Heterogeneity among DN1 Prothymocytes Reveals Multiple Progenitors with Different Capacities to Generate T Cell and Non-T Cell Lineages

Immunity, Vol. 20, 735–745, June, 2004, Copyright 2004 by Cell Press

Heterogeneity among DN1 Prothymocytes Reveals Multiple Progenitors with Different Capacities to Generate T Cell and Non-T Cell Lineages Helen E. Porritt,1,4 Lynn L. Rumfelt,2 Sahba Tabrizifard,1 Thomas M. Schmitt,3 Juan Carlos Zu´n˜iga-Pflu¨cker,2,3 and Howard T. Petrie1,* 1 The University of Miami School of Medicine Department of Microbiology and Immunology Miami, Florida 33101 2 Sunnybrook & Women’s College Health Sciences Centre 3 Department of Immunology University of Toronto Toronto, Ontario M4N 3M5 Canada

Summary The nature of early T lineage progenitors in the thymus or bone marrow remains controversial. Here we assess lineage capacity and proliferative potential among five distinct components of the earliest intrathymic stage (DN1, CD25⫺44ⴙ). All of these express one or more hemato-lymphoid lineage markers. All can produce T lineage cells, but only two of them display kinetics of differentiation, proliferative capacity, and other traits consistent with being canonical T progenitors. The latter also appeared limited to producing cells of the T or NK lineages, while B lineage potential derived mainly from the other, less typical T progenitors. In addition to precisely defining canonical early progenitors in the thymus, this work reconciles conflicting results from numerous groups by showing that multiple progenitors with a DN1 phenotype home to the thymus and make T cells, but possess different proliferative potentials and lineage capacities. Introduction The thymus is a non-self-renewing hematopoietic organ that must be seeded by marrow-derived progenitors for continuous function (reviewed in Bhandoola et al., 2003). The question of whether thymus-seeding cells are committed to the T lineage (or to what extent) prior to thymus homing remains contentious. The lineage potential of early intrathymic progenitors is likewise equivocal. Many studies indicate that early T lineage progenitors have only T, T/NK, or T/dendritic cell (DC) potential, while others would suggest that B cells or even myeloid cells can also be produced (for examples see King et al., 2002; Gounari et al., 2002; Allman et al., 2003; Martin et al., 2003; Matsuzaki et al., 1993; Ikawa et al., 1999; Michie et al., 2000; Radtke et al., 1999; Mori et al., 2001; Kondo et al., 1997; Wang et al., 1994). All of these outcomes are supported by multiple publications from independent groups, suggesting that they are unlikely to be *Correspondence: [email protected] 4 Present address: ProImmune Limited, Oxford BioBusiness Centre, Littlemore Park, Littlemore, Oxford OX4 4SS, United Kingdom.

false and that as yet unidentified details are required to reconcile them. During intrathymic differentiation, immature cells progress through a variety of stages that can be identified by surface marker expression. The least mature cells, representing 2%–3% of all thymocytes, lack CD4 and CD8 markers (double-negative, DN) that characterize ␣␤ T cells in the periphery. The DN subset is also heterogeneous and consists of three sequential stages that can be defined by a variety of additional markers. The most common characterization utilizes CD25 and CD44 (reviewed in Godfrey and Zlotnik, 1993) to discriminate these populations, with the first stage (DN1) being CD25⫺44⫹, the second (DN2) being CD25⫹44⫹, and the third (DN3) being CD25⫹44lo. By virtue of expression of CD25, which is largely specific for T cells, DN2 and DN3 stages are nearly homogeneous, having definitive phenotypes of CD24⫹90⫹117⫹127⫹ or CD24⫹90⫹117lo 127lo, respectively (reviewed in Ceredig and Rolink, 2002). However, numerous non-T lineage cells express CD44 (Ponta et al., 2003), and virtually all non-T cells are consistent with the remaining definitions for DN1 cells (CD25⫺4⫺8⫺). Thus, identification of the component of DN1 cells that represent lymphoid progenitors requires additional distinctions. Recently, the focus has been on CD117 (c-kit) as definitive of lymphoid progenitors within the DN1 subset (reviewed in Ceredig and Rolink, 2002), and CD117⫹ DN1 have been shown to possess most T progenitor potential (Allman et al., 2003). In this manuscript, we show that CD117⫹ DN1 themselves are a heterogeneous mixture that can be distinguished by different levels of CD24 expression. Two of these (CD24⫺ and CD24lo) appear to form a precursor:progeny pair that give rise to DN2 and more mature thymocytes with conventional kinetics and proliferative expansion. The third (CD24⫹) also gives rise to T cells, but without appreciable levels of proliferation, and appears to represent the only CD117⫹ DN1 subset with B lineage capacity. Two additional CD117⫺ DN1 subsets (CD24⫹ and CD24⫺) can also give rise to T cells, but without proliferative expansion, and exhibit other differences from traditional early T progenitors as well. Together, our findings reveal the presence of multiple DN1 subsets that can all give rise to T cells, but that differ in their ability to diverge into other lineages or to produce clonally expanded CD4⫹8⫹ double-positive (DP) cells for positive and negative selection. Results Characterization of Heterogeneity within the DN1 Subset Although DN1 cells are frequently identified solely on the basis of lacking CD4, CD8, and CD25, while expressing CD44, this definition includes a large number of thymocytes that are not T cell precursors, as well as numerous non-T cell lineage cells (see Discussion). Identification of lymphoid progenitors among cells with the CD4⫺ 8⫺25⫺44⫹ phenotype requires additional markers. A va-

Immunity 736

Table 1. DN1 Subset Proportions

DN1a DN1b DN1c DN1d DN1e

% of DN1a

% of Thymusb

Cells/Thymusc

1.4 ⫾ 0.9 11.2 ⫾ 2.7 14.2 ⫾ 5.0 18.5 ⫾ 5.0 54.8 ⫾ 16.0

0.001 0.01 0.01 0.02 0.05

3500 28,000 36,000 46,000 137,000

Figures represent mean ⫾ standard deviation for four experiments. Calculated from the values in the first column, assuming that DN1 cells (in total) comprise 5% of DN thymocytes and that DN thymocytes comprise 2% of all thymocytes. c Assumes 2.5 ⫻ 108 cells per thymus. a

b

Figure 1. Characterization of DN1 Subsets Based on Canonical T Lineage Markers (A) Standard identification of DN subsets in thymocytes depleted of lineage-positive cells. Rectangular gates used to identify DN1 and DN2 cells are shown. (B) shows a representative profile for CD24 x CD117 staining in the DN1 subset; for brevity, these are subsequently defined as DN1 subsets a–e, as indicted. The relative proportions of each subset, derived from several such experiments, can be found in Table 1. In (C), these DN1 subsets are further characterized using the T cell markers CD90 (Thy-1) or CD127 (IL-7R␣). Negative control staining is shown as an unfilled histogram on the CD90 panels. All experiments have been repeated numerous times with nearly identical results.

riety of different markers have been used to identify progenitor thymocytes that differentiate into the lymphoid lineage, including CD24 (heat stable antigen, HSA), CD45 (Ly-5), CD90 (Thy-1), CD117 (c-kit), and/or CD127 (IL-7R␣) (a few examples are Martin et al., 2003; Gounari et al., 2002; Godfrey et al., 1992, 1993; Crispe et al., 1987; Hozumi et al., 1994; Allman et al., 2003; Wilson et al., 1988). The overall consensus seems to point to a phenotype of CD24⫺/lo, CD90lo, CD117⫹, CD127⫹, although there are some notable exceptions (Allman et al., 2003; Martin et al., 2003). One of the factors contributing to these discrepancies, particularly in less recent studies, may have been technical limitations in the ability to simultaneously measure a large number of these parameters. Given the controversy that surrounds the nature of marrow- and thymus-derived

early lymphoid progenitors (reviewed in Bhandoola et al., 2003), a more precise definition of DN1 subsets, both phenotypic and functional, is a necessity. To accomplish this objective, we used six-color flow cytometry to analyze the heterogeneity of DN1 thymocytes (Figure 1). We found the clearest distinctions, in terms of revealing distinct subsets, to be generated by viewing CD24 as a function of CD117 (Figure 1B). Five distinct populations were identified, designated DN1a through DN1e, corresponding to numeric frequency (lowest to highest; see Table 1). Most notably, three of these were clearly CD117⫹ but could be distinguished in terms of being CD24⫺ (DN1a), CD24lo (DN1b), or CD24⫹ (DN1c). Two subsets were CD117⫺ and could be distinguished as either CD24⫹ (DN1d) or CD24⫺ (DN1e). The nature of all of these DN1 subsets is further evaluated as described below. Analysis of other conventional markers of early thymocytes revealed additional distinctions among DN1 cells (Figure 1C). For instance, DN1a and DN1b appear to be completely CD90⫺, while DN1d is almost uniformly CD90⫹. DN1c and DN1e contain both CD90⫹ and CD90⫺ cells. The levels of CD90 on the positive components of these subsets are equivalent to that found on DN3 thymocytes and mature T cells (data not shown). Interestingly, we do not find a true CD90lo population anywhere in the DN1 compartment, which is at odds with some previous reports (Gounari et al., 2002; Michie et al., 2000; Matsuzaki et al., 1993; Godfrey et al., 1992, 1993) but consistent with others (Crispe et al., 1987; Wilson et al., 1988; Perry et al., 2003). The reasons for this are not clear, although mouse strain differences may play some role (discussed in Shortman et al., 1988), as may the uncertain assumption that fetal thymocytes are the same as postnatal ones (reviewed in Kincade et al., 2002). Further, it is not clear for many published studies whether negative staining was determined using all fluorescent stains except CD90 or whether completely unstained cells were used as negative control. The latter gives artificially low backgrounds, while the former is the most accurate control and is used here (note: comparison to completely unstained cells does suggest that DN1a and DN1b are CD90lo; data not shown). CD90 analysis on DN2 cells reveals both low and intermediate populations, consistent with a transition at this stage from negative (or possibly very low level) expression on DN1 to higher expression on DN3 and more mature cells. In contrast to CD90, analysis of

Heterogeneity of Early Intrathymic Progenitors 737

CD127 expression revealed little heterogeneity. Only the major DN1 population, DN1e, expresses this protein (IL7R␣). This finding is consistent with the recent results of others (Allman et al., 2003), which indicated that CD117 and CD127 expression were mutually exclusive on DN1 thymocytes. CD127 expression is subsequently upregulated on DN2 cells in a pattern that is consistent with CD90 expression. Thus, this type of analysis reveals a number of subsets within the DN1 stage, one or more of which are consistent with all of the previous definitions of DN1 progenitors. Recent studies have indicated that a subset of B220⫹CD117⫺CD127⫹lineage⫺ bone marrow cells may represent the earliest thymic homing progenitor (Martin et al., 2003). In this respect, it is important to note that the omission of anti-B220 antibody from our lineage depletion cocktail did not substantially change the distributions of DN1 subsets (data not shown). The only DN1 population exhibiting a detectable B220⫹ component was the DN1d subset, which, being CD127⫺ (Figure 1), does not appear to correspond to the phenotype of the marrow-derived B220⫹ progenitors. Nonetheless, it remains possible that the CD127⫹ component of DN1 (i.e., DN1e) contains a fraction of B220⫹ cells that is too rare to be readily observed but gives rise to all other intrathymic progenitors. Molecular Characterization of Early T Lineage Genes Reveals Genetic Differences among Subsets of DN1 Cells As a first step in characterizing the relationships between DN1 subsets a–e and their relationship to more mature progeny (DN2) or to marrow-derived precursors, a number of molecular characterizations were performed (Figure 2). CD3⑀ is first expressed very early in T cell development (Wilson and MacDonald, 1995) and increases in expression during differentiation through the DN stages. We found that DN1a and DN1b cells did not express detectable levels of CD3⑀, making them similar to marrow-derived stem cells (lin⫺CD117⫹ CD127⫺Sca-1⫹). Trace levels of CD3⑀ were detectable in DN1c, slightly more in DN1d, and DN1e appeared similar in expression to DN2. None of these DN1 populations expressed detectable levels of RAG-1 when 28 PCR cycles were performed (as shown here for CD3⑀ and RAG-1). However, at 35 cycles, faint RAG-1 bands were detectable in DN1d and DN1e cells (data not shown). RAG-1 was readily detectable in DN2 cells. These RAG-1 expression data are thus consistent with the findings of others (Wilson et al., 1994; Plotkin et al., 2003), showing weak but detectable RAG gene expression in total DN1 cells, with stronger expression in DN2 and later progenitors. TCR gene rearrangements are considered to be definitive for T cells. Analysis of early D-J␤ rearrangements in DN1 subsets indicated that only DN1e cells have readily detectable levels of gene rearrangement, while DN1c and DN1d have faint rearranged bands. DN1a and DN1b appear to be mainly in germline configuration. Assuming that all of these DN1 subsets are, in fact, T cell progenitors, these data suggest that DN1a and DN1b would appear to be the least mature cells and DN1e the most mature. These findings are so far consistent with pub-

Figure 2. Molecular Characterization of DN1 Subsets Defined by CD24 and CD117 Staining (A) DN1 subsets a–e, as indicated in Figure 1B, as well as DN2 cells or marrow-derived common lymphoid progenitors (CLP; lineage⫺ CD117⫹CD127⫹Sca-1⫹) were analyzed by RT-PCR for genes expressed early during the intrathymic differentiation process (RAG-1 or CD3⑀). Lane loading was assessed using a housekeeping gene (hypoxanthine guanine phosphoribosyl transferase, HPRT). In (B), the status of TCR␤ gene rearrangement was assessed by PCR using primers flanking the D-J␤ locus. The arrow indicates the relative position of the unrearranged locus.

lished findings that CD117⫹ cells represent canonical T cell progenitors, although the differences between CD24⫺, CD24lo, and CD24⫹ cells remain to be resolved. The possible relationships are further characterized from a functional standpoint as described in the following sections. Only DN1a and DN1b Cells Exhibit a Proliferative Burst Capacity Consistent with Being Canonical T Cell Progenitors Each thymus homing progenitor is estimated to produce approximately one million intrathymic progeny, representing about twenty serial cell divisions (reviewed in Shortman et al., 1990), with the greatest capacity for expansion residing in the earliest populations (DN1). To assess the proliferative capacity of each DN1 subset, a variety of analyses were performed (Figure 3). Relative DNA content can provide an estimate of the proliferative status of any cell population, since dividing cells must replicate their DNA and thus contain hyperdiploid (⬎2n) amounts. Overall, we found that all DN1 cells were cycling to some extent, consistent with previous reports (Tourigny et al., 1997; Penit et al., 1995; Shortman et al., 1990). However, there were marked differences among

Immunity 738

Figure 3. Proliferative Status and Growth Potential of DN1 Subsets In (A), relative cellular DNA content (DAPI staining) was determined by flow cytometry in DN1 subsets identified as shown in Figure 1B. DN2 thymocytes were included as a wellcharacterized control (see text). Values indicate the mean ⫾ standard deviation of hyperdiploid cells (S/G2/M phase) from three independent experiments, as an indicator of proliferative status. DAPI fluorescence is shown on a linear scale. (B) shows photomicrographs of equal numbers of each of these subsets, as well as bone marrow CLP or DN2 controls (precursors and progeny of DN1 cells, respectively), after culture for 6 days on stromal cells capable of supporting T lineage growth (OP9-DL1). Lymphoid cells appear as small phase bright spheres, while stromal cells are fibroblastoid in morphology and phase dull. Lymphoid growth completely obscures the presence of stromal cells in the case of DN1a or DN1b, as well as in control cultures (CLP or DN2). Relative expansion of each subset in such cultures at various time points is summarized in (C). Error bars indicate mean ⫾ standard deviation for two distinct experiments, except day 12, which is one experiment only.

the various DN1 populations. The majority population, DN1e, exhibited the lowest proportion of hyperdiploid cells. Since this population dominates the DN1 stage (Table 1), analysis of cell cycle activity in total DN1 cells would be expected to be heavily influenced by this subset. In contrast, the remaining DN1 subsets all appear to be cycling to a greater extent than DN1e, with the highest activity in DN1b and the second lowest in DN1a. To expand beyond a static assessment of proliferation, the ability of these cells to proliferate when cultured on a stromal line that supports T cell differentiation in vitro (Schmitt and Zu´n˜iga-Pflu¨cker, 2002) was measured. Figure 3B shows a photomicrograph comparing cell density 6 days after initiating such cultures with equal numbers of purified DN1 subsets a–e, as well as bone marrow CLP or DN2 thymocytes (as controls). DN2 and CLP grew well in such cultures, consistent with their known proliferative capacity and lineage potential. Among the DN1 subsets, however, dramatic differences were observed. Only DN1a and DN1b cells grew in a manner similar to DN2 or CLP and consistent with the proliferative capacity that accompanies canonical T cell progenitors (discussed above). Analysis of proliferative expansion at several different culture intervals confirms the differences among DN1 subsets, with only DN1a and

DN1b cells exhibiting substantial growth potential (Figure 3C). No detectable differences in cell viability were noted among the various DN1 populations cultured in vitro at early time points (data not shown), and since viable cells could be recovered from all cultures at all time points, we conclude that the differences in cell recovery were not primarily due to different rates of cell death. Notably, DN1c, despite being CD117⫹, differs substantially from the other CD117⫹ subsets (CD24⫺ and CD24lo, or DN1a and DN1b, respectively), in that it is unable to sustain a proliferative burst consistent with that of a canonical T cell progenitor. Additional differences between these CD117⫹ subsets, as well as the other components of DN1, are further characterized in the experiments described below. Kinetic Analysis of Differentiation along T Lineage Pathways Strengthens an Ordered Relationship between DN1a and DN1b Cells and Marrow-Derived Precursors or DN2 Progeny The experiments illustrated in Figures 1–3 suggest that of the five DN1 subsets identified by CD24 and CD117 staining, DN1a and DN1b appear to be most like canonical T cell progenitors, in terms of combined proliferative capacity, early T lineage gene expression, and TCR rear-

Heterogeneity of Early Intrathymic Progenitors 739

Figure 4. Comparative Kinetics of Differentiation of DN1 Subsets through Various Stages after Culture on Stromal Cells In Vitro The relative progression of each DN1 population (as identified in Figure 1B) was assessed by culture on stromal cells that support T lineage growth (OP9-DL1). CLP, DN1a, DN1b, and DN2 are shown in sequence to emphasize their apparent relationships to each other, as discussed in the text. Quadrants in 4A (CD44 x CD25) indicate the relative positions of DN1 (upper left), DN2 (upper right), DN3 (lower right), and more mature cells (preDP, DP, SP, lower left). Quadrants in 4B (CD4 x CD8) indicate the relative positions of DN (lower left), DP (upper right), and SP cells. In all cases, values shown indicate the mean ⫾ standard deviation for three independent experiments for each starting population and time point, except CLP on day 8, for which only one experiment was completed.

rangements. To further evaluate the capabilities of the various DN1 subsets, kinetic analysis along well-defined T cell differentiation pathways was examined by culturing purified cells on OP9-DL1 stromal monolayers. Figure 4A shows analysis of differentiation through the DN stages, identified by CD44 and CD25, while Figure 4B summarizes differentiation into more mature DP or CD4/

CD8 single-positive (SP) stages by staining with CD4 and CD8. Consistent with the conclusions of the experiments described earlier, DN1a and DN1b exhibited kinetics that trailed those of DN2 thymocytes but were faster than marrow-derived CLP, suggesting that they may lie intermediate to these. In addition, there appeared to be a precursor:progeny relationship between

Immunity 740

these two cell types, with DN1a preceding DN1b. For instance, at day 4 of culture, there were three times as many DN1a cells that remained at the DN1 stage as there were for DN1b; in contrast, there were only about a quarter as many cells that had progressed to DN3. By day 12, very few DN1a cells had exited the DN stage, as indicated by the absence of CD25⫺44lo as well as CD4⫹8⫹ cells. In contrast, about 15% of DN1b had traversed the DN/DP transition by this point (characterized by the presence of CD25⫺44lo cells), and a few percent had even acquired CD4 and CD8 levels consistent with typical DP cells. Both of these populations ultimately differentiate to the DP stage at later time points (see Supplemental Figure S1 at http://www.immunity.com/ cgi/content/full/20/6/735/DC1) and even generate some SP cells, consistent with the original description of this culture system (Schmitt and Zu´n˜iga-Pflu¨cker, 2002). With regard to the other DN1 subsets, the kinetics of differentiation for DN1c appeared very similar to that of DN1a and DN1b. Analysis of gene expression and TCR rearrangement also suggested the possibility that these three CD117⫹ subsets were quite similar. However, as described earlier, DN1c lacks the proliferative potential required of a canonical T cell progenitor (Figure 3). This was not an artifact of the culture system, since two experiments using fetal thymic organ culture (Hare et al., 1999) instead of OP9-DL1 also failed to yield measurable growth from DN1c (or subsets d or e; data not shown), suggesting that the growth potential of these does, in fact, differ from that of the other CD117⫹ DN1 subsets. Given this poor growth, it is also possible that a few contaminating DN1b cells could dominate the analysis of purified DN1c. However, additional characterization of the differences in lineage potential between the various CD117⫹ subsets of DN1, as provided in the subsequent section of Results, suggests that this is not the primary explanation for the differences and that each of these, as well as the other DN1 populations defined in Figure 1, are bona fide subsets with discrete potentials. Consistent with the findings shown in Figures 1–3, the characteristics of DN1d and DN1e subsets do not appear to represent a canonical T cell progenitor intermediate between CLP and DN2. For instance, the kinetics of DN1d differentiation are not only faster than that of other DN1 subsets but are substantially faster than DN2. DN1e are highly atypical in that they do not appear to follow the conventional T cell differentiation pathway at all, in combination with the other differences shown in Figures 1–3. Although these latter cells appear similar to CLP and DN2 in terms of CD127 expression, they differ in their expression of CD117 (Figure 1). These findings are consistent with recent conclusions of others (Allman et al., 2003), showing that CD127⫹ DN1 do not appear to contain canonical T cell progenitors that give rise to an expanded population of thymocytes with normal differentiation kinetics. Overall, these data suggest that DN1a and DN1b form a precursor:progeny pair that is intermediate between thymic homing progenitors and DN2, while the behavior of DN1c, DN1d, and DN1e is inconsistent with such a function. It is important to note that omission of anti-B220 antibody from the lineage depletion cocktail did not change the kinetics of differentiation or the apparent amount of proliferative expansion for any of the DN1 subsets (data not shown). Thus, if a B220⫹ progenitor does give rise to all other intrathymic

progenitors, as indicated by the work of others (Martin et al., 2003), it may be too rare to appear within the time frame used in these experiments. DN1 Subsets that Most Closely Resemble T Lineage Precursors Exhibit the Potential to Generate Cells Expressing NK but Not B Lineage Markers Very early thymocyte progenitors are believed to maintain the capacity to generate non-T lineages. To further assess the relationships of DN1 subsets to each other, their ability to generate cells expressing non-T lineage markers was tested after culture on stromal cells that do not specify the T lineage fate (OP9 control). NK1.1 expression was found on cells derived from CLP, DN1a, and DN1b, and to a limited extent, DN2 (Figure 5). The remaining DN1 subsets appeared to have little potential to generate cells expressing this NK lineage marker. Consistent with previous findings (Kondo et al., 1997), marrow-derived CLP could give rise to cells expressing B lineage markers; in contrast, DN2 cells could not. However, the generation of cells expressing a marker of the B lineage (CD19) differed greatly among DN1 cells and largely contrasted NK potential. DN1a, DN1b, and DN1e did not generate CD19⫹ cells, while cells expressing this marker were generated from DN1c and DN1d. Together with the findings presented previously, this data suggests that canonical T cell progenitors, represented by DN1a and DN1b, may be restricted to T and NK lineages prior to thymus entry, a conclusion that is completely consistent with a number of previous reports (Ikawa et al., 1999; Michie et al., 2000). Our combined findings also suggest that the B cell potential of CD117⫹ cells as described by others (Allman et al., 2003) may derive solely from the noncanonical DN1c subset, which is similar to DN1a and DN1b in many other respects, but does not expand like a normal T cell progenitor. Analysis of Thymic Homing Capacity Supports the Canonical Progenitor Status of DN1a and DN1b Many different types of progenitor cells can generate T lineage progeny when injected intrathymically, but only very early progenitors are capable of homing to the thymus (Pearse et al., 1989). Irradiation or other methods of cytoreduction may render the thymus permissive to homing of cells that home poorly under normal circumstances. Thus, to strengthen the proposed relationship of DN1a and DN1b subsets to each other and to other progenitors, purified subsets of DN1 or DN2 cells were injected intravenously into nonirradiated congenic recipients and analyzed 20 days later for the presence of donor cells in the thymus. This time point was chosen to coincide with the approximate peak of expansion from early progenitors (Porritt et al., 2003; Shortman et al., 1990). No donor progeny were observed when DN1c, DN1d, DN1e, or DN2 cells were transplanted (data not shown). However, both DN1a and DN1b cells were able to home to the thymus and generate detectable levels of progeny (Figure 6) that differentiated with normal kinetics. CLP also homed to the thymus and differentiated in this system (data not shown; see description in Porritt et al., 2003). Given the poor proliferative potential of DN1c, DN1d, and DN1e (Figure 3), it was possible that they homed to the thymus but did not generate adequate numbers to be detected above the overwhelming back-

Heterogeneity of Early Intrathymic Progenitors 741

ground of recipient cells; this is supported by the studies described in the next section. In either case, this and the other data presented in this manuscript make it appear unlikely that these represent typical precursors to normal T cells, even though they apparently can give rise to cells of the T lineage (Figure 4). Further, these data strongly suggest that CD24 expression identifies two subsets of CD117⫹ DN1 cells that do appear to be typical T cell precursors and, further, that the CD24⫺ component, DN1a, is likely to represent the earliest identifiable intrathymic progenitor. Analysis of the Phenotype and Lineage Potential of Bone Marrow-Derived Cells that Rapidly Home to the Thymus The numbers of cells that home to the thymus on a daily basis are extremely limited (Wallis et al., 1975; Kadish and Basch, 1976; Spangrude and Scollay, 1990; Donskoy and Goldschneider, 1992), making analysis of short-term homing progenitors difficult. Nonetheless, when unfractionated bone marrow was administered intravenously to nonirradiated congenic recipients, all DN1 subsets were found within the thymus at 24 hr (Figure 6C), although the proportions were different than that of the steady-state. The CD117⫺ phenotype of the majority population is consistent with the findings of others (Mori et al., 2001), although we do find a very small proportion of CD117⫹ cells (DN1a, DN1b, and DN1c) that those authors did not; the reason for this discrepancy is unclear but may be related to the very small numbers of CD117⫹ cells that home in these short-term assays. The differences in proportion between steady-state and ectopic (intravenous) infusion may reflect the presence of marrow marrow-derived cells with high thymic homing potential that ordinarily do not have access to the bloodstream and/or thymus. To further characterize these short-term thymic homing subsets, they were purified and cultured on OP9DL1 or control OP9 stromal cells. Based on the studies shown in Figures 3 and 4, atypical thymus homing progenitors (i.e., DN1c, d, and e) derived from transplanted marrow would not be expected to generate appreciable progeny such as in vitro cultures, and that was found to be the case (data not shown). Due to the extremely small numbers of cells that home to the thymus in such experiments, it was not possible to purify sufficient DN1a for further testing. However, in several experiments, a small number (8–20) of cells with the pooled DN1a/DN1b phenotype (i.e., CD24⫺/lo117⫹) could be isolated from recipient thymuses. As shown in Figure 6, these cells grew and gave rise largely to T cells (revealed by CD4/8 staining at day 28) as well as some NK cells (also shown at day 28) when cultured on OP9-DL1. When cultured under B cell conditions using OP9 control cells, no detectable donor progeny could be found, suggesting that these cells do not respond to B cell stimuli and probably do not have B cell potential, consistent with the studies shown in Figure 5. Discussion The heterogeneity of DN1 cells is widely known (reviewed in Ceredig and Rolink, 2002; Pear et al., 2004), albeit frequently overlooked. Notably, many nonhemato-

Figure 5. B and NK Lineage Potential of DN1 Subsets The ability of DN1 populations (as defined in Figure 1B) to express markers that characterize B lineage cells (CD19) or NK cells (NK1.1) was tested by culturing them on stromal cells that do not induce T lineage fate (OP9 control). Control cells were bone marrow CLP or DN2 thymocytes. CLP, DN1a, DN1b, and DN2 are shown in sequence to emphasize their apparent relationships to each other, as discussed in the text. Results are for day 6 cultures, and values shown are the mean of two independent experiments. Different experiments analyzed at other time points gave similar results for absolute lineage potential, although the proportions of cells expressing lineage markers differed in some cases.

poietic cells, including thymic stromal cells, express CD44, and since non-T cells are also negative for T lineage markers such as CD4, CD8, and CD25, any CD44⫹ non-T lineage cells in the thymus would be found within the DN1 region. Even when utilizing other markers of the T and/or hematopoietic lineages, the heterogeneity of DN1 is reflected in the variety of phenotypes ascribed to this subset. DN1 T lymphocyte progenitors have been defined by different groups as being positive, intermediate, or negative for a variety of markers, including CD24, CD90, and CD127 (see the first section of the Results). Although most contemporary definitions would suggest that canonical DN1 progenitors are CD117⫹, other recent findings have suggested the possibility that the earliest intrathymic progenitors may be CD117⫺ (Martin et al., 2003). One of the complicating factors in coming to a consensus is the rarity of such cells in the thymus; since each intrathymic progenitor expands approximately one million fold before proliferation stops at the DP stage (Shortman et al., 1990), it is obvious that DN1 cells must be extremely rare, even though there is some amplification of these cells prior to differentiation into DN2 (Shortman et al., 1990; Tourigny et al.,

Immunity 742

1997; Penit et al., 1995; Porritt et al., 2003). Thus, comparative analysis of the lineage and/or proliferation potential of even smaller subsets of such rare cells has been extremely difficult and kinetic analysis of their differentiation nearly impossible. The advent of an in vitro system that supports T cell differentiation, via a bone marrow stromal line (OP9) expressing the Notch ligand DL1, has allowed a more straightforward approach to answer this question (Zu´n˜iga-Pflu¨cker, 2004). To begin this analysis, we identified various populations of DN1 cells expressing hematolymphoid lineage markers, including CD24 (HSA), CD117 (c-kit), and CD127 (IL-7R␣). We found five distinct populations that were best distinguished using a CD24 x CD117 bivariate plot (Figure 1). Importantly, all of these populations appear to be capable of giving rise to T lineage cells, as evidenced by the presence of cells expressing CD4 and/or CD8 when cultured on T cellsupporting stroma (Figure 4). This may help to explain how various progenitors with apparently conflicting phenotypes (as discussed above) might all appear to function as T cell progenitors. However, major differences were seen among these various subsets in their kinetics of differentiation, their lineage potential, and their ability to undergo proliferative expansion, indicating that only some of these DN1 subsets are canonical progenitors to T cells, as discussed in further detail below. As recently described (Perry et al., 2003), orthodox prothymocytes must possess the qualities of rapid thymic engraftment kinetics following intravenous transplantation, dramatic expansion of thymic progeny, and limited production of hematopoietic progeny other than thymocytes. On the basis of these and other related criteria, only two of the three CD117⫹ DN1 subsets, namely DN1a (CD24⫺) and DN1b (CD24lo) appear to be traditional progenitors. These cells have relatively high proliferative expansion potential (Figure 3), and exhibit kinetics of differentiation intermediate between marrow progenitors and DN2 (Figure 4). These are the only DN1 subsets that appear to home to the thymus and give rise to intrathymic progeny after intravenous injection (Figure 6), although given the relatively poor expansion that characterizes other DN1 subsets (Figure 3), it is possible that DN1c-e home to the thymus but fail to give rise to detectable levels of progeny. Not only do DN1a and DN1b fulfill the criteria for canonical T cell progenitors, but they also appear to form a precursor:progeny pair, with DN1a closer in kinetics to marrowderived progenitors and DN1b closer to DN2. DN1a also have slightly higher proliferative burst potential, consistent with being a less mature progenitor than DN1b. Further, the numeric frequency of DN1a and 1b cells supports a precursor:progeny relationship (Table 1), and the number of DN1a is very consistent with what would be expected of the earliest intrathymic progenitors, given what is known about the frequency of thymic homing (Wallis et al., 1975; Kadish and Basch, 1976; Spangrude and Scollay, 1990; Donskoy and Goldschneider, 1992). Finally, although analysis at very early time points was impaired by the very small numbers of cells that can be obtained, culture of DN1a cells on OP9-DL1 stroma indicates that most of them acquire the DN1b phenotype by 4 days of culture (data not shown). Thus, our findings indicate that DN1a cells (CD24⫺117⫹127⫺)

Figure 6. Thymus Homing and Lineage Capacity of Progenitor Cells after Intravenous Injection into Nonirradiated Recipients (A) Purified DN1 subsets were injected intravenously into nonirradiated Ly-5.1⫹ congenic recipients and analyzed 20 days later for the presence of donor progeny. Only DN1a and DN1b cells generated detectable donor progeny in this system. Donor cells at all stages of intrathymic differentiation (as defined by CD4 and/or CD8 expression) were found after 20 days (B), consistent with known kinetics of intrathymic differentiation (see text). When unfractionated marrow cells were used as donors (C), the phenotype of cells homing to the thymus within 24 hr recapitulates the overall DN1 subsets found in normal thymus (see Figure 1B). Culture on OP9 stromal cells expressing DL1 (for T/NK lineage potential) or control OP9 cells (for B lineage potential) showed that canonical T progenitors (CD24⫺/lo117⫹) derived from these short-term chimeras had T and NK potential, but not B potential ([D] and text). Results shown are characteristic of two experiments performed for each type.

may represent the earliest intrathymic progenitor with canonical T lineage potential. We cannot exclude the possibility that such cells arise from an even more rare CD117⫺ progenitor, as suggested by others (Martin et al., 2003); although we do not see substantial production of T lineage cells from the CD117⫺ subsets of DN1 (Figure 3), it is possible that an extremely rare cell might not grow out in time to be detected. Again, however, we would point out that the numbers of DN1a cells found in the thymus (Table 1) are consistent with what might be expected of very early progenitors arriving from the marrow, suggesting that a more rare precursor may not be required. It is also possible that OP9 cells expressing DL1 are insufficient to induce T lineage commitment and expansion from putative CD117⫺ progenitors, although we point out that such conditions are sufficient to induce

Heterogeneity of Early Intrathymic Progenitors 743

even less mature precursors (CLP [Figure 4] or stem cells from bone marrow or fetal liver [Schmitt and Zu´n˜igaPflu¨cker, 2002]) into the T cell pathway. CD117 expression has recently been accepted as definitive of lymphoid progenitors in the thymus (Ceredig and Rolink, 2002), although as mentioned above, this has been challenged (Martin et al., 2003). Our data indicate that canonical T cell progenitors do reside within the CD117⫹ subset of DN1. However, at least half of CD117⫹ DN1 cells are represented by the subset we denote as DN1c (Figure 1 and Table 1). These give rise to T cells with kinetics that are very similar to the other CD117⫹ subsets (Figure 4). However, they have very poor proliferative capacity (Figure 3) and fail to generate detectable thymic progeny after intravenous injection (Figure 6), suggesting that unlike their DN1a and DN1b counterparts, they do not represent traditional thymocyte progenitors. Even more importantly, these appear to be the singular source of CD117⫹ cells with potential to generate cells expressing B lineage markers (Figure 5). DN1c cells would be included among early thymocyte progenitors (ETP) as defined by others (Allman et al., 2003), where B cell potential was found. Thus, our data suggests that isolation of true DN1 progenitors requires exclusion of the CD24hi component of CD117⫹ cells; doing so yields cells with only T and NK potential (dendritic cell potential was not tested), consistent with the findings of others (Michie et al., 2000; Ikawa et al., 1999). Further, our findings suggest that the ideal ETP must be further restricted to CD24⫺ cells that are CD117⫹127⫺ (i.e., DN1a), yielding cells with the slowest kinetics of differentiation, the highest proliferative burst potential, and the ability to home to the thymus after intravenous transplantation. Similar to the findings of others (Igarashi et al., 2002; Allman et al., 2003), we find that most CD127⫹ cells appear to reside within the CD117⫺ population (Figure 1). In particular, they are mainly restricted to the CD24⫺ subset of DN1. These cells are the most atypical of the noncanonical DN1 subsets in their kinetics of differentiation and in their transit (or lack thereof) through the intermediate stages of thymocyte differentiation (Figure 4). These also are the majority population within DN1 (Figure 1 and Table 1) and, since they appear to have relatively few cells in the S/G2/M phases of cell cycle (Figure 3), would be expected to heavily skew the overall cell cycle status of DN1 toward the widely accepted noncycling state. Our findings regarding the relative proliferative rate of DN1 subsets are also consistent with the findings of others (Penit et al., 1995), showing a large proportion of cycling cells among the CD117⫹ population and a relatively low proportion among cells defined as CD24⫺/lo. Other than this, the relative role of the CD117⫺ cells (i.e., DN1d and DN1e) in the thymus is not clear, except that they do not appear to contribute significantly to canonical T cell development. In addition to the definition of a very early intrathymic progenitor (DN1a), the demonstration of a precursor:progeny relationship (DN1a and DN1b) within the DN1 stage, and the identification of a distinct subset of CD117⫹ DN1 with B lineage potential (DN1c), our data generate intriguing insights into the interpretation of mouse models with an apparent DN1 arrest. For instance, Notch-1 strongly specifies T lineage fate (for

examples see Pui et al., 1999; Radtke et al., 1999; Deftos et al., 2000), and thymus-specific deletion of Notch-1 leads to an apparent DN1 arrest, with expression of B cell markers by those DN1 that persist (Radtke et al., 1999). Those cells were also CD24hi, suggesting that they may in fact represent either our DN1c or DN1d subsets (Figure 1), both of which have the potential to generate cells expressing B lineage markers (Figure 5) but probably do not derive from the more typical progenitors represented by DN1a and DN1b. Given the extremely small number of DN1a thymocytes present even in a normal thymus (Table 1), it is possible that these are present among Notch-deleted thymocytes but remain undetected since they cannot expand and/or differentiate into DN1b, DN2, or later stages. This would not change the interpretation of the role of Notch in T cell development but may significantly impact the nature of the B cells that apparently arise in the thymus in the absence of Notch signaling. The nature and source of DN1 cells present in thymuses carrying deficiencies in a number of other genes, such as sfpi1 (PU.1, Anderson et al., 2002), foxn1 (Su et al., 2003), and others may likewise warrant a reexamination, based on the findings presented here. Overall, these considerations reemphasize the concept that definition of a DN1 cell requires additional qualifications and that many cells with a DN1 phenotype, even those expressing CD117, may not be equivalent to those progenitors that represent conventional progenitors to more mature thymocytes and peripheral T cells. Experimental Procedures Cell Purification and Analysis C57BL/6 mice (Jackson Laboratories, Bar Harbor, ME) at 4–5 weeks of age were used to generate starting populations for all experiments. Prior to lineage depletion, thymocyte suspensions were depleted of most small DP cells by centrifugation at 1100 ⫻ g using a gradient of 13.6% Opti-Prep (Greiner Bio-One, Longwood, FL) in mouse-tonicity Hank’s balanced salt solution; bone marrow suspensions were depleted of red blood cells by hypotonic lysis in 0.15 M NH4Cl, 1 mM KHCO3, 0.1 mM EDTA. Lineage depletion was performed by treating cells with a cocktail of rat antibodies recognizing mouse CD3, CD4, CD8, CD11b, CD19, B220, Gr-1, TER-119, followed by goat anti-rat coated paramagnetic beads, as previously described (Porritt et al., 2003). Cells were sorted on a three laser FACS DiVa (BD Biosciences, San Jose, CA). Sort gates were defined as follows. All starting populations were viable as defined by DAPI exclusion (Gordon et al., 2003) and were lineage negative (as described above) as well as side scatter low. Doublets were excluded by discriminating forward scatter (FSC) area X width pulses. DN1 thymocytes were CD25⫺44⫹, as well as various CD117 (clone ACK2) and CD24 (clone M1/69) designations as depicted in Figure 1. DN2 thymocytes were CD25⫹44⫹117⫹. Bone marrow cells were Sca-1⫹ CD117⫹127⫺ (stem cells) or Sca-1⫹ CD117⫹127⫹ (common lymphoid progenitors, CLP). In some cases, fluorochrome-conjugated antirat antibody was used to confirm the lineage⫺ status of starting populations prepared by sorting, but no differences were observed when this step was omitted. Analysis of surface phenotype or cell cycle was performed on an LSR cytometer (BD Biosciences) with custom modifications as described (Gordon et al., 2003). All antibodies were produced, purified, and conjugated to fluorochromes at Memorial Sloan-Kettering Cancer Center. Molecular Analysis RNeasy mini kits (Qiagen, Valencia, CA) were used to isolate total cellular RNA from sorted progenitor cells. Reverse transcription and PCR amplification were carried out using the SuperScript One-Step

Immunity 744

RT-PCR with Platinum Taq (Invitrogen, Carlsbad, CA). cDNA synthesis was performed at 55⬚C for 30 min. PCR amplification step was performed for 28 or 33 cycles at 94⬚C. CD3⑀ primer sequences were as follows (5⬘-3⬘): forward, CTGGGACATTGCTGACTCAA; reverse, AGGGCCAATTAGGAGAGGAA. RAG-1 (Igarashi et al., 2001) or HPRT (Lind et al., 1999) primers were as described. Amplification products were analyzed after electrophoresis in 2.2% metaphor agarose (FMC, Rockland, ME) and staining with ethidium bromide. Image analysis was performed using a Bio-Rad GelDoc 1000 with Molecular Analyst software (Bio-Rad, Hercules, CA). For DJ␤ recombination analysis, DNA was isolated from sorted cells using a GenomicPrep DNA isolation kit (Amersham, Piscataway, NJ). Approximately 100 ng DNA was used as template in PCR reaction with primers as previously described (King et al., 2002): D␤1.1 (forward), GAGGAGCAGCTTATCTGGTG; J␤1.7 (reverse), AAGGGACGACTC TGTCTTAC. DNA was amplified for 30 cycles prior to electrophoresis and staining as described above. Coculture of Progenitor and Stromal Cells OP9-control cells or OP9 cells transfected with delta-like-1 (DL1) were used as previously published (Schmitt and Zu´n˜iga-Pflu¨cker, 2002). Sorted DN1 subsets, DN2 cells, or CLPs were seeded into 24-well tissue culture plates containing a near-confluent monolayer of stromal cells. T cell differentiation was induced by culture on OP9DL1 in the presence of IL-7 (5 ng/ml) and Flt3L (5 ng/ml). Addition of IL-7 was required, since DN2 cells express IL-7R (see Figure 1) and require IL-7 for survival and/or proliferation. For analysis of B potential, control OP9 stromal cells (lacking DL1) were used under the same conditions as T cell cultures. For analysis of NK cell potential, control OP9 cells were used, and IL-6 (1 ng/ml) and IL-15 (25 ng/ml) were also added to the medium. Cocultures were harvested by forceful pipetting at the indicated time points. Stromal cells were excluded by gating on GFP⫺ and, in some cases, CD45⫹ cells with lymphoid FSC/SSC characteristics. All cytokines were purchased from Peprotech (Rocky Hill, NJ). Analysis of Thymus Homing and Generation of Nonirradiation Chimeras Purified donor cells were prepared from C57BL/6 (Ly-5.2⫹) mice as described above and injected intravenously into nonirradiated B6.129Ptprc mice (Ly-5.1⫹, Jackson). Cells were injected in approximately the same ratios found ex vivo (see Table 1), starting with 2.5 ⫻ 104 of the least frequent CD25⫺44⫹117⫹ subset. Recipient mice were used precisely at 4.5 weeks of age to correspond with the peak of responsiveness to thymic homing (Foss et al., 2001). After 20 days, thymuses were harvested and were subjected to a low-stringency depletion of recipient cells, using a mouse antiLy-5.1 antibody followed by rat anti-mouse IgG and goat anti-rat paramagnetic beads (Dynal M450) at a ratio of 0.8 beads per cell. Remaining cells were stained with a combination of fluorochromeconjugated antibodies recognizing donor cells (Ly-5.2) and either CD44 x CD25 or CD4 x CD8. Doublets and dead cells were excluded as described above. Differentiation and Lineage Potential of CD24lo117ⴙ Progenitors from Thymus after Bone Marrow Transplant 2.5 ⫻ 107 whole marrow cells from Ly5.1⫹ mice were injected intravenously into nonirradiated C57BL/6 recipients. After 24 hr, the recipient thymuses were removed and depleted of most recipient cells by complement lysis after treatment with an anti-CD8 antibody, as described (Carlyle and Zu´n˜iga-Pflu¨cker, 1998). The remaining cells were stained for markers to identify donor cells (Ly-5.1) as well as CD24 and CD117. CD24lo117⫹ donor cells were isolated by cell sorting. These cells were then cultured for 14–28 days on OP9-DLL or OP9 control cells, and stained and analyzed for lineage markers, as described above. Acknowledgments The authors express their grateful appreciation to Drs. Gerald Spangrude (University of Utah) and Avinash Bhandoola (University of Pennsylvania) for their comments prior to submission, as well as to other colleagues, too numerous to mention here, for helpful discus-

sion. This work was supported by PHS grants AI33940 and AI53739 (to H.T.P.), as well as by funds from Memorial Sloan-Kettering Cancer Center. J.C.Z.-P. is supported by an Investigator Award from the Canadian Institutes of Health Research. Received: January 21, 2004 Revised: April 1, 2004 Accepted: April 14, 2004 Published: June 15, 2004 References Allman, D., Sambandam, A., Kim, S., Miller, J.P., Pagan, A., Well, D., Meraz, A., and Bhandoola, A. (2003). Thymopoiesis independent of common lymphoid progenitors. Nat. Immunol. 4, 168–174. Anderson, M.K., Weiss, A.H., Hernandez-Hoyos, G., Dionne, C.J., and Rothenberg, E.V. (2002). Constitutive expression of PU.1 in fetal hematopoietic progenitors blocks T cell development at the pro-T cell stage. Immunity 16, 285–296. Bhandoola, A., Sambandam, A., Allman, D., Meraz, A., and Schwarz, B. (2003). Early T lineage progenitors: new insights, but old questions remain. J. Immunol. 171, 5653–5658. Carlyle, J.R., and Zu´n˜iga-Pflu¨cker, J.C. (1998). Requirement for the thymus in alphabeta T lymphocyte lineage commitment. Immunity 9, 187–197. Ceredig, R., and Rolink, T. (2002). A positive look at double-negative thymocytes. Nat. Rev. Immunol. 2, 888–897. Crispe, I.N., Moore, M.W., Husmann, L.A., Smith, L., Bevan, M.J., and Shimonkevitz, R.P. (1987). Differentiation potential of subsets of CD4⫺8⫺ thymocytes. Nature 329, 336–339. Deftos, M.L., Huang, E., Ojala, E.W., Forbush, K.A., and Bevan, M.J. (2000). Notch1 signaling promotes the maturation of CD4 and CD8 SP thymocytes. Immunity 13, 73–84. Donskoy, E., and Goldschneider, I. (1992). Thymocytopoiesis is maintained by blood-borne precursors throughout postnatal life. A study in parabiotic mice. J. Immunol. 148, 1604–1612. Foss, D.L., Donskoy, E., and Goldschneider, I. (2001). The importation of hematogenous precursors by the thymus is a gated phenomenon in normal adult mice. J. Exp. Med. 193, 365–374. Godfrey, D.I., and Zlotnik, A. (1993). Control points in early T-cell development. Immunol. Today 14, 547–553. Godfrey, D.I., Zlotnik, A., and Suda, T. (1992). Phenotypic and functional characterization of c-kit expression during intrathymic T cell development. J. Immunol. 149, 2281–2285. Godfrey, D.I., Kennedy, J., Suda, T., and Zlotnik, A. (1993). A developmental pathway involving four phenotypically and functionally distinct subsets of CD3⫺CD4⫺CD8⫺ triple-negative adult mouse thymocytes defined by CD44 and CD25 expression. J. Immunol. 150, 4244–4252. Gordon, K.M., Duckett, L., Daul, B., and Petrie, H.T. (2003). A simple method for detecting up to five immunofluorescent parameters together with DNA staining for cell cycle or viability on a benchtop flow cytometer. J. Immunol. Methods 275, 113–121. Gounari, F., Aifantis, I., Martin, C., Fehling, H.J., Hoeflinger, S., Leder, P., von Boehmer, H., and Reizis, B. (2002). Tracing lymphopoiesis with the aid of a pTalpha-controlled reporter gene. Nat. Immunol. 3, 489–496. Hare, K.J., Jenkinson, E.J., and Anderson, G. (1999). In vitro models of T cell development. Semin. Immunol. 11, 3–12. Hozumi, K., Kobori, A., Sato, T., Nozaki, H., Nishikawa, S., Nishimura, T., and Habu, S. (1994). Pro-T cells in fetal thymus express c-kit and RAG-2 but do not rearrange the gene encoding the T cell receptor beta chain. Eur. J. Immunol. 24, 1339–1344. Igarashi, H., Kuwata, N., Kiyota, K., Sumita, K., Suda, T., Ono, S., Bauer, S.R., and Sakaguchi, N. (2001). Localization of recombination activating gene 1/green fluorescent protein (RAG1/GFP) expression in secondary lymphoid organs after immunization with T-dependent antigens in rag1/gfp knockin mice. Blood 97, 2680–2687. Igarashi, H., Gregory, S.C., Yokota, T., Sakaguchi, N., and Kincade,

Heterogeneity of Early Intrathymic Progenitors 745

P.W. (2002). Transcription from the RAG1 locus marks the earliest lymphocyte progenitors in bone marrow. Immunity 17, 117–130. Ikawa, T., Kawamoto, H., Fujimoto, S., and Katsura, Y. (1999). Commitment of common T/natural killer (NK) progenitors to unipotent T and NK progenitors in the murine fetal thymus revealed by a single progenitor assay. J. Exp. Med. 190, 1617–1626. Kadish, J.L., and Basch, R.S. (1976). Hematopoietic thymocyte precursors. I. Assay and kinetics of the appearance of progeny. J. Exp. Med. 143, 1082–1099. Kincade, P.W., Owen, J.J., Igarashi, H., Kouro, T., Yokota, T., and Rossi, M.I. (2002). Nature or nurture? Steady-state lymphocyte formation in adults does not recapitulate ontogeny. Immunol. Rev. 187, 116–125. King, A.G., Kondo, M., Scherer, D.C., and Weissman, I.L. (2002). Lineage infidelity in myeloid cells with TCR gene rearrangement: a latent developmental potential of proT cells revealed by ectopic cytokine receptor signaling. Proc. Natl. Acad. Sci. USA 99, 4508– 4513. Kondo, M., Weissman, I.L., and Akashi, K. (1997). Identification of clonogenic common lymphoid progenitors in mouse bone marrow. Cell 91, 661–672. Lind, E.F., Wayne, J., Wang, Q.Z., Staeva, T., Stolzer, A., and Petrie, H.T. (1999). Bcl-2-induced changes in E2F regulatory complexes reveal the potential for integrated cell cycle and cell death functions. J. Immunol. 162, 5374–5379. Martin, C.H., Aifantis, I., Scimone, M.L., von Andrian, U.H., Reizis, B., von Boehmer, H., and Gounari, F. (2003). Efficient thymic immigration of B220⫹ lymphoid-restricted bone marrow cells with T precursor potential. Nat. Immunol. 4, 866–873. Matsuzaki, Y., Gyotoku, J., Ogawa, M., Nishikawa, S., Katsura, Y., Gachelin, G., and Nakauchi, H. (1993). Characterization of c-kit positive intrathymic stem cells that are restricted to lymphoid differentiation. J. Exp. Med. 178, 1283–1292. Michie, A.M., Carlyle, J.R., Schmitt, T.M., Ljutic, B., Cho, S.K., Fong, Q., and Zu´n˜iga-Pflu¨cker, J.C. (2000). Clonal characterization of a bipotent T cell and NK cell progenitor in the mouse fetal thymus. J. Immunol. 164, 1730–1733. Mori, S., Shortman, K., and Wu, L. (2001). Characterization of thymus-seeding precursor cells from mouse bone marrow. Blood 98, 696–704. Pear, W.S., Tu, L., and Stein, P.L. (2004). Lineage choices in the developing thymus: choosing the T and NK pathways. Curr. Opin. Immunol. 16, 167–173. Pearse, M., Wu, L., Egerton, M., Wilson, A., Shortman, K., and Scollay, R. (1989). A murine early thymocyte developmental sequence is marked by transient expression of the interleukin 2 receptor. Proc. Natl. Acad. Sci. USA 86, 1614–1618. Penit, C., Lucas, B., and Vasseur, F. (1995). Cell expansion and growth arrest phases during the transition from precursor (CD4⫺8⫺) to immature (CD4⫹8⫹) thymocytes in normal and genetically modified mice. J. Immunol. 154, 5103–5113. Perry, S.S., Pierce, L.J., Slayton, W.B., and Spangrude, G.J. (2003). Characterization of thymic progenitors in adult mouse bone marrow. J. Immunol. 170, 1877–1886. Plotkin, J., Prockop, S.E., Lepique, A., and Petrie, H.T. (2003). Critical role for CXCR4 signaling in progenitor localization and T cell differentiation in the postnatal thymus. J. Immunol. 171, 4521–4527. Ponta, H., Sherman, L., and Herrlich, P.A. (2003). CD44: from adhesion molecules to signalling regulators. Nat. Rev. Mol. Cell Biol. 4, 33–45. Porritt, H.E., Gordon, K., and Petrie, H.T. (2003). Kinetics of steadystate differentiation and mapping of intrathymic-signaling environments by stem cell transplantation in nonirradiated mice. J. Exp. Med. 198, 957–962. Pui, J.C., Allman, D., Xu, L., DeRocco, S., Karnell, F.G., Bakkour, S., Lee, J.Y., Kadesch, T., Hardy, R.R., Aster, J.C., and Pear, W.S. (1999). Notch1 expression in early lymphopoiesis influences B versus T lineage determination. Immunity 11, 299–308. Radtke, F., Wilson, A., Stark, G., Bauer, M., van Merrwijk, J., Mac-

Donald, H.R., and Aguet, M. (1999). Deficient T cell fate specification in mice with an induced inactivation of Notch1. Immunity 10, 547–558. Schmitt, T.M., and Zu´n˜iga-Pflu¨cker, J.C. (2002). Induction of T cell development from hematopoietic progenitor cells by delta-like-1 in vitro. Immunity 17, 749–756. Shortman, K., Wilson, A., Pearse, M., Gallagher, P., and Scollay, R. (1988). Mouse strain differences in subset distribution and T cell antigen receptor expression among CD4⫺CD8⫺ thymocytes. Immunol. Cell Biol. 66, 423–433. Shortman, K., Egerton, M., Spangrude, G.J., and Scollay, R. (1990). The generation and fate of thymocytes. Semin. Immunol. 2, 3–12. Spangrude, G.J., and Scollay, R. (1990). Differentiation of hematopoietic stem cells in irradiated mouse thymic lobes. Kinetics and phenotype of progeny. J. Immunol. 145, 3661–3668. Su, D.M., Navarre, S., Oh, W.J., Condie, B.G., and Manley, N.R. (2003). A domain of Foxn1 required for crosstalk-dependent thymic epithelial cell differentiation. Nat. Immunol. 4, 1128–1135. Tourigny, M.R., Mazel, S., Burtrum, D.B., and Petrie, H.T. (1997). T cell receptor (TCR)-beta gene recombination: dissociation from cell cycle regulation and developmental progression during T cell ontogeny. J. Exp. Med. 185, 1549–1556. Wallis, V.J., Leuchars, E., Chwalinski, S., and Davies, A.J. (1975). On the sparse seeding of bone marrow and thymus in radiation chimaeras. Transplantation 19, 2–11. Wang, B., Biron, C., She, J., Higgins, K., Sunshine, M.J., Lacy, E., Lonberg, N., and Terhorst, C. (1994). A block in both early T lymphocyte and natural killer cell development in transgenic mice with highcopy numbers of the human CD3E gene. Proc. Natl. Acad. Sci. USA 91, 9402–9406. Wilson, A., and MacDonald, H.R. (1995). Expression of genes encoding the pre-TCR and CD3 complex during thymus development. Int. Immunol. 7, 1659–1664. Wilson, A., D’Amico, A., Ewing, T., Scollay, R., and Shortman, K. (1988). Subpopulations of early thymocytes. A cross-correlation flow cytometric analysis of adult mouse Ly-2⫺L3T4⫺(CD8⫺CD4⫺) thymocytes using eight different surface markers. J. Immunol. 140, 1461–1469. Wilson, A., Held, W., and MacDonald, H.R. (1994). Two waves of recombinase gene expression in developing thymocytes. J. Exp. Med. 179, 1355–1360. Zu´n˜iga-Pflu¨cker, J.C. (2004). Innovation: T-cell development made simple. Nat. Rev. Immunol. 4, 67–72.