To generate T cells throughout adult life, the thymus must import hemopoietic progenitors from the bone marrow via the blood. In this study, we establish that thymus settling is selective. Using nonirradiated recipient mice, we found that hemopoietic stem cells were excluded from the thymus, whereas downstream multipotent progenitors (MPP) and common lymphoid progenitors rapidly generated T cells following i.v. transfer. This cellular specificity correlated with the expression of the chemokine receptor CCR9 by a subset of MPP and common lymphoid progenitors but not hemopoietic stem cells. Furthermore, CCR9 expression was required for efficient thymus settling. Finally, we demonstrate that a prethymic signal through the cytokine receptor fms-like tyrosine kinase receptor-3 was required for the generation of CCR9-expressing early lymphoid progenitors, which were the most efficient progenitors of T cells within the MPP population. We conclude that fms-like tyrosine kinase receptor-3 signaling is required for the generation of T lineage-competent progenitors, which selectively express molecules, including CCR9, that allow them to settle within the thymus.
T cells develop in the thymus (1). However, the thymus contains no long-term self-renewing progenitors. Instead, T lymphopoiesis throughout adult life is maintained by the periodic importation of bone marrow (BM)4 hemopoietic progenitors that reach the thymus via the bloodstream (2, 3, 4, 5, 6). The number of progenitors that enter the thymus each day is estimated to be exceedingly small (6, 7, 8, 9), precluding their direct identification within the thymus. Therefore, which cells physiologically migrate from the BM to the thymus to generate T cells in adult mice is unknown.
Multiple progenitors within the BM have T lineage potential (10) demonstrated experimentally by the ability to generate T cells following intrathymic injection (2). These progenitors include hemopoietic stem cells (HSC), which can produce all blood lineages and have the ability to self-renew (11), multipotent progenitors (MPP), which can generate all hemopoietic lineages but have lost self-renewal capacity (12, 13), common lymphoid progenitors (CLP), which were originally identified as lymphoid committed (14), and cells downstream of the CLP such as the CLP-2 (15, 16). Of these progenitors, HSC and MPP are also known to circulate in the blood of adult mice (17, 18). However, it is unclear whether all of these cells can physiologically settle within the thymus from the bloodstream or if thymus settling is selective.
The molecular basis for progenitor entry into the thymus is poorly understood. This process is likely to be analogous to the homing of mature leukocytes, which involves selectin-mediated weak adhesion to vasculature endothelium, followed by chemokine signaling, strong adhesion through integrins, and transmigration (19, 20, 21). For thymus settling, both CD44 (22, 23) and P-selectin (24) have been shown to be important. Recent work suggests that CCR9 may also play a role in progenitor migration to the thymus. CCR9 is the receptor for CCL25 (25), which is highly expressed by thymic stroma (26). Although CCR9−/− mice have no obvious defect in T cell development or thymic cellularity (27, 28), an early defect in T cell development has been revealed in competitive mixed BM chimeras (29, 30). Migration of CLP-2 cells into the thymus has been found to be CCR9 dependent (31). Furthermore, blocking CCL25 results in reduced migration of progenitors into fetal thymic lobes in vitro (32). However, this same group reported that whereas there is a role for CCR9 in the migration of progenitors into the fetal thymic anlage, there was no role for CCR9 in adult thymic settling (33). Therefore a requirement for CCR9 in the homing of T lineage progenitor to the adult thymus remains controversial.
Recently, an analysis of Ccr9-GFP reporter mice revealed that the earliest identified progenitors in the thymus express high levels of GFP (34), which is consistent with the idea that progenitors that enter the thymus are CCR9+. A similar population was identified in the thymus using the cytokine receptor fms-like tyrosine kinase receptor-3 (Flt3) (35, 36), suggesting that thymus-settling progenitors may also be Flt3+ (37). Flt3 is expressed by multiple BM progenitors with T lineage potential, including MPP, CLP, and CLP-2-like cells, whereas HSC are Flt3− (12, 13, 16, 38). Furthermore, Flt3 is required for efficient generation of CLP and B cells (38, 39). For T cell development, Flt3−/− mice have only a modest reduction in thymic cellularity, but a requirement for Flt3 and Flt3 ligand (Flt3L) was revealed in competitive-mixed BM chimeras (35, 39). The importance of Flt3 signaling was most evident at the early stages of T lymphopoiesis (35). These results suggest that Flt3 and Flt3L may be important for the generation of thymus-settling cells.
In this study, we inquired whether thymus settling is selective. We found that HSC injected into the blood of unirradiated mice were unable to settle within the thymus, whereas downstream MPP and CLP both rapidly generated T lineage cells following i.v. transfer. This selectivity correlated with the expression of CCR9 by a subset of MPP and CLP but not HSC. Furthermore, CCR9 expression was important for thymus settling. We next found that Flt3 signaling was required for the generation of CCR9+ progenitors, including early lymphoid progenitors (ELP) (40). In the absence of prethymic Flt3 signaling, thymus settling by i.v. injected progenitors was impaired. We conclude that thymus settling is selective and is regulated by Flt3 signaling and consequent generation of CCR9-expressing ELP.
Materials and Methods
C57BL/6 (B6) and B6.Ly5.2 (B6.Ly5SJL) mice were purchased from the National Cancer Institute animal facility or Taconic Laboratories. Flt3l−/− mice were purchased from Taconic Laboratories, and Il7ra−/− mice were purchased from The Jackson Laboratory. Flt3−/− mice were obtained from I. Lemischka (Princeton University, Princeton, NJ). NG-BAC mice (41) were obtained from M. Nussenzweig (Rockefeller University, New York, NY) and crossed with Flt3l−/− mice. We previously generated Ccr9−/− mice (29). All mice were backcrossed at least four generations onto the B6 background. Mice used as donors or for analysis were females of 4–9 wk of age. Recipient mice were all 4.5-wk-old females. All live animal experiments were performed according to protocols approved by the Office of Regulatory Affairs of the University of Pennsylvania in accordance with guidelines set forth by the National Institutes of Health.
Cell preparations, flow cytometry, and cell sorting
BM isolated from both femurs and tibias was treated with ACK lysis buffer (Cambrex) to remove RBC. Thymocytes were prepared as a single-cell suspension. To enrich for early progenitors, CD4- and CD8-expressing thymocytes were depleted using subsaturating concentrations of anti-CD4 (GK1.5) and anti-CD8α (53.6-7), followed by removal of Ab-coated cells with magnetic beads conjugated to goat anti-rat IgG (Polysciences).
Cell preparations were stained with optimal dilutions of Ab. Abs in the lineage mixture included anti-B220 (RA3-6B2), anti-CD19 (1D3), anti-CD11b (M1/70), anti-Gr-1 (8C5), anti-CD11c (HL3), anti-NK1.1 (PK136), anti-Ter119 and anti-CD3 (2C11), anti-CD8α (53-6.7), anti-CD8β (53-5.8), anti-TCRβ (H57), and anti-TCRγ (GL-3). Additional Abs used included anti-Ly5B6 (104), anti-Ly5SJL
For progenitor sorts, BM from 10 mice (6 × 108 cells) was used to collect 5 × 104 HSC, 105 MPP, and 105 CLP. Sort gates are indicated in Fig. 1⇓B. For experiments studying MPP subsets, Flt3low, Fltmed and Flt3high MPP were sorted as indicated in Fig. 7⇓A, top panel. Cells were sorted on the FACSAria (BD Biosciences) or analyzed on the LSR-II (BD Biosciences). Cell suspensions were pretreated with 4′,6′-diamidino-2-phenylindole for dead cell exclusion. Doublets were excluded using forward side scatter-height vs forward side scatter-width and side scatter-height vs side scatter-width parameters. Data were analyzed using FlowJo (Tree Star).
Intravenous and intrathymic transfers
Unfractionated BM (5 × 107 cells) or freshly sorted progenitors from B6 BM (5 × 104 cells) were injected i.v., by the retro-orbital route, into unmanipulated B6.Ly5SJL recipients. For experiments examining the effects of gamma irradiation, recipient mice were given 500 rad of irradiation 4–6 h before i.v. injection. To prevent rejection, in transfers of Ccr9−/− BM i.v. into unirradiated recipients, some recipients were treated with anti-CD4 (GK1.5), 0.5 mg i.v. at the day before BM transfer and once a week thereafter. Effective depletion of CD4 splenocytes was confirmed at the time of analysis. Equivalent chimerism between Ccr9−/− and wild-type (WT) cells in the BM of recipient mice further confirmed that rejection had not occurred. Control recipients were either administered anti-CD4 or PBS alone, with no difference in thymic or BM engraftment. Intrathymic injections were done as described previously (2). Unfractionated BM (1 × 106 cells) or freshly sorted progenitors from B6 BM (2000 cells) were injected intrathymically into unirradiated anesthetized B6.Ly5SJL recipients. The number of donor-derived CD4+CD8+ double-positive (DP) thymocytes, following i.v. or intrathymic transfers, was calculated by multiplying the total number of thymocytes with the frequency of donor DP thymocytes determined by flow cytometry. The number of donor-derived early T lineage progenitors (ETP) or CD4−CD8− double-negative (DN)3 thymocytes were determined by multiplying (total thymic cellularity) × (frequency of CD4−CD8− cells) × (frequency of ETP or DN3 within the CD4+CD8+-depleted fraction of thymocytes). To generate mixed BM chimeras, recipient mice were irradiated with 900 rad and then injected i.v. with a mixture of 1–2 × 105 Ly5B6 BM cells from WT or knockout mice and 2 × 105 B6.Ly5SJL BM cells.
RNA was isolated from sorted cells using the RNEasy kit (Qiagen), and cDNA was prepared with the Superscript II kit (Invitrogen Life Technologies). Real-time RT-PCR was performed with TaqMan Universal PCR Master Mix (Applied Biosystems) and analyzed on an ABI Prism 7900 (Applied Biosystems). Primer and probe combinations were purchased from Applied Biosystems to assess Ccr9 (Mm02528165_s1), Rag2 (Mm00501300_m1), and Tdt (Mm00493500_m1) expression. The primers for Hprt were 5′-CTCCTCAGACCGCTTTTTGC-3′ and 5′-TAACCTGGTTCATCATCGCTAATC-3′, and the probe sequence was VIC-CCGTCATGCCGACCCGCAG-TAMRA. Relative expression levels were normalized with Hprt and calculated using the 2−ΔΔCT method.
Selectivity of thymus settling
The ability of BM hemopoietic progenitors to settle within the thymus has been evaluated previously using mainly irradiated recipient mice (4, 7, 8, 42). The advantage of irradiation is that it depletes most host-type hemopoietic cells, vacating hemopoietic niches in the BM and thymus, and allows for efficient generation of donor-derived cells. However, hemopoiesis following irradiation is not physiological. Irradiation causes extramedullary hemopoiesis in the spleen and other sites (43, 44), increased levels of cytokines in the thymus and circulation (45), vascular damage, and the loss of normal competitor cells. Therefore, we attempted to evaluate thymus settling in unirradiated, unmanipulated WT adult recipient mice. For these experiments, we used 4.5-wk-old recipients because this is an age at which the thymus has been reported to be most receptive to settling by progenitors from blood (5, 46). We found that donor progeny could be detected in the thymus after injecting 5 × 107 BM cells into the bloodstream. Donor-derived DP thymocytes were first seen at 2 wk following i.v. transfer, whereas large numbers of DP thymocytes were first generated at 3 wk (Fig. 1⇑A).
We next fractionated BM into progenitor subsets to identify which BM progenitors, when placed in the blood of unirradiated recipient mice, were competent to generate DP thymocytes at 3 wk (Fig. 1⇑B). Candidates included HSC, MPP, and CLP, each of which has T lineage potential in irradiated recipient mice (11, 12, 13, 14, 47). HSC can be identified in the BM by their lack of mature lineage Ag (Lin) markers, high-surface expression of Sca-1, and the cytokine receptor c-Kit (Lin−Sca-1highc-Kithigh; LSK), but lack of surface Flt3 expression (12, 13, 48, 49, 50); MPP are phenotypically LSKFlt3+ (12, 13); and CLP are Lin−IL-7Rα+Flt3+c-KitlowSca-1low (14, 38). HSC, MPP, or CLP, as well as the “other” remaining Lin−c-Kitneg/low cells, were sorted from the BM of 10 B6 donors and 5 × 104 cells from each population were injected i.v. into one unirradiated B6.Ly5SJL-congenic recipient. From each sort, sufficient numbers of progenitors were obtained to inject one recipient with HSC, two recipients with MPP, and two recipients with CLP. Three weeks after i.v. injection of purified progenitors into mice, recipient thymi were analyzed for donor-derived DP thymocytes (Fig. 1⇑, C and D). In all experiments, both MPP- and CLP generated donor thymocytes, most of which were DP, although MPP were 50-fold more efficient than CLP at this time point. Indeed, DP progeny were easily detected from 104 MPP (data not shown). For i.v. delivered MPP ranging from 104 to 105 cells, we found the T lineage chimerism was proportional to the number of cells injected (data not shown). Apart from MPP and CLP, no other BM populations gave rise to T lineage progeny in this assay, including the remaining Lin−c-Kitneg/low fraction of BM or the Lin−c-Kitneg/lowB220+ population, which contains CLP-2 progenitors (15, 16) (Fig. 1⇑D). This suggests that, although CLP-2 cells can efficiently settle within the thymus (31), they are inefficient progenitors of T cells (15, 16). Alternatively, these negative results may be due to the low frequency of T competent progenitors within the donor population. Despite the potent T lineage potential of HSC, this population also failed to produce any DP thymocytes. These data indicate that the early 3-wk wave of T cell production, following the injection of unfractionated BM i.v. into unirradiated recipients, is predominantly driven by the MPP subset.
These results differ from a similar experiment using irradiated recipient mice. Three weeks after i.v. transfers into irradiated recipients, both HSC and MPP generated donor-derived DP thymocytes, as expected from past work (data not shown; Refs. 12 and 47). These results indicate that although HSC have T lineage potential, as revealed using irradiated recipients, HSC do not rapidly generate T cells in unirradiated recipient mice.
We next analyzed the kinetics with which HSC, MPP, and CLP generate early thymic progenitors during the first 4 wk following i.v. transfer (Fig. 2⇓, A and C). Within the thymus, ETP are phenotypically Linneg/lowc-KithighCD25− (47). This is followed by the Linneg/low c-KithighCD25+ DN2 stage and then the Lin−c-Kitneg/lowCD25+ DN3 stage (51, 52). Progenitors next down-regulate CD25 and up-regulate CD4 and CD8 to generate DP thymocytes (10, 53). MPP injected i.v. progressed through this conventional pathway (Fig. 2⇓, A and C). Small numbers of Linneg/low donor progeny were first detected 8 days after transfer, all of which were c-KithighCD25− ETP. By day 15, most progenitors had up-regulated CD25 expression and differentiated into DN3 cells. Large numbers of DP thymocytes were first seen at day 22, which is consistent with the kinetics of unfractionated BM (Fig. 1⇑A), which is dominated by this MPP subset. Similar kinetics were observed following direct intrathymic transfer of 2000 MPP (Fig. 2⇓, B and D).
CLP injected i.v. had accelerated T lineage kinetics but generated fewer peak progeny for a shorter period of time than MPP (Fig. 2⇑, A and C), as expected from past work with irradiated recipient mice (14, 40, 47, 54). Interestingly, although CLP are phenotypically c-Kitlow, they up-regulated c-Kit to generate c-Kithigh ETP and DN2 thymocytes by day 8 after transfer. Significant numbers of DN3 cells were also present at this time. By day 15, all c-Kithigh cells had disappeared, and DN3 and DP thymocyte cellularities reached their peak. The peak number of DP thymocytes derived from CLP was ∼10-fold lower than that of MPP 1 wk later. By day 22, no Linneg/low progeny from CLP remained, and the number of DP thymocytes was declining. Similar kinetics were observed following direct intrathymic transfer of CLP (Fig. 2⇑, B and D). These results indicate that both MPP and CLP, injected into unirradiated mice, can generate T cells through the conventional pathway, but with different kinetics and efficiencies.
For both MPP and CLP, intrathymic transfers were always more efficient than i.v. transfers (Fig. 2⇑). One likely reason is that only a small fraction of i.v. injected progenitors circulates through the thymus and settles within it. Intravenously injected progenitors circulate for a very short period of time, and the thymus is a small organ estimated to receive only 1/400 of the cardiac output (17, 18). Therefore, i.v. injected progenitors may lodge in other sites, including the BM, instead of the thymus. HSC, MPP, and CLP populations each gave rise to donor-derived cells in the BM at each time point analyzed (data not shown). An additional possibility is that the MPP and CLP populations may be heterogeneous with only a small fraction of cells competent to settle within the thymus.
In contrast to MPP and CLP, HSC injected i.v. failed to give rise to any Linneg/low thymocytes during the first 4 wk after transfer (Fig. 2⇑, A and C). This indicates that the inability of HSC to generate DP thymocytes in unirradiated recipients by 3 wk (Fig. 1⇑, C and D) is not due to a delay in intrathymic differentiation. Instead, HSC placed in blood do not establish in the thymus of unirradiated recipients during this time period. This suggests that HSC either cannot directly settle within the thymus or alternatively that HSC that settle cannot compete with endogenous cells in a nonirradiated thymus. To differentiate between these possibilities, we analyzed the ability of HSC to generate T lineage progeny following intrathymic transfer into unirradiated recipient mice (Fig. 2⇑, B and D). Intrathymic transfers differ from i.v. transfers in that they bypass a requirement for thymic entry. HSC injected intrathymically generated large numbers of T lineage progeny, reaching a similar peak number of donor-derived DP thymocytes as MPP at 4 wk after transfer. These results show that HSC, when placed within the thymus, can successfully compete with endogenous progenitors to generate T cells. Therefore, the absence of any T lineage progeny from HSC within the thymus during the first 4 wk following i.v. transfer demonstrates that HSC cannot settle within the thymus.
The i.v. injection of HSC into unirradiated recipients always resulted in HSC engraftment of the BM (data not shown). Therefore, we expected that HSC injected i.v. would indirectly give rise to T lineage progeny in the thymus at later time points by first producing progenitors in the BM capable of directly colonizing the thymus. Indeed, donor-derived DP thymocytes were detected 8 wk after i.v. transfer of HSC (data not shown). These results indicate that HSC generate T cells with a significant delay relative to MPP and CLP, consistent with indirect thymic colonization from HSC via downstream progenitors.
Taken together, these experiments demonstrate that thymus settling is selective. HSC cannot directly settle within the adult thymus, whereas downstream progenitors can. Therefore, the ability of progenitors to settle within the thymus must be acquired in the BM by progenitors downstream of HSC, at the MPP or CLP stage.
CCR9 expression by BM progenitors
We next investigated the molecular basis for selective thymus settling. One molecule proposed to play a role in thymus settling is the chemokine receptor CCR9 (29, 30, 31, 32, 33). Furthermore, a recent analysis of Ccr9 reporter mice demonstrated that a subset of BM LSK progenitors and CLP express CCR9 (34). Therefore, we asked whether CCR9 is selectively expressed by MPP and CLP, which can rapidly generate T lineage cells following i.v. transfer, but not HSC, which cannot settle within the thymus. We found that HSC lacked surface expression of CCR9, whereas a subset of both MPP and CLP expressed surface CCR9 (Fig. 3⇓A). Ccr9 mRNA was also significantly increased in sorted MPP and CLP compared with HSC (Fig. 3⇓B). Altogether, the expression pattern of CCR9 is consistent with the cellular specificity of thymus settling.
Role of CCR9 in thymus settling
We next evaluated the role of CCR9 in early T cell development by generating competitive BM chimeras using a 1:1 mixture of either CCR9−/− or control B6 BM (Ly5B6) and host-type BM (Ly5SJL) to reconstitute irradiated mice. At 10 wk after transfer, the BM and thymi of these mice were analyzed for donor-derived cells (Fig. 4⇓A). CCR9−/− progenitors efficiently engrafted in the BM and generated HSC and MPP. However, the chimerism of all thymic subsets derived from CCR9−/− BM, beginning at the ETP stage, was significantly less than the chimerism in the BM HSC compartment. These data extend previous work (29, 30), indicating an early requirement for CCR9 in mixed BM chimeras, and indicate that CCR9 confers a competitive advantage at or before the ETP stage of T cell development.
To determine whether CCR9 is important for efficient thymus settling or for the intrathymic development of ETP, we injected BM from either WT or CCR9−/− mice i.v. or intrathymically into unirradiated recipients (Fig. 4⇑, B and C). We found that CCR9−/− BM injected i.v. was significantly less efficient at generating ETP than control BM (Fig. 4⇑C, left panel). This difference was evident by day 8 after transfer (Fig. 4⇑B, left panels), a time at which both CLP and MPP contribute to the generation of ETP (Fig. 2⇑). CCR9−/− BM continued to produce less ETP chimerism than WT BM at day 22 (Fig. 4⇑B, right panels), when the generation of ETP is driven by MPP (Fig. 2⇑). This defect was specific to the thymus. LSK progenitors from CCR9−/− BM had no defect in engrafting the BM (Fig. 4⇑C, middle panel). The reduction in donor-derived ETP was not due to an intrathymic defect because there was no significant difference in the ability of WT or CCR9−/− BM injected intrathymically to generate ETP (Fig. 4⇑C, right panel). These results indicate that CCR9 is important for efficient thymus settling. However, CCR9 expression was not absolutely required, as some T lineage progeny were detected in the thymi of mice receiving CCR9−/− BM, although considerably reduced in number relative to WT controls.
If CCR9 is important for thymus settling, why do the thymi of CCR9−/− mice have essentially normal cellularity? To gain insight into this issue, we analyzed the earliest progenitor subsets in CCR9−/− thymi (Fig. 4⇑D). We found a reduced number of ETP and DN2 cells in CCR9−/− thymi compared with aged-match WT controls, whereas the number of DN3 thymocytes was normal. This is similar to the phenotype of P-selectin glycoprotein ligand-1−/− mice, which are also thought to have a thymus settling defect (24). These data suggest that under noncompetitive situations, the thymus may be able to compensate for a reduction in thymus settling by the DN3 stage. Similar to what was seen in the case of P-selectin (24), the requirement for CCR9 in thymus settling was most clearly revealed in competitive assays (Fig. 4⇑, A–C). Taken together, our results demonstrate a role of CCR9 in efficient thymus settling.
Flt3 signaling requirement for generation of CCR9+ ELP
We next investigated the cytokine requirements for the generation of CCR9-expressing progenitors in early hemopoiesis. For lymphocyte development, the cytokine receptors c-Kit, Flt3, and IL-7R are each known to be important at stages before lineage commitment (55). Of these, only Flt3 is expressed by MPP and CLP but not HSC (12, 13, 38), corresponding with the expression pattern of CCR9. Flt3 is first expressed at the MPP stage of hemopoiesis and is important for the generation of CLP from MPP (38). Furthermore, roles for Flt3 in early B lineage differentiation and thymocyte production have been revealed using competitive-mixed BM chimeras (39). Therefore, we asked whether Flt3 signaling was required for the generation of CCR9-expressing BM progenitors. Within the BM LSK compartment of WT mice, CCR9 was selectively expressed by a subset of Flt3+ MPP. However, in Flt3L−/− mice, BM LSK cells lacked CCR9 surface expression (Fig. 5⇓A). The same result was found in Flt3−/− mice (data not shown). This result was specific to Flt3 signaling because mice lacking the α-chain of the IL-7 cytokine receptor had no defect in CCR9 expression (Fig. 5⇓A). Additionally, although Notch signaling is critical for regulating T cell development (36, 56, 57), blocking Notch signaling in BM progenitors with dominant-negative mastermind like-1 (35) had no effect on CCR9 expression (data not shown).
Flt3L−/− mice also had less MPP, evident by the reduced frequency of LSK cells that expressed Flt3 (Fig. 5⇑A). This suggests that Flt3L−/− mice might be missing a subset of MPP that expresses CCR9. One subset of MPP that is known to possess potent T lineage potential is the ELP (40). ELP are the earliest progenitors in hemopoiesis to be lymphoid specified—expressing lymphocyte-specific genes, including Rag1, Rag2, and Tdt (40). Therefore, we asked whether CCR9 expression by MPP correlates with lymphoid specification. We found that CCR9 expression within the MPP population was restricted to a subset of ELP, identified as GFP+ MPP in the BM of NG-BAC reporter mice, which express GFP under the control of Rag2 cis-elements (Fig. 5⇑B). This suggests a common program for lymphoid specification and CCR9 expression within the BM MPP population. This differentiation program precedes lineage commitment, as both ELP and CCR9+ LSK cells have been shown to be multipotent (34, 40). Additionally, both ELP and CCR9+ LSK cells are known to circulate in blood and thus have physiological access to the thymus (18, 34).
Since CCR9 expression within the BM LSK population was restricted to the lymphoid specified ELP subset, we next asked whether Flt3 signaling was important for the generation of ELP in addition to CCR9 expression. We found that Flt3L−/− NG-BAC Rag2-GFP reporter mice lacked the GFP+ ELP subset of MPP (Fig. 5⇑C). Furthermore, the reduced MPP frequency in Flt3L−/− mice could be accounted for by the lack of ELP. These data demonstrate that the defect in lymphocyte development, in the absence of Flt3 signaling (35, 38, 39), is evident at an earlier stage of lymphopoiesis than previously appreciated. Consistent with these results, the mRNA expression of Ccr9, as well as the lymphoid-specific genes Rag2 and Tdt, were significantly reduced within LSK cells from Flt3L−/− BM compared with WT LSK cells (Fig. 5⇑D). Therefore, Flt3 signaling was required for lymphoid specification and generation of ELP, which is the only subset of MPP that expresses CCR9.
Flt3 requirement for thymus settling
Based on the lack of ELP, CLP (38), and CCR9 expression in the absence of Flt3 signaling, we predicted that Flt3 signaling would be required for the generation of efficient thymus settling progenitors. To test this hypothesis, we first analyzed at what stages of T lineage development Flt3 is important by generating mixed BM chimeras using BM from either Flt3−/− or control Flt3+/− mice (Ly5B6) mixed with WT BM (Ly5SJL) and injected into irradiated Ly5SJL hosts (Fig. 6⇓A). At 10 wk after transfer, the BM and thymi of these mice were analyzed for donor (Ly5B6)-derived cells. Whereas Flt3−/− progenitors efficiently engrafted in the BM and generated LSK cells, Flt3−/− progenitors had a competitive disadvantage in the thymus, which was evident at the earliest ETP stage. There was no further requirement for Flt3 downstream of ETP. The chimerism beyond this stage remained stable, which is consistent with the lack of Flt3 expression by thymocytes downstream of ETP (35). This early requirement for Flt3 in T cell development was most clearly revealed under competitive conditions. These results demonstrate a requirement for Flt3 signaling at or before the generation of ETP.
Flt3 signaling may be important either prethymically and/or intrathymically. To determine whether Flt3 signaling was required in the BM for generation of efficient thymus-settling progenitors, we assayed the ability of progenitors from Flt3L-deficient BM, which have developed in the absence of Flt3 signaling, to settle within the thymus (Fig. 6⇑B). Such Flt3L−/− progenitors are competent to receive Flt3 signals when adoptively transferred into Flt3L-sufficient mice. We injected whole BM from either Flt3L−/− or WT B6 mice i.v. into unirradiated WT B6.Ly5SJL recipients. Three weeks later, the recipient thymi were analyzed for donor-derived ETP (Fig. 6⇑B, left). BM from Flt3L−/− mice gave rise to significantly fewer ETP compared with WT BM. All downstream progenitors within the thymus were similarly reduced (data not shown). This defect was specific to the thymus. LSK progenitors from Flt3L−/− BM had no defect in engrafting in the BM (Fig. 6⇑B, middle). Furthermore, the reduction in donor-derived ETP was not due to an intrathymic defect because there was no significant difference in the ability of WT or Flt3L−/− BM injected intrathymically to generate ETP in a Flt3L-sufficient environment (Fig. 6⇑B, right). In the same experiment, generation of downstream DN2, DN3, and DP thymocytes was also comparable between WT and Flt3L−/− BM injected intrathymically (data not shown). These results demonstrate that Flt3 signaling is required in the BM for the generation of efficient thymus-settling cells.
T lineage competence of MPP subsets
Our results suggest that the most efficient thymus-settling progenitors within the MPP population are ELP, as these progenitors are absent in the BM of Flt3L−/−-deficient mice (Fig. 5⇑C). Therefore, we directly compared the ability of ELP and the remaining MPP subsets to generate T cells 3 wk after i.v. transfers into unirradiated recipients. For these experiments, we were unable to use NG-BAC donors, as GFP can cause rejection in immunocompetent recipient mice (58). Instead, we used Flt3 expression to identify ELP. The top 20% of Flt3-expressing LSK progenitors (Flt3high MPP), also referred to as lymphoid-primed MPP (59, 60), were enriched for ELP, whereas the Flt3low and Flt3med fractions of MPP contained fewer ELP (Fig. 7⇓A). Whereas the Flt3med fraction did contain some GFP+ cells, these appeared dull for GFP expression compared with the Flt3high cells. When purified Flt3high, Flt3med, and Flt3low MPP were injected i.v. into unirradiated recipients, Flt3high MPP gave rise to ∼10-fold more DP thymocytes at 3 wk than either of the other subsets (Fig. 7⇓B). We conclude that the early 3-wk wave of T cell development, following i.v. transfer into unirradiated recipients, is driven by the ELP subset of MPP.
Our findings indicate that physiological thymus-settling progenitors express Rag2, Flt3, and CCR9. Therefore, we expected that the earliest progenitors in the thymus would also express these markers. Indeed, we found that ETP are entirely GFP+ in NG-BAC Rag2 reporter mice (Fig. 7⇑C). This differs from a previous result (54) using a different reporter mouse that had GFP knocked into the Rag1 locus, indicating that these two reporter mouse strains are not identical. Recently, a subset of ETP has been shown to express high levels of GFP in a Ccr9-GFP reporter mouse (34) and a parallel study identified the Flt3+ subset of ETP (35). Both studies concluded that their populations contained the earliest progenitors identified in the thymus (34, 35, 36, 37). However, the overlap between these two populations has not been examined previously. Therefore, we analyzed Ccr9 gene expression of sorted ETP Flt3+ cells compared with ETP Flt3low cells and downstream DN2 thymocytes (Fig. 7⇑D). We found that Ccr9 was highly expressed by ETP Flt3+ compared with ETP Flt3low and DN2 thymocytes. This demonstrates that Flt3+ ETP and Ccr9-GFPhigh ETP are overlapping populations, which is concordant with the idea that progenitors enter the thymus expressing Flt3 and CCR9 in addition to Rag2.
In this study, we investigated whether thymus settling is selective. We found that, whereas HSC have potent T lineage potential, revealed by direct intrathymic injection, they fail to settle within the thymus when placed in the blood of unirradiated recipient mice. Downstream MPP and CLP both rapidly generate T cells following i.v. transfer. This selectivity corresponded with the expression of the chemokine receptor CCR9 by a subset of MPP and CLP but not HSC. We demonstrated that CCR9 is important for efficient thymus settling, indicating that CCR9 is one of the molecules that regulate the cellular specificity of thymus settling. Additionally, we found that CCR9 was selectively expressed by ELP but not other MPP subsets. Furthermore, Flt3 expression and signaling was required for the generation of ELP. Consistent with the role of Flt3 in the generation of thymus-setting cells, progenitors from Flt3L−/− BM lacked the ability to efficiently settle within the thymus, whereas sorted Flt3high MPP, enriched for ELP, were efficient progenitors of T cells following i.v. transfer into unirradiated recipients. We conclude that thymus settling is selective. We further conclude that progenitors downstream of HSC acquire competence to settle within the thymus through expression of homing molecules that include CCR9.
Whether HSC directly settle the adult thymus was previously unresolved. HSC have potent T lineage potential and are present in blood, thus having access to the thymus (17, 18). However, self-renewal is the hallmark of stem cell activity, and no self-renewing progenitors have been identified in the thymus (3, 4). Furthermore, when BM was injected i.v. into irradiated mice, donor-derived cells in the thymus lacked the ability to self-renew, indicating that HSC are either excluded from the thymus or that the thymic environment induces the rapid loss of self-renewal from these progenitors (42). In this study, we found that HSC injected i.v. into unirradiated mice failed to generate any donor progeny within the thymus for the first 4 wk after transfer, whereas the same cells injected into the thymus generated large numbers of DP thymocytes within this timeframe. MPP were similarly efficient to HSC upon direct intrathymic injection, but unlike HSC, MPP also rapidly and efficiently generated T cells following i.v. transfer. Even CLP, which were inferior to MPP and HSC upon intrathymic transfer, could be clearly shown to generate donor-derived thymocytes following i.v. transfer. The i.v. injection of HSC did lead to BM engraftment and subsequent T cell development by 8 wk after transfer. These experiments establish that HSC cannot physiologically settle within the thymus of adult mice. One advantage of this may be that the development of specific T competent progenitors downstream of HSC in the BM allows for the regulation of thymus settling independently of development of other lineages. This may be important in situations such as crisis hemopoiesis or in aging in which T lymphopoiesis is decreased relative to myeloid and erythroid development. We conclude that thymus settling is selective and prethymic differentiation steps are necessary to generate physiological thymus-settling progenitors.
The inability of HSC to settle within the thymus was only evident using unirradiated recipient mice. In irradiated recipients, HSC injected i.v. rapidly and efficiently generated donor-derived DP thymocytes by 3 wk after transfer. It is unclear why irradiation has this effect. One possibility is that irradiation makes the thymus receptive to HSC settling, perhaps through vascular damage or the induction of chemokines in the thymus that allow for HSC entry (45). A second explanation is that irradiation can lead to alternative pathways of T cell differentiation. Recent evidence indicates that following irradiation the spleen and lymph nodes become permissive for the development of pre-T cells, which could then settle within the thymus (44, 61). A third possibility is that T cell differentiation occurs normally following irradiation but is accelerated. HSC are normally quiescent, but following irradiation, these cells cycle more rapidly to reconstitute the ablated hemopoietic system (62). This quiescence of HSC may account for the large delay in their ability to generate thymus-settling progenitors in unirradiated recipient mice. Importantly, our findings reveal that the study of physiological thymus settling requires the use of unirradiated recipient mice.
A variety of BM progenitors downstream of HSC may physiologically settle the thymus ranging from MPP subsets to CLP and CLP-2 cells (10). The differential ability of these progenitors to contribute to T lymphopoiesis will depend on their ability to migrate from the BM to the bloodstream (18), their capacity to settle within the thymus from the bloodstream, and their efficiency in generating T cells once within the thymus. MPP and CLP each rapidly generated T lineage progeny in the thymus following i.v. transfer, whereas the remaining lineage-negative BM progenitors did not. MPP were also the most efficient progenitor for reconstituting T lymphopoiesis in both irradiated (44) and unirradiated mice (the present study). MPP are present in the bloodstream and thus have physiological access to the thymus (18). Furthermore, the kinetics of T lineage differentiation from MPP suggests that some of these cells directly enter the thymus (46) (the present study). There is also evidence for a T cell developmental pathway independent of CLP intermediates, again supporting direct thymus settling by MPP (10, 47). Additionally, multipotent cells have been identified in the thymus, at the single-cell level, and this population has been proposed to contain thymus-settling cells or the direct progeny of thymus-settling cells (34, 37). Therefore, it is probable that a subset of MPP can physiologically settle within the thymus.
The ability of CLP to contribute to T lymphopoiesis through the conventional pathway has been questioned recently (63). We found that c-Kitlow CLP, injected either i.v. or intrathymically, rapidly up-regulate c-Kit surface expression to generate c-Kithigh ETP by day 8 after transfer. This was likely due to Notch signaling within the thymus (56) because Notch signals have been shown to up-regulate c-Kit expression on progenitors in vitro (63, 64, 65). Our results indicate that CLP can transit through the conventional pathway of T lineage differentiation. We conclude that CLP placed in blood can settle within the thymus. However, can CLP physiologically enter the bloodstream? We were unable to find CLP in the bloodstream of adult mice (18). However, a new study suggests that low numbers of CLP-like cells may circulate (54). Therefore, CLP may directly contribute to T cell development, although they are much less efficient T lineage progenitors than MPP.
Since thymus settling is selective and HSC cannot physiologically enter the thymus, it follows that progenitors in the BM must acquire the ability to settle within the thymus while still retaining T lineage potential. We refer to this process as the acquisition of T lineage competence. One molecule that has been implicated in the acquisition of T lineage competence is the chemokine receptor CCR9. We found that HSC lack CCR9 expression, whereas downstream progenitors, including a subset of MPP and CLP, express surface CCR9. Therefore, CCR9 expression correlates with the ability of progenitors to rapidly reconstitute the thymus. However, a functional role for CCR9 in thymus settling has been controversial. Recently, two studies used short-term thymus settling assays to determine whether CCR9 is important in adult thymus settling, with conflicting results (31, 33). Furthermore, neither of these studies could determine whether the cells entering the thymus were relevant T lineage progenitors. We instead investigated the development of T lineage progeny from CCR9-deficient progenitors. For these studies, we used unfractionated BM to avoid presumptions about the identity of thymus-settling progenitors. CCR9−/− progenitors had comparable BM engraftment to WT control BM. Additionally, the development of MPP and CLP was normal from CCR9−/− progenitors, and MPP continued to circulate at normal levels in the blood of CCR9−/− mice (data not shown). Instead, CCR9 was found to be important for T lymphopoiesis first at the ETP stage. This was not due to an intrathymic requirement for CCR9 because CCR9-deficient progenitors injected intrathymically generated normal numbers of ETP. Instead, our results are most compatible with a role for CCR9 in thymus settling.
Our work suggests that the acquisition of T lineage competence can occur through the expression of CCR9, as well as other molecules important for migration to the thymus. Therefore we investigated the generation of CCR9-expressing progenitors. We found that CCR9 expression was restricted to a fraction of Flt3high progenitors. Furthermore, all progenitors able to rapidly reconstitute the thymus were Flt3high. Within the MPP population, only Flt3high ELP expressed CCR9 and Flt3 signaling was required for the generation of ELP. Therefore, we predicted that Flt3L−/− BM would be deficient in T competent progenitors. Indeed, we found that Flt3 expression was required at or before the ETP stage of T cell development, and Flt3L−/− progenitors were defective in settling the thymus. Furthermore, ELP were the most efficient progenitors of T cells within the MPP population. We conclude that Flt3 signaling is required for the generation of efficient thymus-settling progenitors. Consistent with this, we found that the earliest subset of ETP in the thymus expresses Flt3, Ccr9, and Rag2 (34, 35, 37).
In summary, we established that thymus settling is selective and have uncovered two molecular determinants for this selectivity. The chemokine receptor CCR9 is important for the homing of progenitors to the thymus. The cytokine receptor Flt3 regulates the generation of T competent progenitors, including CCR9-expressing ELP. However, other molecules, which remain to be discovered, may also play a role in these processes. Further characterization of these molecular determinants will help refine our understanding of thymus-settling progenitors.
We thank V. Zediak, M. Cancro, B. Kee, and D. Allman for critical comments; M. Vélez for technical support; and R. Schretzenmair, R. Wychowanec, and H. Pletcher of the Abramson Cancer Center Flow Cytometry and Cell Sorting Shared Resource for technical expertise.
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.
↵1 This work was supported by the National Institutes of Health Grants AI059621 (to A.B.) and T32-AI-055428 (to B.A.S.) and Damon Runyon Cancer Research Foundation Grant DRG-102-05 (to I.M.). A.B. is the recipient of a Career Development Award from the Leukemia and Lymphoma Society.
↵2 B.A.S. and A.S. contributed equally to this work.
↵3 Address correspondence and reprint requests to Dr. Avinash Bhandoola, Department of Pathology and Laboratory Medicine, 264/266 John Morgan Building, 37th and Hamilton Walk, University of Pennsylvania School of Medicine, Philadelphia, PA 19104. E-mail address:
↵4 Abbreviations used in this paper: BM, bone marrow; CLP, common lymphoid progenitor; DN, CD4−CD8− double negative; DP, CD4+CD8+ double positive; ELP, early lymphoid progenitor; ETP, early T lineage progenitor; Flt3, fms-like tyrosine kinase receptor-3; Flt3L, Flt3 ligand; HSC, hemopoietic stem cell; Lin, lineage Ag; MPP, multipotent progenitor; Sca-1, stem cell Ag-1; WT, wild type; LSK, Lin−Sca-1highc-Kithigh.
- Received October 26, 2006.
- Accepted December 1, 2006.
- Copyright © 2007 by The American Association of Immunologists