Human B-1 and B-2 B Cells Develop from Lin⁻CD34⁺CD38^lo Stem Cells

Acting as an important bridge between the innate and adaptive immune responses, B-1 cells and the Abs they produce provide a rapid response against pathogens during the lag period required for adaptive Ab production by B-2 cells, and prevent autoimmunity through their capacity to rapidly clear noxious molecules and cellular debris (1–4). Murine B-1 and B-2 cells are derived from two distinct lineages (5, 6). However, in humans, in part because it has been difficult to discriminate between them phenotypically (7–10), efforts to pinpoint the developmental origins of human B-1 and B-2 cells have lagged.

The Lin⁻CD34⁺ phenotype is a hallmark marker of hematopoietic stem cells (HSCs) in humans (11, 12), and CD34⁺ enriched cell populations are widely used in human HSC transplantation (HSCT) (13, 14). Changes in the CD38 expression level indicate a reduction in the multilineage potential of an HSC population (15). In the early 1990s, Terstappen et al. (16) used three-channel flow cytometric analysis and in vitro blast colony formation culture systems to show that Lin⁻CD34⁺ HSCs lost pluripotency as they acquired CD38 expression, suggesting that the increase in CD38 expression reflects differentiation of CD34⁺ HSCs into a more lineage-committed status. In xenogeneic transplant studies, Bhatia et al. (17) and Ishikawa et al. (18) independently showed that only Lin⁻CD34⁺CD38^lo/− cells gave rise to multilineage blood cells, including B cells, whereas Lin⁻CD34⁺CD38⁺ cells were unable to generate any blood cells after being transplanted into NOD/SCID and NOD/SCID/β₂-microglobulin–null (NOD/SCID/BMG^null) mice. These data indicate that the Lin⁻CD34⁺CD38^lo/− population includes B cell progenitors. It is not known whether this population contains a single progenitor for all B cell subsets or contains distinct progenitors for each.

Much progress has been made using different immune-deficient mouse models to study human hematopoiesis. NOD/SCID and NOD/SCID/BMG^null mice are the most widely used; however, these immune-deficient models have limitations. The NOD/SCID mouse environment favors human B cell, but not T cell, engraftment (19). In this respect, NOD/SCID/BMG^null mice, which support the development of a greater variety of blood cells, including T cells and B cells, have an advantage over the NOD/SCID model (20). NOD/SCID and NOD/SCID/BMG^null mice exhibit a shortened lifespan (6−8.5 mo) due to thymic lymphomagenesis (20–22). Limited lifespan is not an issue with NOD.Cg-Prkdc^scid Il2rg^tm1wjl/SxJ (NSG) mice, which have a disease-free lifespan > 16 mo (23). NSG mice were shown to be excellent recipients for engrafting human HSCs. They support the reconstitution of greater numbers of cells and a wider variety of blood cell lineages (24) than do the other models (25, 26).

Despite controversy (27–35), human B-1 cells were recently defined as CD20⁺CD27⁺CD43⁺CD38^lo/int cells with clinically relevant potential (36, 37). This population exhibits repertoire skewing toward expression of the Ig VH4-34 gene (37), which encodes autoreactive Ab (38, 39), and produces natural Abs (36), characteristics of mouse B-1 cells. In this article, we report that human Lin⁻CD34⁺CD38^lo cells from cord blood (CB) and bone marrow (BM) give rise to B-1 and B-2 cells, whereas Lin⁻CD34⁺CD38^hi cells do not give rise to B cells. In patients with hematologic malignancies who underwent autologous and allogeneic transplantation of mobilized HSCs (CD34⁺ enriched mononuclear cells [MCs]), B-1 and B-2 cells were reconstituted. Thus, our data demonstrate that, in humans, B-1 and B-2 B cell populations can be generated from Lin⁻CD34⁺CD38^lo stem cells derived from CB or BM.

Umbilical CB samples (n = 44) were obtained from healthy neonate cords immediately following uncomplicated delivery. BM tissues (n = 12) were obtained from otherwise healthy adults undergoing hip surgery, and peripheral blood samples were obtained from patients undergoing HSCT for treatment of hematologic malignancies. All human materials were obtained in accordance with protocols approved by the Northwell Health Institutional Review Board.

NSG mice were obtained from the Jackson Laboratory and were bred and maintained in ventilated cages with irradiated chow and sterile acid water (pH 3.2). All mice were cared for and handled in accordance with Institutional Animal Care and Use Committee guidelines at the Feinstein Institute for Medical Research.

MCs were obtained from CB and BM by density gradient separation using lymphocyte separation medium (Cellgro). MCs were washed (2 mM EDTA in PBS), resuspended in cell isolation/sort buffer (0.5% BSA in PBS), and subjected to lineage cell depletion using a Lineage Cell Depletion Kit (Miltenyi Biotec). Lin⁻ cells were stored short term at −80°C or long term in liquid nitrogen in freezing medium (10% DMSO in FBS) until use.

BM, spleen, peritoneal cells, and serum were collected from xenotransplanted NSG mice 10−17 wk posttransplantation. Isolated cells were stained with fluorophore-conjugated Abs, as described below. Stained cells were sort purified and/or analyzed using a BD Influx cell sorter or a Beckman Coulter Gallios flow cytometer.

Before transplantation, thawed Lin⁻ cells were treated with normal mouse serum and stained with a lineage-specific Ab mixture (containing Abs distinct from those used in the lineage depletion kit), HSC markers (CD34 and CD38), and Aqua LIVE/DEAD dye in cell sorting buffer. After washing, cells were resuspended in cell sorting buffer with 10 U/ml DNase. Stained cells were subjected to sort purification using a BD Influx cell sorter.

For analysis, single-cell suspensions (1−2 × 10⁶ cells/sample) of spleen cells from xenotransplanted NSG mice and PBMCs from patients undergoing HSCT were stained with predetermined optimal concentrations of fluorophore-conjugated mAbs. The following mouse anti-human Abs were used: anti-CD19–Alexa Fluor 700 or PE-Cy7, anti-CD20–Pacific Blue or allophycocyanin-Cy7, anti-CD3–ECD, anti-CD4–ECD, anti-CD7–ECD, anti-CD27–allophycocyanin, anti-CD38–PerCP–Cy5.5 or PE-Cy7, anti-CD43–APC–Alexa Fluor 750 or FITC, CD22–allophycocyanin–Alexa Fluor 700, CD5–PE–Cy7, CD24–allophycocyanin–Alexa Fluor 750, Aqua LIVE/DEAD dye, and MitoTracker Green (MTG) dye (Life Technologies). Events (3 × 10⁵–1.5 × 10⁶) were collected on a Gallios (Beckman Coulter) flow cytometer and/or a BD Influx cell sorter. Data were analyzed using FlowJo software v. 9.7.6 (TreeStar, San Carlos, CA). Abs were purchased from Beckman Coulter, BD Biosciences, or BioLegend. Cells were stained and run in FACS buffer (2.5% FBS, 1 mM EDTA, 0.02% NaN₃). B cell subsets were gated as we described earlier (37), unless otherwise noted.

The transplantation protocol was adapted from Vuyyuru et al. (40). Sort-purified Lin⁻CD34⁺CD38^lo or Lin⁻CD34⁺CD38^hi stem cells were transplanted into NSG neonates (24−48 h after birth) by intrahepatic injection of 0.7−1 × 10⁵ cells in 25 μl PBS using a 30-gauge needle; 57 neonates received Lin⁻CD34⁺CD38^lo CB cells, 9 received Lin⁻CD34⁺CD38^hi CB cells, 20 received Lin⁻CD34⁺CD38^lo BM cells, and 5 received Lin⁻CD34⁺CD38^hi BM cells. Mice were weaned at 3 wk of age and randomly distributed among different experimental groups. Hereafter, NSG mice transplanted with human stem cells are designated human immune system (HIS) mice, and cells obtained from HIS mice are so designated (e.g., HIS B-1 cells).

Single-cell sorting, cDNA synthesis, PCR, and VH usage analysis were performed as described (37). Briefly, following initial sorting, single cells of each subset were resorted at one cell per well into 96-well PCR plates containing 20 μl of lysis buffer, immediately frozen at −20°C, and stored at −80°C. cDNA was synthesized, and cDNA products were subjected to two rounds of PCR to amplify Ig H chains. All PCR products were checked for concentration and efficiency using the QIAxcel Advanced system (QIAGEN), sequenced (GENEWIZ), quality checked (4Peaks software), and analyzed (IMGT database). Identical sequences from different wells were counted as a single clone.

All statistical analyses were done using GraphPad Prism and R (http://www.r-project.org/)

Engraftment of xenotransplanted human Lin⁻CD34⁺ cells into immunodeficient mouse systems was shown to generate a functional HIS (23, 25). In vitro and in vivo studies that fractionated Lin⁻CD34⁺ cells into Lin⁻CD34⁺CD38^−/low and Lin⁻CD34⁺CD38^+/high populations (hereafter called Lin⁻CD34⁺CD38^lo and Lin⁻CD34⁺CD38^hi cells, respectively) showed that the Lin⁻CD34⁺CD38^lo population includes HSCs with multilineage potential (18, 25, 41–44). In contrast, although in vivo xenotransplant study models showed that Lin⁻CD34⁺CD38^hi cells do not generate any blood cells in vivo (45), a study by Galy et al. (46) showed that Lin⁻CD34⁺CD38^hi cells are able to generate B cells during in vitro culture in the presence of cytokines (IL-3, IL-6, and LIF). To determine whether NSG mice support hematopoiesis by Lin⁻CD34⁺CD38^lo and Lin⁻CD34⁺CD38^hi cells, Lin⁻CD34⁺ HSCs were sort purified and intrahepatically injected into 1–2-d-old NSG neonates (Fig. 1A, 1B). We found that transplanted Lin⁻CD34⁺CD38^lo cells (52 of 57 transplanted mice, 91%) led to human B cell (CD19⁺, mean ± SD, 14.5 ± 13.6% of total live spleen cells) development, whereas Lin⁻CD34⁺CD38^hi cells did not lead to CD19⁺ B cell development (0 of 9 transplanted mice) (Fig. 1C). Similar to previous reports (16, 18, 47), these data suggest that Lin⁻CD34⁺CD38^hi cells are lineage-committed progenitor cells and have limited engraftment capability in our xenotransplantation model.

Of interest, CD43 is expressed in mice at early lineage commitment stages, and the expression is lost as pre-B cells develop (48). In humans, it was shown that CD43 is expressed on some transitional B cells (9) and is highly expressed on terminally differentiated plasmablast/plasma cells. Recently, we showed that CD43 is a hallmark marker for human B-1 cells, in addition to CD27, CD20, and CD38 (37). Interestingly, although CD43 is expressed on Lin⁻CD34⁺CD38^lo and Lin⁻CD34⁺CD38^hi cells, it was shown that only Lin⁻CD34⁺CD38^loCD43⁺ cells have multilineage potential and self-renewal capacity (43). To investigate the expression of CD43 in the context of the B cell lineage, we examined the expression of CD19 and CD43 on CD34⁺ cells and found that the frequency of cells coexpressing CD19 and CD43 increased as CD34⁺ cells lost CD34 and gained CD38 expression in CB but not in BM. More importantly, CD43 expression was reduced as these cells matured into CD34⁻CD38^loCD19⁺ B cells in CB and BM (Supplemental Fig. 1). These data suggest that CD43 expression is characteristic of human B cell development, although our data do not speak to the functional significance of CD43 expression in the development of human B-1 and B-2 cells.

We evaluated B cell subsets within the engrafted CD19⁺ B cell compartment in mice that were transplanted with human CB Lin⁻CD34⁺CD38^lo cells (hereafter, data presented are from transplanted CB Lin⁻CD34⁺CD38^lo cells unless otherwise indicated). Splenocytes from 12–17-wk-old HIS mice were examined for surface marker expression. We found that, similar to the B cell compartment in human peripheral blood samples (37), the engrafted CD19⁺ B cell compartment consisted of B-1 (CD20⁺CD27⁺CD43⁺CD38^lo/int) and B-2 cells, including memory (CD20⁺CD27⁺CD43⁻) and preplasmablast (CD20⁺CD38^hi) cells (Fig. 2A).

To determine the maturation status of B-1 and B-2 cell subsets, human B cell developmental markers (49), including CD24 and CD38, were evaluated. We found that engrafted B-1 and memory B cells exhibited mature phenotypes, as evidenced by low-to-intermediate expression levels of CD24 and CD38, similar to CD43⁻ mature cells, whereas CD43⁺ mature cells exhibited a partial transitional phenotype, as reflected by high expression levels of CD24 and CD38 (Fig. 2B). Preplasmablasts exhibited a differentiated phenotype, having lost CD24 expression and expressing very high levels of CD38 (Fig. 2B). Supplemental Fig. 2A shows representative CD24 and CD38 expression levels for examined B cell subsets from CB and adult peripheral blood. To further explore the similarities and differences between human B-1 cells developing in murine systems and the human system, we investigated CD5 expression, which is a hallmark marker of murine B cells yet is expressed on transitional and activated human B cells. We found that >50% of B cells from HIS mice express CD5, similar to that of B cells in CB but very different from B cells in peripheral blood (Supplemental Fig. 2B−D). The high frequency of CD5 expression in B cells from HIS mice and native human CB might indicate that 50% of B cells from these two sources are in an activated or a less mature state than those from adult peripheral blood, and it raises the question of whether using CD5 alone is sufficient to measure the maturity/activation status of human B cells. Additionally, another unexplained phenomenon was observed. In the human system, CD22, an inhibitory coreceptor of BCR (50), is expressed on most B cells (Supplemental Fig. 2F), with the exception of plasmablasts and preplasmablasts (37); however, <50% of cells from each B cell subset in HIS mice express CD22 (Supplemental Fig. 2E). The data suggest that other extrinsic developmental factor(s) in the local environment could be important in regulating the surface expression of CD22 on human B cells.

In mice, B-1 cells of fetal origin reside in the spleen, peritoneal cavity, and BM (6, 51). We determined the distribution of CB-derived engrafted B cells in spleen, BM, and peritoneal cavity of HIS mice. Like murine B-1 cells, engrafted human B-1 cells were present in the spleen, BM, and peritoneal cavity (Fig. 2C−E), and they were present at greater frequencies in spleen and peritoneal cavity than are found in CB (Fig. 2F). Altogether, our data show that engrafted human B cells from Lin⁻CD34⁺CD38^lo cells found in different HIS mouse tissues can be characterized phenotypically as B-1 or B-2 cells; although the majority of CD43⁺ B-2 cells displayed transitional phenotypes, most engrafted B-1 and other B-2 cells displayed a more mature B cell phenotype based on their low levels of expression of CD38 and CD24.

To examine the ability of HIS B-1 cells to spontaneously secrete Ab, as seen with B-1 cells from adult peripheral blood (37), we sort purified B-1, memory, and CD20⁺CD38^hi preplasmablast cells from HIS mice and assessed IgM Ab secretion using ELISPOT. Similar to previously published data (37), on average, 0.3% of HIS B-1 cells secreted Ab compared with 0% of HIS memory cells and 7.9% of CD20⁺CD38^hi preplasmablast cells (Supplemental Fig. 3). These data suggest that HIS B-1 and CD20⁺CD38^hi phenotype preplasmablast cells behave like adult peripheral blood B-1 and CD20⁺CD38^hi preplasmablast cells.

High-frequency VH4-34 usage is a hallmark feature that distinguishes human B-1 cells from memory B cells and preplasmablasts (37). To determine whether HIS B-1 cells display VH usage similar to that of CB B-1 cells, engrafted B-1 cells from HIS mouse spleens and B-1 cells directly isolated from CB were single-cell sort purified, and their IgM H chains were amplified and sequenced (Fig. 3). We found that HIS B-1 cells and CB B-1 cells exhibited a largely similar distribution of Ig VH, DH, and JH family use (Fig. 3A). Detailed analysis of VH gene usage indicated a similar diversity between HIS B-1 cells and CB B-1 cells (Fig. 3B, 3C). The analysis also demonstrated that HIS B-1 cells and CB B-1 cells exhibited a comparably high frequency of VH4-34 use (Fig. 3B, 3C). On average, 8% of analyzed sequences from HIS B-1 cells used VH4-34 similar to CB (Fig. 3D) and adult peripheral blood B-1 cells (37) and quite different from HIS preplasmablasts (Fig. 3D) and adult peripheral blood preplasmablasts, memory B cells, and plasmablasts (37). In keeping with the checkpoint paradigm (52), HIS transitional (Trans, CD19⁺CD20⁺CD27⁻MTG⁺) B cells express relatively frequent use of VH4-34 compared with naive (CD19⁺CD20⁺CD27⁻CD43⁻MTG⁻) B cells (37, 53, 54), consistent with the autoreactive nature of VH4-34. Detailed VH subfamily usage of HIS transitional, naive, and preplasmablast B cells is shown in Supplemental Fig. 4.

N-region addition is an important mechanism for generating Ig diversity; together with combinatorial variation and somatic mutation, it results in numerous Ig specificities (55–57). In mice, B-1, in particular B-1a, cell-derived IgM Abs are characterized by low numbers of N-additions (58), which are added by the enzyme TdT (59). TdT expression is restricted to adult mice (60), after the development of fetal-derived B-1a cells (58, 61). Therefore, in mice, fetal-derived B-1a cells lack N-additions for the most part (61), whereas adult BM-derived B-1a cells exhibit high numbers of N-additions (62–65). In contrast, TdT is expressed during fetal and adult life in humans (66); as a result, fetal- and adult-derived B cells, including B-1 cells, express Ig with numerous N-additions (37, 67). It was shown that the repertoires of human and mouse fetal Igs exhibit similarities and differences (68). For example, it was demonstrated in humans that B cells from preterm and term infants exhibit fewer N-nucleotide additions and shorter CDR-H3 length compared with adults (69). To determine whether N-additions at VH-DH and DH-JH junctions in human B-1 cells are affected by development in the murine environment, we compared N-additions at VH-DH and DH-JH junctions of HIS B-1 cells and CB B-1 cells. We found that, similar to CB B-1 cell sequences, HIS B-1 cell Ig sequences contained N-additions at both junctions (Fig. 3E). We found N-additions at one or more junctions in 93.0% of HIS B-1 cells and in 97.3% of CB B-1 cells. Overall, these data show comparable VH-DH-JH composition in HIS and CB B-1 cells.

After birth, BM is the major habitat for HSCs, Lin⁻CD34⁺CD38^lo cells, and Lin⁻CD34⁺CD38^hi cells. We evaluated whether transplantation of BM Lin⁻CD34⁺CD38^lo or Lin⁻CD34⁺CD38^hi cells into NSG neonates resulted in human B cell engraftment and, more specifically, B-1 cell engraftment. Adult BM Lin⁻CD34⁺CD38^lo and Lin⁻CD34⁺CD38^hi cells were sort purified and intrahepatically transplanted into 1−2-d-old NSG neonates (Fig. 4A, 4B). We found that transplantation of Lin⁻CD34⁺CD38^lo cells led to human B cell (CD19⁺) engraftment (7 of 24 injected mice), whereas transplantation of Lin⁻CD34⁺CD38^hi cells did not (0 of 5 injected mice) (Fig. 4C). Similar to the results with transplanted CB Lin⁻CD34⁺CD38^lo cells, the BM-derived B cell compartment contained cells of B-1 and B-2 phenotypes. B-1 cells were found in BM and spleen of HIS mice (Fig. 4D, 4E) and were present at a higher frequency (1.6 and 1.8% of total B cells, respectively) than in native adult BM (0.7%) (Fig. 4F). Together, these data demonstrate that Lin⁻CD34⁺CD38^lo stem cells from human adult BM are also capable of reconstituting B-1 and B-2 cell subsets.

Of interest, we also observed that the overall efficiency of B cell engraftment in HIS mice after transplantation with stem cells from BM (29%) was significantly lower than after transplantation with stem cells from CB (91%), and, of total live HIS splenocytes, the frequency of engrafted B cells was much lower in mice transplanted with BM versus CB (0.4 ± 0.3% and 14.5 ± 13.6%, respectively) (Figs. 1C, 4C). However, this result is consistent with a previous report that CB is a better stem cell source than BM for human CD45⁺ cell engraftment in a xenotransplantation model (NOD/SCID) (70).

Recently, it was shown that, following clinical transplantation of allogeneic CD34⁺-enriched stem cells, recipients reconstituted B cells, some of which may be B-1 cells (71). Although this report lacked functional analysis and did not fully discriminate between B-1 and B-2 cells (the reported CD19⁺CD27⁺CD43⁺ population includes B-1 cells, preplasmablasts, and plasmablasts), the data suggested that reconstitution of human B-1 and B-2 cells had occurred in HSC recipients. We collected peripheral blood samples from HSC recipients before and after transplantation (see 2Materials and Methods) and found that, irrespective of the type of transplantation, as early as 8 wk posttransplantation B-1-phenotype cells (CD20⁺CD27⁺CD43⁺CD38^lo/int) developed in the recipients as soon as, or shortly after, the appearance of CD19⁺ B cells (∼1% of total CD19⁺ B cells were B-1 cells, Fig. 5). Our data suggest that, similar to xenogeneic transplantation, mobilized HSCs used to transplant patients with hematologic malignancies also reconstitute B-1 and B-2 cell populations. The significance, if any, of the observed changes in the frequencies of B-1 cells during follow-up remains to be determined.

Our results demonstrate that CB and adult BM HSCs with a Lin⁻CD34⁺CD38^lo phenotype have the ability to generate B-1 and B-2 cells. The experiments were conducted using a xenogeneic transplantation system in which human HSC populations were transplanted into neonates of immune-deficient NSG mice. Engrafted B-1 cells expressed rearranged Ig genes with features recapitulating those of B-1 cells from human CB. Our results demonstrate that a population of human Lin⁻CD34⁺CD38^lo stem cells reconstituted B-1 and B-2 cells, suggesting that human B-1 and B-2 cells derive from the same progenitor cell population; however, it remains possible that adoptively transferred Lin⁻CD34⁺CD38^lo stem cells consisted of a mixture of progenitors separately committed to B-1 and B-2 cell differentiation at this early stage. To rule out this latter possibility, it will be necessary to identify markers that subdivide this population or to assess the differentiation potential of individual cells.

In mice, there are two major subsets of B-1 cells that are distinguished on the basis of CD5 expression (B-1a cells express CD5, and B-1b cells lack CD5 expression) (72) and function (B-1a cells produce natural IgM Abs and provide innate-immune responses, and B-1b cells produce isotype-switched Abs and provide adaptive-immune responses) (3, 73). Insofar as they have been characterized, human B-1 cells display characteristics of murine B-1a and B-1b cells, including having natural and adaptive Ab features (30, 36, 37, 40); therefore, we anticipate that the development of human B-1 cells could have characteristics that mimic murine B-1a and B-1b development. In mice, it was shown in an adoptive-transfer experiment with single HSC that adult BM failed to give rise to B-1a cells, whereas HSC-deficient yolk sac gave rise to B-1a cells but not conventional B-2 cells (74–76). These data suggest that murine B-1a cells are HSC independent. However, other studies also showed that adult BM cells give rise to B-1a cells, B-1b cells, and B-2 cells (63–65, 77). Consistent with the latter, our study focused on the development of human B-1 cells from human CB and adult BM stem cells; however, it remains to be determined whether distinct progenitors of human B-1 and B-2 cells occur during further development, as found in mice (5).

In a study of Borrellia hermsii infection, Vuyyuru et al. (40) showed that transplantation of the HSC CD34⁺ (positive selection by magnetic beads) population into NSG mice led to engraftment of human B cell subsets, including CD20⁺CD27⁺CD43⁺ B cells, proposed B-1 cells of Griffin et al. (27). These HIS mice were able to clear B. hermsii infection, whereas the nontransplanted control NSG mice failed to do so. Because clearing borrelliosis infection is a T cell–independent process (78) and correlates with B-1b cell expansion (79), Vuyyuru et al. (40) suggested that CD20⁺CD27⁺CD43⁺ cells (27, 37) are the key source for clearing the infection. These data suggested that progenitors for B-1 cells are present in the human CD34⁺ stem cell compartment. Consistently, when we separate the HSC Lin⁻CD34⁺ cell compartment into two HSC populations based on CD38 expression, Lin⁻CD34⁺CD38^lo cells give rise to B cells but Lin⁻CD34⁺CD38^hi cells do not, suggesting that Lin⁻CD34⁺CD38^lo cells are the source of B cells. In particular, they are also the source of CD20⁺CD27⁺CD43⁺ B cells reported by Vuyyuru et al. previously (40). Of note, on average, the total engrafted B cells in the spleen, BM, and peritoneal cavity of our HIS mice transplanted with Lin⁻CD34⁺CD38^lo cells consisted of relatively low combined frequencies of B-1 cells and preplasmablast (CD20⁺CD38^hi) cells (Fig. 2C−E), whereas NSG mice transplanted with total CD34⁺ cells in our previous study (40) developed a higher frequency of CD20⁺CD27⁺CD43⁺ cells (∼35% of total CD19⁺ cells). We speculate that the discrepancy in the frequency of B-1 cells in human HSC–reconstituted NSG mice between this study and our earlier one (40) is due to irradiation of NSG recipients in the previous study (40), which might contribute to a host environment that is more favorable to CD20⁺CD27⁺CD43⁺ B-1 cell development or less favorable to B-2 cell development, or is due to the loss of support for the development of CD20⁺CD27⁺CD43⁺ B-1 cells provided by Lin⁻CD34⁺CD38^hi stem cells, which were included in adoptive transfers carried out in our previous study (40) but were treated separately in this study.

In summary, our analysis of the development potential of human Lin⁻CD34⁺CD38^lo and Lin⁻CD34⁺CD38^hi stem cell populations from CB and adult BM in an NSG xenotransplant model demonstrated that Lin⁻CD34⁺CD38^lo cells reconstituted B-1 and B-2 B cells. We also found that the Ig repertoire of B-1 cells reconstituted from Lin⁻CD34⁺CD38^lo stem cells was consistent with distinctive features of the B-1 cell Ig repertoire found in native human B-1 cells. Thus, it appears that NSG mice constitute a favorable host environment for xenogeneic development of human B-1 cells, as well as B-2 cells, and that early human HSCs give rise to B-1 and B-2 cells in this model, as well as during clinical transplantation. It remains to be determined whether B-1 cell–specific progenitors appear later in development or in other tissues.

We thank members of the Rothstein laboratory for helpful discussions and technical support; the R.-L.B. team for recruiting and collecting patient samples; and the Tissue Donation Program at the Feinstein Institute for Medical Research and its staff for helping with BM and CB collections. We give special thanks to our healthy donors and patient volunteers for donations of peripheral blood, CB, and BM.

The authors have no financial conflicts of interest.