Requirement for RORγ in Thymocyte Survival and Lymphoid Organ Development

See allHide authors and affiliations

Science  30 Jun 2000:
Vol. 288, Issue 5475, pp. 2369-2373
DOI: 10.1126/science.288.5475.2369


Most developing thymocytes undergo apoptosis because they cannot interact productively with molecules encoded by the major histocompatibility complex. Here, we show that mice lacking the orphan nuclear hormone receptor RORγ lose thymic expression of the anti-apoptotic factor Bcl-xL. RORγ thus regulates the survival of CD4+8+ thymocytes and may control the temporal window during which thymocytes can undergo positive selection. RORγ was also required for development of lymph nodes and Peyer's patches, but not splenic follicles. In its absence, there was loss of a population of CD3CD4+CD45+ cells that normally express RORγ and that are likely early progenitors of lymphoid organs. Hence, RORγ has critical functions in T cell repertoire selection and lymphoid organogenesis.

During selection of a T cell repertoire within the thymus, CD4+CD8+(double positive, DP) thymocytes with T cell antigen receptors (TCRs) that recognize complexes of host major histocompatibility complex (MHC) proteins and peptides survive and are exported to the periphery (1). Cells undergo apoptosis if they have strongly self-reactive TCRs or fail to interact productively with MHC/peptide complexes (2). The latter process, termed “death by neglect,” accounts for more than 90% of thymocyte death and weeds out useless cells from the developing thymus. It has been proposed that activation of the glucocorticoid receptor by endogenous corticosteroids regulates apoptosis of DP thymocytes (3), but other members of the nuclear receptor superfamily have also been implicated in thymocyte selection (4). We studied the role of a retinoic acid receptor-related orphan receptor, TOR or RORγ, which is expressed in thymus and skeletal muscle but not in mature T cells (5, 6). A thymus-specific isoform of RORγ, RORγt, which shares the DNA and ligand-binding domains with RORγ but differs in the NH2-terminus, protects hybridomas from TCR-induced apoptosis by inhibiting expression of Fas ligand and interleukin-2 (IL-2) (7), probably by its ability to block the function of NFAT (nuclear factor of activated T cells) transcription factors (8).

The ability of RORγ to inhibit FasL and cytokine gene expression suggested that this nuclear receptor may regulate apoptosis during thymocyte development. We therefore generated RORγ-deficient mice by gene targeting (9). RORγ −/− mice were born healthy in a Mendelian distribution, had no discernible physical defect, and were fertile. RORγ mRNA and protein were undetectable in the thymus of homozygous mutant mice, confirming the complete inactivation of the RORγ gene and loss of expression of RORγ and the thymus-specific RORγt isoform (10).

The thymi of RORγ −/− animals appeared normal in size, but total cell numbers recovered were ∼30 to 50% of wild-type levels. Flow cytometric analysis showed a reduction in DP thymocyte number (Fig. 1A). Thymocytes also displayed higher forward scatter (FSC), indicative of larger cell size, than did wild-type cells (Fig. 1B), and most FSChighcells were confined to the DP compartment (11). The amount of CD4 and TCR expressed was reduced in DP cells fromRORγ −/− mice (Fig. 1, C and D). However, CD4+CD8 single-positive (CD4 SP) cells, although reduced in number, expressed normal amounts of both CD4 and TCR, suggesting that they had undergone positive selection (Fig. 1, C and E). Earlier developmental stages appeared normal in RORγ −/− mice, as assessed by the number of double-negative (CD4CD8) thymocytes and by CD44 and CD25 expression (11).

Figure 1

Phenotypic analysis ofRORγ −/− thymocytes. (A) Reduced thymic cellularity of RORγ −/− mice. Cells within each thymic subset were enumerated by fluorescence-activated cell sorting (FACS) analysis. Results from five mice of each genotype were averaged, and standard deviation is indicated. Solid and open histograms indicate mutant and control heterozygous littermates, respectively. (B) FSC profile of thymocytes from heterozygous (gray trace) and homozygous mutant (open trace) mice. (C) Surface expression of CD4 versus CD8 on thymocytes. (D) TCR levels in DP thymocytes from wild-type and mutant mice [as gated in (C)]. Percentages of DP thymocytes expressing low, intermediate, and high levels of TCR are indicated. (E) TCR levels in CD4 single-positive thymocytes, as gated in (C).

To determine whether reduced DP thymocyte cellularity was caused by increased cell death, we performed an ex vivo apoptosis time course analysis (Fig. 2A). About 80% of the RORγ −/− thymocytes underwent spontaneous apoptosis after 5 hours, compared to only 20% of thymocytes from heterozygous littermates, and the accelerated apoptosis was confined to DP cells (11). TUNEL analysis (2, 12) showed localization of apoptotic RORγ −/− cells within cortical regions, where the DP thymocytes reside (Fig. 2B). Multiprobe ribonuclease (RNase) protection assays ofRORγ −/− thymocytes showed that the amount of Bcl-xL mRNA was one-tenth that in normal cells, but no differences were found in expression of other death-related molecules, including FasL (11). Bcl-xL protein was almost undetectable by immunoblot analysis of RORγ −/− thymocyte extracts (Fig. 2C). These results suggested that RORγ regulates DP thymocyte survival by enhancing expression of the anti-apoptotic molecule Bcl-xL.

Figure 2

RORγ−/−thymocytes undergo massive apoptosis and have abnormal cell cycle progression. (A) Survival of freshly isolated thymocytes, cultured in medium for various time intervals. The level of apoptosis was determined by FACS analysis of cells stained with Annexin V and propidium iodide (12). Percent of cells that were Annexin V PI is indicated. The CDK inhibitor roscovitine (A.G. Scientific Inc.) was used at a concentration of 5 μM. (B) TUNEL analysis of thymic sections from wild-type and RORγ-deficient mice (12). Apoptotic cells stain brown. C, cortex; M, medulla. (C) Immunoblot analysis of Bcl-xL levels in wild-type and mutant thymocytes. The blot was stripped and reprobed with antisera recognizing PKC-θ as a control for protein loading. (D) Reduced amount of p27kip1 and enhanced CDK2 activity in RORγ−/− thymocytes. CDK2 was immunoprecipitated and subjected to in vitro kinase assay using Histone H1 as substrate (13) (top panel). Expression of p27 and CDK2 was assayed by immunoblotting of total cell extracts (lower two panels).

The FSC profile of DP thymocytes fromRORγ −/− mice (Fig. 1B) suggested that a malfunction of the cell cycle may accompany accelerated apoptosis. About 30% of thymocytes from RORγ −/− mice, but only 4 to 5% from heterozygous littermates, had greater than 2N DNA content, indicating that they were in the S/G2 phases of the cell cycle (Fig. 3) (12). Thymocytes from RORγ −/−mice had reduced amounts of p27kip1, which negatively regulates the transition from G1 to S phase by inhibiting the activity of the cyclin-dependent kinase CDK2 (Fig. 2D). Correspondingly, CDK2 activity was increased 10-fold in mutant compared to wild-type cells (Fig. 2D) (13). When these thymocytes were cultured in the presence of the CDK inhibitor roscovitine, apoptosis was prevented, even though Bcl-xL expression was not restored (Fig. 2A) (11). Thus, CDK function is required for thymocyte apoptosis in the absence of RORγ, consistent with recent suggestions that anti-apoptotic Bcl-2 family members regulate CDK2 activity (14, 15).

Figure 3

Rescue by a Bcl-xL transgene of the apoptosis and cell cycle phenotype inRORγ −/− thymocytes. (A) Rescue of p27kip1 expression inRORγ −/− thymocytes expressing the Bcl-xL transgene (17). Immunoblot analyses of Bcl-xL and p27kip1 expression are shown. (B) Effect of the Bcl-xL transgene expression on thymic cellularity inRORγ −/− mice. (C) Restoration of normal thymocyte development, but not the CD4 level, by expression of Bcl-xL. Flow cytometric analysis of surface expression of CD4, CD8, and TCRαβ was performed. (D) Bcl-xL rescuesRORγ −/− thymocytes from apoptosis. Apoptosis analysis was as described in Fig. 2A. (E) Bcl-xL rescuesRORγ −/− thymocytes from abnormal S-phase entry (12). Freshly isolated cells were stained with propidium iodide, and flow cytometric analysis was performed to detect DNA content in each cell. The bracketed regions demark cells possessing >2N amounts of DNA.

Thymocytes with homozygous disruption of the Bcl-xL gene exhibit reduced survival similar to that inRORγ −/− mice (16). The thymic phenotype observed in RORγ −/− mice may thus be explained by the loss of positive regulation of Bcl-xL. To test this hypothesis, we prepared RORγ −/−mice that expressed Bcl-xL under the regulation of thelck gene proximal promoter, which directs expression in immature thymocytes (17). Bcl-xL expression in theBcl-xLTg/ RORγ −/− animals restored most aspects of normal thymocyte development (Fig. 3). Thymocyte number inBcl-xLTg/RORγ −/−animals was similar to that in RORγ+ mice expressing theBcl-xL transgene (about 1.5-fold higher than normal thymus) (Fig. 3B), and both DP and CD4 SP cells were present in normal numbers (Fig. 3C). TCR expression and the size of DP thymocytes were normal in the Bcl-xLTg/ RORγ −/− mice, although the level of CD4 expression remained low (Fig. 3C). Both thymocyte survival and cell cycle regulation were corrected by the forced expression of Bcl-xL (Fig. 3, D and E). The latter was correlated with rescue of p27kip1 expression to wild-type levels (Fig. 3A). RORγ thus appears to act genetically upstream of Bcl-xL to prolong DP thymocyte survival and promote G1 phase cell cycle arrest.

We crossed mice with a disrupted TCR-Cα gene, which does not express TCR on DP cells and, hence, cannot undergo negative selection, with RORγ −/− mice.TCRα −/− /RORγ −/−thymocytes had the same phenotype asRORγ −/− thymocytes, indicating that negative selection signals do not initiate premature apoptosis in the absence of RORγ (10). This suggests that accelerated apoptosis is due to enhanced death by neglect. We also crossed RORγ-null mice to gld/gld mice, which are defective for FasL function (18). Thymocyte apoptosis ingld/gld/RORγ −/− mice was identical to that in RORγ −/− animals (10). Thus, the Fas:FasL system appears not to be regulated by RORγ to enhance thymocyte survival.

T cells in RORγ-deficient mice appeared to be exported normally to the periphery and to have normal proliferative functions (11). However, RORγ-deficient mice completely lacked lymph nodes and Peyer's patches (Fig. 4A) (11, 19). Development of lymph nodes and Peyer's patches is dependent on several tumor necrosis factor (TNF) family members (lymphotoxin α and lymphotoxin β, TRANCE/OPGL/RANKL) and their receptors (TNF receptor p55, lymphotoxin β receptor) (20–25). Mutations of these genes, with the exception of TRANCE, also result in disruption of splenic follicular structure and the absence of follicular dendritic cells. In contrast, splenic architecture and follicular dendritic cells were normal in RORγ −/− mice (11), suggesting that the membrane-bound form of lymphotoxin, LTαβ, and TNFα are produced normally in the follicles of mutant mice (26). Splenocytes fromRORγ −/− mice contained one-third the normal number of T cells, but the proportion of CD4 and CD8 single-positive cells was unaffected (11). Lymphoid organ development was also defective in gld/gld/RORγ −/− mice, indicating that this phenotype is unrelated to loss of repression of FasL expression.

Figure 4

Absence of lymph node development and of CD3CD4+CD45+ putative lymphoid organ precursors in RORγ −/− mice. (A) Lymph nodes were visualized by Evan's blue dye, which identifies both superficial inguinal and mesenteric lymph nodes (19). (B) Loss of CD3CD4+CD45+ cells from mesentery and intestines of E18.5 mutant animals. Flow cytometry was used to analyze live cells for expression of CD4 and CD45 (19). Gated CD4+CD45+ cells lacked expression of CD3. (C through E) Expression pattern of RORγ in E12.5 mouse embryos (10). (C) Detection of RORγ mRNA by in situ hybridization of transverse section. Arrows mark expression in regions flanking the esophagus and a region dorsal to the fetal liver. (D) Expression of RORγ protein in a section serial to that shown in (C). (E) Expression of CD4 (green) in some, but not all, RORγ-expressing cells (red). Magnification: (C and D): ×5; (E): ×63.

The absence in mutant mice of lymph nodes and Peyer's patches despite normal splenic B cell follicle development suggested that, although LTαβ and TNFα can be produced by splenic B cells, there may be a defect in other cells that produce these cytokines. WhenRORγ −/− fetal liver stem cells were transferred into RAG2 −/− mice, normal reconstitution of all secondary lymphoid organs of the recipient mice occurred, indicating that defective lymph node development inRORγ −/− animals cannot be attributed to defects in T or B cell homing (11). We analyzed cells with the cell-surface phenotype CD3CD4+CD45+IL-7Rα+, which are associated with early lymph nodes in fetal mesentery, produce lymphotoxin (19), and may function in lymphoid organ development (27). These cells express large amounts of RORγ mRNA (10). CD3CD4+CD45+ cells were absent from both mesentery and intestines of E16.5 and E18.5 mutant embryos and of 2-day-old mutant mice (Fig. 4B) (11). In normal embryos, expression of RORγ was detected in cell clusters flanking the pharynx, intestines, and pericardium as early as E12.5 (Fig. 4, C and D) (10, 28). Many of the cells that expressed nuclear RORγ also expressed cell-surface CD4 (Fig. 4E). In E14.5 embryos, near the limb buds we observed bilateral clusters of cells expressing RORγ, where axillary and inguinal lymph nodes develop (11). These results suggest that RORγ is required for the development of CD4+ progenitor cells involved in the morphogenesis of lymph nodes and Peyer's patches.

Our results indicate that RORγ is essential for survival of immature CD4+CD8+ thymocytes and for development of lymphoid organs. The defect in thymocyte development in mice lacking RORγ can be attributed directly to its regulation of Bcl-xL expression in DP thymocytes. Anti-apoptotic effects of Bcl-xL provide these cells the opportunity to interact with MHC/peptide complexes and undergo positive selection. Regulated shut-off of RORγ expression, or ligand-mediated regulation of RORγ function, could attenuate Bcl-xL expression and result in thymocyte death by neglect. Bcl-xL also promotes maintenance of high p27kip1 levels; in its absence, the increased CDK2 activity contributes to cell death. Thymocytes lacking Bcl-xL display accelerated apoptosis, similar to that observed in the RORγ-null animals, but it is not known whether they also have a defect in cell cycle progression (16).

Disruption of the gene encoding the transcriptional repressor Id2 results in a lymphoid organ phenotype similar to that ofRORγ −/− mice (29). Interestingly, both Id2 and RORγ transcripts were detected in CD3CD4+CD45+IL-7Rα+cells (10, 29), and this cell population was undetectable in animals lacking either Id2 or RORγ (Fig. 4B) (29). Thus, Id2 and RORγ may both be required for survival or differentiation of these hematopoietic progenitors, which may be essential at an early stage of lymphoid organ development.

  • * To whom correspondence should be addressed. E-mail: littman{at}


Stay Connected to Science

Navigate This Article