Identification of a bone marrow precursor of the earliest thymocytes in adult mouse

Abstract

The thymus requires continuous replenishment of progenitors from the bone marrow (BM) to sustain T cell development. However, it remains unclear which hematopoietic progenitors downstream from hematopoietic stem cells in the BM home to the thymus in adult mice. In this work, we demonstrate that although multiple BM populations have intrinsic T lineage differentiation potential, a small subset of multipotent progenitors (MPPs) expressing CCR9 preferentially homes to the thymus. These CCR9⁺ MPPs are phenotypically similar to the most immature early T lineage progenitors (ETPs) in the thymus and are present in the peripheral blood. Similar to ETPs, CCR9⁺ MPPs undergo Notch signaling, as indicated by higher expression of Notch1 and downstream target Hes1 genes compared with other MPP subsets. Furthermore, CCR9⁺ MPPs possess differentiation potential similar to that of ETPs, with very limited granulocyte/macrophage differentiation potential, but they can differentiate into T, B, and dendritic cells. These characteristics implicate CCR9⁺ MPPs as the BM precursors of the earliest thymic progenitors. In addition, our data suggest that before transition from BM to thymus, MPPs are lymphoid-specified and primed for T lineage differentiation.

T cells are continuously produced from the thymus (for review, see ref. 1). The most immature thymocyte population has been characterized in the CD4^lo population or in the CD4 and CD8 double negative (DN) population as c-Kit^highCD44⁺CD25⁻ cells (2–8). These most immature thymocytes, namely early T lineage progenitors (ETPs), have no self-renewal abilities (9, 10). Therefore, hematopoietic progenitors in the bone marrow (BM) derived from hematopoietic stem cells (HSCs) need to seed the thymus, and they give rise to ETPs to initiate intrathymic T cell development in adults.

The mechanism underlying the mobilization of progenitors from the BM and subsequent homing to the thymus, however, remains largely unclear. Because BM cells transit to the thymus through blood flow, it is important to characterize hemato/lymphopoietic progenitors in the peripheral blood (PB), as well as the most immature T cell progenitors in the thymus and thymic immigrants in BM, to understand the connection between BM and thymopoiesis (11, 12). T cell progenitors significantly expand during T cell development, and the thymus is filled with various stages of developing T cells. Therefore, the number of thymus-seeding cells at any given time should be extremely low in the nonstress steady state in adults (12, 13), making the characterization of these recent thymic immigrants challenging. A recent study in chemokine receptor CCR9-GFP knockin (KI) mice demonstrated that immature ETPs express CCR9, because only GFP^hi ETPs in CCR9-GFP KI mice have T and B differentiation ability from the same progenitor (14). Therefore, CCR9 is likely to be expressed on thymic immigrants in the BM.

In BM, multiple progenitor populations have been shown to possess T lineage differentiation potential. Several studies suggest that subsets of multipotent progenitors (MPPs) are thymus-seeding cells (15, 16). We recently demonstrated that BM MPPs are quite heterogeneous and can be subdivided into three subsets by using Flt3 and vascular cell adhesion molecule 1 (VCAM-1) expression (17). Because each of these MPP subsets contains different lineage differentiation potentials (17, 18), it is unclear which subset contributes to thymopoiesis. In this work, we found that a small subfraction of Flt3^hiVCAM-1⁻ MPPs (hereafter simply VCAM-1⁻ MPPs) is positive for CCR9. CCR9⁺VCAM-1⁻ MPPs and ETPs exhibit similar characteristics in cell surface phenotype, gene expression pattern, and differentiation potential. Thus, CCR9⁺VCAM-1⁻ MPPs represent a subset of thymic immigrant from the BM and are likely the immediate upstream precursors of ETPs.

Results

Comparison of T Cell Differentiation Potential of MPPs and Common Lymphoid Progenitors (CLPs).

Multiple BM progenitor populations can give rise to T cells; however, the frequency of cells with T cell potential within each population is not clear. To determine which progenitor population is most enriched with T lineage differentiation potential, we first compared the T cell readout frequency of VCAM-1⁺ MPPs [containing both Flt3^loVCAM-1⁺ and Flt3^hiVCAM-1⁺ MPPs (ref. 17)], VCAM-1⁻ MPPs, and CLPs by culturing cells on OP9-DL1 stromal cell layers (19). We found that all three populations were equally potent in giving rise to T cells, each with a limiting number of fewer than two cells for T cell differentiation potential (Fig. 1A). Furthermore, the expansion of cells in culture was similar, suggesting that all three populations have comparable T cell potential in vitro.

Fig. 1.

Comparison of T cell differentiation of early BM hematopoietic progenitors. (A) VCAM-1⁺ MPPs (Left), VCAM-1⁻ MPPs (Center), and CLPs (Right) all potently differentiated into T cells in vitro. Multiple wells of 2, 5, and 10 cells of each cell population were cultured on OP9-DL1 layers in the presence of IL-7 and Flt3L. Positive wells with T cell growth were determined by the presence of Thy-1.1⁺CD25⁺ cells after 7–14 days in culture. At least 20 wells were analyzed for each of 2-, 5-, and 10-cell culture wells. The limiting number of cells with T cell potential is shown in parentheses in each graph. (B) VCAM-1⁺ MPPs (Left) and VCAM-1⁻ MPPs (Center) but not CLPs (Right) potently developed into T cells in vivo by i.v. injection. Purified VCAM-1⁺ MPPs (40, 80, or 100 cells), VCAM-1⁻ MPPs (20, 40, or 80 cells), and CLPs (500, 750, or 1,000 cells) from WT mice (CD45.1) were injected i.v. into sublethally irradiated RAG2 KO hosts (CD45.2). Mice with positive T cell readouts were determined by detecting CD45.1⁺CD4⁺CD8⁺ double positive thymocytes. Five to 10 mice were analyzed for each data point.

In our initial characterization of CLPs, we noticed that T cell readout frequency of CLPs is quite different in vivo between i.v. and intrathymic injections (20). Because of different thymic-homing capability in progenitors, intrinsic T lineage differentiation potential may not necessarily equate to physiological contribution. Therefore, we also determined the in vivo T lineage potency of each progenitor population by i.v. injection. Because the number of thymocytes peaks at 4 weeks from VCAM-1⁺ MPPs and 3 weeks from VCAM-1⁻ MPPs and CLPs (18, 20), recipient mice were analyzed at these time points after injection. Both VCAM-1⁺ and VCAM-1⁻ MPP subsets were equally potent in giving rise to T cells in vivo by i.v. injection; the limiting numbers for thymocyte readout were 1/80 and 1/75 for VCAM-1⁺ and VCAM-1⁻ MPPs, respectively (Fig. 1B). On the other hand, the limiting number of CLPs that gave rise to thymocytes was ≈1/1,000 by i.v. injection (Fig. 1B).

To determine whether MPPs and CLPs normally have access to the thymus during the steady-state condition, we examined the PB for the presence of these progenitors. MPPs have been previously shown to be circulating in PB, whereas the presence of CLPs in PB is controversial (12, 21). In our work, both MPPs and CLPs were detectable in the PB, with the same phenotype and differentiation potential as their BM counterparts [supporting information (SI) Fig. 6]. These results suggest that the discrepancy between intrinsic and in vivo T cell differentiation potential of CLPs by i.v. injection (but not intrathymic injection) is caused by lower thymic-homing capacity of the cells from the PB.

Expression of CCR9 on VCAM-1⁻ MPPs in Early Hematopoietic Progenitors in BM.

It has been suggested that thymic homing of progenitors involves, at least in part, chemokine-mediated cell attraction (22). As shown in SI Fig. 7, inhibition of chemokine receptor signaling in BM progenitors led to a significant reduction of cells homing to the thymus, especially c-Kit⁺ BM progenitors that include MPPs. Given that MPPs can more superiorly home to the thymus compared with CLPs, we next attempted to search for chemokine receptors that are expressed on MPPs but not on CLPs. CCR9 has been implicated in thymic colonization and homing of progenitors in both fetal and adult T cell development (23, 24). Because ETPs express CCR9 (14), we examined the expression of CCR9 on early hematopoietic progenitors in the BM that express c-Kit. As shown in Fig. 2A, MPPs, but none of HSCs (defined as Thy-1.1^loKLS cells), CLPs, or myeloid progenitors, expressed CCR9 (Fig. 2A).

Fig. 2.

CCR9 is expressed predominantly on VCAM-1⁻ MPPs among early hematopoietic progenitors in BM. (A) CCR9 expression on HSCs, myeloid progenitors (CMP+GMP+MEP), VCAM-1⁺ MPPs, VCAM-1⁻ MPPs, and CLPs. (B) Quantitative RT-PCR analysis of CCR9 expression in VCAM-1⁺ MPPs, CCR9⁻VCAM-1⁻ MPPs, CCR9⁺VCAM-1⁻ MPPs, and CLPs. (C) Migration of CCR9⁺VCAM-1⁻ MPPs in response to CCL25. Triplicate samples of 6,000 purified VCAM-1⁺, CCR9⁻VCAM-1⁻, CCR9⁺VCAM-1⁻ MPPs, and CLPs were deposited into the upper chamber of each well in Transwell plates with or without CCL25 in the lower chamber. ∗, P < 0.05, calculated by Student's t test. (D) Comparison of thymus-repopulating ability by CCR9⁻VCAM-1⁻ and CCR9⁺VCAM-1⁻MPPs. Purified CCR9⁻VCAM-1⁻ MPPs (20, 40, 80) and CCR9⁺VCAM-1⁻ MPPs (20, 40, 60) were injected i.v. into a RAG2 KO host and analyzed as described in Fig. 1B. ∗, 0% negative readout from mice injected with 60 CCR9⁺VCAM-1⁻ MPPs. Five to 15 mice were analyzed for each data point. The difference between the two groups was calculated by Student's t test, where P < 0.05. (E) Direct homing of CCR9⁺VCAM-1⁻ MPPs to thymus upon i.v. injection. Ten thousand purified CCR9⁺VCAM-1⁻ MPPs (CD45.2 and Thy-1.2) were injected into sublethally irradiated (300 rads) WT mice (CD45.1 and Thy-1.1). Donor-derived CD45.2⁺c-Kit⁺CCR9⁺ cells were detected in the recipient thymus after 24 h of injection.

Within the MPP population, CCR9 expression was observed predominantly in the VCAM-1⁻ fraction (5–10%) (Fig. 2A). Purified CCR9⁺VCAM-1⁻ MPPs also expressed a much higher level of CCR9 mRNA compared with other MPPs and CLPs (Fig. 2B). In addition, CCR9⁺VCAM-1⁻ MPPs, but not other MPPs or CLPs, migrated significantly in response to CCL25, the known ligand for CCR9, in an in vitro chemotaxis assay (≈45%; Fig. 2C). Although Scimone et al. (22) showed that CLPs also express CCR9 (22), the use of different markers in defining CLPs may have caused this discrepancy. Scimone et al. defined CLPs as Lin⁻c-Kit⁺hCD25⁺ in pre-Tα-hCD25 transgenic mice, whereas we noted CLPs by their original definition (IL-7Rα⁺Lin⁻Thy-1.1⁻c-Kit^loSca-1^lo; ref. 20). Our results here suggest that CCR9 is expressed preferentially on lymphoid-biased VCAM-1⁻ MPPs in the c-Kit⁺ BM fraction.

Next, we compared the thymic repopulation ability of CCR9⁻VCAM-1⁻ and CCR9⁺VCAM-1⁻ MPPs by i.v. injections by limiting dilution assay as performed in Fig. 1B. As shown in Fig. 2D, CCR9⁺VCAM-1⁻ MPPs were superior to CCR9⁻VCAM-1⁻ MPPs in thymic repopulation, with limiting number of 1/19 and 1/54, respectively, in T cell readout. Engrafted thymi from both CCR9⁻VCAM-1⁻ MPPs and CCR9⁺VCAM-1⁻ MPPs did not differ in the number of donor-derived cells (data not shown). To determine whether CCR9⁺VCAM-1⁻ MPPs can directly home to the thymus, we performed in vivo thymic homing assay of purified CCR9⁺VCAM-1⁻ MPPs. As shown in Fig. 2E, we could detect CD45.2⁺ donor derived c-Kit⁺CCR9⁺ cells in the recipient thymus 24 h after injection. Collectively, these results suggest that CCR9⁺VCAM-1⁻ MPPs have the most robust thymic repopulating ability among the MPP subpopulations, and their direct thymus-homing ability implicates CCR9⁺VCAM-1⁻ MPPs as thymic immigrants from BM.

CCR9⁺ MPPs Are Released Preferentially into the PB.

Next, we examined for the presence of CCR9⁺ MPPs in the PB. We found that CCR9 is expressed on a significant portion of phenotypic MPPs in the PB, ≈25% of all KLS cells (including both HSCs and MPPs), and nearly 50% of PB MPPs (Flt3⁺Thy-1.1⁻ KLS) (Fig. 3A). Although CCR9⁺ MPPs have only 2- to 4-fold advantage over CCR9⁻ MPPs in thymic repopulation (Figs. 1B and 2D), there are 10- to 20-fold more CCR9⁻ MPPs in the BM. The higher percentage of CCR9⁺ MPPs in the PB compared with the BM (50% vs. 1–2% of total MPPs, respectively) suggests that either CCR9⁺ MPPs are preferentially mobilized and/or CCR9⁻ MPPs have a tendency to be retained in the BM. Because the chemokine CXCL12 (or SDF-1) has been shown to mediate the retention of hematopoietic progenitors including HSCs in BM (25, 26), we investigated whether in vivo administration of PTX would result in increased mobilization of CCR9⁻ MPPs from BM into the PB. As shown in Fig. 3B, the percentage of CCR9⁻ MPPs in the PB increased after PTX treatment. This effect peaked at 3 days after PTX injection, where ≈90% of MPPs in the PB were CCR9⁻ (Fig. 3B). Because both CCR9⁻ and CCR9⁺VCAM-1⁻ MPPs in BM expressed similar levels of CXCR4 (data not shown), the retention of CCR9⁻ and CCR9⁺ MPPs in the BM may not be mediated solely by CXCL12/CXCR4. Nonetheless, the superior thymic repopulating ability, coupled with the preferential mobilization of CCR9⁺ MPPs into the circulation during steady-state condition, further validate these specialized MPPs as candidates of physiological thymus-seeding progenitors in adult mice.

Fig. 3.

A significant percentage of PB MPPs expressed CCR9. (A) CCR9 expression on PB MPPs (Flt3⁺ KLS). (B) Change in the percentage of CCR9⁻ MPPs in PB after administration of PTX (500 ng per mouse) by i.v. injection at the indicated time points. PB from five mice was pooled for analysis at each data point. Error bars denote SD from two experiments.

CCR9⁺VCAM-1⁻ MPPs in BM and ETPs in the Thymus Have Similar Differentiation Potentials.

To characterize further the full lineage differentiation potential of CCR9⁺VCAM-1⁻ MPPs, we compared B cell, granulocyte/macrophage (GM), and dendritic cell (DC) differentiation from CCR9⁻VCAM-1⁻ and CCR9⁺VCAM-1⁻ MPPs in BM as well as ETPs. Both subsets of MPPs and ETPs were capable in B cell and DC differentiation on OP9 stromal cells in culture (Fig. 4A), although the frequency of ETPs that gave rise to CD19⁺ B cells was much lower compared with MPPs. Although both MPPs and ETPs also gave rise to some Mac-1⁺ and Gr-1⁺ cells on OP9 stromal cells in bulk cultures, GM colony-forming ability from CCR9⁺VCAM-1⁻ MPPs was almost completely abolished in a fashion similar to that of ETPs (Fig. 4B). CCR9⁻VCAM-1⁻ MPPs, on the other hand, retained significant GM colony-forming activities (Fig. 4B). These data suggest that, like the most immature ETPs, CCR9⁺ MPPs have T and B cell bipotent differentiation ability and can give rise to DCs while retaining only minimal GM differentiation ability.

Fig. 4.

Lineage potential of CCR9⁺ MPPs. (A) B cell (CD19⁺) and DC (CD11c⁺) differentiation from CCR9⁻VCAM-1⁻ MPPs, CCR9⁺VCAM-1⁻ MPPs, and ETPs on OP9 stromal cell culture. (B) GM colony formation from CCR9⁻VCAM-1⁻ MPPs, CCR9⁺VCAM-1⁻ MPPs, and ETPs on methylcellulose cultures. (C) Phenotypic characterization of CCR9⁺ MPPs. CD62L, PSGL-1, RAG1 (GFP in RAG1-GFP KI mice), and CD27 expression on CCR9⁻VCAM-1⁻ MPPs, CCR9⁺VCAM-1⁻ MPPs, and ETPs (open histogram). The dotted histogram represents isotype control staining.

CCR9⁺VCAM-1⁻ MPPs Are Phenotypically Similar to ETPs in the Thymus and Overlap with Previously Characterized T Lineage Progenitors in BM.

Next, we compared the expression of molecules that were previously shown to be correlated to, or indispensable for, T lineage potential and thymic homing of BM progenitors on CCR9⁻VCAM-1⁻ MPPs, CCR9⁺VCAM-1⁻ MPPs, and ETPs. Previous reports suggest that CD62L⁺ MPPs may have superior thymic-repopulating capability and T lineage differentiation potential (15, 26). Although most CCR9⁺VCAM-1⁻ MPPs and ETPs expressed CD62L, CCR9⁻VCAM-1⁻ MPPs and VCAM-1⁺ MPPs also expressed CD62L heterogeneously (Fig. 4C and data not shown). We also examined the expression of P-selectin glycoprotein ligand 1 (PSGL-1) because mice deficient in this molecule have less immature thymocytes (27). Both CCR9⁻VCAM-1⁻ and CCR9⁺VCAM-1⁻ MPPs expressed PSGL-1 similarly to its expression in ETPs (Fig. 4C).

Early lymphoid progenitors (ELPs), which express RAG1 and CD27, demonstrated potent in vivo T cell potential after i.v. injection (21, 28). In RAG1-GFP KI mice (29), both MPP subsets uniformly expressed CD27, and the majority of CCR9⁺VCAM-1⁻ MPPs were RAG1⁺ (82%) (Fig. 4C). Some CCR9⁻VCAM-1⁻ MPPs also expressed RAG1, although at a lesser proportion, which suggests that CCR9⁺VCAM-1⁻ MPPs significantly overlap with ELPs. However, only a small part of the ELP population is positive for CCR9 (≈10%). It is possible that the high thymic homing capacity of ELPs demonstrated by Perry et al. (21) could be enriched in the CCR9⁺ fraction. We should note that although some GFP⁺ ETPs were detected in RAG1-GFP KI mice by FACS, ETPs significantly down-regulated RAG1 mRNA expression (data not shown). It is possible that the stability of GFP protein allows its continued expression in ETPs after transit from BM. Our data here show that CCR9⁺VCAM-1⁻ MPPs and ETPs share many phenotypic similarities, although they are not identical in expression levels.

Initiation (or Priming) of the T Cell Differentiation Program in CCR9⁺VCAM-1⁻ MPPs in BM.

Notch signaling, particularly by Notch1, plays a major role in T lineage commitment and differentiation (30, 31). Recently, it has been demonstrated that Notch signaling is important before the ETP stage during T cell development (32). We found that compared with CCR9⁻VCAM-1⁻ MPPs or VCAM-1⁺ MPPs, CCR9⁺VCAM-1⁻ MPPs expressed higher Notch1 but not Notch2 on mRNA level (Fig. 5A). In addition, CCR9⁺VCAM-1⁻ MPPs expressed significantly higher levels of Hes1 (Fig. 5A), a direct downstream target of Notch signaling (33, 34), as well as up-regulated Deltex1 (Fig. 5A). Furthermore, CCR9⁺VCAM-1⁻ MPPs in transgenic Notch reporter mice, which carry GFP as a reporter gene for Notch activity (34), expressed higher GFP mRNA level compared with CCR9⁻VCAM-1⁻ MPPs. These data demonstrate that CCR9⁺VCAM-1⁻ MPPs are undergoing active Notch signaling.

Fig. 5.

Priming of the T lineage program in CCR9⁺ MPPs. Quantitative RT-PCR analysis of expression level of Notch signaling components and targets in VCAM⁺ MPPs, CCR9⁻ VCAM-1⁻ MPPs, CCR9⁺ VCAM-1⁻ MPPs, and ETPs (A) and T lineage and B lineage-affiliated genes in these cell populations (B) is shown. Error bars denote SD from expression levels in three independent samples. ∗, P < 0.05, calculated by Student's t test.

Expression of several T lineage-affiliated genes, including GATA3, pre-Tα, CD3ε, and CD25, were also examined (Fig. 5B). We found that although CCR9⁺VCAM-1⁻ MPPs up-regulated GATA3, pre-Tα, and CD25, CD3ε was not detected (Fig. 5B). Although most of the T lineage-affiliated genes were drastically up-regulated in ETPs, ETPs expressed a lower level of pre-Tα compared with CCR9⁺VCAM-1⁻ MPPs. It is possible that pre-Tα expression may be regulated differentially in the BM and the thymus microenvironment. Alternatively, ETPs are heterogeneous, and pre-Tα expression may be enriched in a small population within ETPs.

Because CCR9⁺VCAM-1⁻ MPPs readily differentiated into B cells (Fig. 4A), we asked whether B lineage priming has also occurred at this stage. Both CCR9⁻VCAM-1⁻ and CCR9⁺VCAM-1⁻ MPPs similarly up-regulated EBF, a transcription factor involved in B lineage specification (35), whereas upstream VCAM-1⁺ MPPs did not express EBF (Fig. 5B). In addition, both CCR9⁻VCAM-1⁻ and CCR9⁺VCAM-1⁻ MPPs were equally potent in B cell differentiation in vivo (data not shown). These data collectively suggest that CCR9⁺VCAM-1⁻ MPPs are a specialized subset of MPPs which is similar to ETPs phenotypically and functionally. Although they do not appear to be committed or biased toward the T lineage, CCR9⁺VCAM-1⁻ MPPs have initiated or primed for a T lineage differentiation program in the BM before thymus homing.

Discussion

Differentiation Status of T Lineage Progenitors in BM.

It has been a subject of debate as to when and how BM progenitors home to the thymus, whether this process is contributed by early hematopoietic progenitors in multiple developmental stages, or whether a specific subset of progenitors contributes to thymopoiesis (1, 11). It is also unclear whether T lineage commitment can occur in the BM or only in the thymus and whether T and B lineages diverge from a common progenitor (11).

Our characterization of CCR9⁺ MPPs, coupled with our previous subfractionation of MPPs using VCAM-1 and Flt3 expression, indicate that these thymic immigrants have already lost most myeloid lineage differentiation potential in the BM while retaining T and B lineage bipotent differentiation potentials (18). The minimal myeloid differentiation potential in CCR9⁺VCAM-1⁻ MPPs suggests that they are lymphoid-specified (but not committed) progenitors upstream from CLPs. These results shed light on the hierarchy of lineage restriction during the course of HSC maturation and define a branching point for T and B cells at a developmental stage immediately upstream from the CLP stage (SI Fig. 8).

In contrast to the report by Benz and Bleul (14), our data suggest that during T lineage differentiation, B cell developmental potential is turned off after the loss of GM differentiation ability. It is possible that these progenitors lose B cell differentiation potential rapidly upon entry to the thymus, likely a consequence of increased Notch signaling. However, the residual GM differentiation ability may not be immediately affected upon thymic entry but rather completely shut down at a more advanced maturational stage such as DN2 by a different mechanism.

Mechanisms of Progenitor Homing to the Thymus.

The expression of CCR9 on the most immature ETPs and MPPs suggests an involvement of CCR9/CCL25-mediated attraction of bone marrow progenitors to the thymus. Although CCR9 knockout (KO) mice do not have defects in T cell development, competitive reconstitution of CCR9 KO BM cells with wild type (WT) BM cells revealed that CCR9 enables superior thymic repopulation by BM progenitors (23, 36). Therefore, CCL25/CCR9-mediated cell attraction, as well as other chemokines such as CCL21, may synergistically attract CCR9⁺ MPPs to the adult thymus as demonstrated in fetuses (37). However, we did not observe surface CCR7 expression on CCR9⁺ MPPs, unlike their fetal counterparts (data not shown).

Homing of cells to the thymus involves not only chemokines but also adhesion molecules such as selectins, as demonstrated by the impairment of thymic homing in PSGL-1 KO mice (27). The involvement of α4β1 and αLβ2 in lymphoid progenitor homing to the thymus is also reported (22). In addition, it was also reported that MPPs expressing CD62L yield better thymic engraftment, although mice deficient in CD62L do not have any defect in intrathymic T cell development (15). Because CCR9⁺VCAM-1⁻ MPPs overlap significantly with CD62L⁺ MPPs, it is possible that thymic progenitors are further enriched in the CD62L⁺ fraction (Fig. 4C). All of these molecules are likely to participate cooperatively or redundantly during thymic homing and thymic retention of progenitors. It is also plausible that because the niche size for thymic-seeding cells is very small (38), CCR9⁺ MPPs can more competitively home to the thymus compared with other MPPs and CLPs in PB.

Possible Contribution to T Cell Development by Other BM Progenitor Populations.

Recently it has been reported that CLP-2, a BM progenitor population downstream from CLPs, also expresses CCR9 and can efficiently home to the thymus (22). CLP-2, which phenotypically overlaps with prepro-B cells, is characterized by pre-Tα expression using pre-Tα-hCD25 reporter transgenic mice (39). Because we cannot purify CLP-2 from WT mice, direct comparison of CCR9⁺VCAM-1⁻ MPPs and CLP-2 in the degree of contribution to T cell progenies in the steady state is not clear. When the CLP-2 population contributes to thymic T cell development, CLP-2 seems to give rise to DN2 cells directly, bypassing the ETP stage (40). CCR9⁺VCAM-1⁻ MPPs, on the other hand, are more likely the immediate BM precursor of ETPs in the thymus because the two populations share similar phenotypic and functional characteristics.

Our preliminary data suggest that CCR9⁺ MPPs do home to the thymus more efficiently than CCR9⁻ MPPs when coinjected in vivo (data not shown). In this case, the potent in vivo T cell differentiation from CCR9⁻ MPPs could be the result of a precursor–progeny relationship between CCR9⁻ and CCR9⁺ MPPs, where CCR9⁻ MPPs first home to the BM and differentiate into thymic immigrants such as CCR9⁺ MPPs. We also cannot exclude the possibility that the use of irradiated hosts in our thymic-repopulating and homing experiments (Fig. 2 D and E) can enhance the homing of CCR9⁻ MPPs as suggested in ref. 41. However, because we could see a clear difference in the thymic-repopulating ability by MPPs and CLPs in irradiated hosts (Fig. 1B), comparison of thymic-homing ability of progenitors in irradiated host should be informative. Future competitive homing experiments should be performed in nonirradiated hosts to increase the specificity of thymic homing of progenitors. Regardless, multiple BM progenitors may initiate thymic T cell development at the different DN subpopulations, such as the ETP and DN2. Because progenitors can rapidly change cell surface phenotypes upon thymus entry (40), it is necessary to purify each progenitor population and compare their thymic-homing capacities in a competitive setting. We should emphasize that in addition to thymic-homing capacity, it is also important to compare the T cell number generated from each BM progenitor population within the thymus to understand the major T cell developmental pathway.

T Lineage Priming in MPPs Before Thymic Homing.

Because Notch signaling is necessary for the generation of ETPs in the thymus (32), our observation of Notch activity in CCR9⁺VCAM-1⁻ MPPs further supports our notion that CCR9⁺VCAM-1⁻ MPPs in BM are the immediate precursors of ETPs in the thymus. It is unclear at this moment how CCR9⁺VCAM-1⁻ MPPs have higher Notch activity than CCR9⁻VCAM-1⁻ MPPs. One possibility is that CCR9⁺ and CCR9⁻VCAM-1⁻ MPP subsets are localized in different niches in the BM and thus exposed to different levels of Notch ligands. Activated Notch signals might in turn up-regulate CCR9 and other T lineage gene expression. It also needs to be clarified whether Notch signaling, other stimuli, or cell stochastic mechanisms trigger T lineage differentiation program in MPPs. A recent study suggests that Flt3/Flt3L signaling may be involved in the generation and/or the maintenance of CCR9⁺ thymic immigrants in the BM (36). Further characterization and gene expression profiling of these CCR9⁻VCAM-1⁻ and CCR9⁺VCAM-1⁻ MPP subpopulations may give us more insights into how T lineage differentiation is initiated at the MPP stage in BM.

Materials and Methods

Mice.

All mouse strains used in this work are listed in SI Materials and Methods. 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.

Flow Cytometry, Cell Cultures, in Vivo Injection, and Gene Expression Analysis.

FACS sorting and analysis, in vitro and in vivo cell differentiation assay, and gene expression analysis by quantitative RT-PCR were done as reported previously (17, 18), also described in SI Materials and Methods. Cell populations were double sorted to >99% purity (SI Fig. 9).

Transwell Migration Assays.

Dual-chamber chemotaxis assays were performed as described previously with modification (42). Purified cell populations were resuspended in Iscove's modified Dulbecco's medium (Invitrogen, Carlsbad, CA) with 10% FCS that was equilibrated at 37°C for 24 h, supplemented with SCF (50 ng/ml; R&D, Minneapolis, MN) and Flt3L (30 ng/ml, R&D). The cells were then pipetted into the top chambers of 96-well HTS Transwell plates with 5.0-μm pore size inserts (Corning, Corning, NY). Cytokine-supplemented medium with CCL25/TECK (300 nM; R&D) or the same volume of PBS as a control was added to the lower chamber. After a 2-h incubation, cells from the bottom chamber were harvested for quantification by FACS. A known quantity of fluorescent beads (Flow Check APC; Polysciences, Warrington, PA) was added to the bottom chamber for normalization of migrated cells.

Abbreviations

BM

bone marrow

CLP

common lymphoid progenitor

DC

dendritic cell

DN

double negative

ELP

early lymphoid progenitor

ETP

early T lineage progenitor

GM

granulocyte/macrophage

GMP

granulocyte/macrophage progenitor

HSC

hematopoietic stem cell

KI

knockin

KO

knockout

MPP

multipotent progenitor

PB

peripheral blood

PSGL-1

P-selectin glycoprotein ligand 1

VCAM-1

vascular cell adhesion molecule 1.