We have identified a distinctive lymphoid-restricted progenitor population in adult mouse bone marrow based on a unique c-Kit−Sca-1highFlt3+ AA4+ surface phenotype. These cells are highly lymphoid biased and rapidly generate B and T cells after adoptive transfer. However, whereas previously described lymphoid progenitors such as common lymphoid progenitors express TdT and relatively high levels of RAG2, and are enriched for cells with an active V(D)J recombinase, Flt3+ AA4+ cells within the c-Kit−Sca-1high bone marrow fraction are TdT−, are RAG2low, and do not display evidence for ongoing or past recombinase activity. Furthermore, unlike common lymphoid progenitors that readily generate B cells upon stimulation with IL-7, c-Kit−Sca-1highFlt3+ precursors do not express abundant levels of the IL-7R, and require costimulation with Flt3 ligand and IL-7 to generate B cells in vitro. Moreover, these findings suggest that hematopoietic stem cells in adults generate an array of lymphoid-biased progenitor populations characterized by distinct gene expression and cytokine response profiles.
During the earliest phases of lymphopoiesis, lymphoid-biased multipotent progenitors (MPPs)3 are generated from hematopoietic stem cells (HSCs). Lymphoid progenitors are receptive to environmental cues such as stromal cell-derived cytokines, and signals derived by specific cytokine receptors are thought to promote survival as well as differentiation of lymphoid-biased precursors toward the B or T cell lineages. Despite intensive interest in the cellular basis for lymphoid lineage commitment, the precise relationships among lymphoid MPPs and their ultimate lineage-committed progeny remain unclear.
The identification of common lymphoid progenitors (CLPs) (1), defined by their capacity to yield predominantly B and T cells, lends support to the notion that all lineage-committed B and T lineage precursors derive from a single progenitor pool. However, numerous observations call this viewpoint into question. First, bipotent myeloid and B lineage progenitor populations lacking T cell potential have been described in the fetal liver and in adult bone marrow (BM) (2, 3), and more recently early T lineage precursors in the adult thymus were shown to generate T and myeloid progeny without substantial B cell potential (4, 5). Second, CLPs have recently been shown to exhibit myeloid potential under certain culture conditions (6, 7). Third, a subset of common myeloid progenitors (CMPs), defined in part by surface expression of Flt3/Flk2, the receptor for the cytokine Flt3 ligand, exhibits residual B lineage but not T lineage potential and express low levels of lymphoid-affiliated genes such as RAG2 and TdT (8, 9, 10). Finally, early T lineage precursors likely arise independently of CLPs (11, 12). These observations argue against the notion that all B and T cells derive from a single lymphoid-restricted precursor population.
Defining lineage relationships in early lymphocyte development is confounded by potential differences between fetal and adult hematopoiesis. Fetal lymphoid precursors are more likely to generate CD5+“B1” B cells and γδ T cells (13), and TdT expression occurs in adult but not fetal precursors (14, 15, 16, 17). Furthermore, whereas adult lymphopoiesis requires the cytokine IL-7, B cell development occurs normally in fetal and perinatal IL-7-deficient mice (18). Interestingly, although precursors dedicated to generating B1 B cells are highly enriched in fetal life, these cells persist in small numbers in adults (19). Therefore adult HSCs may generate an array of B lineage or lymphoid MPPs throughout life. Consequently, all B cells generated by adult HSCs may not derive from CLPs.
In this study, we describe a novel lymphoid progenitor population in adult mouse BM. These cells are characterized by low to intermediate levels of a RAG2 reporter transgene and the unexpected c-Kit−Sca-1highIL-7Rαlow cell surface phenotype. Despite low-level expression of c-Kit and IL-7Rα, these precursors generate B and T cells both in vitro and following transfer into lethally irradiated hosts. Although both c-Kit−Flt3+ precursors and CLPs generate B and T cells without overt potential for nonlymphoid lineages, these populations differ in that unlike CLPs, c-Kit−Sca-1highFlt3+ precursors are TdT− and generate fewer myeloid progeny relative to cells within the CLP fraction. Cells within each pool also differ with respect to relative expression of several transcription factors affiliated with early B lymphopoiesis including EBF1, PAX5, and E2a, and unlike CLPs Flt3+ precursors are highly dependent on Flt3 ligand stimulation for IL-7-induced B lineage differentiation in vitro. Our data show that lymphoid-restricted precursors exist in multiple distinct progenitor pools that vary with respect to expression of cytokine receptors and transcription factors commonly associated with early lymphoid precursors.
Materials and Methods
C57BL/6 mice and B6.Ly5.2 (referred to as B6.Ly5SJL) mice were purchased from the National Cancer Institute animal facility (Frederick, MD) or The Jackson Laboratory. Both NG-BAC (RAG2/GFP) transgenic reporter mice (20) and H2-SVEX (21) transgenic mice are maintained in our animal colony and were provided by Dr. M. Nussenzweig (Rockefeller University, New York, NY) and Dr. R. Gerstein (University of Massachusetts, Worcester, MA), respectively.
Abs and analytical flow cytometry
For flow cytometric analyses, BM, spleen, and thymus suspensions were prepared and stained with optimal dilutions of Abs directly conjugated to fluorescein, PE, PE-Cy5.5, PE-Cy7, allophycocyanin, allophycocyanin-Cy7, or biotin-Cy7, as previously described (22). The following specific Abs were used for the detection of cell surface Ags: B220 (RA3-6B2), CD11b (M1/70), Gr-1 (8C5), Ter119, and CD3 (2C11) comprising lineage (Lin); CD127/IL-7Rα (A7R34) (23), CD135/Flt3 (A2F10), CD117/c-Kit (2B8), C1qR/AA4 (AA4.1) (24), Sca-1/Ly6 A/E (E13-161.7), CD19 (1D3), and CD11c (HL3). Biotin-conjugated Abs were revealed with streptavidin coupled to Pacific blue (Molecular Probes), streptavidin-PE Texas red, streptavidin-PE-Cy5.5, streptavidin-PE-Cy7, or streptavidin allophycocyanin-Cy7 (Caltag Laboratories). All directly conjugated Abs were purchased from eBioscience except for CD11c, which was purchased from BD Pharmingen, and B220, CD19, AA4, and Ter119, which were purified and conjugated by standard methods in our laboratory. For anti-TdT intracellular staining, cells were stained with Abs to cell surface Ags before treatment with Fix/Perm solutions A and B (Caltag Laboratories). Cells were then incubated with rabbit anti-TdT (SuperTechs) or an isotype control, followed by a biotinylated anti-rabbit goat secondary (Jackson ImmunoResearch Laboratories). TdT expression was then visualized using streptavidin allophycocyanin-Cy7 (Caltag Laboratories). Analysis was performed on an LSRII flow cytometer (BD Biosciences) equipped with four lasers for excitation of UV, violet, blue, and red-excited dyes. Nonviable cells were excluded from all analyses with the UV-excited DNA dye DAPI (4′,6-diamidino-2-phenylindole). All flow cytometric data were analyzed by uploading files into FlowJo 8.3 (Tree Star).
Cell suspensions were applied to one of two 12-parameter FACSDiva high-speed cell sorters or a 12-parameter FACSAria (BD Biosciences). The FACDiva sorters are equipped with Coherent Innova argon lasers tuned to 488 nm, or UV (351/363) excitation and with a Coherent Spectrum argon/krypton laser tuned to 647 nm for excitation of allophycocyanin and its derivatives. The alternative FACSDiva is equipped with argon and argon/krypton lasers for excitation in the blue and red wavelengths as described and a Coherent Innova 302C krypton laser tuned to 407 nm for violet excitation. For both FACSDiva analyses, cells were applied at a sheath pressure of 40 p.s.i. and a drop delay frequency of ∼70,000 drops/sec resulting in sort trigger rates of 20–25,000 cells/sec. The FACSAria is equipped with solid-state blue (488 nm) and violet (407 nm) lasers (Coherent) and a helium-neon red (633 nm) laser (JDS Uniphase). Cell suspensions were applied at a sheath pressure of 70 p.s.i. with a drop delay of ∼98,000 drops/s. This resulted in sorting rates of 28–30,000 cells/s with abort rates of 10–12%.
Intravenous and intrathymic transfers
Hosts were maintained on water containing a Bactrim suspension (400 mg sulfamethoxazole and 80 mg trimethoprim per 500 ml of water) for 1 wk before, and at least 3 wk following, irradiation. One day before i.v. injection, hosts were lethally irradiated (900 rad). Subsequently, a mixture of 500-1000 sorted progenitor cells and 2 × 105 unfractionated host-type BM cells were transferred via the retro-orbital sinus. Intrathymic transfers were performed as previously described (11). Briefly, 500 freshly sorted BM progenitors from female C57BL/6 (Ly5B6) adults were injected into thymi of anesthetized female B6.Ly5SJL mice given 500 rad for 6 h previously.
Cells were sorted directly into RNA lysis buffer consisting of 4.23 mM guanidinium isothiocyanate, 0.67% sarcosyl, 33.3 μM citrate buffer, and 134 mM 2-ME, and RNA precipitated as described (25). cDNA for each sample was synthesized using the First Strand cDNA synthesis kit (Roche Applied Science). For real-time RT-PCR analysis, cDNA from ∼1000 cell equivalents was mixed with Taqman Universal PCR Mastermix and the specific Taqman probes (Applied Biosystems) in 20-μl reactions, and cells were analyzed using a 7300 Real Time PCR System and Sequence Detection software (Applied Biosystems). All cDNA samples were analyzed in triplicate using standard Taqman cycling parameters.
Methylcellulose culture conditions were as previously described (10). Triplicate cultures were established by plating sorted cells at 500 cells/plate according to the manufacturer’s instructions. On day 12, discrete colonies were counted and typed by the morphology.
OP9-GFP and OP9-DL1 stromal cells were a gift of Dr. J.-C. Zuniga-Pflucker (University of Toronto, Toronto, Ontario, Canada) and were maintained as previously described (26). Stromal cells were plated in 24-well plates and grown overnight. The following day plates containing confluent OP9 monolayers were irradiated with 3000 rad before 100 sorted progenitors were added to each well. Medium was supplemented with stem cell factor (SCF), IL-7, and Flt3 ligand (5 ng/ml each) for B and T cell assays or SCF, IL-3, IL-6, G-CSF, M-CSF, GM-CSF (all 10 ng/ml), Flt3 ligand (5 ng/ml), and IL-7 (1 ng/ml) for myeloid-biased assays. Triplicate cultures were harvested at the indicated time points. Cell numbers were calculated by comparison to samples containing known cell numbers collected at the same flow rate.
For stromal-free cultures, 2000 sorted progenitor populations were cultured in 96-well round-bottom plates for 13–14 days. Medium was supplemented with the appropriate cytokines at 50 ng/ml and was changed halfway through the culture period.
The RAG1/2 locus is active in very early lymphoid progenitors (27). Specifically, Igarashi et al. (27) used RAG1-GFP reporter mice to identify lymphoid-biased HSC-proximal progenitors termed early lymphoid progenitors (ELPs). The earliest ELPs are further defined by the cell surface phenotype IL-7Rα−Flt3+c-KithighSca-1+, and constitute a subset of cells within the Lin−Sca-1+c-Kithigh (LSK) progenitor pool that also contains self-renewing RAG1− Flt3− HSCs and nonrenewing or transiently renewing RAG1−/low Flt3+ lymphoid-primed MPPs (28, 29, 30, 31). Using 8-wk-old NG-BAC (RAG2/GFP) BAC transgenic mice (20), we noted that a fraction of Lin−IL-7Rα−c-Kit−Sca-1high BM cells expressed low to intermediate levels of GFP (data not shown). Further analysis of the Lin−IL-7Rαlow/−c-Kit−Sca-1high BM fraction in these mice showed that GFP/RAG2low cells within this pool coexpressed the C1qR homolog AA4 and the cytokine receptor Flt3. The 18–22% of Lin−IL-7Rα−c-Kit−Sca-1+ BM cells were Flt3+ AA4+ RAG2/GFPlow (Fig. 1,A). We did not detect GFP/RAG2+ cells among Lin−IL-7Rα−/lowc-Kit−Sca-1high BM cells that were Flt3− AA4−, and these cells did not generate detectable progeny cells in any of the assays described below (data not shown). Significantly, GFP/RAG2low c-Kit−Sca-1highFlt3+ cells were clearly distinguished from cells within the BM “CLP” population by relative expression levels of the GFP/RAG2 transgene, whereas GFP/RAG2+ c-Kit−Sca-1highFlt3+ cells expressed uniformly low levels of GFP/RAG2, CLPs (Lin−IL-7Rα+c-KitlowSca-1low Flt3+ AA4+) were GFP/RAG2high (Fig. 1,A). Furthermore, by backgating on the CLP and c-Kit−Sca-1highFlt3+ populations, we found that these populations differ substantially with respect to the relative surface levels of Sca-1 and AA4, with cells within the c-Kit−Flt3+ population expressing substantially higher levels of both cell surface proteins (Fig. 1 B).
c-Kit−Sca-1highFlt3+ precursors preferentially generate B and T cells
Previously described RAG+ precursors in adult BM preferentially generate lymphocytes over myeloid cell types following adoptive transfer. We therefore compared the differentiative potential of c-Kit−Sca-1highFlt3+ AA4+ cells to Flt3+ LSK cells and CLPs defined as IL-7Rα+Flt3+Sca-1lowc-Kitlow AA4+. Cells from 10-wk-old C57BL/6 mice were sorted and transferred at 500 cells per recipient into irradiated B6.Ly5SJL congenics. Intravenous and intrathymic transfers were performed to assess B and myeloid lineage or T lineage potential, respectively, and reconstitution of each lineage was assessed at several time points to allow us to infer the relative degree of maturity of potential precursors within each population. This latter strategy is supported by data showing that Flt3+ LSK cells and CLPs differ with respect to the rate at which they first generate T cells and their capacity to sustain thymopoiesis following intrathymic transfer (11). These studies were extended by experiments to assess lymphoid and myeloid lineage potential under several culture conditions.
For i.v. transfer experiments, host splenocytes were harvested at 14, 22, and 36 days posttransfer and analyzed for surface expression of CD11b (Mac1), CD11c, CD3ε, and CD19. As shown in Fig. 2, whereas Flt3+ LSK cells generated B cell (CD19+), myeloid cell (CD11b+ CD11c−), and dendritic cell (CD11b+/−CD11c+) progeny, only donor-derived B lineage cells were consistently detected in recipients given either CLPs or c-Kit−IL-7Rα−Sca-1+Flt3+ AA4+ cells. Furthermore, consistent with previous data, within this time frame donor-derived T lineage cells (CD3ε+) were only readily detected in mice given Flt3+ LSK cells (Fig. 2 B).
Donor-derived T lineage development was assessed at 14, 22, and 29 days postintrathymic transfer (Fig. 2, C and D). Consistent with previous findings (11), Flt3+ LSK cells generated a greater number of CD4+CD8+ thymocytes at day 14 and effectively sustained the CD4+CD8+ pool for at least 29 days compared with an equivalent number of CLPs. Strikingly, reconstitution kinetics for c-Kit−IL-7Rα−Sca-1+Flt3+ AA4+ precursors paralleled CLPs in terms of both overall numbers of donor-derived T lineage cells at each time point as well as their failure to sustain the CD4+CD8+ pool beyond 22 days posttransfer (Fig. 2 B).
We also assessed myeloid potential with two in vitro systems. Using methylcellulose cultures supplemented with IL-3, SCF, and erythropoietin, we found that, in contrast to Flt3+ LSK cells, neither CLPs nor c-Kit−Sca-1highFlt3+ AA4+ precursors yielded a significant number of myeloid colonies (Fig. 3,A). In addition, we tested whether c-Kit−Sca-1highFlt3+ AA4+ precursors generated myeloid progeny when added to cytokine-supplemented OP9 stromal cultures. These cultures were designed to strongly bias toward myeloid differentiation in that they contained minimal concentrations of IL-7 while also containing IL-3, IL-6, SCF, M-CSF, G-CSF, GM-CSF, and Flt3. A total of 100 cells were added in triplicate for each population, and cultures were analyzed at day 7. As shown, whereas Flt3+ LSK cells and CMPs only generated myeloid cell types (CD11b+ GR-1low/high) in this system, despite the large number of myeloid-affiliated cytokines the bulk of cells derived from CLPs and c-Kit−Sca-1highFlt3+ AA4+ precursors were B220+CD19+ B lineage cells (Fig. 3, B and C). However, it should be noted that under these conditions we discerned a small number of myeloid cell types in cultures initiated with CLPs, and this was not the case for c-Kit−Sca-1highFlt3+ AA4+ precursors (Fig. 3 C). We conclude that c-Kit−Sca-1highFlt3+ AA4+ precursors are a largely lymphoid-restricted progenitor population. We refer to these cells as lymphoid-biased progenitors or LBPs.
Although the data in Fig. 1 suggest that CLPs and LBPs are separate populations, we thought it important to further rule out potential overlap between these populations. Indeed, although the disparate RAG2/GFP levels between these cells (Fig. 1,A) and our back gating approach (Fig. 1,B) suggested little if any overlap, the latter approach is directly influenced by upstream “parent” gates that might magnify subtle differences in surface expression of key cell surface Ags. Given that adult lymphopoiesis is highly dependent on IL-7, we thought it particularly important to compare relative IL-7Rα expression on CLPs vs c-Kit−Flt3+ precursors without parent gates based on IL-7Rα. Accordingly, we first gated on Lin− NG-BAC BM cells with high vs low GFP/RAG2 levels, then resolved the relevant populations based on potential differences in c-Kit and Sca-1 surface levels. As shown (Fig. 4 A), whereas c-KitlowSca-1low AA4+ CLPs were highly enriched among Lin− GFP/RAG2high cells, the c-Kit−Sca-1high AA4+ population was restricted to the GFP/RAG2low gate. Furthermore, c-Kit−Sca-1high AA4high precursors clearly expressed lower levels of IL-7Ra compared with CLPs.
To test the functional relevance of these differences in cytokine receptor expression, we exploited our previous finding that IL-7 stimulation is sufficient to promote B cell differentiation from CLPs. We sorted CLPs vs c-Kit−Sca-1highFlt3+ AA4high precursors into culture medium containing all potential combinations of IL-7, SCF, and Flt3, then assayed viability and B cell differentiation by up-regulation of CD19 and B220 4 days later. As shown (Fig. 4,B), IL-7 is sufficient to promote B cell differentiation from CLPs in this system. In contrast, to induce B cell differentiation from c-Kit−Sca-1highFlt3+ AA4high precursors it was necessary to stimulate with IL-7 and FL (Fig. 4 B). Furthermore, in contrast to CLPs, c-Kit−Sca-1highFlt3+ AA4high precursors failed to generate B lineage cells when stimulated with IL-7 and SCF. We conclude that the c-Kit−Sca-1highFlt3+ AA4high population contains novel IL-7Rlow RAG2low B lineage competent precursors with a unique cytokine response profile.
Burst size and kinetics
Given the unique cytokine responsiveness of LBPs, we sought to better quantify burst sizes and the kinetics of B and T cell differentiation in well-controlled culture assays. Accordingly, we sorted and cultured 100 LSK Flt3+, CLP, or LBPs from 8-wk-old C57BL/6 mice on OP9 stromal cells previously transduced with retrovirus expressing the Notch ligand DL-1 or the control construct MigR1-GFP. Previous work has shown that MPPs in adults will generate early T lineage cells when cocultured on OP9-DL1 cells. In contrast the absence of DL-1 expression on OP9-GFP cells allows an efficient measure of B lineage precursor activity (26, 32). As shown in Fig. 5, all three populations including LBPs yielded Thy-1high c-kithigh T lineage cells and CD19+B220+ B lineage cells when cocultured on OP9-DL-1 and OP9-GFP cells, respectively. Notably, however, each population exhibited distinct expansion kinetics in these cultures.
Following 13 days in culture, cells within all three progenitor pools produced a significant number of CD19+B220+ B cell progeny on OP9-GFP stromal cells (Fig. 5,C). Despite their CLP-like characteristics, LBPs were more similar to MPPs in terms of their efficiency for B lineage development in vitro; both MPPs and LBPs expanded ∼500-fold in OP9-GFP culture, whereas CLPs expanded ∼2500-fold (Fig. 5,C). In contrast, all three populations yielded similar numbers of T-lineage progeny when cultured on OP9-DL1 stromal cells, although Thy1+ cells derived from MPPs exhibited greater expansion in the last 3 days of culture compared with CLPs and LBPs (Fig. 5 D). Therefore, although T cell developmental capacity was similar in LBPs compared with CLPs, CLPs were substantially more efficient at producing B cells under these conditions.
We reasoned that the lower number of CD19+B220+ cells recovered from cultures seeded with LBPs reflect either reduced frequency of B lineage precursors or less cellular expansion in response to these culture conditions. To distinguish between these possibilities, we sorted a limiting number of LBPs or CLPs onto pre-established OP9-GFP monolayers supplemented with SCF, IL-7, and Flt3, and enumerated wells containing B220+CD19+ cells 21 days later. As shown in Table I, relative to CLPs, LBPs exhibited 2- to 4-fold reduced cloning efficiency on OP9-GFP stromal cells, and 3- to 6-fold decreased cloning efficiency on OP9-DL1 cells.
|Stromal Cell Line .||No. of Cells Seeded/Well .||Progenitor Populationb .||.|
|.||.||CLPs .||LBPs .|
|OP9-GFPc||1||12/360 (3.3%)||15/360 (4.2%)|
|2||10/96 (10.4%)||5/96 (5.2%)|
|5||12/42 (28.6%)||3/38 (7.9%)|
|20||4/7 (57.1%)||4/8 (50%)|
|OP9-DL1d||1||40/72 (55.6%)||12/72 (16.7%)|
|2||16/16 (100%)||6/16 (37.5%)|
|20||8/8 (100%)||7/8 (87.5%)|
|Stromal Cell Line .||No. of Cells Seeded/Well .||Progenitor Populationb .||.|
|.||.||CLPs .||LBPs .|
|OP9-GFPc||1||12/360 (3.3%)||15/360 (4.2%)|
|2||10/96 (10.4%)||5/96 (5.2%)|
|5||12/42 (28.6%)||3/38 (7.9%)|
|20||4/7 (57.1%)||4/8 (50%)|
|OP9-DL1d||1||40/72 (55.6%)||12/72 (16.7%)|
|2||16/16 (100%)||6/16 (37.5%)|
|20||8/8 (100%)||7/8 (87.5%)|
Cells were cultured for 21 days in medium supplemented with SCF, IL-7, and Flt3 (5 ng/ml each). Data were combined from two independent experiments.
Values indicate number of positive wells per total number of seeded wells (and the percentage) for each population at the indicated dose.
Wells scored positive if containing CD19+B220+ progeny.
Wells scored positive if containing Thy-1high progeny.
TdT expression and V(D)J recombinase activity
We next sought to identify additional characteristics that might distinguish LBPs from ELPs or CLPs. In this regard, Igarashi et al. (27) showed that TdT expression does not strictly correlate with RAG1 expression among BM LSK cells. We therefore compared TdT expression among defined progenitor pools in the BM and thymus by flow cytometry. As shown in Fig. 6 A, whereas TdT+ cells were clearly observed among Flt3+ LSK cells and CLPs, with nearly 100% of CLPs being TdT+, we did not detect TdT+ cells within the LBP pool.
In adults initial TdT and RAG1/2 expression and initiation of DH-JH rearrangements on the Ig H chain locus occur within HSC-proximal Flt3+ LSK precursors with myeloid potential (Fig. 3) (27). However the paucity of TdT+ cells together with the relatively low levels of RAG2 expression among highly LBPs, (Fig. 1,A), suggested that loss of myeloid potential can also precede initiation of V(D)J recombination. To address this possibility further, we determined the frequency of LBPs and CLPs exhibiting ongoing or past V(D)J recombinase activity using H2-VEX recombinase reporter mice. This strategy is advantageous over PCR assays for DH-JH rearrangements because it reveals the precise frequency of cells within a given population with functional recombinase activity (33). As shown (Fig. 6 B), consistent with our previous data (21), 20–40% of CLPs were recombinase (VEX) positive. In contrast, few if any LBPs were recombinase-positive. Therefore, unlike CLPs, LBPs have yet to express an active V(D)J recombinase.
B cell subset potential
Given that TdT expression is unique to adult progenitors, the paucity of TdT+ cells among LBPs suggested that these cells exhibit certain properties of fetal lymphoid precursors. Another property that distinguishes adult from fetal lymphoid precursors is the preferential nature with which fetal precursors generate B1 B cells. Many studies suggest that adult precursors do not effectively reconstitute the B1 B cell pool (13, 34, 35), and B1 B cells often bear Ag receptor rearrangements that lack TdT-mediated N-additions (36, 37). Similarly marginal zone B cells are enriched for B cells lacking N-additions (38, 39). However in other studies, undefined adult progenitors contributed measurably to the B1 B cell pool (40), and marginal zone B cells are readily reconstituted by both fetal liver and adult BM progenitors (41) (our unpublished data). To test whether LBPs preferentially generate B1 or splenic marginal zone B cells, sorted LBPs, CLPs, and Flt3+ LSK cells were transferred i.v. into irradiated Ly5-congenic hosts, and the subset composition for donor-derived B cells in the peritoneal cavity and spleen determined 12 wk later. We found that under these conditions each donor population including CLPs and LBPs generated all B cell subsets at comparable frequency. Regardless of donor population used, the donor-derived B cell pool in each chimera contained near normal frequency of peritoneal cavity B1 B cells (20–25% CD19+ IgMhigh CD11b+CD5+/−) and splenic marginal zone B cells (5–10% IgMhigh CD21highCD23−) (data not shown). We conclude that LBPs do not preferentially generate B1 or marginal zone B cells.
Gene expression profile
The lack of c-Kit and IL-7Rα surface expression by LBPs led us to further compare expression of genes typically associated with early lymphoid progenitors in LBPs vs HSCs, Flt3+ LSK cells, CLPs, and pro-B cells. Accordingly, we subjected cDNA prepared from each population to real-time PCR using primers and probes specific for the genes indicated in Fig. 7. As expected LBPs exhibited lower message levels for TdT, and IL-7Rα compared with CLPs. We should note, although we were at first surprised that TdT message levels were reduced in pro-B cells compared with CLPs, these data are consistent with our unpublished flow cytometric analyses showing that although 100% of CLPs are TdT+, only 50% of B220+CD43+CD19+ AA4+ pro-B cells are TdT+ (data not shown).
Additional gene expression patterns of interest concerned Notch1, E2a, and the B lineage-affiliated genes EBF1, Ig-α, and PAX5. Given that LBPs readily generated T lineage cells upon intrathymic transfer, we were surprised to observe that Notch1 message levels among LBPs were 8-fold lower compared with Flt3+ LSK cells and roughly 11-fold lower compared with CLPs (Fig. 7). Likewise, compared with CLPs, LBPs also exhibited lower message levels for EBF1, PAX5, and E2a, but roughly 5-fold higher for Igα. Interestingly, LBPs and CLPs expressed equivalent levels for CCR9, a chemokine receptor recently proposed to play a significant role in homing of BM progenitors to the thymus (42). Together, these data illustrate that LBPs are characterized by relatively low expression of many genes typically associated with early lymphoid development including Notch1 and EBF1, yet exhibit substantial expression of other genes associated with early lymphocyte development such as Igα and CCR9.
Standard models of early lymphocyte development predict that, in adults, lymphoid-specified precursors arise from within the Flt3+ LSK pool and ultimately progress into the CLP populations before yielding B or T cells. This model further predicts that progression of lymphoid-biased cells toward the CLP populations is accompanied by the down-regulation of c-Kit and Sca-1 expression and relative increases in RAG1/2 and IL-7Rα expression (Fig. 1) (1). The notion that all B and T cell precursors arise from c-KitlowSca-1low CLPs has been challenged by studies suggesting that thymic T lineage precursors arise directly from Flt3high LSK cells or ELPs, and from more recent data suggesting that T lineage-restricted precursors in the blood are c-Kit− (43). Our data characterizing unique CLP-like lymphoid precursors further challenge this model, and suggest that B lineage- and T lineage-biased precursors are resident within several functionally distinct progenitor pools characterized by differential expression of several gene products associated with early lymphocyte development including TdT, IL-7R, and c-Kit.
Past studies characterizing potential c-Kit− BM cells have come to markedly different conclusions. Earlier work (44) concluded Lin−/low Sca1+c-Kit− BM cells contained little progenitor activity in standard adoptive transfer experiments. Given that Flt3+ AA4+ RAG2int LBPs constitute only 15–20% of the Lin−/lowSca-1+c-Kit− BM pool (Fig. 1), and precursor frequency for the T cell and B cell lineages are below or at best comparable to cells within the CLP population (Table I), obtaining measurable lymphoid precursor activity for Lin−/lowSca-1+c-Kit− BM cells likely requires clean isolation of Flt3+ AA4+ cells within this population. More recently Krueger and von Boehmer (43) characterized c-Kit− T lineage-restricted precursors in the blood. In this study, blood T lineage precursors were further identified via their expression of a reporter gene for the T lineage-affiliated gene pre-Tα, and were shown to arise independently of Notch activation (43). Significantly, blood c-Kit− precursors appear to differ from BM LBPs in that, whereas the blood population lacks B lineage potential on OP9-GFP stromal cells, BM LBPs yield B cells in vitro and in vivo. Nonetheless, it is tempting to speculate that BM LBPs serve as a progenitor pool for c-Kit− pre-Tα+ T lineage precursors in the blood. However, it should also be stressed that surface levels of c-Kit may oscillate dramatically under different conditions and in different microenvironments. For instance c-Kit surface expression is higher in fetal CLPs compared with adult, and c-Kit levels on adult-derived CLPs increase upon stimulation with Notch ligands (45, 46).
It should be emphasized that LBPs are responsive to IL-7 provided that they are also stimulated with Flt3. Therefore, although LBPs are enriched in the IL-7Rα− gate illustrated in Fig. 1,A, LBPs must express sufficient numbers of IL-7R heterodimers to confer responsiveness. This notion is consistent with past data show that progenitors may respond to cytokines despite undetectable surface levels of the appropriate receptor (47). Moreover, our past data indicate that IL-7 and Flt3 signaling synergize to maximize proliferation of CLPs (22). We suggest that costimulation with IL-7 and Flt3 activates similar signaling pathways within the CLP and LBP populations. Therefore the unique feature of LBPs, with regard to cytokine responsiveness, is that LBPs do not express a sufficient number of IL-7R heterodimers to respond to IL-7 without coengagement of additional cytokine receptors. The low IL-7R surface expression on LBPs (Fig. 4,A) may also explain the reduced clonal burst size observed for LBPs under B lineage inducing conditions in vitro. Consistent with the capacity of Notch stimulation to enhance survival and proliferation, when placed in Notch ligand-driven T-lineage promoting conditions LBPs overcome this defect (Fig. 5).
Our data indicate that RAG2 expression defines several distinct lymphoid-biased BM precursor populations. These populations vary with respect to several attributes typically associated with lymphopoiesis in adults including TdT expression and V(D)J recombinase activity. Thus, whereas canonical LBPs within the LSK and CLP populations express relatively high levels of RAG2 and are TdT+, noncanonical progenitors within the c-Kit−Sca-1highFlt3+ LBP population express low to intermediate levels of RAG2 and are TdT− (Figs. 1 and 6). Although lack of TdT expression is a feature of lymphoid precursors in fetal life, our data suggest that adult HSCs continue to generate TdT− progenitors throughout life. This possibility is further supported by experiments in which donor-derived LBPs were readily detected in BM chimeras established with adult HSCs (data not shown). Together these observations strengthen and extend the notion advanced by Igarashi et al. (27) that initial expression of RAG1/2 and TdT occur asynchronously in nonoverlapping progenitors.
It is generally thought that B1 B cells derive predominately from TdT− fetal liver precursors. This notion is consistent with many experiments, performed mostly with BALB/c mice, showing that whereas adoptive transfer of fetal liver progenitors results in robust reconstitution of peritoneal cavity B1 B cells, adult progenitors fail to effectively generate these cells (13, 35). Given that LBPs in adults are TdT−, we tested whether LBPs preferentially generated B1 B cells. We found that adult CLPs and LBPs derived from C57BL/6 adults yielded similarly small numbers of B1 B cells (data not shown). Although this result may appear surprising, a lack of TdT-mediated N-additions is by no means the single key factor for whether a given progenitor adopts the B1 B cell fate. Indeed, Ig receptors expressed by many B1 B cells show evidence of N-additions (48). Likewise, whereas B cells lacking Ig N-additions may also preferentially colonize the splenic marginal zone (38, 39), the BCR repertoire of marginal zone B cells is also exceptionally complex, and contains B cells derived from both TdT+ and TdT− precursors (38). Given these considerations it is perhaps not surprising that CLPs and LBPs also generated normal ratios of follicular and marginal zone B cells (data not shown).
Warren and Rothenberg (49) proposed that stochastic interactions among multiple transcription factors initiate the generation of multiple functionally similar progenitor populations characterized by disparate gene expression profiles. This model argues against the notion that a single specific sequence of molecular events is required for adoption of a particular fate. Consequently, expression of different surface receptors and transcription factors typically used to identify functionally defined precursors may vary across functionally analogous progenitor populations. Several previous observations fit this model. For instance, although the cell surface receptor CD19 is clearly a direct target of PAX5, a key transcription factor for promoting B-lineage commitment, Montecino-Rodriguez and Dorshkind (3) identified CD19+ precursors with potential for the myeloid as well as the B cell lineage. Therefore activating the PAX5/CD19 pathway may not always reflect firm B lineage commitment, and additional transcription factors such as EBF1 may play a central role in driving B cell commitment from these progenitors. This possibility is strengthened by our recent data showing that EBF1 can suppress myeloid fates independently of PAX5 (50).
Current models of early lymphopoiesis further predict that lymphoid priming, defined by initial expression of RAG1/2, TdT, and additional lymphoid-affiliated genes occurs within myeloid-lineage competent Flt3+ LSK cells (27). These cells have also been termed lymphoid-primed MPPs (29, 51). Recent data indicate that initiation of lymphoid-affiliated gene expression including RAG1 and TdT expression in lymphoid-primed MPPs likely requires the E2a and Ikaros transcription factors (52, 53). Significantly Ikaros and the E2a proteins have also been connected to the mechanisms underlying loss of myeloid potential in early lymphoid precursors (52, 54). Given that LBPs have markedly reduced RAG2 expression relative to CLPs, we suggest that loss of myeloid potential does not always coincide with increased RAG and TdT expression and initiation of V(D)J recombinase activity. Altogether these observations suggest that the molecular machinery underpinning loss of myeloid potential can be separated from those responsible for activating the V(D)J recombinase. One possibility is that the transcriptional events required to initiate the lymphoid program in LBPs differ from those required to generate CLPs. Indeed, the unique cell surface and gene expression profile exhibited by LBPs suggests that many transcriptional events associated with early lymphocyte development in canonical progenitors such as lymphoid-primed MPPs and CLPs either do not occur or occur with unique kinetics and through alternative molecular mediators in LBPs. This viewpoint is further supported by data suggesting that vigorous expression of the canonical EBF1 target Igα is achieved in LBPs despite less than robust expression of EBF1 (Fig. 7). In this regard, LBPs may constitute an alternative cellular pathway for generating early T and B cell precursors in adults. Therefore in our view attempts to place LBPs in the cellular pathway at some point between lymphoid-primed MPPs and CLP or CLP2 precursors may be pointless.
Finally, a subset of CMPs identified by Flt3 expression exhibit residual B cell potential (8, 9, 10). Whereas Flt3+ CMPs, CLPs, and LBPs each express Flt3 and AA4 (10), they differ with respect to surface expression of Sca-1, c-Kit, and IL-7Rα. We suggest that better strategies for identifying and manipulating lymphoid-related precursors in the BM and elsewhere will be necessary to generate a comprehensive and testable model for all precursor-product relationships in lymphopoiesis. In this regard, the generation of several additional reporter mouse lines for identifying lymphoid precursors based on the expression of key transcription factors and their relevant target loci would be highly advantageous. Integration of these novel experimental tools with current strategies to resolve progenitors based on expression of specific cell surface markers should lead to a unifying model for all relevant precursor-product relationships in fetal and adult lymphopoiesis.
We thank Drs. Avinash Bhandoola, Rachel Gerstein, and Hans Snoeck for helpful discussions. We also gratefully acknowledge the expert technical support in flow cytometry provided by the Abramson Cancer Center Flow Cytometry and Cell Sorting Shared Resource and in particular the efforts of Richard Schretzenmair, William Murphy, and Ryan Wychowanec.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Grant AG546776 from the National Institutes of Health. D.A. is the recipient of a Career Development Award from the Leukemia and Lymphoma Society.
Abbreviations used in this paper: MPP, multipotent progenitor; HSC, hematopoietic stem cell; CLP, common lymphoid progenitor; CMP, common myeloid progenitor; LBP, lymphoid-biased progenitor; ELP, early lymphoid progenitor; LSK, Lin−Sca-1+c-Kithigh; BM, bone marrow; SCF, stem cell factor.