Mechanisms of lymphoid and myeloid lineage choice by hemopoietic stem cells remain unclear. In this study we show that the multipotent progenitor (MPP) population, which is immediately downstream of hemopoietic stem cells, is heterogeneous and can be subdivided in terms of VCAM-1 expression. VCAM-1+ MPPs were fully capable of differentiating into both lymphoid and myeloid lineages. In contrast, VCAM-1− MPPs gave rise to lymphocytes predominately in vivo. T and B cell development from VCAM-1− MPPs was 1 wk faster than that from VCAM-1+ MPPs. Furthermore, VCAM-1+ MPPs gave rise to common myeloid progenitors and VCAM-1− MPPs in vivo, indicating that VCAM-1− MPPs are progenies of VCAM-1+ MPPs. VCAM-1− MPPs, in turn, developed into lymphoid lineage-restricted common lymphoid progenitors. These results establish a hierarchy of developmental relationship between MPP subsets and lymphoid and myeloid progenitors. In addition, VCAM-1+ MPPs may represent the branching point between the lymphoid and myeloid lineages.
Hemo/lymphopoiesis is a developmental program that is sustained throughout the life span of an animal. This continuous process is made possible through the balance between hemopoietic stem cell (HSC)3 renewal and differentiation (1). Residing in the bone marrow, HSC is the common ancestor of all classes of lymphoid and myeloid lineage cells. Upon committing to differentiation, HSCs progressively lose their self-renewal ability and can be separated into long-term (LT-) and short-term (ST-) HSCs based on the duration for which they can sustain hemopoiesis in vivo (2). LT-HSCs can fully reconstitute the hemopoietic system for at least the life span of an animal, whereas reconstitution by the more differentiated ST-HSCs is only transient, for ∼6 wk (3, 4). ST-HSCs give rise to multipotent progenitors (MPPs) that have no or very limited self-renewal capability, but have multilineage differentiation potential (3, 4, 5). After developmental stage, the differentiation potential of MPPs becomes restrictive to either the lymphoid or myeloid lineage. Common lymphoid progenitors (CLPs) and common myeloid progenitors (CMPs), of which differentiation potential is exclusive to the lymphoid and myeloid lineage, respectively, are the earliest lineage committed progenitors identified and isolatable from the mouse bone marrow (6, 7). The developmental relationship between MPPs and CLPs or CMPs, however, has not been determined.
Although CLPs are the most immature progenitors identified with restricted differentiation potential within the lymphoid lineage, the branching point between the lymphoid and myeloid lineages probably takes place at an earlier developmental stage. The lymphocyte developmental program is suggested to initiate upstream of CLPs, because their differentiation potential is skewed toward the B lymphocyte lineage despite having the potential to give rise to T and NK cells at the clonal level (6, 8). Furthermore, progenitors with the cell surface phenotype of CLPs are not found in the thymus, where T cell development takes place. Instead, cells with the phenotype of the more upstream MPPs have been identified in the thymus as well as in the circulation with a high T cell differentiation potential (9, 10).
The MPP population, which is upstream of CLPs, has not been well characterized. Within the c-Kit+lineage (Lin)−/lowSca-1+ (KLS) population in the mouse bone marrow, which contains both HSCs and MPPs, MPPs have been characterized using different combinations of surface markers, e.g., Thy-1.1lowMac-1lowCD4low, Flt3+, or Thy-1.1−Flt3+, all of which overlap with one another, but not totally identical (3, 5, 11). In vivo reconstitution with limiting numbers of MPPs (10–20 cells) in some animals do not give rise to myeloid lineage cells (3, 4), suggesting that the MPP population may be heterogeneous and contain lymphoid- or myeloid-restricted progenitors. Additionally, progenitors with lymphoid and GM, but not erythroid, differentiation potential has been identified in the fetal liver (12, 13), pointing to the possibility that an adult counterpart might be present upstream of CLPs in the bone marrow.
In this report we show that VCAM-1 is a useful cell surface marker to subdivide the Flt3+ MPP population. Although both VCAM-1+ and VCAM-1− MPPs can give rise to lymphocytes, VCAM-1− MPPs have significantly less GM differentiation potential and almost no erythroid differentiation potential compared with the VCAM-1+ subset. These data suggest that MPPs with multilineage differentiation potential are contained within the VCAM-1+ subset, whereas VCAM-1− MPPs are enriched for lymphocyte precursors. We propose a more precise hemopoietic tree, which shows the potential branching point of lymphoid and myeloid lineages.
Materials and Methods
Wild-type mice used in this study were C57BL/Ka-Thy1.1 and C57BL/Ka-Thy1.1-Ly5.2 (CD45.1). C57BL/Ka-Thy1.1 (CD45.1/CD45.2) recipients were generated by crossing C57BL/Ka-Thy1.1 and C57BL/Ka-Thy1.1-Ly5.2. RAG2/common γ-chain (γc) double-knockout (DKO) mice were backcrossed more than eight generations with C57BL/6 mice. Enhanced GFP (EGFP) mice were generated as described previously (14) and were backcrossed onto C57BL/Ka-Thy1.1 background for more than eight generations. For characterization and purification of bone marrow progenitor populations, mice at 5–8 wk of age were used. Recipient mice used for reconstitution studies were 8–12 wk of age. All mice were maintained under specific pathogen-free conditions at the Duke University animal care facility. All studies and procedures were approved by the Duke University Animal Care and Use Committee.
To detect HSCs and MPPs, bone marrow cells from C57BL/Ka-Thy1.1 mice were incubated with anti-c-Kit MACS beads (Miltenyi Biotec) for 20 min. After washing with staining medium (HBSS containing 2% FCS and 0.02% NaN3), c-Kit+ cells were enriched by autoMACS with Possels mode. After blocking with 1 mg/ml normal rat IgG (Sigma-Aldrich), cells were stained with FITC-anti-Thy-1.1 (HIS51), PE-anti-Flt3 (A2F10.1), PE/Cy5-anti-lineage (Lin; B220 (RA3-6B2), Mac-1 (M1/70), Gr-1 (RB6-8C5), TER119, CD3ε (145-2C11), CD4 (RM4-5), and CD8α (53-6.7)), Alexa Fluor 594-anti-Sca-1 (E13-161.7), and allophycocyanin-anti-c-Kit (2B8). For purification of MPPs in EGFP mice, after enrichment of c-Kit+ bone marrow cells, cells were stained with PE-anti-Flt3 (A2F10), PE/Cy5-anti-Lin with anti-Thy1.1, Alexa Fluor 594-anti-Sca-1, and allophycocyanin-anti-c-Kit. HSCs and MPPs were defined as c-KithighThy-1.1lowLin−/loSca-1+ and Flt3+c-KithighThy-1.1−Lin−/loSca-1+, respectively, as described previously (3). LT-HSC and ST-HSC were Flt3− and Flt3+ HSCs, respectively. To detect VCAM-1 expression, cells were stained with biotinylated anti-VCAM-1 (clone 429), followed by streptavidin-PE/Cy7.
To purify CMPs, we stained c-Kit+ cell-enriched bone marrow with FITC-anti-CD34 (RAM34), PE-anti-FcγR (2.4G2), PE/Cy5-anti-Lin-containing anti-IL-7Rα (A7R34) and anti-Thy-1.1, Alexa Fluor 594-anti-Sca-1, and allophycocyanin-anti-c-Kit. CMPs were defined as Lin−Sca-1−c-KithighCD34+FcγRlowThy-1.1− (7). CLPs were purified as described previously (15).
Analysis of reconstituted mice was performed as previously described (6, 16). To analyze peripheral blood, thymus, and spleen, cells were stained with FITC-anti-CD45.2 (clone 104) and PE/Cy7-anti-CD45.1 (A20) as well as PE- or allophycocyanin-conjugated anti-lineage markers in combinations as described in the figure legends. For analysis of EGFP donor-derived cells, cells were detected with Abs for MPPs and CMPs as described above, except PE-anti-CD34 (RAM34) was used in place of FITC-CD34.
Dead cells were gated out from analyses and cell sortings as propidium iodide-positive cells. All Abs were purchased from eBioscience, except for anti-CD34 and FcγR, which were obtained from BD Pharmingen. Anti-Sca-1 Abs were purified from hybridoma culture supernatant and conjugated in our laboratory with Alexa Fluor 594 using the protein labeling kit (A-10239) available from Molecular Probes. Cell sorting and cell surface phenotyping were performed on a FACSVantage SE with a DiVa option (488 nm argon, 599 nm dye, and 408 nm krypton lasers; BD Bioscience Flow Cytometry Systems), which is available in the FACS facility of Duke Comprehensive Cancer Center. Data were analyzed with FlowJo software (TreeStar).
One hundred cells of each population were sorted directly onto 1 ml of methylcellulose containing stem cell factor (SCF), IL-3, IL-6 (MethoCult 3534; StemCell Technologies) supplemented with 10 ng/ml GM-CSF (R&D Systems) in 35-mm dishes (Falcon 3001; BD Biosciences) to detect GM colonies. To detect erythroid differentiation potential, cells were sorted onto methylcellulose containing SCF, IL-3, IL-6, erythropoietin (Epo; MethoCult 3434; StemCell Technologies), supplemented with 10 ng/ml thrombopoietin (Tpo; R&D Systems). After 5–7 days of culture, the number of GM (CFU-GM) and mixed (CFU-GEMM) colonies was enumerated under the microscope. Erythroid colonies (CFU-E) were enumerated 3–4 days after culture. Colony types were confirmed by cytospin preparation, followed by Giemsa staining and microscopic observations.
In vivo injections
Two thousand VCAM-1+ or VCAM-1− MPPs (CD45.1) were i.v. injected into lethal dose (950 rad)-irradiated host (CD45.1/CD45.2) with 2 × 105 whole bone marrow cells (CD45.2), or injected into sublethally irradiated (400 rad) congenic RAG2/γc DKO mice (CD45.2) for analysis of T and B cell development kinetics. Peripheral blood and thymocytes were obtained at various time points from reconstituted mice for FACS analysis as described above. The percentage of chimerism was calculated by (% donor-derived population/(% donor-derived population + % rescue bone marrow-derived population)) × 100. Donor-derived thymocytes were enumerated as % CD45.1+ cells (obtained by FACS) × total number thymocytes counted using a hemocytometer. For analysis of the relationship between MPP subsets and CMPs and CLPs, 5–10 × 104 VCAM-1+ or VCAM-1− MPPs from EGFP mice were injected into lethally irradiated host (CD45.1). Bone marrow cells were harvested from femurs and tibiae 5–7 days after injection for detection and purification of downstream progenitor populations.
Stromal cell cultures
For clonal analysis of GM and B cell differentiation potential, single VCAM-1+ or VCAM-1− MPPs were sorted by the automatic cell deposition unit (BD Biosciences) on FACSVantage SE into each well of 96-well plate on OP9 stromal cell layers. After 2 days of culture in the presence of SCF (50 ng/ml; R&D Systems) and Flt3 ligand (Flt3L; 30 ng/ml; R&D Systems), additional cytokines including IL-3 (10 ng/ml; R&D Systems), GM-CSF (10 ng/ml; R&D Systems), and IL-7 (10 ng/ml; R&D Systems) were added to each well to drive both GM and B cell differentiation. Positive wells were identified by microscopic observation. Positive wells were harvested by rigorous pipetting after 8–11 days and were analyzed by FACS to evaluate their GM and B cell differentiation potential.
For B and T cell differentiation analysis of CMPs on OP9 and OP9-DL1 cocultures, 100 CMPs were sorted into 96-well plates with OP9 stromal cell layers supplemented with SCF (50 ng/ml), Flt3L (30 ng/ml), IL-3 (10 ng/ml), GM-CSF (10 ng/ml), and IL-7 (10 ng/ml) to evaluate GM and B cell differentiation. To evaluate T cell differentiation potential, cells were similarly sorted into 96-well plates with OP9-DL1 stromal cell layers supplemented with Flt3L (1 ng/ml) and IL-7 (0.5 ng/ml). After 7–14 days in culture, cells were harvested and analyzed by FACS.
For culture of VCAM-1+ MPPs for gene expression analysis, 5000 cells were sorted into 96-well plates with OP9 stromal cell layers. After 1–2 days of culture in the presence of SCF (50 ng/ml), Flt3L (30 ng/ml), and IL-7 (10 ng/ml), cells were harvested from the plates and stained with allophycocyanin-anti-c-Kit and Alexa Fluor 594-anti-Sca-1 for sorting of the c-Kit+Sca-1+ population into TRIzol (Invitrogen Life Technologies) for subsequent RNA isolation and first-strand cDNA synthesis for RT-PCR analysis.
Total RNA was purified from freshly isolated MPP subpopulations and cultured VCAM-1+ MPPs, and whole bone marrow cells were cultured with TRIzol. Oligo(dT)-primed cDNAs were amplified by PCR with GeneAmp PCR system 9700 (Applied Biosystems) for 30–35 cycles using gene-specific primers as previously described (15, 17, 18, 19, 20). Amplified products were subjected to electrophoresis on 1.5% agarose gel and visualized under UV light after ethidium bromide staining. The amount of input DNA was standardized with GAPDH.
Subdivision of MPP population by VCAM-1 expression
To identify early-intermediate progenitors of the lymphoid or myeloid lineage within the MPP population, we performed gene expression profiling of HSCs, CMPs, and CLPs to identify cell surface molecules differentially expressed on these early hemopoietic progenitors. We identified VCAM-1, a cell surface adhesion molecule which was expressed in HSCs and myeloid progenitors, but its expression was absent in CLPs (data not shown). FACS analysis was performed to confirm cell surface expression of VCAM-1 on these populations. All HSCs, defined as Thy-1.1low KLS cells in bone marrow, expressed high levels of VCAM-1, whereas the entire CLP population did not express VCAM-1 (Fig. 1,A). Similar to HSCs, most CMPs expressed VCAM-1 (Fig. 1 A).
Because VCAM-1 was expressed on both HSCs and CMPs, but not on CLPs, we hypothesized that VCAM-1 expression might be down-regulated during early hemopoiesis before lymphoid lineage commitment at the MPP stage. To explore this possibility, we examined VCAM-1 expression on this population. Within the KLS population in mouse bone marrow, LT-HSC, ST-HSC, and MPPs are prospectively distinguishable by the expression of Thy-1.1 and Flt3 (Fig. 1,B). HSCs express low levels of Thy-1.1, whereas MPPs are Thy-1.1 negative (3). All Thy-1.1low cells, comprised of both LT- and ST-HSCs, expressed VCAM-1 (Fig. 2,A, left panel). Although the majority of Thy-1.1− and Flt3+ KLS MPPs expressed VCAM-1 as in HSCs, 10–15% of this population did not express VCAM-1 (Fig. 2,A). Upon two rounds of FACS purification, we were able to isolate distinct VCAM-1+ and VCAM-1− subsets of Thy-1.1−Flt3+ KLS MPPs with purity >99% (Fig. 2,B). The levels of Flt3 expression on VCAM-1+ and VCAM-1− MPPs were not identical. Although all VCAM-1− MPPs expressed high levels of Flt3, there were Flt3high and Flt3low VCAM-1+ MPPs. The mean fluorescent intensities for the Flt3 expression level of VCAM-1+ and VCAM-1− MPPs in Fig. 2,B are 33.5 and 61.4, respectively. Because CLPs are VCAM-1− and express Flt3 (21), we compared the surface phenotype of the MPP subsets and CLPs to ensure that there were no overlaps between these populations. Both MPP subsets expressed high levels of c-Kit and Sca-1, whereas CLPs expressed both markers at lower levels (Fig. 2,C). In addition, the majority of either MPP subset did not express IL-7Rα on the cell surface as in CLPs. Although a small subset of VCAM-1− MPPs appeared to express a low level of IL-7Rα, their higher c-Kit and Sca-1 expressions indicate that they were distinct from the CLP population (Fig. 2 C).
Myeloid differentiation potential of VCAM-1+ and VCAM-1− MPPs
To characterize VCAM-1+ and VCAM-1− subsets of MPPs, we first compared the myeloid differentiation potentials of VCAM-1+ and VCAM-1− MPPs in in vitro cultures as well as in vivo injections. Purified VCAM-1+ and VCAM-1− MPPs were plated onto methylcellulose-containing, myeloid-driven cytokines to evaluate GM and erythroid differentiation potentials. After 5–6 days of culture in the presence of GM-CSF without Epo and Tpo, 40–50% of VCAM-1+ MPPs formed GM colonies. In contrast, only 15–20% of VCAM-1− MPPs initiated GM colony formations (Fig. 3,A). The addition of Epo and Tpo without GM-CSF in methylcellulose supports the growth of all myeloid colonies, including erythroid, GM, and mix colonies. Mix colonies are also called CFU-GEMM, which consists of both GM and erythroid-type cells. Under this culture condition, VCAM-1+ MPPs formed erythroid colonies at 8–11% plating efficiency, whereas 23–26% formed GM colonies and 7–9% formed mix colonies (Fig. 3,A). VCAM-1− MPPs, in contrast, had nearly no erythroid differentiation potential. In addition, VCAM-1− MPPs did not give rise to mix colonies and formed GM colonies at 5% plating efficiency in the absence of GM-CSF (Fig. 3 A).
Next, we examined myeloid differentiation potential of VCAM-1+ and VCAM-1− MPPs in vivo by competitive reconstitution assays. In this experiment we injected 2 × 103 VCAM-1+ or VCAM-1− MPPs (CD45.1) with 2 × 105 whole bone marrow rescue cells (CD45.2) into lethally irradiated congenic mice (CD45.1/CD45.2). Accordingly, we could distinguish donor-derived cells (CD45.1), competitors (CD45.2), and host-type cells (CD45.1/CD45.2) on FACS. Because the burst size of hemopoietic cells from MPPs is limited (3), we examined the reconstitution pattern in the peripheral blood of recipient mice 2 wk after injection of cells. The chimerism of donor-derived myeloid cells was ∼18% when VCAM-1+ MPPs were injected. In contrast, only 4% donor-derived cells were observed in the Gr-1+Mac-1+ myeloid population from VCAM-1− MPPs (Fig. 3 B). These results indicate that VCAM-1− MPPs have significantly less myeloid differentiation potential both in vitro and in vivo compared with the VCAM-1+ subset.
T and B cell differentiation potential of VCAM-1− MPPs
Because VCAM-1− MPPs have lower myeloid lineage differentiation potential, we hypothesized that the differentiation potential of VCAM-1− MPPs might be skewed toward the lymphoid lineage. The reconstitution assay with VCAM-1+ and VCAM-1− MPPs in lethally irradiated mice revealed that although both subsets were potent in differentiating into T and B cells, the kinetics of lymphoid differentiation were different between the two populations (Fig. 4,A). The peaks of CD19+ mature B cells generated from VCAM-1+ MPPs and VCAM-1− MPPs in the periphery occurred 3 and 2 wk after injection, respectively, suggesting that B cell ontogeny is earlier from VCAM-1− MPPs than from VCAM-1+ MPPs (Fig. 4 A, left panel).
Similarly, the appearance of donor-derived T cells in the periphery of reconstituted mice was earlier from VCAM-1− MPPs. The peaks of chimerism of CD3+ T cells from VCAM-1+ MPPs and VCAM-1− MPPs occurred 5 and 4 wk after injection (Fig. 4,A, middle panel). In addition, we observed a delay in thymocyte development when i.v. injecting either subset of MPPs into sublethally irradiated RAG2/γc DKO mice. Because RAG2/γc DKO mice have no mature lymphocytes and very few thymocytes, we can easily detect donor-derived cells in the thymi from these mice. In the thymus, T cell development proceeds in the order of CD4−CD8−, CD4+CD8+, and CD4+CD8− or CD4−CD8+ populations. Fig. 4,B shows the sequential changes in CD4 and CD8 expression from the progeny of VCAM-1+ MPPs and VCAM-1− MPPs in the thymi of RAG2/γc DKO mice. This analysis showed that the onset of T cell development from VCAM-1+ MPPs appears to have a 1-wk delay compared with that from VCAM-1− MPPs. The peak total number of donor-derived thymocytes occurred 3 and 4 wk after injection from VCAM-1+ MPPs and VCAM-1− MPPs, respectively, confirming a 1-wk difference between the development of the two MPP subsets (Fig. 4 A, right panel). Together with the results of B cell differentiation kinetics, we concluded that VCAM-1− MPPs have more rapid lymphoid lineage differentiation potential and are developmentally more advanced than VCAM-1+ MPPs toward the lymphoid lineage.
Majority of both VCAM-1+ and VCAM-1− MPPs are lymphoid and myeloid bipotent progenitors
The lower myeloid colony formation activity from VCAM-1− MPPs prompted us to investigate whether this population is a mixture of lymphoid- and myeloid-restricted progenitors or whether they still retain lymphoid and myeloid bipotency on a clonal level. Single VCAM-1+ and VCAM-1− MPPs gave rise to both Mac-1+/Gr-1+ myeloid cells and CD19+ B cells 9–12 days after coculture with stromal cells at similar frequencies (Fig. 5,B). Although the frequencies of single GM or B lineage readout from VCAM-1+ and VCAM-1− MPPs were obviously different, it is clear that not only VCAM-1+ MPPs, but also VCAM-1− MPPs, have myeloid and lymphoid bipotent differentiation potential. Because VCAM-1− MPPs did not have erythroid differentiation potential, as shown in Fig. 3, the majority of VCAM-1− MPPs might be GM-lymphoid bipotent progenitors with no erythroid differentiation potential. Although both VCAM-1+ and VCAM-1− MPPs comprised similar frequencies of lymphoid and myeloid bipotent progenitors, the difference in myeloid colony formation activities between the two MPP subsets shown in Fig. 3,A could be attributed to the different percentages of GM-committed progenitors in each population (Fig. 5 B). Alternatively, the more primitive progenitors, in this case VCAM-1+ MPP, have a greater capacity to form colonies.
VCAM-1+ MPPs give rise to VCAM-1− MPPs and CMPs in vivo
Because VCAM-1+ MPPs could give rise to all types of myeloid lineage cells as in CMPs, we hypothesized that VCAM-1+ MPPs are the precursors of CMPs. In addition, our kinetics data of lymphocyte development suggest a linear relationship between the two MPP subsets. To test the ability of VCAM-1+ MPPs to develop into VCAM-1− MPPs and CMPs, we injected 0.5–1 × 105 VCAM-1+ MPPs isolated from EGFP mice into congenic lethally irradiated hosts without whole bone marrow rescue cells. Because our results of lymphocyte development suggest a 1-wk difference between the maturation of the two subsets of MPPs, we analyzed the mice 5–7 days after injection. As shown in Fig. 6 A, donor-derived (GFP+) cells comprised of 30% of host bone marrow 7 days after injection with VCAM-1+ MPPs. We were able to detect and isolate Lin−Thy-1.1− IL-7Rα−c-Kit+Sca-1+ phenotypic MPPs (1.8%) as well as Lin−IL-7Rα−Thy-1.1−c-KithighSca-1− myeloid progenitors (13%) by FACS.
As shown in the top panel of Fig. 6,A, we were able to detect both VCAM-1+ and VCAM-1− phenotypic MPPs (Thy-1.1−Flt3+KLS) from donor-derived VCAM-1+ MPPs after 7 days in vivo. We sorted these MPPs onto methylcellulose to evaluate their myeloid differentiation potentials. In two separate experiments, purified VCAM-1+ MPP gave rise to GM colonies at 7 and 12% plating efficiencies, whereas the VCAM-1− population forms colonies at 0 and 4% (Fig. 7 A). These results demonstrated that VCAM-1+ MPPs can give rise to VCAM-1− MPPs with less GM differentiation potential, suggesting a linear relationship between VCAM-1+ and VCAM-1− MPPs.
To confirm the hierarchy of VCAM-1+ and VCAM-1− MPPs in development, we cultured purified VCAM-1+ MPPs in vitro and examined their gene expression changes. We found that after 2 days of culture, up-regulation of RAG1 and induction of IL-7Rα and early B cell factor expression were observed from VCAM-1+ MPPs (Fig. 8). Importantly, these lymphoid-related genes were expressed in freshly isolated VCAM-1− MPPs. It has been shown previously that CLPs, but not earlier progenitors, express Pax5 (22). In agreement with this report, we did not detect Pax5 expression in freshly isolated VCAM-1− MPPs (Fig. 8). These results also demonstrate that VCAM-1+ MPPs can give rise to a downstream intermediate with a similar gene expression pattern as that in VCAM-1− MPPs.
We also purified CMPs upon in vivo injection of VCAM-1+ MPPs, which were CD34+FcγR−/low and VCAM-1+ (Fig. 6,A, bottom panel). These CMPs derived from VCAM-1+ MPPs gave rise to GM colonies at 20% plating efficiency in the presence of SCF, IL-3, IL-6, and GM-CSF. In the absence of GM-CSF, but with the addition of Epo and Tpo, these phenotypic CMPs gave rise to 27% erythroid colonies, 7% GM, and 5% mix colonies. Plating efficiencies of each type of colonies were comparable to those of freshly isolated CMPs from the bone marrow, with slightly higher percentages of erythroid colonies (Fig. 7,B). We also examined the lymphoid differentiation potential of CMPs derived from VCAM-1+ MPPs. B and T cell differentiation from these CMPs were compared with those of CLPs on OP9 and OP9-DL1 cocultures, respectively. Although CLPs efficiently gave rise to B220+ B cells and Thy-1+ T cells (Fig. 7,C), CMPs derived from VCAM-1+ MPPs gave rise exclusively to Mac-1+ myeloid cells on OP9 cell layers and did not proliferate on OP9-DL1 cells in our culture system (Fig. 7 C). These results demonstrate that VCAM-1+ MPPs can give rise to phenotypic and functional CMPs in vivo.
VCAM-1− MPP is an intermediate between VCAM-1+ MPP and CLP
Because the development of VCAM-1− MPPs is skewed toward the lymphoid lineage, we next determined whether VCAM-1− MPPs can give rise to CLPs in vivo. Five days after injection of 5 × 104 purified VCAM-1− MPPs into an irradiated host, we detected donor-derived (GFP+) cells in the bone marrow at 3.7% chimerism (Fig. 6,B). The lower chimerism from VCAM-1− MPPs compared with VCAM-1+ MPPs suggests that perhaps VCAM-1+ MPPs have higher proliferative capacity. We were able to detect 20% phenotypic CLPs (Lin−c-KitlowSca-1lowIL-7Rα+Thy-1.1−) from donor-derived cells (Fig. 6,B). Furthermore, within the Lin− population, we did not detect donor-derived c-Kithigh cells as in VCAM-1+ MPPs, also suggesting that VCAM-1+ MPPs give rise to CMPs and VCAM-1− MPPs, but not vice versa, in vivo. Collectively, these results demonstrated a linear relationship among VCAM-1+ MPPs, VCAM-1− MPPs, and CLPs during the earliest stages of lymphocyte development (Fig. 9).
HSCs lose life-long self-renewal activity and multipotent differentiation potential during the course of maturation. To understand the mechanisms of lymphoid and myeloid lineage commitment from HSCs, it is important to characterize various maturational stages of cells in early hemopoiesis. We show in this paper that VCAM-1 is a useful marker to subdivide the MPP population. Based on the results presented in this paper and other reports, we propose a revised version of the hemopoietic tree (Fig. 9). VCAM-1+ MPPs gave rise to CMPs and VCAM-1− MPPs (Fig. 6,A). In addition, VCAM-1− MPPs differentiated into CLPs, but not into VCAM-1+ MPPs or CMPs, in vivo (Fig. 6,B), suggesting that hemopoiesis may branch into myeloid and lymphoid lineages at the VCAM-1+ MPP stage. The phenotype of VCAM-1− MPPs, which homogeneously expressed high levels of Flt3, seems to overlap significantly with the Flt3high KLS population previously characterized by others (11, 23, 24). Similar to Flt3high KLS cells, VCAM-1− MPPs exhibited accelerated B and T cell reconstitution and reduced GM differentiation and had virtually no erythroid differentiation potential. The VCAM-1− MPP population might be an intermediate toward the lymphoid lineage while maintaining myeloid differentiation potential at a low level. On a population level, VCAM-1− MPPs are multipotent for T, B, and GM differentiation. We also demonstrated that the majority of these cells are B and GM bipotent on a clonal level (Fig. 5 B), suggesting that these cells are GM and lymphoid bipotent progenitors. Because of the lack of direct evidence of T, B, and GM differentiation on a clonal level due to the limitation of the experimental approach, we cannot rule out the possible presence of T/B- and T/GM- or T lineage-restricted progenitors in the VCAM-1− MPP population.
Branching point of lymphoid and myeloid lineage from VCAM-1+ MPP
In this report we used Flt3 as a marker for multipotent hemopoietic progenitors. Flt3+ KLS cells have been shown to be a lymphoid skewed population, however, this population can give rise to myeloid lineage cells in vitro and in vivo at frequencies similar to Flt3− KLS cells (3, 23). Although a recent report suggests a lack of megakaryocyte/erythroid (MegE) potential in KLS cells that express high levels of Flt3 (24), VCAM-1+ MPPs, a subset of the Flt3+ KLS population, can give rise to phenotypic and functional CMPs in vivo. Although the more upstream HSCs can contribute to the development of CMPs, all HSC activities were shown to be in the Thy-1.1+ KLS fraction (25). After double sorting of our MPP populations, we did not detect any Thy-1.1+ cells. We therefore ruled out the possibility of HSC contamination in our sorting. In addition, the VCAM-1+ MPP population expressed GATA-1 (data not shown), indicating their potential to develop into the MegE lineage.
As we show in Fig. 2, the VCAM-1+ MPP population is heterogeneous, containing both Flt3low- and Flt3high-expressing cells, whereas the VCAM-1− MPP subset contains exclusively Flt3high cells. In agreement with the recent report by Adolfsson et al. (24), we did not observe erythroid differentiation from VCAM-1− MPPs. It is possible that only Flt3low VCAM-1+ MPPs retain erythroid differentiation potential and contribute to the development of CMPs. Nutt et al. (26) recently showed that the CMP population is heterogeneous, and only Flt3− CMPs have MegE differentiation potential. At this moment it is not clear whether VCAM-1+ MPPs preferentially give rise to only Flt3− CMPs, where Flt3 expression is down-regulated during the course of development from VCAM-1+ MPPs to CMPs. It is also possible that these MPPs give rise to both Flt3− and Flt3+ CMPs. This issue needs to be determined in the future. Nonetheless, we show in this study that VCAM-1+, but not VCAM-1−, MPPs can give rise to CMPs and have the potential to develop into all classes of myeloid lineage cells. The transition of VCAM-1+ to VCAM-1− MPPs or CMPs thus represents the branching point of lymphoid and myeloid lineages.
Lymphoid progenitors before lineage commitment
It has been shown previously that multiple steps of culture are required for HSCs to develop into lymphocytes (27, 28). Because VCAM-1+ MPPs immediately up-regulated lymphoid-related genes, as shown in Fig. 8, it is possible that lymphocytes are more easily induced from MPPs in in vitro cultures than from HSCs. Among the lymphoid-affiliated genes we examined, RAG1 was up-regulated first in VCAM-1+ MPPs in vitro, suggesting that RAG1 might be one of the earliest lymphoid genes expressed in the course of lymphocyte development. Kincade et al. (22) have shown that RAG1+ cells in the KLS population (ELPs) derived from RAG1/GFP knockin mice have efficient lymphocyte differentiation potential with low myeloid differentiation activity. Because VCAM-1− MPPs had a differentiation pattern similar to that of ELPs, it is intriguing to examine the relationship between VCAM-1− MPPs and ELPs. Preliminary analysis in RAG1/GFP knockin mice suggests that VCAM-1− MPPs can be further subdivided into RAG1+ and RAG1− fractions (A. Y. Lai and M. Kondo, unpublished observation). We are in the process of characterizing the relationship between MPP subsets distinguished by both VCAM-1 and RAG1 expression patterns.
Stepwise loss of myeloid differentiation potential during early lymphocyte development
Lineage commitment during hemopoiesis has been thought to involve multiple steps of progressive restriction of differentiation potential (29). Compared with VCAM-1+ MPPs, VCAM-1− MPPs and CLPs have less and no myeloid differentiation abilities, respectively, suggesting that early lymphocyte development supports this model of differentiation. Although MPPs have limited or no self-renewal potential, we could detect VCAM-1+ MPPs in vivo even 1 wk after injection. VCAM-1 expression was evident, but its expression level was lower than that on freshly isolated VCAM-1+ MPPs in bone marrow. GM colony formation of VCAM-1+ MPPs 1 wk after injection was lower than that of VCAM-1+ MPPs from normal mice, suggesting that VCAM-1+ MPPs observed after injection might be in the course of maturation to VCAM-1− MPPs.
Importantly, VCAM-1− MPPs have virtually no erythroid differentiation potential while still possessing significant GM differentiation potential. This observation suggests that during early lymphocyte development from multipotent hemopoietic progenitors, developing cells first shut down erythroid differentiation potential and then turn off GM differentiation ability before lineage commitment at the CLP stage. In support of this idea, although DN1 cells have low macrophage differentiation potential, they do not have erythroid colony formation activity (30).
At the population level, MPPs were previously shown to promiscuously express lymphoid and myeloid lineage-related genes (31). Perhaps during early lymphocyte development, MPPs sequentially turn off erythroid and GM-affiliated genes and up-regulate lymphoid-related genes. Gene expression profiling of lymphoid- and myeloid-related genes in the MPP subsets supports this mechanistic model of lymphoid differentiation (A. Y. Lai and M. Kondo, manuscript in preparation), although it is still unclear when and how the lymphocyte developmental program is initiated downstream of HSCs. Nonetheless, more refined characterization of MPP subsets in the future gives us insights into the earliest stages of lymphoid lineage differentiation during hemopoiesis.
We greatly appreciate the help of Lynn Martinek and Mike Cook (FACS facility in the Duke University Comprehensive Cancer Center) with the maintenance of cell sorters and advice in FACS sorting.
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 in part by National Institutes of Health Grants T32AI52077 (to A.Y.L.) and R01AI056123 and R01CA098129 (to M.K.). M.K. is a scholar of the Sidney Kimmel Foundation for Cancer Research.
Abbreviations used in this paper: HSC, hemopoietic stem cell; CLP, common lymphoid progenitor; CMP, common myeloid progenitor; DKO, double knockout; EGFP, enhanced GFP; ELP, early lymphoid progenitor; Epo, erythropoietin; Flt3L, Flt3 ligand; γc, common γ-chain; GM, granulocyte/macrophage; KLS, c-Kit+lineage−/lowSca-1+ bone marrow cell; LT, long term; MegE, megakaryocyte/erythroid; MPP, multipotent progenitor; SCF, stem cell factor; ST, short term; Tpo, thrombopoietin.