Complex regulation of CCR9 at multiple discrete stages of T cell development



We have conducted a comprehensive assessment of CCR9 expression and function at the important milestone stages of murine thymocyte development. We reveal an unusually complex regulatory pattern, in which CCR9 influences T cell development at several widely dispersed stages. We find that CCR9 is not expressed within the thymus until the double-negative (DN)3 stage, although it appears to contribute to T cell precursor development prior to residence in the thymus. CCR9 expression is influenced by pre-T cell receptor signals, and is dramatically up-regulated in a population that appears to be transitional between the DN4 and double-positive stages. In the periphery, functional CCR9 is expressed by all naive CD8 T cells, but not by naive CD4 T cells. To our knowledge, this latter finding is the first difference observed in homing receptor expression between naive lymphocyte populations. This suggests that naive CD8 T cells might have access to lymphoid microenvironments from which naive CD4 T cells are excluded.


bone marrow chimera








reconstitution ratio


subcapsular zone




The chemokine receptor CCR9 1 is prominently expressed by most human and mouse thymocytes 2, 3. Its only known ligand, TECK/CCL25 46, is expressed within the thymic cortex 7. In vitro migration assays show CCL25 to be a highly efficient thymocyte chemoattractant 4, 8. Despite the abundance of CCR9 and CCL25 within the normal thymus, CCR9-deficient mice show only a mild delay in embryonic thymic maturation 9. However, in vivo competitive assays show CCR9-deficient cells to be dramatically disadvantaged with respect to WT cells during thymic development 10. Nevertheless, the regulation and biological roles of these molecules remain incompletely understood, and are in many ways controversial.

Uehara et al. 10 found that adoptively transferred CCR9-deficient BM-derived precursors were significantly disadvantaged at developing into early CD25+, CD4/CD8 double-negative (DN) thymocytes when competing with normal WT BM. These authors concluded that a competitive effect within such an early thymocyte population would suggest a role for CCR9 in targeted homing of BM-derived T cell precursors to the thymus. The DN/CD25+ population studied by these authors is primarily (>90%) of the “DN3” stage (DN/CD117/CD44lo/CD25+) 11, 12. There are, however, at least two developmental stages in the thymus prior to the DN3 stage. These include the “Shortman” (CD4lo/CD117+/CD44hi/CD25) population and the “DN2” (DN/CD117+/CD44hi/ CD25+) population 11, 12. Thus, the interpretation of Uehara et al. 10 depends upon the testable hypothesis that the competitive effect observed at the DN3 stage will also exist at the earlier stages.

RAG–/– thymocytes are an enriched source of early thymocytes, as they do not progress beyond the DN3 stage. Their developmental block is caused by an inability to rearrange TCRβ genes and the resulting failure to express a functional pre-TCR complex 11, 12. Interestingly, Norment et al. 2 found CCR9 to be quite poorly expressed in the thymus of RAG–/– mice. Taken by itself, this finding is not consistent with the conclusions of Uehara et al. 10: If CCR9 were indeed expressed by the earliest thymic populations, one would expect both RAG–/– and WT thymocytes to express similar levels of CCR9. Norment et al. 2 further demonstrated that RAG–/– thymocytes could be induced to strongly express CCR9 mRNA when the need for pre-TCR expression was artificially bypassed by cross-linking CD3ϵ. This suggested that CCR9 might be induced during normal T cell development at the time when signaling occurs through the pre-TCR 13.

To shed light on this seemingly contradictory body of data, we have set out to study CCR9 and its functional effects over the entire course of thymic development. We have utilized three independent approaches to assess thymic CCR9 regulation and the biological role of CCR9 in T cell development: We (1) examined CCR9 function using in vitro chemotaxis assays for precisely defined early thymic populations; (2) assessed CCR9 expression on these same populations using novel anti-mouse CCR9 mAb; and (3) performed in vivo competitive BM reconstitution assays to determine whether T cell development is disadvantaged prior to the DN3 stage for CCR9-deficient thymocytes.


Chemotaxis of early thymocytes

CCR9 is the only known receptor for CCL25 4, 6. To confirm CCR9 as the sole mediator of CCL25-induced thymocyte chemotaxis, we compared in vitro migration of WT and CCR9–/– murine thymocytes over a wide range of CCL25 concentrations (10–3000 nM). The titration yielded a typical chemotactic bell curve for WT thymocytes (Fig. 1A, squares), with the optimal dose at 300 nM. Checkerboard assays proved this migration to be direction-dependent (J.J.C., unpublished data and 8). In contrast, no response was detected to any CCL25 concentration tested for CCR9–/– thymocytes (Fig. 1A, triangles). Both WT and CCR9–/– cells responded equally well to CXCL12 (Fig. 1C, lower right), whose receptor, CXCR4, is present and functional throughout thymic development 8, 14. WT and CCR9–/– thymocytes both responded well to CXCL12 in ICAM-1 adhesion experiments performed in parallel to the chemotaxis assays, but only WT cells responded to CCL25 (M.A.W. and J.J.C., unpublished data). We conclude that CCR9 accounts for all observed CCL25-induced chemotactic and adhesive responses of murine thymocytes.

Figure 1.

Thymocyte chemotaxis. (A) Titration of CCL25 concentrations. Black squares: WT; gray triangles: CCR9–/–. (B) Immunophenotypic definitions. Left: gates used to identify Shortman and DN2; right: DN3 [Lineage cocktail (Lin) is described in Multicolor flow cytometry]. (C) Chemotaxis of thymic subsets. Gray bars: migration in the absence of chemokine; black bars: migration to 300 nM CCL25; open bars: migration to 30 nM CXCL12. Means and SEM are shown for at least four experiments with at least six chemotactic wells for each data point. (D) Chemotaxis of DN4 and DP subsets. Means and SD are shown for seven experiments with at least six chemotactic wells for each data point. The Mann-Whitney U-test was used to calculate the statistical significance between basal migration and chemokine-induced migration; ns = not significant, **p<0.01.

We next evaluated the CCL25-induced migration of individual early thymic subpopulations. Gating criteria were established for the Shortman, DN2 and DN3 populations (Fig. 1B), and migration to CCL25 (Fig. 1C, black bars), CXCL12 (open bars), or medium alone (gray bars) is shown for each population. Response to CXCL12 was included as an approximation of “maximal” chemotactic potential for each cell type. Parallel experiments were performed with thymocytes from mice with developmental blocks that cause enrichment of the populations that precede the block: T cell maturation is blocked at the double-positive (DP) stage for TCRα–/– mice and at the DN3 stage for RAG–/– mice. Thymocytes from CCR9–/– versions of WT, TCRα–/– and RAG–/– mice were also included as negative controls. None of the subpopulations examined from any of these CCR9-deficient strains responded to CCL25.

The Shortman and DN2 thymocytes did not respond to CCL25 for any of the genotypes tested. The earliest population to display CCL25-mediated migration was DN3, but this occurred only for WT and TCRα–/– mice. The CCR9-mediated response of DN3 thymocytes was approximately one third that of their CXCL12 response. Interestingly, DN3 cells from RAG–/– mice (which differ from WT DN3 cells only in their ability to present a complete pre-TCR complex on the cell surface 12) did not respond to CCL25. Thymic stages subsequent to DN3 (i.e. DN4 and DP) from WT and TCRα–/– mice displayed strong chemotaxis to CCL25, migrating at levels comparable to CXCL12-induced chemotaxis (Fig. 1D, upper panel).

CCR9 expression as assessed by flow cytometry

We generated a panel of mAb to native murine CCR9. Flow cytometry is shown for one representative member of this panel, CW-1.2 (Fig. 2). CCR9 is present on the vast majority of CD4+ and CD8+ thymocytes from WT but not CCR9–/– mice (Fig. 2A). Treatment of thymocytes with CCL25 down-regulates the CW-1.2 epitope, but does not significantly affect binding of an anti-CXCR4 mAb (Fig. 2B), which confirms that CW-1.2 is CCR9-specific.

Figure 2.

CCR9 expression by thymic subsets. (A) Two-color flow cytometry of CCR9 vs. CD4 or CD8 by WT and CCR9–/– thymocytes. (B) Specific down-regulation of chemokine receptors CCR9 (left) and CXCR4 (right) after chemokine treatment. Mean fluorescence intensity and SD as % of untreated brightness is shown for three experiments. (C) CW-1.2 cytometry for WT (left) and CCR9–/– (right) thymic subsets. CW-1.2 (black line) overlaid on isotype-matched control (gray line). (D) CW-1.2 cytometry for RAG–/– (left column) and RAG–/–/CCR9–/– (right column). CW-1.2 (black line) overlaid on isotype-matched control (gray line). (E) Flow cytometry of CCR9, CD4 and CD8 induction and CD25 loss after in vivo pre-TCR-complex cross-linking via CD3ϵ. All cytometry presented in this figure shows a single experiment representing at least five repeats.

Cytometry data for CCR9 using the CW-1.2 mAb were consistent with our analysis of CCL25-mediated thymic chemotaxis: CW-1.2 did not recognize the Shortman or DN2 populations, but did recognize the entire DN3 population at a low level (Fig. 2C, left column). The bulk of the DN4 population expressed CCR9 with intensity similar to that of DN3, but a subset was ∼10× brighter than DN3, suggesting a new way to identify a transitional population within the DN4 population. It should be noted that we used a “Lineage” cocktail to exclude cells expressing TCRγδ, Gr-1, B220, NK1.1, CD11b and CD11c from our DN4 gate. DP cells were universally bright for CCR9. No thymic subsets from CCR9–/– mice stained positively with CW-1.2 (Fig. 2C, right column). The other mAb from our anti-CCR9 panel gave essentially identical staining patterns (not shown).

The Shortman and DN2 populations from RAG–/– mice were also negative for CW-1.2 (Fig. 2D). Interestingly, RAG–/– DN3 cells (which did not yield significant CCL25 chemotactic responses) expressed detectable CCR9, albeit at levels threefold lower than WT DN3 cells (Fig. 2C, D, and M.A.W. and J.J.C., unpublished data). Cross-linking the RAG–/– DN3 pre-TCR complex (by in vivo injection of anti-CD3ϵ mAb) artificially induced DN-to-DP transition (Fig. 2E, top panels). This treatment also induced high CCR9 expression (Fig. 2E, bottom panels). Significantly, induction of CCR9 correlated with down-regulation of CD25 (Fig. 2E, bottom panels).

Competitive BM reconstitution

We next asked whether the developmental disadvantage of CCR9–/– DN3 cells observed by Uehara et al. 10 was also present at thymic stages earlier than DN3. In the original study 10, mixtures of WT and CCR9–/– BM were transferred into irradiated RAG–/– recipients. After engraftment, the CD45.1/CD45.2 congenic system was used to identify cells derived from each donor type. This BM adoptive transfer model allowed WT and CCR9-deficient thymocytes to develop side-by-side within the same host, forcing them to compete with each other for limited developmental niches within the thymus.

We modified this method to allow adoptive transfer of BM mixtures into WT instead of RAG–/– hosts, because of concerns that development might be altered in the hypotrophic RAG–/– thymus. To accomplish this, we used irradiated WT CD45.1/CD45.2 F1 hybrids as the hosts for our BM chimeras (BMC). By this strategy, radio-resistant CD45.1+/CD45.2+ doubly positive host cells were distinguished from CD45.1+ WT and CD45.2+ CCR9–/– singly positive donor cells (see Fig. 3A, B). For each “experimental” BMC mouse (i.e. CD45.1+ WT donor BM plus CD45.2+ CCR9–/– donor BM), a “control” BMC mouse (i.e. CD45.1+ WT donor BM plus CD45.2+ WT donor BM) was created in parallel. BM was allowed to engraft for 6–12 wk before harvesting thymus, blood, spleen and peripheral lymph nodes for analysis.

Figure 3.

Competitive BM reconstitution. (A, B) Flow cytometry of CD45.1 and CD45.2 on whole thymus from control (A) and experimental (B) BMC. Each panel displays equal numbers of events. (C) Development D of CCR9–/–vs. WT splenic lymphocytes. Means and SEM are shown for eight matched pairs of control (gray bars) and experimental BMC (black bars), each from three independent batches. (D) D of early thymic subsets. Means and SEM are shown for four matched pairs from two independent batches. (E) D of late thymic subsets. Means and SEM are shown for five matched pairs from three independent batches. [Note: Data in (D, E) are from entirely separate experiments]. (F) Data from (E) graphed as change in D, i.e. ΔD, with respect to DN3. The Mann-Whitney U-test was used to calculate the statistical significance between control and experimental chimeras; ns = not significant, **p<0.01.

CCR9–/– BM-derived precursors were profoundly disadvantaged at populating the host thymus when competing with their WT BM counterparts (compare Fig. 3A, B). In order to quantify the competition between WT-derived and CCR9–/–-derived donor cells within various populations, we normalized for the reconstitution ratio (R), indicative of the relative efficiency with which CD45.1 vs. CD45.2 BM cells engrafted individual BMC mice.

R was attained by normalizing the CD45.1:CD45.2 ratio for a given population to a separate population expected to be unaffected by the presence or absence of CCR9. We chose peripheral blood neutrophils as the normalizing population because of their high turnover rate, and because CCR9 is not expressed by neutrophils 3. Thus:

equation image(1)

Data are presented as disadvantage (D) for transferred CCR9–/– BM cells competing with WT BM cells. D is calculated by comparing R of a given population from an experimental BMC mouse to R of the corresponding population in a control BMC mouse harvested in parallel. Thus,

equation image(2)

D for any control population is therefore “1” by definition, and a D of >1 for a given cell type implies that CCR9–/– cells are developmentally disadvantaged. As shown in Fig. 3C, the D for splenic NK and B cells was not significantly different between control (gray bars) and experimental (black bars) BMC mice, suggesting that development of these lymphocyte types did not require CCR9. As reported by Uehara et al. 10, we observed a large D for DN3 cells in the thymus (Fig. 3D). Interestingly, this effect was equally strong for the earlier Shortman and DN2 populations, even though neither of these populations expressed functional CCR9 (Fig. 1, 2). CCR9 expression must therefore be important at some point prior to the Shortman stage to engender this disparity.

The functional role of CCR9 induction during the DN3 stage

Although the D for CCR9–/– cells was similar among the Shortman, DN2 and DN3 populations, the D began to increase after the DN3 stage (Fig. 3E). To cancel any component of D engendered prior to DN3 stage from our calculations, we assessed the change in DD) with respect to DN3, and graphed this value for thymic populations subsequent to DN3 (Fig. 3F). D was significantly greater for each subsequent population, culminating in a threefold disadvantage for CCR9-deficient CD4 single-positive (SP) and a twofold disadvantage for CCR9-deficient CD8SP cells. This demonstrates that CCR9–/– cells, already disadvantaged at the DN3 stage, were further out-competed by WT cells after the DN3 stage.

In addition to the strong effect within the thymus, the competitive assays demonstrate a profound influence of CCR9 on composition of the peripheral T cell pool. Adoptively transferred CCR9–/– BM precursors are more than tenfold less likely than WT BM precursors to complete a full thymic developmental program and become mature circulating T cells (Fig. 3C).

CCR9 expression at thymic maturity

Flow cytometry with CW-1.2 demonstrates that many medullary CD4SP stage thymocytes continue to express CCR9 (Fig. 4A, top panel). However, in agreement with our previously reported functional data 8, CCR9 is low to negative on approximately one third of these cells. The CD4SP cells lacking CCR9 were primarily of the CD69/CD62L+ “late medullary” phenotype (M.A.W. and J.J.C., unpublished data). Late medullary CD4SP thymocytes are mature enough to exit the thymus 15, and are known to have lost CCR9 chemotactic function 8.

Figure 4.

CCR9 expression at later stages of T development. (A–C) Flow cytometry of CCR9 expression (black line) overlaid on IgG2a isotype-matched control (gray line) for thymic (A) and splenic (B, C) populations from WT and CCR9–/– mice. (A) SP thymocyte subsets were defined as CD4+/CD8 or CD4/CD8+. (B) Naive CD4 T splenocytes = CD4+/CD45RBhi/CD44lo; naive CD8 T splenocytes = CD3+/CD8+/PNAlo/CD44lo. (C) Memory CD4 T splenocytes = CD4+/CD45RBlo/CD44hi; memory CD8 T splenocytes = CD3+/CD8+/PNAhi/CD44hi. Single experiments shown are representative of at least five repeats. (D) Chemotaxis of naive splenocytes to CCL25. Means and SEM are shown for four experiments with at least four chemotaxis wells per data point. The Mann-Whitney U-test was used to calculate the statistical significance between basal migration and chemokine-induced migration; ns = not significant; **p<0.01.

In marked contrast to CD4SP thymocytes, CD8SP thymocytes remained uniformly positive for CCR9 (Fig. 4A, lower panels). In fact, naive splenic CD8 T cells continued to strongly express CCR9, in marked contrast to naive CD4 T cells, which did not express detectable CCR9 (Fig. 4B lower panels). The CCR9 expressed by naive CD8 T splenocytes was functional, as shown in chemotaxis assays (Fig. 4D). Naive CD8 T cells (but not naive CD4 T cells) isolated from blood, peripheral lymph nodes, mesenteric lymph nodes and Peyer's patches also expressed CCR9 (data not shown). Further, we attempted to assess CCR9 expression by naive CD8 and CD4 T cells from small intestinal lamina propria, where CCR9 is believed to play a role in memory cell homing. However, memory T cells dominated this tissue, and naive T cells were not present in sufficient quantities for characterization (data not shown). Thus in mice, functional CCR9 is expressed by essentially all peripheral naive CD8 T cells, but not by peripheral naive CD4 T cells.

Although small subsets of splenic CD4 and CD8 memory T cells expressed CCR9 (presumably intestinal-homing memory T cells 3), the majority neither expressed CCR9 (Fig. 4C) nor responded to CCR9 ligand (not shown). Thus (with the exception of “intestinal” memory CD8 T cells 3), naive CD8 T cells apparently lose CCR9 only upon antigen-dependent differentiation into memory cells.


Complexity of CCR9 regulation

Our multi-approach assessment of CCR9 expression and function during early T cell development has pointed towards a much more complex regulatory program than previously anticipated, explaining why several previous findings initially appeared contradictory. Such complex regulation suggests that a single chemotactic receptor may contribute to the development of a given cell lineage at multiple discrete stages. It is important to note that only minor phenotypic differences are observed between unmanipulated WT and CCR9–/– mice 9. However, by creating in vivo conditions in which cells must compete during T cell development, we and others 10 have seen dramatic differences between cells from these two genotypes. This suggests that the mixed environment within the BMC mice restores competitive conditions that normally exist in WT mice, but are absent from unmanipulated CCR9-deficient mice.

Effect of CCR9 on T cell development prior to the thymus

Our competitive BM reconstitution assays agree with a previous finding that CCR9–/– BM-derived precursors are dramatically less likely to reach the DN3 thymic stage than their WT counterparts 10. We have now extended this observation to demonstrate a similar effect within earlier thymic stages, including the CD117+ “Shortman” and “DN2” populations.

Taken by itself, this finding is consistent with a role for CCR9 in populating the thymus with BM-derived precursors, as suggested by Uehara et al. 10. However, this interpretation becomes unwieldy if one also considers our present finding that CCR9 is absent from early thymocytes until after they reach the DN3 stage (discussed further below). This would require one to assume that CCR9 is rapidly down-regulated after BM-derived precursors arrive at the thymus, only to be up-regulated again at the DN3 stage.

Although rapid down-regulation of a chemokine receptor is certainly not impossible, our data offer an alternative hypothesis that does not require such rapid modulation of CCR9. We suggest that CCR9 may be required for the development of T cell precursors within the BM. In this scenario, CCR9 may be directly expressed by the T cell precursors themselves, or by some other cell type upon which T precursors depend for survival or continued development. This hypothetical CCR9-dependent event would occur after the T cell developmental pathway has branched off from the neutrophil, NK cell and B cell developmental pathways, as these latter cell types were not affected by CCR9 deficiency (Fig. 3C). This would imply that CCR9–/– BM is simply less efficient at producing thymic precursors than WT BM. Our hypothesis thus makes the testable prediction that CCR9–/– mice would possess relatively fewer CLP-2 cells in the BM 16 or LSK cells in the circulation 17 than WT mice. Of course, our model only addresses thymic seeding in the adult, as thymic seeding during fetal development may potentially involve alternative homing mechanisms.

Induction of CCR9 expression at the DN3 stage

The results of our flow cytometry and functional chemotaxis assays both agree that CCR9 is not expressed by Shortman and DN2 thymic stages. DN3 was the earliest stage at which CCR9 expression and function were detectable within WT thymus. Interestingly, DN3 is the stage at which thymocytes are selected for a successful TCRβ gene rearrangement 11, 12. Cells that fail to productively rearrange TCRβ at this stage are unable to assemble a functional pre-TCR complex, and are therefore unable to receive the signals required for further development 11, 12. As a result of such signaling, we discovered that high levels of CCR9 expression might be a new marker identifying a possible transitional subset within the resultant DN4 population (Fig 2C). This subpopulation may be useful for studying intracellular events that occur as DN thymocytes progress to the DP stage.

Induction of CCR9 during the DN3 stage in WT thymus is consistent with the findings of Norment et al. 2. This group showed that artificial cross-linking of CD3 complex could dramatically increase CCR9 transcription by RAG–/– thymocytes. We have further demonstrated that CD3ϵ cross-linking on RAG–/– thymocytes induces high expression of CCR9 protein on the cell surface (Fig. 2E).

It is tempting to speculate that, during normal development, selective induction of CCR9 enables only those cells that have successfully rearranged TCRβ to enter microenvironments required for further development. This makes teleological sense: Only ∼50% of normal DN3 cells successfully rearrange TCRβ 12, and these are the only cells for which further development would be useful. The nature of this hypothetical permissive microenvironment remains to be seen. However, recent studies have identified previously unrecognized separate niches for CD4 vs. CD8 SP cells in the thymic medulla 18, making it possible that novel microenvironments will also be discovered in the cortex.

In fact, Benz et al. 19 have recently identified a migratory event occurring at the DN3 stage that may require CCR9 function. This group compared thymic sections from WT and CCR9-deficient mice. CD25+ thymocytes (mostly DN3) were found (as expected) tightly associated with the subcapsular zone (SCZ) of the WT cortex, with very few found in the cortex proper. In marked contrast, CD25+ cells were not enriched in the SCZ of CCR9-deficient thymus, and were found randomly scattered about the cortex.

Benz et al. 19 speculated that perhaps CCR9 is necessary for CD25+ thymocytes to home to the SCZ. However, our current findings suggest an alternative scenario: We propose that CCR9 actually helps thymocytes to escape from the SCZ, to continue their further development. In this scenario, escape from the SCZ would consist of a rate-limiting, competitive migratory event through a barrier between the SCZ and the cortex proper. Further, cells within the SCZ must compete with each other to exit, but the existence of a CCL25 chemotactic gradient originating from the cortex proper would enhance the escape of those few cells bearing functional CCR9.

Those DN thymocytes that have successfully rearranged TCRβ, and have therefore begun to express high levels CCR9, would be more competitive at escaping the SCZ than their counterparts lacking successfully rearranged TCRβ. We have shown that gain of CCR9 expression correlates with loss of CD25 expression (Fig. 2E). Thus, in the WT thymus, CD25 cells would become more competitive at leaving the SCZ as they gain CCR9 expression along with their loss of CD25 expression. This would explain why CD25+ cells are greatly enriched in the SCZ as opposed to the remainder of the cortex.

In contrast, CCR9-deficient thymocytes with successfully rearranged TCRβ do not express CCR9. In these mice, the CD25+ and CD25 cells are therefore equally efficient at escaping the SCZ because there are no competing CCR9+ cells. CD25+ cells are therefore found randomly scattered about the cortex 19. If our proposed scenario proves true, this will mean that escape from the SCZ (enhanced by CCR9 function) is a competitive event that eliminates dead-end thymic lineages, by preventing them from entering an environment necessary for their continued viability.

CCR9 expression on medullary thymocytes and naive peripheral T cells

Flow cytometry with our new anti-CCR9 antibodies confirmed that CCR9 was down-regulated by the most mature CD4SP thymocytes prior to exit from the thymus 8. In contrast, CCR9 was expressed by all medullary CD8SP thymocytes, and remained expressed by naive CD8 T cells from spleen (Fig. 4), lymph nodes and peripheral blood (not shown).

It has been suggested that CCR9 may be a marker for peripheral CD8 T cells that have recently emigrated from the thymus 20, 21, or for a specialized CD8 T cell population that expresses CD103 21, 22. We, however, find that functional CCR9 expression is a general marker for all naive peripheral CD8 T cells in mice. In fact, we found that essentially all naive CD8 T cells continue to express CCR9 in WT mice greater than 1 year old, after the thymus has become largely atrophic. Further, essentially all CD8 T cells from OT-1 TCR-transgenic mice (which have never encountered their cognate peptide antigen and are therefore naive by definition) express CCR9 (M.A.W. and J.J.C., unpublished data).

The expression of functional CCR9 by naive CD8 T cells, but not by naive CD4 T cells, raises the possibility that naive CD8 T cells have access to previously unrecognized niches in peripheral lymphoid organs that are not accessible to naive CD4 cells. This appears to be a difference between mice and humans, as currently available anti-human CCR9 mAb show CCR9 to be absent from most naive human CD8 T cells 3.

In addition to its role in T cell development discussed here, it should be noted that CCR9 expression and function are believed to participate heavily in the cascade by which memory and effector CD4 and CD8 T cells home to intestinal tissue 23. Thus, the up-regulation of CCR9 on T cells dedicated to intestinal homing 24 constitutes yet another complexity in the controlled expression of this molecule that we did not address in this study.


CCR9 is not expressed in the thymus until the DN3 stage, although it appears to contribute to development of T cell precursors prior to their residence within the thymus. Cell surface CCR9 protein expression can be induced on RAG-deficient DN3 thymocytes by cross-linking CD3ϵ, suggesting that pre-TCR signaling may be responsible for functional CCR9 induction during normal thymic development. High CCR9 expression identifies a subset of DN4 cells that may be transitional between the DN4 and DP stages. CCR9 expression distinguishes naive CD8 T cells from naive CD4 T cells in mice, suggesting that naive CD8 T cells may have access to lymphoid microenvironments from which naive CD4 T cells are excluded.

Material and methods


WT C57BL/6N (CD45.2+) mice were obtained from Charles River (Wilmington, MA). RAG-1–/–, TCRα–/– and BoyJ (CD45.1+) mice were obtained from Jackson Laboratories (Bar Harbor, ME). CCR9–/– mice 9 were maintained on C57BL/6N background over 15 generations in our own facility. CCR9–/– mice were also backcrossed onto the RAG-1–/– and TCRα–/– background to obtain negative control thymocytes for CCL25 migration assays. Animal experiments were approved by the Animal Care and Use Committee of Children's Hospital Boston.

Chemotaxis assays

Migration was assessed as described 8, 25. Single-cell suspensions were obtained from thymus and spleen and incubated 2×30 min in RPMI 1640 medium with 10% FBS in a T-175 flask (Nunc, Naperville, IL) to remove adherent cells. Cells were migrated through 5-μm pore inserts (Costar, Cambridge, MA) at 5×105 cells/100 μL/insert. Chemokines were used at previously determined optimal chemotactic concentrations, unless otherwise indicated: murine CCL25 (R&D Systems, Minneapolis, MN), optimal concentration = 300 nM; human CXCL12 (PeproTech, Rocky Hill, NJ), optimal concentration = 30 nM. Migrated cells and input populations were stained with fluorescent mAb (see below) and quantitated by an internal standard of 15-μm beads 8.

Generation of anti-mouse CCR9 mAb

CCR9–/– mice were repeatedly immunized i.p. with both irradiated WT murine and human thymocytes. Seropositive animals were boosted i.p. and i.v. with WT murine thymocytes 3 days prior to harvest. Fusion and screening was conducted by standard methods 26. Six positive hybridomas (CW-1 through 6) were subcloned and one representative (CW-1.2) used for flow cytometry presented here.

In vivo CD3ϵ treatment

RAG–/– mice were injected i.p. with 50 μg anti-CD3ϵ mAb (145-2C11), as described 2, and thymi were harvested at day 5.

Competitive BM reconstitution

BM was isolated from femur and tibia of donor mice. A 1:1 BM suspension of 2×106 WT (CD45.1+/+) cells plus CCR9+/+ (CD45.2+/+) cells for “control BMC” mice or WT (CD45.1+/+) cells plus CCR9–/– (CD45.2+/+) cells for “experimental BMC” mice were injected retro-orbitally into 4-wk-old, 1200-rad-irradiated F1 WT mice (CD45.2+/–/CD45.1+/–). At 6–12 wk, thymi, blood and spleens were harvested, and cells were analyzed by multicolor flow cytometry. The percentage of CD45.1+ cells and CD45.2+ cells was determined for each population. Radio-resistant host-derived cells (CD45.1+/CD45.2+) were excluded from analysis.

Multicolor flow cytometry

The antibodies used for the flow cytometry studies are detailed below according to the experimental procedure.

Chemotaxis studies: CD45.2-FITC, CD117-PE, CD25-PE-Cy7, CD44-allophycocyanin, unconjugated CD4, SAV-allophycocyanin-Cy7 (Pharmingen, San Diego, CA), PE-TR-goat F(ab’)2 anti-rat IgG(H+L) (Caltag, Burlingame, CA), and a “Lineage” cocktail of biotinylated mAb: TCRγδ, NK1.1, B220, Gr-1, CD11b, CD11c and Ter119 (all from Pharmingen).

BMC analysis: (1) Early DN cocktail = CD45.1-FITC, CD117-PE, unconjugated CD4, CD25-PE-Cy7, anti-CD44-allophycocyanin, CD45.2-biotin, PE-TR-goat F(ab’)2 anti-rat IgG(H+L) and SAV-allophycocyanin-Cy7; (2) Late DN cocktail = CD45.1-FITC, CD45.2-PE, CD25-PE-Cy7, CD44-allophycocyanin and “Lineage” (“Lineage” was as above but with CD4 and CD8 included); (3) DP and SP cocktail = CD45.1-FITC, CD4-PE, unconjugated CD8α, CD69-PE-Cy7, CD3ϵ-allophycocyanin or CD62L-allophycocyanin, CD45.2-biotin, PE-TR-goat F(ab’)2 anti-rat IgG(H+L) and SAV-allophycocyanin-Cy7; (4) Naive CD4 cocktail = CD45.1-FITC, CD45RB-PE, CD4-PE-Cy7, CD44-allophycocyanin, CD45.1-biotin and SAV-allophycocyanin-Cy7; (5) Naive CD8 cocktail = PNA-FITC, CD45.2-PE, unconjugated CD8α, CD3-PE-Cy7, CD44-allophycocyanin, CD45.2-biotin, PE-TR-goat F(ab’)2 anti-rat IgG(H+L) and SAV-allophycocyanin-Cy7; (6) B + NK cocktail = CD45.1 FITC, B220-PE, NK1.1-PE-Cy7, CD3ϵ-allophycocyanin, CD45.2-biotin, PE-TR-goat F(ab’)2 anti-rat IgG(H+L) and SAV-allophycocyanin-Cy7; (7) neutrophil cocktail = CD45.1-FITC, CD11b-PE-Cy7, Gr-1-allophycocyanin, CD45.2-biotin and SAV-allophycocyanin-Cy7.

CCR9 staining: Unconjugated mAb CW1.2 or mouse isotype-matched IgG2a control (MPC-11) replaced CD45.1-FITC in the above BMC cocktails, followed by Cy2-conjugated goat F(ab’)2 anti-mouse IgG-Fcγ (Jackson, West Grove, PA).

Cytometry was acquired on dual-laser MoFlo cytometer (DakoCytomation, Ft. Collins, CO) configured for six colors: (1) FITC, (2) PE, (3) PE-TR, (4) PE-Cy7, (5) Cy5, (6) allophycocyanin-Cy7, and analyzed using Summit 3.1 software (DakoCytomation).

CCR9 down-regulation

WT thymocytes (1×106) were incubated in RPMI/10% FBS and treated 1 h at 37ºC with 2 μg CCL25 or CXCL12. Cells were put on ice immediately after treatment and stained for CCR9 or CXCR4. Anti-CXCR4 mAb (2B11) and its isotype-matched rat IgG2b negative control (R35-38) were both from BD Pharmingen.


We thank E. S. Baekkevold and L. A. Jopling for contributions to early experiments. We thank Drs. J. Manis and C. M. Wain for critical reading of the manuscript. This study was funded by NIH grant R01AI46784 to J.J.C. M.A.W. was supported in part by a postdoctoral fellowship from the Foundation pour la Recherche Medical (FRM), Paris, France, and in part by a postdoctoral fellowship from Crohn's and Colitis Foundation of America (CCFA), New York, NY.


  1. 1


  2. 2


  3. 3


  4. 4