IL-7 Receptor Drives Early T Lineage Progenitor Expansion

This article is featured in Top Reads, p.

T lymphocytes differentiate from hematopoietic progenitors of bone marrow origin that seed the thymus (1). The estimated number of thymus-seeding progenitors in adult mice is in the range of 150–200, with only ∼10 open niches available for seeding at steady state (2). The few thymus-seeding progenitors proliferate mostly as result of Notch1 signaling and give rise to the most immature thymocyte population, the early T lineage progenitor (ETP) (3). The ETP is a heterogeneous population composed of sequential subsets that follow a highly dynamic process of gene regulation toward T lineage commitment (4). Within the ETP, the most immature cells express Ccr9 and Flt3 (5, 6). These receptors are lost in few days, together with alternative cell fate potentials, as result of Notch1/Deltex4 signaling that instructs T cell lineage commitment (7–9). Further differentiation of the ETP into the CD4⁻CD8⁻ double- negative (DN) 2 stage completes commitment toward the T cell lineage (10, 11).

Mice deficient for IL-7 or either of the IL-7R chains (IL-7Rα or common γ-chain [γ_c]) cannot generate T lymphocytes (12–15). In immunodeficient patients with loss-of-function mutations in the homologous genes, T lymphocytes also fail to differentiate (16, 17). This reflects the absolute requirement of IL-7/IL-7R signaling for T lymphocyte differentiation (18). Although IL-7/IL-7R signaling in the adult thymus is best described in DN2 and DN3 thymocytes (19–21), its role at former stages has yet to be addressed. Lineage-tracing studies in vivo have shown that the ETP has a history of IL-7Rα expression (22). Although the label in the ETP was considered to report recombination in upstream bone marrow progenitors, it is difficult to completely exclude IL-7R expression by ETPs (22). In this line, IL-7R expression in the newborn thymus has been reported in 50–55% of Flt3⁺ ETPs (23). These data led us to ask whether the ETP requires IL-7/IL-7R signaling.

We show that, in the adult thymus, the ETP population requires IL-7/IL-7R signaling. Mice lacking IL-7R have a marked loss of ETPs that is mostly caused by defective proliferation and survival. A fraction of ETPs phosphorylated Stat5 ex vivo in response to IL-7 stimulation. Furthermore, IL-7R expression was enriched in the most immature ETPs identified by Flt3 and Ccr9. Consistently, IL-7 promoted the expansion of Flt3⁺ but not Flt3⁻ ETPs on OP9-DLL4 cocultures, without affecting differentiation at either stage. Altogether, our data show that IL-7/IL-7R is required following thymus seeding to promote expansion and survival of the most immature thymocytes, thereby contributing to establish the normal cell number of ETPs in the adult thymus.

C57BL/6J mice (referred to as wild-type, CD45.2⁺) were maintained and bred at the Instituto Gulbenkian de Ciência (IGC) in a colony frequently refreshed by mice imported from Charles River Laboratories. B6.SJL-Ptprc^a Pepc^b/BoyJ mice were purchased from The Jackson Laboratory (stock no. 002014). IL-7Rα^−/− (stock no. 002295) and γ_c^−/− (stock no. 003174) mice were purchased from The Jackson Laboratory. IL-7^−/− mice (12) were backcrossed to C57BL/6 mice (24) and kept at the IGC. Ccr7^−/−Ccr9^−/− mice were provided by T. Boehm after crossing the original single mutants (25), and kept at the IGC. The original Ccr7^−/− mice (26) had been purchased from The Jackson Laboratory, and the Ccr9^−/− mice (27) were generated by targeting the Ccr9 endogenous locus with enhanced GFP (eGFP), generating a knockin/knockout at the Max Planck Institute of Immunobiology and Epigenetics (Freiburg, Germany). Ccr7^−/−Ccr9^−/− mice were crossed with C57BL/6J mice to obtain Ccr7^+/−Ccr9^eGFP/+ mice (here referred to as Ccr9^eGFP). Both males and females were used at 3–6 wk of age as indicated, and were age and sex matched in each experiment. The mice were kept in individually ventilated cages under specific pathogen-free conditions. All animal experiments were approved by the Ethics Committee of the IGC and the Direção Geral de Alimentação e Veterinária and followed the Portuguese and Europeans laws for animal experimentation.

Thymus transplants were performed as described previously (28–33). Briefly, thymi were harvested from newborn mice, the lobes were separated, and each lobe was transplanted into one extremity of the kidney, under the capsule. Mice were kept anesthetized with xylazine (16 mg/kg) and ketamine (100 mg/kg). Recipient mice were 5–8 wk of age, and groups were age and sex matched in each experiment. For the experiment comparing wild-type and IL-7^−/− mice grafted with wild-type thymi, donors were CD45.1⁺CD45.2⁺ or CD45.1⁺ and recipients were CD45.2⁺. For the experiments comparing wild-type and IL-7^−/− thymus donors, each wild-type recipient received one wild-type and one IL-7^−/− thymus graft. Wild-type donors were CD45.1⁺CD45.2⁺, IL-7^−/− donors were CD45.1⁺ or CD45.2⁺, and recipients were CD45.2⁺ or CD45.1⁺, respectively.

Bone marrow cells from 5- to 6-wk-old B6.SJL-Ptprc^a Pepc^b/BoyJ (CD45.1⁺) and wild-type or IL-7Rα^−/− mice (both CD45.2⁺) were depleted of Lin⁺ cells by magnetic separation using Biotin Binder Dynabeads (Thermo Fisher Scientific) after staining with biotinylated Abs against CD3, CD4, CD8, CD11b, CD11c, CD19, Gr-1, and Ter119. Lineage-depleted CD45.1⁺ cells were mixed with competitor CD45.2⁺ cells at a 1:1 ratio, and a total of 1 × 10⁶ cells per recipient were administered i.v. via the tail vein. Recipient mice were wild-type (CD45.1⁺CD45.2⁺) with 5–6 wk of age and irradiated (700 rad) 10–12 h before reconstitution. Chimeric mice were analyzed 9–10 wk after reconstitution.

Organs were harvested in PBS with 10% FBS and cell suspensions were prepared from thymus by gently smashing against and passing through a 40-μm cell strainer, and from bone marrow by flushing and disaggregating the marrow with a 26G syringe before passing through a 40-μm cell strainer. Cells were blocked with mouse IgG (Jackson ImmunoResearch Laboratories) before staining with Abs listed below as target Ag (clone; fluorophore): Ccr9 (CW-12; AF647), CD3 (145-2C11; allophycocyanin-Cy7), CD4 (GK1.1; AF647, BV421, BV605, FITC, PE, or PE-Cy7), CD8 (53-6.7; allophycocyanin/Fire750, BV711, FITC or PerCP-Cy5.5), CD25 (PC61; BV421, BV605, FITC, or AF594), CD44 (IM7; BV711, eFluor 450, FITC, or PerCP-Cy5.5), c-Kit (2B8; allophycocyanin or allophycocyanin-Cy7), Flt3 (A2F10; PE), IL-7R (A7R34; PE-Cy7), Sca-1 (D7; PercP-Cy5.5), CD45.1 (A20; allophycocyanin, BV421, or PE-Cy7), and CD45.2 (104; BV711 or Pacific Blue), all from BioLegend. The lineage mixture contained CD3 (145-2C11), CD4 (GK1.1), CD8 (53-6.7), CD11b (M1/70), CD11c (N418), CD19 (6D5), Gr-1 (RB6-8C5), NK1.1 (PK136), and Ter119 (TER-119), all conjugated to biotin or PE and from BioLegend. Of note, CD4 was omitted from the lineage when analyzing ETPs. Streptavidin conjugated to BV785 or allophycocyanin-Cy7 (BioLegend) was used to detect biotinylated Abs. Dead cells were excluded by SYTOX Blue (Invitrogen) or Zombie Aqua (BioLegend). Intracellular staining to Ki-67 (16A8; a700) from BioLegend and Bcl-2 (3F11; PE) and active caspase-3 (C92-605; FITC), both from BD Biosciences, was performed with a True-Nuclear transcription factor buffer set (BioLegend). Samples were acquired in a BD LSRFortessa X-20 analyzer. Sorts were performed in a BD FACSAria II, and sorted populations were Flt3⁻ ETPs (Lin⁻CD25⁻CD44⁺c-Kit^hiFlt3⁻) and Flt3⁺ ETPs (Lin⁻CD25⁻CD44⁺c-Kit^hiFlt3⁺). Populations identified in this study are defined as follows: LSK hematopoietic progenitors (Lin⁻Sca1⁺c-Kit⁺IL-7Rα⁻), lymphoid-primed multipotent progenitors (LMPPs; Lin⁻Sca1⁺c-Kit⁺IL-7Rα⁻c-Kit^hi), common lymphoid progenitors (CLPs; Lin⁻Flt3⁺IL-7Rα⁺), ETP (CD4⁻CD8⁻Lin⁻CD25⁻CD44⁺c-Kit^hi), DN2a (CD4⁻CD8⁻Lin⁻CD25⁺CD44⁺c-Kit^hi), DN2b (CD4⁻CD8⁻Lin⁻CD25⁺CD44⁺c-Kit^lo), DN3 (CD4⁻CD8⁻Lin⁻CD25⁺CD44⁻), DN4 (CD4⁻CD8⁻Lin⁻CD25⁻CD44⁻), CD8 immature single-positive (ISP; Lin⁻CD4⁻CD8⁺), CD4 CD8 double-positive (DP; CD4⁺CD8⁺), CD4 single positive T cells (SP4; CD3⁺CD4⁺CD8⁻), and CD8 single-positive T cells (SP8; CD3⁺CD4⁻CD8⁺).

Thymi from 4-wk-old wild-type mice were harvested and 5 million thymocytes per condition were cultured in IMDM (Life Technologies) with 2% FBS (HyClone/Thermo Fisher Scientific) for 15 min with or without murine IL-7 (50 ng/ml, PeproTech). Following stimuli, cells were immediately fixed with prewarmed fixation buffer (BioLegend) for 15 min at 37°C, permeabilized with chilled True-Phos Perm buffer (BioLegend) for 1 h at −20°C, and finally stained for phosphorylated Stat5 (BD Biosciences, clone 47/Stat5; a647) in PBS/10% FBS for 30 min at room temperature. For each thymus three samples were prepared as follows: treated with IL-7 but not stained for phospho-Stat5 (fluorescence minus one [FMO] control), cultured without IL-7 and stained for phospho-Stat5, and treated with IL-7 and stained for phospho-Stat5.

Three- to 4-wk-old mice were injected i.p. twice with 0.25 mg of 5-ethynyl-2′-deoxyuridine (EdU; Sigma-Aldrich) 2 h apart, and thymi were harvested 4 h after the first injection (experiments in (Fig. 4). For EdU incorporation assays comparing wild-type to IL-7Rα^−/− and IL-7^−/− thymocytes, 3- to 5-wk-old mice were injected i.p. once with 0.5 mg of EdU and thymi were harvested 2 h later (experiments in (Fig. 5). Thymocytes were stained for extracellular Ags before fixation in 4% paraformaldehyde and EdU detection using a Click-iT Plus EdU Alexa Fluor 488 flow cytometry assay kit (Invitrogen) according to the manufacturer’s instructions.

OP9-DLL4 cells were maintained in IMDM (Life Technologies) supplemented with 10% FBS (HyClone), 0.03% Primatone (Sigma-Aldrich), 5 μg/ml insulin (Sigma-Aldrich), and 0.05 mM 2-ME (Life Technologies) before coculture with thymocytes. On the evening before the sort, OP9-DLL4 cells were irradiated (3000 rad) and 5000 cells/well were plated on 96-well plates. Thymocytes from 3- to 4 wk-old wild-type mice were stained with biotinylated Abs for lineage markers (excluding CD4) and depleted of lineage-positive cells by magnetic separation using Biotin Binder Dynabeads (Thermo Fisher Scientific). Flt3⁺ or Flt3⁻ ETPs were presorted for yield and then 200 cells/well were sort-purified directly onto the OP9-DLL4 culture plate. The cocultures were maintained in IMDM supplemented as above but with 2% FBS, murine Flt3L (5 ng/ml), and with or without murine IL-7 (10 ng/ml), both from PeproTech. Cytokines were refreshed every other day.

A D’Agostino–Pearson normality test was applied to each dataset. Mann–Whitney, Wilcoxon signed rank, repeated measures one-way ANOVA with Tukey’s multiple comparison, and paired and unpaired t and paired t tests were used, as indicated in the figure legends. Data normality and statistical significance were calculated with Prism 7, and significance is represented in the figures as follows: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001; n.s., not significant (p > 0.05).

To reproduce the analysis from Zhou et al. (4), we downloaded the sequence data from Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/) with the accession number GSE130812. The standard Cell Ranger (v6.1.2) pipeline was used to generate fastq files and map the reads against the GRCm38 reference genome. The final output of the Cell Ranger pipeline is a feature-barcode sparse matrix (10x hdf5), which we took into R (v3.4.2) to get a unique molecular identifier count matrix. Values in the matrix represent the number of molecules for each gene that are detected in each cell. The generated count matrix was then used to create a Seurat (v3.0.1) object for further downstream analysis. The standard preprocessing Seurat workflow was used for quality control, data normalization, and scaling. Importantly, the feature selection step was skipped and the analysis proceeded with the curated list of 63 genes used in the original publication (4), plus the inclusion of IL-7R. Also, supervised clustering and preparation of the data for construction of single-cell trajectories (Monocle v2) were done after computational removal of DN3b and granulocyte precursor clusters (clusters 11 and 13 in our analysis). The workflow for ordering cells along a trajectory in Monocle was done in three main steps: 1) choose genes that define progress (same curated list of genes used in the original publication [4] was used here), 2) perform dimensionality reduction of the data using a reversed graph embedding algorithm, and 3) order cells in pseudotime to measure how much an individual cell has progressed through cell differentiation. All visualizations were performed using the R package ggplot2 (v2.2.1). Code can be found in https://github.com/DeFariaJ/Interleukin-7-receptor-drives-Early- T-lineage-Progenitor-expansion.

We noted that the ETP population is greatly reduced in IL-7Rα^−/− adult thymi when compared with wild-type controls (Fig. 1A). Although a reduction in the absolute number of ETPs in thymi deficient for IL-7 or IL-7R (IL-7Rα^−/− or γ_c^−/−) could be due to general thymic atrophy (Fig. 1B), the reduction in frequency (Fig. 1C) hints at a specific mechanism. To test whether the phenotype could be caused by an effect prior to thymus seeding, we quantified the bone marrow progenitors proposed to contribute to thymopoiesis, that is, CLPs and LMPPs. Both progenitor populations were increased in percentage in IL-7–deficient bone marrow relative to wild-type controls (Supplemental Fig. 1A, 1B), which reflects the absence of other cells requiring IL-7/IL-7R signaling, such as mature lymphocytes and B cell precursors. However, the absolute cell numbers of both CLPs and LMPPs did not differ between mutant and control mice (Supplemental Fig. 1C, 1D).

The identity of the physiological bone marrow progenitor populations contributing directly to thymopoiesis remains unresolved (34, 35), and it was therefore possible that the bone marrow analyses (Supplemental Fig. 1) might have missed a minor, but relevant, subpopulation. To test whether the loss of ETPs was caused by a defect prior to thymus homing, we transplanted wild-type thymi into either wild-type or IL-7^−/− recipients (Fig. 2A). In these experiments, progenitors developed either in the presence or absence of IL-7 and were exposed to IL-7 only after thymus seeding. In both conditions, cells are IL-7R proficient and can, therefore, respond to IL-7 after thymus colonization that is produced by the wild-type thymus stroma. Thymus grafts were analyzed 5 wk later, ensuring complete repopulation by bone marrow–derived (host) cells. A sizable ETP population was detected in both groups of recipients (Fig. 2B). Although there was on average a 1.5-fold reduction in percentage (Fig. 2B) and absolute number (Fig. 2C) of ETPs when the host was deficient for IL-7, the total cellularity of the thymus grafts did not differ (Fig. 2D). Importantly, no differences in the percentage of cells in cycle (Supplemental Fig. 2A) or apoptosis (Supplemental Fig. 2B) were noted in ETPs regardless of the host genotype. These data suggest that there is a minor role for IL-7/IL-7R signaling in prethymic progenitors that affect ETP cellularity. Nevertheless, the prethymic defect uncovered with these thymus transplantation experiments could not account for the magnitude of the phenotype observed in IL-7 and IL-7R deficient thymi (Fig. 1A–C). Therefore, to test for a putative effect of IL-7/IL-7R signaling in the ETP solely after thymus seeding, we grafted wild-type and IL-7^−/− thymi into wild-type recipients (Fig. 2E). In these conditions, prethymic progenitors developed in an IL-7 proficient environment, and the effect of IL-7 signaling on ETPs could be revealed following thymus graft seeding. Interestingly, despite having a continuous supply of wild-type bone marrow progenitors, the ETP population was almost absent in the IL-7^−/− thymus grafts in percentage (Fig. 2F) and absolute cell numbers (Fig. 2G). As expected, thymocyte cellularity in the IL-7^−/− grafts was much lower when compared with wild-type thymus grafts (Fig. 2H). Because differences in thymus cellularity could impact the effective number of niches available for colonization by hematopoietic progenitors, we generated competitive bone marrow chimeras at a 1:1 ratio, whereby wild-type progenitors could reconstitute the thymus to its normal size, and IL-7Rα^−/− versus control ETP could be compared (Fig. 2I). Both wild-type and IL-7Rα^−/− progenitors engrafted the bone marrow to similar levels (Fig. 2J). However, although the thymocyte populations in wild-type:wild-type chimeras reflected the level of bone marrow chimerism, IL-7Rα^−/− thymocytes had a clear disadvantage that started at the ETP (Fig. 2J), even though thymus cellularity was similar between the two groups (Fig. 2K). Although the disadvantage of IL-7Rα^−/− thymocytes relative to the wild-type counterparts was expected, the fact that it started in ETPs supported the hypothesis that the population directly depends on IL-7Rα. Indeed, the size of the ETP population was similar in the two experimental groups, with a difference only in the relative contribution of IL-7Rα–proficient or –deficient ETPs (Fig. 2L). Altogether, our data indicate that the reduction of ETPs in IL-7/IL-7R–deficient thymi results mostly from an intrathymic, cell-autonomous defect.

IL-7/IL-7R signaling in the thymus is known to drive proliferation and survival of DN2 and DN3 thymocytes. Thus, using the DN2 as an internal control, we asked whether these parameters differed between wild-type and IL-7Rα^−/− ETPs. IL-7Rα^−/− ETPs had lower levels of Ki-67 than did their wild-type counterparts, suggestive of defects in cell cycle (Fig. 3A, 3B). Additionally, the IL-7Rα^−/− ETP was enriched for dying cells, as determined by the levels of cleaved caspase-3 (Fig. 3C, 3D). These results were in accordance with those measured in DN2 thymocytes (Fig. 3A–D). However, although IL-7R–deficient DN2 expressed lower levels of Bcl-2, no differences were detected in Bcl-2 protein level between wild-type and IL-7Rα^−/− ETPs (Fig. 3E, 3F). Finally, we tested whether wild-type ETPs are responsive to IL-7, and whether signal transduction could be mediated by Stat5. Indeed, we found that a fraction of ETPs phosphorylated Stat5 in response to IL-7 (Fig. 3G, 3H). Although this differed from the homogeneous response of the DN2, it shows that a subset of ETPs is responsive to IL-7. Altogether, these data show that the lack of IL-7/IL-7R signaling in ETPs impairs cell cycle and cell survival, although the latter is not mediated by Bcl-2 expression.

Because only a fraction of ETPs responded to IL-7, we reasoned that receptor expression might be heterogeneous and addressed whether this could be related to differentiation. Within the ETPs, we used Flt3 expression to discriminate the most immature cells (5) (Fig. 4A), and found that they were enriched in IL-7R–expressing cells (Fig. 4B, 4C). IL-7R expression in the ETPs was lower than in DN2, and did not correlate with the differentiation from ETPs to DN2a (Supplemental Fig. 3A, 3B). Indeed, IL-7Rα expression was downregulated from Flt3^hi to Flt3⁻ ETPs and was only upregulated as thymocytes reached the DN2a stage (Supplemental Fig. 3A, 3B). These data were consistent with bulk RNA sequencing (RNA-seq) data generated by others (4), and reanalyzed in the present study, that showed a similar expression pattern for IL-7R mRNA (Supplemental Fig. 3C). We further assessed heterogeneity within the ETPs based on Ccr9 protein expression and reporter expression using Ccr9^eGFP mice (6) (Fig. 4D) and consistently found that IL-7R expression is enriched in the most immature ETPs (Fig. 4E, Supplemental Fig. 3D). Next, as Flt3⁺ ETPs were larger than Flt3⁻ ETPs (Fig. 4A), we asked whether this reflected differences in proliferation. Indeed, proliferation was highest in Flt3^hi ETPs and became progressively lower as ETPs downregulated Flt3 (Fig. 4F, 4G). These results were in line with single-cell (sc)RNA-seq data by others (4) that were reanalyzed in this study (Fig. 4H). Pseudotime analysis (Supplemental Fig. 3E) revealed the differentiation trajectory (Supplemental Fig. 3F) that captured the expected expression dynamics of well-known differentiation-associated genes (Supplemental Fig. 3G). These analyses enabled the identification of 2011 ETPs included in the most immature state (state 4) that were negative for Il2ra (or Cd25) and Bcl11b (Fig. 4I). Within the ETPs, the most immature Flt3⁺ cells (Fig. 4J) were broadly overlapping with cluster 0 (Fig. 4H). We identified IL-7R transcripts in 57 out of 435 Flt3⁺ ETPs (Fig. 4K), a lower percentage (13%) than the corresponding cells identified by flow cytometry (Fig. 4B, 4C). This may be explained because scRNA-seq data can miss transcripts expressed at low levels and will not inform on protein-expressing cells that may have already lost the coding mRNA. Altogether, these data show that IL-7R⁺ cells are enriched in the most immature Flt3^hi ETPs.

As IL-7R expression was enriched in Flt3^hi ETPs (Fig. 4), we wondered about the effect of IL-7/IL-7R deficiency in this population and found that both IL-7Rα^−/− and IL-7^−/− thymi almost completely lacked Flt3⁺ ETPs (Fig. 5A, 5B). Next, we compared the proliferation levels of ETPs by one pulse of EdU incorporation between wild-type, IL-7Rα^−/−, and IL-7^−/− ETPs. Consistent with former experiments (Fig. 4), the proliferation rate of wild-type ETPs reduced as they lost Flt3 expression (Fig. 5C, 5D). Interestingly, both IL-7Rα^−/− and IL-7^−/− ETPs (mostly Flt3⁻) proliferated at levels that were comparable to those of wild-type Flt3⁻ ETPs (Fig. 5D). Furthermore, the combined analysis of Ki-67 and EdU incorporation in IL-7– and IL-7Rα–deficient ETPs revealed an increase in the percentage of quiescent ETPs, that is, in G₀ (Fig. 5C, 5E), hinting at a requirement of IL-7/IL-7R signaling for the cell cycle. Taken together, these results suggest that IL-7/IL-7R deficiency impacts specifically the most immature ETPs, whereas Flt3⁻ ETPs proliferate independently of IL-7R signaling. To test this hypothesis, we sorted Flt3⁺ versus Flt3⁻ wild-type ETPs (Supplemental Fig. 4) and cocultured them on OP9-DLL4 stromal cells in the presence or absence of IL-7 (Fig. 5F). IL-7 indeed promoted the expansion of Flt3⁺ ETPs, but no effect was detected on Flt3⁻ ETPs (Fig. 5G). In addition, we observed that Flt3⁺ ETPs generated CD25⁺ cells (transitional to DN2) slower than Flt3⁻ ETPs (Fig. 5H), as expected from their respective developmental stages. Nevertheless, the differentiation kinetics from either Flt3⁺ and Flt3⁻ ETPs was independent of the presence of IL-7 in the cultures (Fig. 5F, 5H), suggesting that IL-7 does not impact differentiation at the ETP stage. Altogether, our data indicate that IL-7/IL-7R signaling is restricted to the most immature Flt3⁺ ETPs where it is essential for their expansion following thymus seeding.

We show that IL-7/IL-7R signaling is an essential and nonredundant driver of survival and proliferation at the ETP stage, specifically for the most immature Flt3⁺ progenitors. We propose that the rare hematopoietic progenitors seeding the thymus use IL-7/IL-7R together with Notch1/Deltex4 signaling to survive and expand, thereby contributing to the generation of ETPs in physiological numbers. We show that genetic abrogation of either the cytokine or the receptor results in a dramatic reduction in the number of ETPs, which is in line with previous reports (36, 37). We go further to demonstrate that this phenotype resulted mostly from an intrathymic and cell-autonomous defect, with only a minute prethymic contribution. Interestingly, the cells expressing the highest levels of IL-7 in the thymus are thymic epithelial cells at the corticomedullary junction (38), where hematopoietic progenitors enter the thymus and ETPs reside for ∼11 days before differentiating into the next (DN2) stage (39). This suggests that the most immature ETPs might require close proximity to IL-7–producing cells, where the cytokine promotes the first steps of intrathymic differentiation. Consistently, our data show that Flt3⁺ ETPs require IL-7/IL-7R signaling. This was evidenced both by the lack of Flt3⁺ ETPs in IL-7– and IL-7Rα–deficient thymi, as well as via expansion of wild-type Flt3⁺ ETPs in vitro in response to IL-7. As ETPs mature and lose Flt3 expression, IL-7/IL-7R signaling appears to become dispensable for proliferation. Nevertheless, IL-7/IL-7R–deficient ETPs express Ki-67 at lower levels than their wild-type counterparts, which suggests a defect in leaving G₀ and entering cell cycle.

Our data indicate that, although with a minor role, IL-7/IL-7R signaling in the bone marrow also contributes to the population size of the ETPs. This might involve instructive signals to hematopoietic progenitors toward the T cell lineage, and/or favoring their migration to the thymus. Nevertheless, those aspects remain to be tested. Along this line, Notch1 signaling has been proposed to prime bone marrow progenitors toward the T lymphocyte lineage (40).

The reduced number of ETPs in IL-7R–deficient thymi might explain why these thymi are permissive to direct colonization by hematopoietic progenitors injected into the bloodstream (41). Although classical experiments hinted at a regulation of thymus seeding by competition for free niches, it is plausible that the niches are unoccupied in IL-7R–deficient thymi because ETPs fail to expand. In this context, IL-7/IL-7R signaling might be explored as a potential means to improve thymus function following bone marrow transplantation. In patients undergoing bone marrow transplantation, infections, relapse, and graft-versus-host disease are common complications that associate with the long recovery time of the T lymphocyte compartment (42, 43). The thymus has therefore received considerable attention as to how it could be treated to reduce the refractory period of thymopoiesis after bone marrow transplantation (44). While administration of IL-7 in bone marrow–transplanted patients led mostly to an expansion of effector memory T cells (45), studies in mice show that thymus activity is improved in similar settings (46, 47). It will therefore be important to test whether thymus function can be improved specifically, and whether the regulation of thymus seeding by IL-7/IL-7R signaling improves thymus reconstitution and function following bone marrow transplantation in patients.

The Ccr7^−/− Ccr9^−/− mice that were crossed to obtain the Ccr9^eGFP mice generated by C. Benz and C.C. Bleul (6) were a kind gift from T. Boehm (see Materials and Methods for details). We thank A. Cumano for the OP9-DLL4 cells, and H.J. Fehling, J. Demengeot, and T. Boehm for critical reading of the manuscript. We thank the Animal House Facility and the Bioinformatics and the Flow Cytometry Units of the IGC in supporting this work.