Lin^-Sca1⁺Kit^- bone marrow cells contain early lymphoid-committed precursors that are distinct from common lymphoid progenitors

Abstract

The significance of a population in mouse bone marrow of lineage negative, Sca1 positive, c-kit negative (LSK^-) cells, which is reported to be devoid of long-term repopulation capacity or myeloid potential, is unknown. Here, we show that the LSK^- population is composed of several subsets defined by the expression of flt3, CD25 and IL7Rα. The first subset was CD25^- and more than 90 % expressed either flt3, IL7Rα or both. The CD25^-LSK^- population had T, B and NK cell potential in vivo, and most of this activity was localized in the flt3⁺ subset, irrespective of the expression of IL7Rα. While lymphoid potential of flt3⁺LSK^- cells in vivo was threefold lower than that of lin^-Sca1^lokit^loIL7Rα⁺ common lymphoid progenitors (CLPs), their cloning efficiency in vitro was tenfold lower than that of CLPs. Furthermore, while the myeloid potential of flt3⁺LSK^- cells was tenfold lower than that of CLPs in the absence of M-CSF, the relative myeloid potential of both population was similar in its presence. These observations suggest differential growth factor requirements of both populations. The second subset of LSK^- cells was homogeneously CD25⁺⁺flt3^-IL7Rα⁺ and could be generated from both CD25^-LSK^- cells and from CLPs, but did not engraft in immunodeficient Rag1^-/- or Rag1^-/-γc^-/- hosts. This population, of which the significance is unclear, was increased in Rag1^-/- mice and in old mice. Thus, the LSK^- population is phenotypically and functionally heterogeneous and contains early lymphoid-committed precursors. Our findings imply that the early stages of lymphoid commitment are more complex than was thus far assumed.

INTRODUCTION

Hematopoietic stem cells (HSCs) can self renew and generate all lineages of the hematopoietic system¹. In the mouse, they are enriched in a population that does not express markers of mature hematopoietic cells (i.e. they are lineage^-), and expresses Sca1 and c-kit, but is devoid of flt3 expression (flt3^-lin^-Sca1⁺kit⁺ or flt3^-LSK cells)^2-5. Upon differentiation into multipotential progenitors (MPPs), which do not possess extensive self renewal capacity, HSCs acquire the expression of flt3^4,5. During lymphoid commitment of HSC myeloid, erythroid and megakaryocytic potential progressively decreases, although whether or not erythroid/megakaryocytic and myeloid potential are sequentially lost during the very early stages of differentiation is still a matter of debate^6-8. Lymphoid commitment of MPPs is accompanied by the sequential upregulation of the RAG genes in early lymphoid progenitors (ELPs)⁹ and of interleukin 7 receptor-α (IL7Rα) in common lymphoid progenitors (CLPs)¹⁰. While CLPs have T-cell potential in vitro and in vivo, they may not be obligate precursors for T cells and function predominantly as early B cell progenitors in vivo^11,12. An earlier progenitor, probably a HSC/MPP that expresses CCR9, is likely the main T-cell precursor that seeds the thymus^13,14. This contention is further supported by the recent finding that, while CLPs are mostly devoid of myeloid potential, early thymic precursors (ETPs) retain myeloid differentiation capacity^15,16. In addition, a lin^-IL7Rα⁺flt3^-kit^lo population with predominant T cell potential has been identified in the blood¹⁷. Expression of c-kit is maintained throughout the early stages of T and B cell development¹⁸.

In addition to LSK cells, a cell population has been described in the bone marrow with a similar lin^-Sca1⁺ phenotype but lacking the expression of c-kit (lin^-Sca1⁺kit^- or LSK^- cells)¹⁹. LSK^- cells could be generated from transplanted LSK cells in vivo and express the pan-hematopoietic marker CD45, suggesting that they are hematopoietic cells. Their function was unknown, however, as they did not possess long-term repopulation capacity and could not be grown in vitro. LSK^- cells were furthermore present in Rag1^-/- mice, indicating that they are not a mature lymphocyte subpopulation that requires gene rearrangement of antigen receptors. Quite intriguingly, LSK^- cells are rare in fetal liver and accumulate with age in the bone marrow¹⁹.

Here, we show that LSK^- cells contain early lymphoid committed precursors with both T and B cell potential that are functionally and phenotypically distinct from CLPs. In addition, a subpopulation of LSK^- cells expresses high levels of CD25, expands with age and has no lymphoid precursor activity. A similar population can be generated from CD25^-LSK^- cells and from CLPs, however, suggesting that, although its function is unknown, the CD25⁺⁺LSK^- population belongs to the lymphoid lineage.

METHODS

Mice

4 to 8-week-old C57BL/6 (CD45.2) mice and B6.Ly5.2 (B6.Ly5^SJL)(CD45.1) mice were purchased from the National Cancer Institute animal facility. Rag1^-/- (B6.129^{S7-Rag1tm1Mom}/J) (CD45.2) mice were purchased from Jackson Laboratory (Bar Harbor, ME). 18-month-old C57BL/6 were aged at the Mount Sinai animals facility. Animals were kept in a specific pathogen free facility. Experiments and animal care were performed in accordance with the Mount Sinai Institutional Animal Care and Use Committee (IACUC).

Antibodies

Unlabeled anti-CD2, CD3ε, CD4, CD8α, Mac-1, Gr1, B220, IgM, FITC-conjugated anti-CD45.1, SPRD-conjugated anti-B220 and biotinylated anti-Thy1 were purchased from Southern Biotech (Birmingham, AL). FITC-conjugated anti-TER119, CD2, CD3ε, CD4, CD8α, B220, Mac1, Gr1 and TCR-β, APC-conjugated CD25, PE-conjugated anti-Flt3 and IL7Rα, PECy7-conjugated anti-IL7Rα, and PE-Cy7-streptavidin were from eBioscience (San Diego, CA). Unlabeled anti TER119, APC-conjugated anti-IgM, c-kit and CD44 , PECy7-conjugated anti-CD19, PE-conjugated anti-CD25, FITC-conjugated anti-CD19, APCCy7-conjugated anti-CD8, biotinylated and PE-conjugated, anit-CD4, anti-Sca1 and PerCP-Cy5.5-streptavidin were from Pharmingen (San Diego, CA). Flt3 ligand and IL7 were purchased from R&D Systems (Minneapolis, MN).

Preparation of hematopoietic cells

Bone marrow cells were prepared by flushing the femura and tibia of mice with cold DMEM (Cellgro, Mediatech, VA) containing 2% FBS and 100 ng/ml penicillin/streptomycin. Thymus and spleen cell suspensions were prepared by mincing the organs through nylon mesh. Mononuclear cells were obtained after gradient centrifugation using lymphocyte separation medium (Cellgro, Mediatech, VA).

Flow cytometry

Low-density bone marrow cells, thymus or spleen cells were stained with appropriate antibodies. The lineage cocktail used for the isolation of CLPs and LSK^- cells contained: anti-CD2, -CD3ε, -CD4, -CD8α, -Ter119, -B220, -CD19, -Mac1 and -Gr1. Cells were sorted on three-laser a FACSVantage SE with DiVa software (Becton Dickinson), a MoFLo (Cytomation) or a Influx (Cytopeia) flow cytometer. Doublets were excluded by plotting forward scatter pulse area versus width. Purities > 97 % were routinely achieved, and were required to use the cells in adoptive transfer experiments. Analytical flow cytometry was performed on a three-laser LSR II with diva software (Becton Dickinson). Data were analyzed using FlowJo software.

Reconstitution of Rag1^-/- mice

Sorted CLPs or LSK^- cells and subpopulations thereof from bone marrow of CD45.2⁺ mice were injected in the tail vein of sublethally irradiated (500cG) Rag1^-/- mice. Cell doses ranged from 500 to 1,500 cells per mouse. Donor derived cells were distinguished by expression of CD45.1.

Quantitative PCR

Absolutely RNA Nanoprep kit (Stratagene) according to the manufacturer’s instructions. RNA was treated with Dnase I and reverse transcribed into cDNA using random hexamers with SuperScript first-strand synthesis system for RT-PCR (Invitrogen).Realtime quantitative PCR was performed on ABI 7900HT thermocycler (Applied Biosystems), with a 10-minute step at 95° C followed by 40 cycles of 95° C for 15 seconds and 60° C for 1 minute, 95° C for 15 seconds, 60° C for 15 seconds and 95° C for 15 seconds.All experiments were done in triplicate with SYBR GreenER qPCR SuperMix (Invitrogen).Primers sequences used were the following:

Rag1: 5′-ACCCTGAGCTTCAGTTCTGC-3′ (sense); 5′-GCCTTTTCAAAGGATCTCACC-3′ (antisense); Rag2: 5′- TGAACCCAGATACGGCCATTCCAT-3′ (sense); 5′-TGGTTCTCTGGGTAGAAGGCATGT-3′(antisense); Notch1 5′-TAACAGTGCCGAATGTGAGTGGGATG-3′ (sense); 5′-CCGCAGAAAGTGGAAGGAGTTGT-3′ (antisense); GAPDH: 5′-TGAGCCCTTCCACCATGCCAAA-3′ (sense); 5′-GTGATGGGTTGAACCACGAGAAA-3′ (antisense). Relative quantification was obtained in relation to a standard curve. The standard curve was created using total RNA from sorted DN thymic progenitors, through a 10-fold dilution series of cDNA standards ranging from 100 ng/ml to 0.1 ng/μl. quantified values for each gene of interest were normalized against the input determined by the housekeeping gene GAPDH. Combined data from three independent triplicate experiments were normalized to the data obtained for CLPs.

D_H-J_H gene rearrangements

Genomic DNA from 10,000 sorted LSK^-CD25^- cells and CLPs cells was extracted using QIAmp DNA micro kit (QIAGEN), following the manufacturer’s instructions. D_H-J_H rearrangements were analyzed by nested PCR following the protocol previously described by Borghesi et al.²⁰.

OP9 cultures

OP9-Mig R1 (OP9) cells and OP9-DL1 were provided by J.C. Zuniga-Pflucker (University of Toronto, Ontario, Canada). 1,000 sorted cells were seeded in 6-well culture plates containing a monolayer of OP9 cells. Culture medium was AMEM (Cellgro, Mediatech, VA) containing 20% FBS (Hyclone, Utah), 100ng/ml penicillin/streptomycin and 5ng/ml recombinant mouse IL-7 and Flt3L and, in some experiments, M-CSF (50 ng/ml). Cultures were harvested after 7 to 14 days for analysis by flow cytometry. Hematopoietic cells were identified as cells with a low FSC/SSC that did not express GFP.

Statistical analysis

Student’s t-test for unpaired samples was used, unless otherwise indicated. P < 0.05 was considered indicative of statistical significance.

RESULTS

Lymphoid potential of LSK^- cells

A population of lin^-c-kit^- cells isolated from the bone marrow by elutriation has been suggested to contain precursors of long-tem repopulating HSC by Ortiz et al.²¹. The LSK^- cells described by Randall et al. are likely different, however, as the cells described by Ortiz et al. expressed variable levels of Sca1, whereas LSK^- cells express high levels of Sca1 (Fig. 1a). Furthermore, in our hands, LSK^- cells had no long-term competitive repopulation capacity, even when recipient mice were analyzed 12 months after transplantation (not shown), confirming previously published data¹⁹.

Figure 1. Lymphoid potential of LSK^- cells.

(a) Sort windows used for the isolation of LSK^- cells from bone marrow. (b) Generation of B (upper panels) and T (lower panels) lineage cells 10 days after plating of 1000 LSK^- cells in the presence of OP9 cells, IL7 ad flt3 (upper panels) or of OP9-DL1 cells (lower panels). (c) Repopulation of sublethally irradiated Rag1^-/- mice 4 weeks after transfer of 1,000 LSK^- cells. Representative of 8 experiments.

LSK^- cells do not grow in colony assays, suggesting that they have limited or no myeloid or erythroid potential. Their lymphoid potential was never tested however. To begin to examine the lymphoid potential of LSK^- cells, we plated these cells in the presence of OP9 stromal cells, IL7 and flt3L, conditions that support B cell development in vitro²². After 7 to 14 days of culture B220^loCD19^- (compatible with prepro- B cells) and B220⁺CD19⁺ (compatible with pro- and pre-B cells)²³ populations were detected (Fig. 1b, upper panels). Next, we tested T cell potential in vitro. After plating in the presence of OP9-DL1 cells, which support the development of T lineage cells²⁴, LSK^- cells generated Thy-1⁺ cells (Fig. 1b, lower panels), some of which expressed CD8 (2 to 4 % after 2 weeks, not shown). The earliest T cell precursors do not express CD3, CD4 and CD8 and are called double negative (DN). The DN population can be further divided into 4 populations based on the expression of CD25 and CD44. The earliest T cell precursors are CD25^-CD44⁺, and are termed DN1 cells. These cells proceed to express CD25 (CD25⁺CD44⁺ or DN2 cells), then lose the expression of CD44 (CD25⁺CD44^- cells or DN3 cells) and finally become CD25^-CD44^- cells (DN4), which begin to express both CD4 and CD8 (double positive or DP cells)²⁵. LSK^- cells plated on OP9-DL1 cells and analyzed after 10 days for the expression of CD25 and CD44 faithfully traversed these consecutive early stages of T cell development (Fig. 1b, lower panels). Thus, the LSK^- population contains cells with T and B cell potential in vitro.

To evaluate the lymphoid potential of LSK^- cells in vivo, we injected 1,000 CD45.1⁺ LSK^- cells from CD45.1⁺ congenic C57BL/6 mice into sublethally (500 cG) irradiated Rag1^-/- mice (CD45.2⁺ and backcrossed for minimum 15 generations onto C57BL/6). Donor-derived cells were detected in the spleen and thymus after 4 weeks. No myeloid cells were observed and the majority of cells in the spleen were either T or B cells, while a small fraction of NK cells was also detected (Fig. 1c). Thus, LSK^- cells have T, B and NK cell potential in vivo.

Subpopulations of LSK^-cells defined by IL7Rα, flt3 and CD25

As LSK^- cells contain early lymphoid progenitor activity but appear devoid of myeloid potential, they are functionally similar to CLPs. CLPs are defined by the lin^- Sca1^lokit^loIL7Rα⁺ phenotype¹⁰ and express flt3²⁶. LSK^- cells are phenotypically distinct from CLPs. Although both populations occur with similar frequency in the bone marrow (Fig. 2a) and CLPs are, like LSK^- cells, by definition negative for lineage markers, but express c-kit and have a lower expression of Sca1 compared than LSK^- cells (Fig 2a). Furthermore, LSK^- cells contain cells that are smaller than CLPs, suggesting that they may be a heterogeneous population (Fig. 2a).

Figure 2. Subpopulations of LSK^- cells.

(a) Comparison of the phenotype of LSK^- cells and CLPs. Lower middle panel: comparison of forward and side scatter of LSK^- cells and CLPs. (b) Expression of IL7Rα and flt3 on LSK^- cells (black line) and on total BM cells (shaded) (see Figs 1a and 2a for representative analysis gates). Representative of 8 experiments. (c) Expression of CD25 on LSK^- cells (upper panel), and expression of IL7Rα and flt3 on CD25^- and CD25⁺⁺LSK^- cells (lower panels). The shaded histogram in the upper panel is the expression of CD25 on the total BM population. Representative of 7 experiments. (d) Sca1 (upper panel) and FSC (lower panel) profile of CD25⁺⁺ and CD25^- LSK^- cells. Representative of 7 experiments.

Given the lymphoid potential of LSK^- cells and the fact that CLPs express flt3 and IL7Rα^10,26, we analyzed the expression of these cytokine receptors in LSK^- cells. 60 to 80 % (n = 8) of LSK^- cells expressed IL7Rα (Fig. 2b, upper panel). 40 to 50 % of LSK^- cells also expressed flt3 (Fig. 2b, lower panel). In addition, we observed that 50 to 90 % (n=7) of LSK^- cells expressed high amounts of CD25, the α-chain of the IL2 receptor (Fig. 2c, upper panel). As LSK^- cells expressed more CD25 than any other cell type in the bone marrow (shaded area in Fig. 2c, upper panel), we call these cells CD25⁺⁺LSK^- cells. CD25⁺⁺LSK^- expressed much more Sca1 than CD25^-LSK^- cells, and were smaller as judged by their forward scatter profile (Fig. 2d). They are not regulatory T cells however as they expressed neither CD4 nor Foxp3 (not shown). Combined analysis of the expression of CD25, flt3 and IL7Rα on LSK^- cells showed that > 85 % of CD25^-LSK^- cells were either flt3⁺IL7Rα⁺ or flt3⁺IL7Rα^-, while CD25⁺⁺LSK^- cells were homogeneously flt3^-IL7Rα⁺ (Fig. 2c, lower panels).

To investigate how expression of IL7Rα, flt3 and CD25 affects the lymphoid potential of LSK^- cells in vivo, we injected 1,000 IL7Rα⁺, IL7Rα^-, flt3⁺, flt3^-, CD25⁺ and CD25^- LSK^- cells from CD45.1⁺ congenic C57BL/6 mice into sublethally (500 cG) irradiated Rag1^-/- mice. Both IL7Rα⁺ and IL7Rα^- populations generated donor-derived cells after adoptive transfer into Rag1^-/- hosts (Fig. 3a). The percentage of donor-derived CD45.1⁺ cells tended to be higher after adoptive transfer of IL7Rα^- than of IL7Rα⁺ LSK^- cells, although the difference did not reach statistical significance (Fig. 3a). Flt3⁺LSK^- cells generated approximately 30-fold more donor cells than flt3^- cells (Fig. 3a). CD25^-LSK^- cells consistently yielded donor-derived cells. In all populations tested, more than 90 % of the cells generated were B and T cells, while myeloid cells (Mac1⁺Gr1⁺) were never observed (not shown). In contrast, CD25⁺⁺LSK^- cells did not engraft after three weeks (Fig. 3a). However, when we injected 10⁴ CD25⁺⁺LSK^- cells, a donor-derived CD25⁺⁺CD19^- population was detectable in the BM up to one week after transfer (Fig. 3b). These cells were not regulatory T cells as they did not express CD4 (Fig. 3b). These observations indicate that CD25⁺⁺LSK^- cells are capable of homing to the BM. Similar data were obtained after transfer into Rag1^-/-γc^-/- mice (not shown). As CD25⁺⁺LSK^- did not have lymphoid potential, but do express lymphoid-associated markers such as CD25 and IL7Rα, we examined if they could be generated from CLPs or from CD25^-LSK^- cells. Indeed, one week after plating of either CLPs or CD25^-LSK^- cells in the presence of OP9 cells, flt3L and IL7, a population of B220^-CD25⁺⁺ cells, ranging from 0.5 % to 8 % of the total population, was invariably generated (Fig. 3c). These cells were CD19^-Sca1^hi (not shown). These findings indicate that cells with a phenotype very similar to that of CD25⁺⁺LSK⁺ cells can be derived from both CLPs and CD25^-LSK^- cells and that CD25⁺⁺LSK^- very likely belong to the lymphoid lineage. Thus, most lymphoid potential of LSK^- cells is located in the CD25^-flt3⁺ fraction of LSK^- cells, while the CD25⁺⁺ fraction of LSK^- cells has no lymphoid progenitor potential, but very likely belongs to the lymphoid lineage.

Figure 3. Potential of subpopulations of LSK^- cells.

(a) Percent donor-derived cells in the spleen 3 weeks after transfer of 1,000 IL7Rα⁺, IL7Rα^-, flt3⁺, flt3^-, CD25⁺⁺ or CD25^- LSK^- cells into sublethally irradiated Rag1^-/- mice. (b) Detection of donor-derived cells 3 days after transfer of 10⁴ CD25⁺⁺LSK^- cells into sublethally irradiated Rag1^-/- mice. (c) Generation of CD25⁺⁺B220^-CD19^- cells 7 days after plating of CD25^-LSK^- cells or CLPs in the presence of OP9 cells, IL7 and flt3L.

Flt3⁺LSK^- cells and CLP differ functionally

As the flt3⁺ fraction of LSK^- cells, which is uniformly CD25^-, contained most of the lymphoid activity of this population, we directly compared CD25^-flt3⁺LSK^- cells to CLPs. Approximately 50 % of CLP have D_H-J_H rearrangement of the immunoglobulin heavy chain gene^21,27. The D_H-J_H rearrangement status was similar in flt3⁺LSK^- cells and CLPs (Fig. 4a), suggesting that both cell populations are at a similar developmental stage. Nevertheless, expression of Rag1 and Rag2 was lower in flt3⁺LSK^- cells than in CLPs (Fig. 4b). We isolated CLPs (see Fig. 2a for sort windows) and the flt3⁺IL7Rα⁺ and flt3⁺IL7Rα^- fractions of CD25^-LSK^- cells (see Fig. 2c for sort windows), and injected 1,000 cells of those populations into sublethally irradiated Rag1^-/- mice. In vivo, the lymphoid potential of both flt3⁺IL7Rα⁺LSK^- and flt3⁺IL7Rα^-LSK^- populations was approximately one third of that of CLPs (Fig. 4c). The distribution of donor-derived B, T, and NK cells was similar after transfer of CLPs, flt3⁺IL7Rα⁺LSK^- and flt3⁺IL7Rα^-LSK^- cells (Fig. 4d). As the highly lymphopenic environment of Rag1^-/- mice may overestimate the lymphoid potential of progenitor cell populations, we co-injected 1,500 CD45.1⁺ CLPs and CD45.1⁺CD45.2⁺ flt3⁺LSK^- cells into Rag1^-/- mice (CD45.2⁺). We did not distinguish between IL7Rα⁺ and IL7Rα^- fraction in these experiments, as both subpopulations behaved similarly after transfer into Rag1^-/- mice. In these experiments, the progeny of CLPs was on average 4.1 ± 0.81 —fold larger than flt3⁺LSK^- cells (P = 0.016, n = 4, see Fig. 4e for a representative example), confirming that flt3⁺LSK^- cells have lymphoid potential, but that their proliferative capacity in vivo is significantly lower than that of CLPs.

Figure 4. Functional distinction between CLPs and flt3⁺LSK^- cells.

(a) D_H-J_H gene rearrangements in purified CLPs and flt3⁺LSK^- cells analyzed by nested PCR²⁰. (b) Quantitative PCR of the expression of Rag1 and Rag2 in CLPs and in flt3⁺LSK^- cells (n = 3 triplicate paired experiments, data for CLPs normalized to 1 in each experiment). (c) Percent donor-derived cells in the spleen after transfer of 1,500 CLPs, IL7Rα⁺ or IL7Rα^- flt3⁺LSK^- cells into sublethally irradiated Rag1^-/- mice (n = 4, *P<0.01, Wilcoxon signed rank test). (d) Contribution of T, B and NK cells to donor-derived cells in the spleen of after transfer of 1,500 CLPs, IL7Rα⁺ or IL7Rα^- flt3⁺CD25^- LSK^- cells into sublethally irradiated Rag1^-/- mice. (e) Representative example of repopulation of the spleen three weeks after co-transplantation of 1,500 CD45.1⁺CD45.2⁺ flt3⁺LSK^- cells and CD45.1⁺ CLPs. Gated on cells with low to intermediate FSC and low SSC (see Fig. 1a). (f) Best fitting curve in limiting dilution experiments of CLPs and CD25^-flt3⁺LSK^- cells in the presence of OP9 cells, IL7 and flt3L. (g) Cloning efficiency of CLPs, IL7Rα⁺ or IL7Rα^- flt3⁺LSK^- cells in single cell sorting and culture experiments in the presence OP9 cells, IL7 and flt3L, and of OP9-DL1 cells (* P < 0.0001, n = 3 independent experiments). (h) Ratio of B cells (CD19⁺) to myeloid cells (Mac1⁺Gr1⁺) cells in cultures of CLPs or of flt3⁺LSK^- cells in cultures supported by OP9 cells in the presence or absence of M-CSF.

Next, we compared the proliferative capacity of flt3⁺LSK^- cells and CLPs in vitro by determining the cloning efficiency of these populations in limiting dilution cultures supported by OP9. Surprisingly and in contrast to previously reported data¹⁰, both for CLPs and flt3⁺LSK^- cells the best fitting curve when the logarithm of the fraction of negative wells was plotted against plated cell number was bicubic (R² = 0.985 for CLPs and R² = 0.903 for flt3⁺LSK^- cells) and not linear (R² = 0.761 for CLPs and R² = 0.543 for flt3⁺LSK^- cells) (Fig. 4f). Similar observations were made in cultures supported by OP9-DL1 cells (not shown). These data suggest complex density-dependent interactions among the plated cells. Although the curves indicate a much lower proliferative capacity of flt3⁺LSK^- cells than of CLPs (Fig. 4f), this precludes estimation of the cloning efficiency of these populations using this method, which is based on the premise of single hit kinetics^28,29. Therefore, we performed single cell sorting experiments. The cloning efficiency of flt3⁺IL7Rα⁺LSK^- was at least 10-fold lower than that of CLPs, both in cultures supported by OP9 and by OP9-DL1 cells, while in a total of >400 wells, no colonies derived from flt3⁺IL7Rα^-LSK^- cells were ever observed (Fig. 4g). There were no obvious differences in the size of the colonies generated from CLPs or from flt3⁺IL7Rα⁺CD25^-LSK^- cells (not shown). While the lower expression of c-kit, and in the flt3⁺IL7Rα⁺LSK^- fraction of IL7Rα, might provide an obvious explanation for the lower proliferative capacity of flt3⁺LSK^- cells, we also measured expression of Notch1, as this receptor is essential for T cell development²⁴. By quantitative PCR, Notch expression in flt3⁺LSK^- cells was 42 ± 0.15 % of that in CLPs (n = 3 independent triplicate independent experiments, P = 0.059, P < 0.05 in each individual experiment). Thus, proliferative capacity of flt3⁺LSK^- is lower in vivo and particularly in vitro compared top CLPs. Furthermore, flt3⁺LSK^- cells do not only express less c-kit than CLPs, they also express lower levels of Notch1 mRNA.

The CLP population has been shown to retain some myeloid potential²⁷. Therefore, we compared the myeloid potential of CLPs and of flt3⁺LSK^- cells by plating the cells on OP9 cells in the presence of IL7, flt3L with or without M-CSF. Myeloid potential was expressed as the ratio of B cells to myeloid cells (defined as cells expressing Gr1 and Mac1). In the absence of M-CSF, the B/myeloid ratio was strikingly higher in flt3⁺LSK^- cells than in CLPs (Fig. 4h). However in the presence of M-CSF, the relative myeloid potential of both populations was similar (Fig. 4h). Because of the very low cloning efficiency of flt3⁺LSK^- cells, it was difficult to impossible to establish whether any flt3⁺LSK^- cells posses both lymphoid and myeloid potential. We conclude that the myeloid potential of flt3⁺LSK^- is lower than that of CLPs, but that difference is caused by hyporesponsiveness to other myeloid factors than M-CSF. These findings again support the notion that CLPs and flt3⁺LSK^- cells are functionally distinct.

LSK^- cells are increased in lymphopenic and in aged mice

If LSK^- cells contain physiologically relevant early lymphoid precursors, their number may increase in conditions of lymphopenia. In Rag1^-/- mice, the frequency of CLPs and of CD25⁺⁺LSK^- cells but not of CD25^-LSK^- cells was two- to threefold higher than in wt mice (Fig. 5a). The distribution of flt3 and IL7Rα was similar on CD25^-LSK^- cells from Rag1^-/- mice compared to wt mice (not shown). As bone marrow cellularity was similar in Rag1^-/- and wt mice (not shown), the increases in frequency represent increases in absolute number. Next, we examined LSK^- cells and CLPs in aged mice, as it has been shown that the frequency of LSK^- cells increases with age¹⁹, while the frequency and function of CLPs declines^30,31. Confirming these previously published findings, we observed that the frequency of LSK^- cells was approximately 5-fold higher in 18-month than in 2-month-old mice, while the frequency of CLPs was 2- to 3-fold lower in old compared to young mice (Fig. 5a). However, similar to the situation in Rag1^-/- mice, the increase in LSK^- cells was entirely due to an increase in the CD25⁺⁺ fraction. While the expression of IL7Rα was similar in LSK^- cells from old and young mice (not shown), the expression of flt3 on LSK^- cells was lower in old than in young mice (Fig. 5b). A similar decline in flt3⁺ cells was observed among LSK⁺ cells (Fig. 5b), indicative of an age-related decline in MPPs, as has been demonstrated previously^30,32. Consistent with the fact that most lymphoid potential of LSK^- cells resided in the flt3⁺ fraction, LSK^- cells from aged mice generated sevenfold fewer donor-derived cells in the spleen after transfer into Rag1^-/- mice than LSK^- cells from young mice (Fig. 5c).

Figure 5. Age and lymphopenia regulate LSK^- and CLP numbers.

(a) Frequency of CLPs, CD25^-LSK^- and CD25⁺⁺LSK^- cells in the bone marrow of 2-month-old wt mice, 2-month-old Rag1^-/- mice and 18-month-old wt mice (*P< 0.05, n = 5). (b) Expression of flt3 on lin^-Sca1⁺ cells from 2-month and 18-month-old wt mice. Representative of 7 experiments. (c) Percent donor-derived cells 3 weeks after transfer of LSK^- cells from 2-month and 18-month-old wt mice into sublethally irradiated Rag1^-/- mice (n = 3, P < 0.001).

We conclude that CD25⁺⁺LSK^- cells expand in both lymphopenic and aged mice. On the other hand, the fact that in lymphopenic mice CLPs are increased while CD25^-LSK^- cells remain unchanged suggests differential regulation of these populations. Finally, aging is associated with a decline in CLPs, MPPs and flt3⁺LSK^- cells.

DISCUSSION

We have shown here that the LSK^- population, previously dubbed a ‘mystery’ population¹⁹, contains a novel subset of early lymphoid precursors with T, B and NK potential. In addition, we have identified a subset of LSK^- cells with a very high expression of CD25 and Sca1 that has, using currently available assays, no detectable potential in vivo. The significance of this CD25⁺⁺LSK^- population therefore remains a mystery.

Our findings raise three questions: what is the role of LSK^- cells in vivo, what is origin of LSK^- cells and what are the lineage relations between the various subpopulations of LSK^- cells? Most, if not all lymphoid potential of the LSK^- population is located in the flt3⁺LSK^- fraction, which is uniformly CD25^-. As this population is approximately 4-fold less frequent and one third as potent as CLPs, it is likely that their relative importance as lymphoid precursors is minor compared to CLPs. However, lymphoid potential is measured as the capacity to engraft in immunodeficient, sublethally irradiated hosts. These conditions are not necessarily reflective of steady-state, but represent a situation of maximally active feedback mechanisms to compensate for lymphopenia³³. While the lymphoid potential of flt3⁺LSK^- cells is three to fourfold lower than that of CLPs in vivo, it is at least an order of magnitude lower in vitro, and undetectable for the IL7Rα-subfraction, suggesting differential growth factor requirements of both populations. The lower expression of c-kit, Notch1 and IL7Rα may all contribute to the lower proliferative potential of flt3⁺LSK^- cells in vitro. Differential growth factor requirements of flt3⁺LSK-cells and CLPs are also suggested by the fact that the myeloid potential of both populations is vastly different in the absence, but not in the presence of M-CSF. Given the differential response of CLPs and flt3⁺LSK^- cells to in vitro versus in vivo microenvironments, it is not excluded that these populations will behave differently in steady-state as opposed to lymphopenic conditions. The fact that in Rag1^-/- mice CLPs, but not CD25^-LSK^- cells, are increased compared to wt mice argues in favor of this idea. However, only experiments where specific lineage tracing of the progeny of CLPs vs. LSK^- cells can be performed will clarify the relative roles of CLPs and LSK^- cells in lymphoid development in steady-state.

The phenotype and function of flt3⁺LSK^- cells suggests that they are early lymphoid precursors at a differentiation stage similar to that of MPPs (IL7Rα^-) and CLPs (IL7Rα⁺)^4,5,9,10,34. These data are consistent with the observation that the status of D_H-J_H rearrangements is similar in CLPs and flt3⁺LSK^- cells, and that flt3⁺LSK^- cells, CLPs and MPPs, defined as lin^-Sca1⁺kit⁺flt3⁺ cells, all decline with age^30,32. flt3⁺LSK^- cells are distinct from CLPs, however. First, while CLPs express c-kit^4,5,10, flt3⁺LSK^- cells do not. Second, flt3⁺LSK^- cells express less Notch1 than CLPs. Third, the proliferative potential of flt3⁺LSK^- cells is lower than that of CLPs in vivo and much lower than that of CLPs in vitro, suggesting differential growth factor requirements of both populations. Fourth, Rag expression is lower in flt3⁺LSK^- cells compared to CLPs. Fifth, the myeloid potential of flt3⁺LSK^- cells is much lower than that of CLPs when the cells are cultured in the presence of OP9 cells in the absence of M-CSF, but not in its presence, again indicating differential cytokine responsiveness of both populations. Finally, while CLPs are increased in Rag1^-/- mice compared to wt mice, flt3⁺LSK^- cells are not, suggesting differential regulation of both populations. Thus, CLPs and flt3⁺LSK^- cells are functionally distinct lymphoid progenitor populations.

The lineage relation between CLPs and flt3⁺LSK^- cells is unclear. It is possible that multiple early lymphoid differentiation pathways exist and that CLPs, defined as lin^- Sca1^lokit^loIL7Rα⁺ cells¹⁰, do not contain all early committed lymphoid precursors activity in the BM as was previously assumed. Alternatively, it is possible that flt3⁺LSK^- are lymphoid progenitors that for some reason failed to develop further. The lymphoid potential of these cells in vivo argues against this possibility however. Our findings suggest that the early lymphoid progenitor compartment should probably be viewed more broadly than it is currently being defined.

The lineage relations among subpopulations of LSK^- cells are unclear as well. Cooperative signaling through flt3, IL7Rα and c-kit is essential for early lymphoid development, but none of these three receptors individually is absolutely required.^34-40. Within the CD25^-LSK^- fraction, > 90 % of the cells express either flt3 or IL7Rα or both. The existence of lymphoid committed precursor populations that do not simultaneously express IL7Rα, c-kit and flt3 might be explained by the fact that upon differentiation of HSC into lymphoid-committed progenitors, differentiation markers are acquired in a way that is neither abrupt nor synchronous. This notion is supported by the observation by Igarashi et al. that within the LSK⁺flt3⁺Rag^GFP+ ELP compartment, there is initial heterogeneity in the expression of Rag1 and TdT individual cells⁹. Similarly, not all cells defined as lymphoid primed multipotential progenitors (LSK^hiflt3⁺CD34⁺) uniformly express early lymphoid transcripts⁶. The same principle may apply to cytokine receptors. c-kit is not essential for lymphopoiesis, although the requirement for this receptor increases with age⁴⁰. Interestingly, in young viable c-kit-deficient (Vickid) mice, CLP numbers are severely reduced, while prepro- and pro-B cells are not affected⁴⁰. These findings suggested that linear development from HSCs over CLPs to prepro-B cells is not obligatory⁴⁰, and that kit-independent pathways may exist. It is therefore possible that flt3⁺LSK^- cells might be one of the alternative intermediaries between HSCs and committed B and T cell progenitors.

The role and significance of CD25⁺⁺IL7Rα⁺LSK^- cells is unclear. No engraftment was observed beyond the first week after transfer into sublethally irradiated, immunodeficient hosts, and the engrafted cells did not show any sign of differentiation in vivo. It is interesting to note however that this population increases with age and in conditions of lymphopenia. In particular the latter finding may suggest a link between this subset and lymphopoiesis. One explanation for their lack of engraftment potential may be that CD25⁺⁺LSK^- cells are not capable of homing from the bloodstream to the bone marrow. However, the fact that these cells could be detected in the BM up to one week after transfer argues against this idea. An alternative explanation may be that CD25⁺⁺LSK^- cells are in fact not lymphoid precursors. However, the finding that the majority of CD25⁺⁺LSK^- cells express IL7Rα , a receptor required for B-cell development in adult mice^35-37, in addition to CD25, a lymphoid lineage associated receptor, argues in favor of the lymphoid nature of this cell population. Furthermore, the observation that both CLPs and CD25^-LSK^- cells can generate a cell type that is phenotypically very similar to CD25⁺⁺LSK^- cells strongly suggest that these cells belong to the lymphoid lineage. It is possible that CD25⁺⁺LSK^- are in fact a novel type of mature lymphoid cell of which we failed to assay the function.

In conclusion, our observations for the first time assign a function to at least a subfraction of LSK^- cells, and suggest that the early stages of lymphoid commitment are more complex than was previously assumed.