity (Figure S2H), consistent with previous data (Kim et al., 2006). Furthermore, consistent with a previous report (Christensen and Weissman, 2001), we demonstrated both phenotypic (Figure S3E) and functional heterogeneity within the fetal KLS compartment, with LTMR capability found in both Flk2+ and Flk2- fetal KLS cells from WT (non-FlkSwitch) mice (Figure S3F-H). These data further corroborate the FlkSwitch model, as cell surface expression of Flk2, alone, is capable of distinguishing a phenotypically distinct fetal HSC. Collectively, we show that both Tom+ and GFP+ KLS cells isolated from the FlkSwitch FL were capable of LTMR upon transplantation. As all adult HSCs in the FlkSwitch model are Tom+, GFP+ HSCs appear to exist transiently during development.

To determine whether both populations with LTMR capability of the PB were also capable of repopulating the HSC compartment, we examined BM chimerism in recipients of Tom+ or GFP+ E14.5 FL KLS cells 18 weeks post-transplantation. This analysis revealed donor-derived cells of all phenotypes examined, including HSCs, in the BM of recipient mice (Figure 2F). In addition, we determined whether both populations of fetal HSCs displayed the self-renewal capability necessary to support serial reconstitution, the ‘gold standard’ for defining functional HSCs (Figure 2A). WBM cells from individual primary recipients demonstrating LTMR were transplanted into one or more secondary recipients, and donor-derived contribution to the PB was monitored over an additional 16 weeks. Remarkably, the majority of secondary recipients of either Tom+ or GFP+ FL KLS cells displayed LTMR (Figure 2G-I and Table S2). As in primary recipients, reconstitution was robust for all mature lineages (averaging 10% and above) and sustained over time, with remarkably similar reconstitution kinetics from Tom+ and GFP+ cells. Serial reconstitution was also observed when fewer cells were transferred from primary to secondary recipients, with persistence for longer than 6 months (Figure S4E-H). Consistent with the dogma that HSCs are the only cell population capable of serial reconstitution (Lemischka et al., 1986), LTMR in secondary recipients, including recipients of GFP+ KLS cells,, was accompanied by sustained BM chimerism of stem, progenitor, and mature cell compartments 18+ weeks after secondary transplantation (Figure S5), whereas recipients without sustained LTMR did not display evidence of BM chimerism (Figure S5). These data identify two coexisting populations of fetal HSCs: a Tom+ HSC that likely contributes to the adult HSC compartment, and a transient GFP+ HSC that can be distinguished from co-existing Tom+ HSCs and from adult HSCs by development via a Flk2-expressing pathway. GFP+ HSCs therefore represent a functional HSC with sustained self-renewal and multilineage engraftment potential upon transplantation, but whose existence in situ is restricted to a limited developmental window.

RNA-Seq reveals unique molecular regulators of GFP+ FL HSCs

A fetal HSC capable of LTMR upon transplantation that does not persist into adulthood in situ has never before been identified. To gain further insight into the molecular regulation of this transient HSC population, we performed RNA-sequencing (RNA-seq) analysis of GFP+ and Tom+ FL HSCs (CD150+ KLS), and also compared these populations to adult HSCs. Overall, Tom and GFP+ FL HSCs had a very similar gene expression profile, as expected of two HSC populations. We identified 398 genes that were differentially expressed at least 2-fold between Tom+ and GFP+ FL HSCs, many of which were expressed at similar levels in Tom+ FL HSCs and adult HSCs (Figure 3A and (Table S3). Principal component analysis (PCA) of the same 398 differentially expressed genes identified three distinct HSC populations (Figure 3B). These analyses yielded a unique molecular profile of the distinct properties of the GFP+ HSC, despite their high degree of similarity to Tom+ FL HSCs. Furthermore, hierarchical clustering analysis revealed that Tom+ FL HSCs clustered more closely to adult HSCs (Figure 3A and data not shown), consistent with Tom+ HSCs giving rise to adult HSCs.

Cell-extrinsic and cell-intrinsic mechanisms regulate the lifespan of the GFP+ HSC

RNAseq analysis revealed that genes regulating cell migration and location were differentially regulated between Tom+ and GFP+ HSCs (Figure 3C and Table S3). We therefore investigated whether GFP+ HSCs perish post-birth due to an inability to respond to CXCR4 ligands to seed the BM. However, GFP+ and Tom+ FL HSCs expressed similar levels of CXCR4 and showed equivalent capacity to migrate towards an SDF1 gradient in vitro (Figure 4A). Consistent with normal homing ability, GFP+ HSCs were capable of seeding the BM, as GFP+ FL HSCs were detected within the KLS fraction of the neonate (P14) BM by phenotypic (Figure 1I) and functional analyses (Figure 4B, C). Transplantation of 2000 GFP+ or 500 Tom+ KLS cells from P14 BM led to long-term reconstitution of all myeloid and lymphoid lineages (Figure 4B,C), in a pattern similar to that observed for FL cells (Figure 2D). GFP+ HSCs therefore arise as early as E10.5 (Figure 1D), and are capable of homing to the fetal liver and BM. However, they disappear from the BM between 2 and 8 weeks of age, coinciding with a previously described “switch” in hematopoiesis that occurs after 3 weeks of age in mice (Benz et al., 2012; Bowie et al., 2007).

Despite their inability to persist into adulthood in situ, GFP+ FL HSCs were able to persist when transplanted into an adult environment (Figure 2). We examined the impact of the microenvironment on the persistence of GFP+ FL HSC by determining the ability of the GFP+ FL HSCs to engraft when transplanted into a fetal environment. We performed in utero transplantation (IUT) of Tom+ or GFP+ FL or adult KLS cells into the fetal liver of unconditioned WT E14.5 embryos (Figure 4D-F). Adult BM cells performed poorly in IUT assays as compared to FL cells (Figure 4D-F), as previously described (Hayashi et al., 2003). Similarly, LTMR was only observed in a small fraction (<10%) of fetal recipients of GFP+ FL KLS cells, with the majority of fetal recipients exhibiting no or short-term reconstitution. In contrast, >40% of fetal recipients of Tom+ FL KLS cells demonstrated LTMR (Figure 4D; Table S4). The inefficiency of GFP+ FL HSCs to sustain LTMR when transplanted into an unirradiated, fetal environment is reflective of their inability to persist in this same setting in situ. These data suggest that exposure to cell-extrinsic factors within the conditioned adult environment confer long-term persistence of the GFP+ FL HSC.

To determine whether differences in quiescence may underlie the inability of the GFP+ HSC to persist in situ, we compared cell cycle parameters between Tom+ and GFP+ FL HSCs and adult HSCs. Propidium iodide staining revealed a greater proportion of GFP+ HSCs in S/G2/M phase as compared to Tom+ HSCs (Figure 4G,H). Additionally, costaining for Hoechst and EdU incorporation revealed a significantly decreased proportion of GFP+ HSCs in G0/G1 as compared to both fetal Tom+FL and adult HSCs (Figure 4I,J). Our data support previously established differences in the proliferation state between fetal HSCs and adult HSCs (Figure 4J) (Bowie et al., 2007) and further show that within the fetal HSC fraction, GFP+ HSCs are more proliferative than Tom+ HSCs. Consistent with increased proliferation, GFP+ FL HSCs also expressed higher levels of the cell cycle regulator cyclin D1 as compared to Tom+ FL HSCs (Figure 4K). The less quiescent state of GFP+ HSCs within the fetal liver suggests that they may exhaust more rapidly during development as compared to Tom+ FL HSCs, as recently suggested by studies tracing unperturbed fetal HSCs in vivo (Busch et al., 2015).

GFP+ FL HSCs are functionally distinct from Tom+ FL HSCs

We noted that despite robust and persistent generation of myeloid cells, the percent contribution to the myeloid, but not lymphoid, lineages was significantly lower in primary recipients of GFP+ FL KLS cells as compared to recipients of Tom+ FL KLS cells (Figure 2D). These differences were also reflected in significantly different distribution amongst nucleated blood cells (Figure 5A) and lower ratio of donor-derived myeloid to lymphoid cells in individual recipients of GFP+ KLS cells as compared to recipients of Tom+ KLS cells 16 weeks after transplantation (Figure 5B). Differences in lineage output could not be attributed simply to persistence of long-lasting lymphoid cells from GFP+ HSCs, as we also observed a lower ratio of myeloid to lymphoid progenitors in the BM of recipients of GFP+ HSCs as compared to recipients of Tom+ HSCs (Figure 5C). Comparison of the levels of phenotypic BM HSC chimerism in individual recipients further suggested that differences in lineage output were due to higher output of lymphoid cells, as opposed to lower myeloid output, from GFP+ FL HSCs (Figure 5D). The same conclusion was reached by using the limiting dilution data (Figure 2E) to calculate the numbers of functional Tom+ versus GFP+ HSCs delivered to each transplanted recipient (see Experimental methods). Differences in lymphoid readout between Tom+ and GFP+ cells were further exaggerated when Flk2 was included as an additional marker to account for inefficient Cre-mediated floxing (Figure S3B-D). Evidence for a lymphoid bias of GFP+ HSCs was also found when they were isolated from the P14 neonate BM (Figure 5E,F), when they were transplanted in utero (Figure 5G,H), when CD150+ GFP+ HSCs were transplanted (Figure S2F,G), and when Flk2 was used as a surface marker alone (Figure S3G,H). Remarkably, the enhanced lymphoid cell production from GFP+ FL HSCs was maintained upon serial transplantation, even when secondary transplantations were performed from primary donors with similar levels of GM chimerism (Figure 5I,J), and was also observed when transplanted in competition with adult BM cells in a lethal irradiation setting (Figures S4C-D) and in secondary transplants beyond 40 weeks post-transplantation (Figure S4G-H). Importantly, GFP+ FL HSCs exhibited transcriptional lymphoid priming, as evidenced by upregulated expression of Rag1, Rag2, IL7rα, TCR-β, and CCR9 as compared to CD150+ Tom+ FL HSCs (Figure 5K and Table S3). Increased mRNA levels of lymphoid-associated genes suggest that the lymphoid bias was intrinsic to the GFP+ HSCs themselves, and not mediated by co-transplanted progenitor cells. Thus, transient GFP+ FL HSCs are functionally distinguished from their Tom+ counterparts as they displayed transcriptional lymphoid priming and consistently exhibited greater lymphoid cell production under a variety of experimental conditions.

GFP+ FL HSCs have enhanced capacity to generate innate-like T-cells within a fetal thymic microenvironment

As the existence of GFP+ FL HSCs during a limited developmental window coincides with production of specialized hematopoietic cell types, we examined their capacity to generate developmentally regulated mature cell subsets. Expression of distinct globin genes distinguishes primitive from definitive erythropoiesis. However, we detected only definitive, adult-type globin gene expression in erythroid progenitors derived from transplanted GFP+ or Tom+ FL HSCs or adult HSCs (Figure S6), suggesting that GFP+ FL HSCs do not contribute to primitive erythropoiesis. These data are consistent with the switch from primitive to definitive erythropoiesis occurring prior to E14.5 (Sankaran et al., 2010). Based on the strong lymphoid bias of the GFP+ FL HSCs, we instead compared the ability of GFP+ and Tom+ FL HSCs to generate fetal-specified lymphoid cell subsets. Specialized B1-type B-cells, and certain γδ T-cells are efficiently generated from fetal progenitor cells, whereas adult BM cells have mostly lost this capacity (Ghosn et al., 2012; Ikuta et al., 1990). These “innate-like” lymphocyte subsets differ from conventional B- and T-cells in that they are tissue-resident and express limited immunoglobin receptors that respond primarily to conserved self-antigens.

To ensure that donor-derived cells originated from HSCs and not from co-transplanted progenitor cells, we examined the generation of innate-like lymphocytes in secondary, as opposed to primary, recipients of Tom+ or GFP+ FL KLS cells (Figure 2A) and in comparable transplants of adult HSCs. These three recipient cohorts displayed no differences in the frequency of donor-derived thymic T-cell subsets (Figures 6A-C). The vast majority of donor-derived CD3+ T-cells in the thymus expressed the adult-type T-cell receptor (TCR)-β, whereas <1% of donor-derived thymic CD3+ T-cells expressed TCR-γδ (Figures 6B and 6C), regardless of whether the transplanted HSCs were Tom+ or GFP+ FL cells or from adult donors.

In mice, T-cells expressing the developmentally limited TCR-Vγ3 are generated only from FL, but not adult, HSCs and contribute to innate-like T-cells within the skin (Havran and Allison, 1988). Consistent with a previous report that the adult thymus does not support the generation of developmentally limited TCR-Vγ3+ cells (Ikuta et al., 1990), we observed negligible (< 0.1%) generation of TCR-Vγ3+ cells in the thymi of transplanted adult recipients (Figure 6C). Examination of donor-derived TCR-Vγ3+ cells within the epidermis in secondary recipients revealed minimal levels of chimerism (Table S5). While these results show that FL HSCs do retain the capability to generate TCR-Vγ3+ cells in the skin of adult recipients, the requirement for very high levels of overall T-cell chimerism and the absence of consistent chimerism among transplanted individuals prevented conclusive quantitative comparisons of the TCR-Vγ3+ capability of the different HSC populations within the skin of secondary recipients.

We next examined the generation of another innate-like T-cell, invariant natural killer T cells (iNKTs), which are thought to be developmentally regulated (Benlagha et al., 2002; Gapin et al., 2001). While we observed some contribution of both Tom+ and GFP+ FL KLS cells to splenic iNKT cells in secondary recipients, we did not observe significant differences in chimerism between the two donor cell types (Figure 6D-G). As the iNKT compartment did not fully recover after irradiation despite transplantation of BM cells (compare iNKT frequencies in Figures 6E and 6F), it is possible that potential differences in iNKT potential between the two HSC populations are under-appreciated in a transplantation setting.

As a previous study reported that TCR-Vγ3+ cells could be generated within a fetal thymic microenvironment (Ikuta et al., 1990), we examined differences in developmental T-cell potential using fetal thymic organ cultures (FTOCs). Adult HSCs consistently failed to contribute robustly to CD3+ T-cells in the FTOC assay, possibly because they competed poorly with resident fetal progenitors that survived the pre-conditioning by irradiation (Figure 6H). In contrast, fetal CD150+ KLS cells efficiently differentiated into T cells (Figure 6H). Tom+ or GFP+ HSCs cells yielded similar profiles of T-cell differentiation (Figures 6I-K), with the notable exception of more efficient generation of TCR-Vγ3+ T-cells by GFP+ HSCs as compared to their Tom+ counterparts (Figure 6L-N). Whereas more than 1/3 of Tom+ cultures failed to yield any TCR-Vγ3+ T-cells, all GFP+ cultures contained TCR-Vγ3+ T-cells (Figure 6L), and each GFP+ Vγ3-containing culture had significantly higher proportion of Vγ3+ T-cells compared to Tom+ FTOCs (Figures 6M and 6N). These findings show that both the progenitor cell and thymic environment must be of fetal origin to allow generation of TCR-Vγ3+ T-cells. In addition, these data suggest that GFP+ FL HSCs are an important source of TCR-Vγ3+ T-cells during development.

GFP+ FL HSCs efficiently generate innate-like B-cells in vivo

We next investigated the ability of Tom+ and GFP+ FL HSCs to give rise to innate-like B-cell subsets. As with in vivo T cell experiments, all analyses were conducted in secondary recipients. We confirmed that there were virtually no B1 progenitors (B1P) within the originally transplanted KLS fractions (Figure S7A,B), excluding these cells as a differential source of B1 cells in the primary transplant.

Supporting the reported fetal origin of peritoneal cavity (PerC) B-cells (Barber et al., 2011; Hardy and Hayakawa, 1991), we observed strikingly higher donor-derived contribution to peritoneal IgM+ B-cells in secondary recipients of fetal compared to adult HSCs (Figures 7A, 7B, and S7G). GFP+ FL HSCs gave rise to significantly higher peritoneal IgM+ B-cell chimerism than Tom+ FL HSCs, even in cohorts of mice with similar B-cell contribution in the PB (Figure 7B). To determine whether the low contribution to PerC B-cells by adult HSCs could be attributed to their overall lower lymphoid cell generation, we quantified peritoneal B cell output in recipient mice of adult HSCs or FL HSCs that displayed comparable B-cell chimerism in the PB (Figure S7I). This analysis reinforced the relative myeloid bias of adult versus fetal HSCs, and attenuated the differences in peritoneal B cell contribution between adult HSCs and Tom+ FL HSCs. Strikingly, the superior ability of GFP+ FL HSCs to generate peritoneal B-cells was still apparent (Figure S7I). Furthermore, not only was the contribution of GFP+ FL HSCs higher to all peritoneal B-cell subtypes (Figures 7B, S7G, and S7I), they also gave rise to a significantly higher proportion of B1a and B1b type B-cells compared to both adult HSCs and Tom+ FL HSCs (Figures 7C and 7D). The few donor-derived peritoneal B-cells found in recipients of adult HSCs were predominantly B2 cells (IgM+ CD5-CD11b), with very few B1a (IgM+CD5+CD11b+) or B1b (IgM+CD5-CD11b+) cells (Figures 7C and 7D). While Tom+ FL HSCs gave rise to a significantly higher proportion of B1a and B1b cells compared to adult HSCs, GFP+ FL HSCs were even more efficient in B1 cell generation (Figures 7C and 7D). Furthermore, despite lower overall reconstitution levels and limited long-term reconstitution upon in utero transplantation (IUT) (Figure 7H and Table S4), GFP+ KLS cells still retained a greater capability to seed the peritoneal IgM+ B-cell compartment and particularly B1a cells as compared to Tom+ KLS cells upon IUT (Figure 7I). Although a recent report suggested that fetal HSCs are not a major source of B1a cells (Ghosn et al., 2016), we did not observe appreciable regeneration of innate-like B-cells in the absence of LTMR and BM reconstitution (Figure S5). Secondary recipients that did not demonstrate LTMR showed limited ability to reconstitute innate-like lymphocytes in vivo (Figure S5D). Similarly, secondary transplantation of BM B-cells could not sustain PB lymphopoiesis (Figure S7C,D), nor yield appreciable innate-like B-cell reconstitution of the peritoneal cavity (Figure S7E) or spleen (Figure S7F). Collectively, our data reveal the unique capacity of the transient GFP+ HSCs to generate innate-like B cells, and in particular fetal-restricted B1 B-cells, in the peritoneum.

Like peritoneal B1 B-cells, marginal zone (MZ) B-cells within the spleen exhibit innate-like immune responses and have a suggested fetal origin (Carey et al., 2008). We therefore examined the contribution of Tom+ and GFP+ FL HSCs and adult HSCs to MZ B-cells in the spleen by strategies described above. In mice with comparable PB B-cell chimerism, we observed a significantly lower contribution by adult HSCs to MZ B-cells (Figures 7E-G, and S7H),supporting their proposed fetal origin. GFP+ FL HSCs exhibited higher chimerism within MZ (CD21^hiCD23^lo), but not follicular (FO, CD21^-CD23^hi), splenic B-cells subsets B-cells as compared to Tom+ FL HSCs (Figure 7E and S7H). Furthermore, within the donor-derived fraction of splenic B-cells, GFP+ FL HSCs gave rise to a significantly higher ratio of MZ B-cells as compared to both Tom+ FL HSCs and adult HSCs (Figures 7F and 7G). Similarly, GFP+ KLS cells to preferentially seed the MZ of the spleen when transplanted in utero, despite yielding significantly lower overall chimerism within the spleen as compared to Tom+ KLS cells (Figures 7J and 7K). Together, these data reveal that the transient GFP+ HSCs have a unique capability to generate fetal-restricted innate-like B-cells. Retention of this ability upon transplantation into both fetal and adult recipients demonstrates that this capability is mainly regulated cell-intrinsically and does not require a fetal environment.

Mice, cell isolation, and analysis

All mice were maintained in the UCSC vivarium according to IACUC-approved protocols. FlkSwitch mice (Boyer et al., 2012; Boyer et al., 2011; Epelman et al., 2014; Hashimoto et al., 2013) were used as donors for cell isolation and 8-12 week old WT C57Bl/6 mice were used as recipients. Adult and fetal cells were isolated and processed as previously described (Perez-Cunningham et al., 2016; Smith-Berdan et al., 2015) and analyzed using a four-laser FACSAria or LSRII flow cytometer (BD Biosciences, San Jose, CA).

Transplantation assays

Transplantations were performed as previously described (Forsberg et al., 2006; Smith-Berdan et al., 2011). Recipient mice were either sublethally irradiated (750 rad, single dose) or lethally irradiated (1000 rad, split dose). Fetal intrahepatic injections were performed as previously described (Nijagal et al., 2011). Briefly, under isofluorane-induced general anesthesia, a midline laparotomy was performed on pregnant dams, the uterus was exposed, and 5 μL of cell suspension was injected into the fetal liver of each pup using pulled glass micropipettes. The abdominal incision was closed in multiple layers. PB chimerism was analyzed by flow cytometry.

In vitro analyses

RNA-sequencing, qRT-PCR, cell cycle analysis, EdU incorporation, and fetal thymic organ culture assays were all performed as described previously (Beaudin et al., 2014); (Ugarte et al., 2015) and in the Supplemental Experimental Procedures.

Statistical analysis

Transplantation assays were randomized across age, sex, and cage for any given experiment. Investigator was blinded to treatment during preparation and analysis of samples. Unless otherwise noted, all experiments were performed in at least three independent replicates. Statistically significant differences between groups with similar variance as determined by the standard error of the mean were assessed by unpaired, two-tailed Student’s t-test, unless otherwise noted.