Stat3 in Thymic Epithelial Cells Is Essential for Postnatal Maintenance ...

10 downloads 2388 Views 1MB Size Report
molecular signals are involved in the maintenance and ited a higher susceptibility of the thymus to suboptimal involution of thymic architecture remains unclear.
Immunity, Vol. 15, 261–273, August, 2001, Copyright 2001 by Cell Press

Stat3 in Thymic Epithelial Cells Is Essential for Postnatal Maintenance of Thymic Architecture and Thymocyte Survival Shigetoshi Sano,1,4,6 Yousuke Takahama,4 Takehiko Sugawara,4 Hiroshi Kosaka,1 Satoshi Itami,1 Kunihiko Yoshikawa,1 Jun-ichi Miyazaki,3 Willem van Ewijk,5 and Junji Takeda2 1 Department of Dermatology 2 Department of Social Environmental Medicine 3 Department of Nutrition and Physiological Chemistry Osaka University Graduate School of Medicine 2-2 Yamadaoka, Suita Osaka 565-0871 4 Institute for Genome Research University of Tokushima 3-18-15 Kuramoto Tokushima 770-8503 Japan 5 Department of Immunology Erasmus University Rotterdam PO Box 1738 3000 DR Rotterdam The Netherlands

Summary This study describes abnormalities of the thymus in mice in which the Stat3 gene has been specifically disrupted behind the keratin 5 promoter. In these mice, virtually all of the thymic epithelial cells (TEC) were deficient for Stat3 activation. Adult mutant mice developed severe thymic hypoplasia, which included alterations in the cortical TEC architecture that coincided with the loss of thymocytes. Even during the asymptomatic period of preadolescence, these mice exhibited a higher susceptibility of the thymus to suboptimal doses of dexamethasone or ␥-irradiation, while their thymocytes per se were no more sensitive than controls. These results indicate that Stat3 in TEC plays an essential role in maintaining thymic architecture and thymocyte survival. Introduction T cell development is dependent on two-way interactions between thymocytes and stromal cells in the thymus (Boyd et al., 1993; van Ewijk, 1991; van Ewijk et al., 1994). The thymic stroma consists of epithelial cells, macrophages, dendritic cells, and fibroblasts, all of which play specific roles at different stages of T cell differentiation (Anderson et al., 1996). Thymic epithelial cells (TEC) have been demonstrated to provide a major microenvironment in which T cell precursors mature and differentiate through interaction with MHC/peptide complexes expressed by TEC, resulting in positive or negative selection of the T cell receptor repertoire (Capone et al., 2001; Hugo et al., 1993). TEC are divided into two 6

Correspondence: [email protected]

major subsets, namely, cortical TEC and medullary TEC. Although the origin, development, and differentiation of each of the subsets of TEC have not been fully determined (Klug et al., 1998; Ropke et al., 1995), recent studies have demonstrated that thymocyte/TEC interactions not only promote thymocyte differentiation, but also determine TEC differentiation (Hollander et al., 1995; Klug et al., 1998; Shores et al., 1991; van Ewijk et al., 2000). Thymocyte/TEC interactions are apparent in mice with blocked T cell development (Hollander et al., 1995; Shores et al., 1991), indicating that TEC and differentiating thymocytes mutually influence each other in a stepwise fashion to build up the thymic architecture (Klug et al., 1998; van Ewijk et al., 2000). Even in adult life, after the completion of the programmed generation of the thymic architecture, the thymus is highly susceptible to environmental influences and undergoes atrophy with age (Mackall and Gress, 1997). However, it is not yet well understood how the architectural integrity of the thymus is maintained in adult life and how it degenerates with aging. Thymic involution is a physiological process of aging, and it is thought to be a consequence of simultaneous changes in the bone marrow, the extrathymic milieu, and the thymic microenvironment (Mackall and Gress, 1997; Mackall et al., 1998). For example, changes in the level of systemic hormones, such as sex hormones, may be involved in thymic involution (Windmill and Lee, 1999), so that a progressive decline in the thymic mass begins at the age of 6 weeks, which approximates sexual maturity in mice (Hirokawa and Makinodan, 1975). It has recently been determined that changes in the thymic microenvironment are responsible for thymic involution (Mackall et al., 1998), presumably through changes in the signals between TEC and thymocytes. However, which molecular signals are involved in the maintenance and involution of thymic architecture remains unclear. Signal transducer and activator of transcription (Stat) 3 belongs to a family of cytoplasmic proteins that are activated by a large number of extracellular signaling molecules including cytokines, growth factors, and hormones (Leonard and O’Shea, 1998; Schindler and Darnell, 1995), e.g., cytokines of the gp130 family, epidermal growth factor (EGF), platelet-derived growth factor (PDGF), hepatocyte growth factor (HGF), and many other molecules (Akira, 1997; Zhong et al., 1994). Stat3 also plays an antiapoptotic role through upregulation of the Bcl-2 family molecules (Grad et al., 2000; Hirano et al., 2000), while systemic ablation of the Stat3 gene in the germline leads to embryonic lethality (Takeda et al., 1997). To examine the tissue-specific role of Stat3, we previously generated mice in which the floxed Stat3 gene could be truncated by Cre recombinase under the control of a tissue-specific promoter (Akira, 2000; Sano et al., 1999; Takeda et al., 1998). The keratin 5 (K5) promoter was used to ablate the Stat3 gene exclusively in the stratified epithelia (Sano et al., 1999). It was found that, although the initial morphogenesis of the skin appeared normal, wound healing and hair cycling were severely impaired in K5-specific Stat3-disrupted mice

Immunity 262

Figure 1. Stat3 in TEC Is Ablated by Cre-Mediated Recombination under Control of the K5 Promoter (A) Expression of the K5 antigen in the thymus. K5 protein expression in the medulla (right). A bright field of the same section (left). The scale bar represents 250 ␮m. (B) GFP reporter gene expression under the K5 promoter in the thymus. GFP-positive cells both in the medulla and the cortex (left). GFPpositive cells in a network pattern at high magnification (right). Thymi of mice carrying either a GFP reporter gene alone or a K5-Cre gene alone are negative for GFP (data not shown). The scale bar represents 250 ␮m (left) and 40 ␮m (right). (C) GFP expression in cultured TEC. Cultured TEC isolated from 2-week-old K5-Cre:STOPflox-GFP mice are positive for GFP (right). A bright field image (left). Some spindle cells, presumably fibroblasts, are negative for GFP (right). TEC isolated from mice carrying either a GFP reporter gene alone or a K5-Cre gene alone are negative for GFP (data not shown). The scale bar represents 100 ␮m. (D and E) Disruption of Stat3 in TEC. Thymocytes (D, left) and thymic stroma (D, right) separated from newborn mice with the genotype of K5-Cre:Stat3flox/⫺ are subjected to genomic PCR as previously described (Sano et al., 1999). A truncated Stat3 allele (D, arrowhead) is seen in the stroma, but not in the thymocytes, in which only the floxed Stat3 gene (D, asterisk) is observed. Floxed Stat3 found in the stroma is presumably derived from fibroblasts, macrophages, endothelial cells, dendritic cells, or contaminating thymocytes. Lysates from cultured TEC of K5-Cre:Stat3 flox/⫹ (E, ⫹/⫺) and K5-Cre:Stat3 flox/⫺ (E, ⫺/⫺) mice are Western blotted with antiphosphorylated Stat3 (705Y), Stat3 (C-20), and ␤-actin.

(Sano et al., 1999). Their skin phenotype, which included spontaneous development of skin ulcers and alopecia, became more severe as they aged, suggesting that epidermal Stat3 plays a crucial role in the maintenance of postnatal interactions between epithelia and mesenchymal compartments (Sano et al., 1999; Sano et al., 2000). It has been shown that the K5 promoter is also active in thymic epithelia (Klug et al., 1998). The study presented here demonstrates that the Cre-loxP technology using the K5 promoter allows for gene ablation in TEC without affecting the genomes of the thymocytes. We found that the ablation of Stat3 in TEC resulted in severe thymic atrophy with age and the hypersusceptibility of the thymus to environmental stress, such as treatment with glucocorticoid or ␥-irradiation. The thymocytes themselves from mutant mice were, however, no more sensitive than controls. Our results suggest that epithelial Stat3 is essential for maintaining the thymic architecture and thus for the survival of thymocytes. Results Disruption of Stat3 Gene in TEC Consistent with the result of a previous report (Klug et al., 1998), K5 was predominantly expressed in the majority of medullary TEC and cortico-medullary junction TEC (Figure 1A). By contrast, the majority of cortical

TEC do not express K5 (Klug et al., 1998). K5 is strongly expressed in the proliferating compartment of stratified squamous epithelia such as epidermal basal cells and is downregulated along with epidermal differentiation (Fuchs, 1995). It is thus possible that mature TEC, including cortical K5⫺ TEC as well as medullary K5⫹ TEC, may be derived from K5⫹ TEC precursor cells (Klug et al., 1998; Klug et al., 2000). To confirm this, K5-Cre transgenic mice (Tarutani et al., 1997) were mated with mice carrying a CAG-loxP-STOP-loxP-GFP (STOPflox-GFP) transgene (Kawamoto et al., 2000). Under the control of the K5 promoter, Cre-mediated deletion of loxP sites gave rise to the expression of GFP in stratified epithelia. GFP-positive cells were seen diffused all over the thymus (Figure 1B), suggesting that TEC had once activated the K5 promoter, irrespective of K5 expression. In fact, GFP-positive cells were costained in situ with either antimedullary or anticortical TEC (ER-TR5 or ER-TR4, respectively; data not shown). In vitro-cultured TEC isolated from these mice showed positive for GFP (Figure 1C). These results are consistent with the hypothesis that K5⫺ TEC in the mouse thymus are derived from K5⫹ cells. To ablate Stat3 in TEC, we mated mice carrying loxPcontaining (floxed) Stat3 alleles (Stat3flox/flox) with K5-Cre transgenic mice carrying either one Stat3null allele (K5Cre:Stat3⫹/⫺) or one floxed Stat3 allele (K5-Cre:

Role of Stat3 in Thymic Epithelia 263

Stat3flox/⫹). Offspring carrying either the K5-Cre:Stat3flox/⫺ or K5-Cre:Stat3flox/flox genotype showed ablation of the Stat3 gene in the epithelia of the skin, also known as the epidermis (Sano et al., 1999). Allele-specific PCR revealed that a truncated Stat3 allele (Figure 1D, arrowhead) was found in thymic stroma taken from K5Cre:Stat3flox/⫺ mice, whereas only a floxed Stat3 allele was detected in their thymocytes (Figure 1D, asterisk), indicating that Cre-mediated ablation of floxed Stat3 occurred in TEC, but not in thymocytes. Exon 21 of the Stat3 gene, which is the target of Cre, encodes the tyrosine to be phosphorylated. Western blot analysis of lysates from TEC revealed that a reduced amount of the Stat3 protein (Sano et al., 1999; Takeda et al., 1998) and no phosphorylated Stat3 was observed in TEC from K5Cre:Stat3flox/⫺ mice (⫺/⫺) as a result of stimulation with EGF, an activator of Stat3 (Zhong et al., 1994), while Stat3 was phosphorylated in control TEC (⫹/⫺) (Figure 1E). Thus, epithelium-specific ablation of the Stat3 gene resulted in functional disruption of Stat3 in TEC. Thymic Hypoplasia in Epithelium-Specific Stat3-Disrupted Mice Epithelium-specific Stat3-disrupted mice were born normal, and their thymi appeared normal in the neonates in terms of size, thymocyte numbers, and morphological architecture (data not shown). Thymocytes of newborn Stat3⫺/⫺ mice developed normally with regard to surface antigens, including CD4, CD8, CD3, and the TCR␤ chain, as well as differentiation-associated markers such as J11d and c-kit (data not shown). Fetal thymus organ culture (FTOC) showed no impairment in the fetal development of T cells in the thymic microenvironment of mutant mice (data not shown). Thus, epithelium-specific Stat3 ablation did not affect the primary development of the thymus. However, the number of thymocytes of these mice after adolescence significantly decreased with age compared with controls (Figure 2A; 5–7 and 10–12 weeks), even though there was no difference in preadolescence (Figure 2A; 0–7 days). Stat3⫺/⫺ mice at 6 weeks of age developed thymic hypoplasia (Figure 2B, top panel). FACS analyses of these adult mice suggested that immature CD4⫹CD8⫹ double-positive (DP) thymocytes were most severely affected (Figure 2B, bottom panel). It should be noted that the thymi of Stat3⫹/⫺ mice (Stat3flox/⫺ and K5-Cre:Stat3flox/⫹) were essentially indistinguishable from wild-type thymi in size or thymocyte distribution (Figures 2A and 2B). Stat3⫺/⫺ mice older than 4 months harbored progressively atrophic thymi with only small thymic rudiments (Figure 2C). Histological examination revealed that thymi of postadolescent Stat3⫺/⫺ mice showed greatly reduced cellularity, while the separation between the cortex and medulla was no longer obvious (Figure 2E), in contrast to the agematched control thymi with a cortex containing densely packed thymocytes and a well-demarcated medulla (Figure 2D). Higher magnification of the mutant thymus frequently detected pale-stained epithelial cells with foamy cytoplasm (Figure 2F, arrowheads). These results indicate that epithelium-specific Stat3⫺/⫺ mice exhibit severe thymus hypoplasia as the number of thymocytes decreases with age, although TEC of these mice are fully competent to support T cell development during the fetal and neonatal stages.

T Cell Function in Epithelium-Specific Stat3-Disrupted Mice Next, we examined the function of peripheral T cells generated in the K5-specific Stat3⫺/⫺ mice. Lymph node cells (LNC) from 9-week-old Stat3⫺/⫺ mice with severe hypoplasia of the thymus were used to examine if they had normal T cell function. The number of LNC and the ratios of CD3-, CD4-, and CD8-positive cells in these mice were comparable to those in age-matched controls (data not shown). Proliferative responses of LNC from the mutant mice to various T cell stimulants, such as IL-2, ConA, PHA, anti-CD3 antibody, and staphylococcal enterotoxin (SE) B were essentially comparable to those from wild-type mice (medium only: 244 and 147 cpm; IL-2: 3,232 and 3,698 cpm; ConA: 48,349 and 46,920 cpm; PHA: 9,867 and 12,124 cpm; anti-CD3: 39,960 and 47,952 cpm; SEB: 5,089 and 5,306 cpm incorporated in LNC of Stat3⫹/⫺ and Stat3⫺/⫺ mice, respectively). It was also found that aged mutant mice normally elicited contact hypersensitivity to in vivo sensitization with dinitrofluorobenzene (data not shown). These results suggest that epithelium-specific Stat3⫺/⫺ mice are capable of generating normal T cells and maintaining competent T cells in vivo, even after severe involution of the thymus. Morphological Changes in TEC of Stat3-Disrupted Mice Monoclonal antibodies specific for cortical TEC (ERTR4) and medullary TEC (ER-TR5) were used for morphological examination of TEC. There were no morphological changes in TEC of Stat3-disrupted mice at the preadolescent stage (data not shown). As the mutant mice aged, the size of the cortex was greatly reduced, while cortical TEC were found condensed in a thin subcapsular layer (Figure 3B, arrow). They were aligned parallel to the capsule, in contrast to the cortex in wildtype control mice, in which the cortical TEC were oriented perpendicularly to the capsule (Figure 3A, arrows). In addition, there was a high proportion of cells resembling thymic nurse cells (TNC) in Stat3⫺/⫺ mice (Figure 3B, asterisks in inset). Interestingly, these morphological changes in cortical TEC were very similar to those in human CD⑀ transgenic mice or RAGnull mice, in which T cell development was arrested at an early stage (Hollander et al., 1995; van Ewijk et al., 2000). In contrast, medullary TEC of the mutant mice showed a relatively normal appearance (Figure 3B, right). Strikingly, in aged mutant mice, K5 was diffusely expressed in both the cortex and medulla (Figure 3D), while control thymi exhibited K5 predominantly in medullary TEC (Figure 3C). Again, this feature of Stat3⫺/⫺ thymi appears to be very similar to that seen in human CD⑀ transgenic mice, in which TEC diffusely express K5 (Klug et al., 1998). Apoptosis in the Thymi of Adult Epithelial-Specific Stat3-Disrupted Mice To explore the mechanisms underlying age-dependent thymic hypoplasia in epithelium-specific Stat3⫺/⫺ mice, in situ TUNEL analysis was carried out. We established K5-specific Stat3⫺/⫺ mice that also carried a GFP reporter gene (K5-Cre:STOPflox-GFP: Stat3flox/flox) so that

Immunity 264

Figure 2. Thymic Hypoplasia in Epithelium-Specific Stat3-Disrupted Mice (A) Age-related reduction in thymocytes in Stat3⫺/⫺ mice. Values represent the mean cell numbers per thymic lobe ⫾ SEM. The number of mice examined (⫹/⫹, ⫹/⫺, and ⫺/⫺): 0–7 days old, n ⫽ 10, 10, and 12, respectively; 5–7 weeks old, n ⫽ 5, 8, and 7, respectively; 10–12 weeks old, n ⫽ 11, 8, and 7, respectively. An asterisk indicates that p ⬍ 0.05; three asterisks indicated that p ⬍ 0.0005. (B) The gross appearance of thymi (top) and FACS analysis of the corresponding thymus (bottom). Stat3 wild-type (⫹/⫹), hemizygous (⫹/⫺, of K5-Cre:Stat3 flox/⫹), and disrupted (⫺/⫺, of K5-Cre:Stat3 flox/⫺) 6-week-old littermates. Percentages of CD4⫹CD8⫹(DP) cells are shown. This figure shows a representative result of at least ten independent experiments with reproducible observations. (C) Severe thymic atrophy in Stat3⫺/⫺ mice of advancing age (5 months old). This macrograph shows the representative data of at least 9 independent Stat3⫺/⫺ mice older than 2 months with similar results. The scale digit represents 1 mm. (D–F) The histological appearance of thymi from a (D) 9-week-old Stat3 heterozygous and a (E) Stat3⫺/⫺ mouse. The scale bar represents 200 ␮m. (F) A higher magnification of (E) discloses a number of pale-stained epithelial cells with foamy cytoplasm (arrowheads). This micrograph shows the representative results for 11 independent Stat3⫺/⫺ mice older than 2 months with reproducible results. Hematoxylin-eosin stain. The scale bar represents 25 ␮m.

TEC could be identified. We found that adult Stat3⫺/⫺ thymi showed a major increase in the number of apoptotic cells (Figures 4C and 4D), compared with controls (Figures 4A and 4B). Most, if not all, TEC were TUNELnegative (Figures 4C and 4D), whereas most TUNELpositive cells coexpressed CD4 and CD8 (data not shown). This indicates that apoptotic cells consisted of dying thymocytes, not TEC. These apoptotic thymocytes formed clusters and were frequently found adjacent to TEC (Figure 4D). These results indicate that many thymocytes in the adult epithelium-specific Stat3⫺/⫺ mice were undergoing apoptosis, which probably contributed to thymic hypoplasia.

Increased Sensitivity of Stat3-Disrupted Thymi to Apoptotic Stimuli Thymic hypoplasia accompanied by thymocyte apoptosis is acutely induced by various stimuli such as glucocorticoid (Cohen, 1992) and ␥-irradiation (Clarke et al., 1993). To examine whether ablation of Stat3 in TEC might affect the susceptibility of the thymus to apoptotic stimuli, suboptimal doses of dexamethasone (DEX) or ␥-irradiation were administrated to epithelium-specific Stat3⫺/⫺ mice at the preadolescence stage, when changes in their thymi were not yet evident. Strikingly, DEX-administration of suboptimal doses to 17-day-old Stat3⫺/⫺ mice resulted in a marked histological change

Role of Stat3 in Thymic Epithelia 265

Figure 3. Morphology of TEC in Adult Stat3⫺/⫺ Mice (A and B) Immunohistochemical staining of thymi with antibodies to cortical TEC (ER-TR4) and medulla TEC (ER-TR5). Cortical TEC in adult Stat3⫺/⫺ mice (8 weeks old) are condensed in alignment when parallel to the capsule (B, arrow), whereas wild-type cortical TEC are arranged perpendicularly to the capsule and radially to the medullae (A, arrows). The Stat3⫺/⫺ cortex contains a number of cells resembling thymic nurse cells (TNC) (B, asterisks in inset). The medullary TEC of Stat3⫺/⫺ mice (B, right) do not differ in morphology from those of wild-type mice (A, right). The scale bar represents 100 ␮m, and it represents 20 ␮m for the inset of (B). (C and D) Expression of K5 in adult mice (7 weeks old). K5 is largely expressed in the medulla of wild-type mice (C), but K5⫹ cells in Stat3⫺/⫺ mice are diffusely distributed in both the cortex and medulla (D). The scale bar represents 100 ␮m.

in their thymi (Figure 5A, ⫺/⫺; 3–30 ␮g), whereas control thymi were only marginally (Figure 5A, ⫹/⫹; 3 and 10 ␮g) or mildly (Figure 5A, ⫹/⫹; 30 ␮g) affected. The changes in the mutant mice included a major reduction in thymocyte numbers and disorganization of the architecture, including enlargement of medullae even in the subcapsular region (Figure 5A, asterisks). DEX-treatment resulted in a higher ratio of TUNEL-positive apoptotic thymocytes in Stat3⫺/⫺ mice than in control mice (data not shown). At high magnification, Stat3⫺/⫺ thymi treated with DEX (10 ␮g) frequently exhibited TEC carrying ample nuclei (Figure 5A, arrowheads in inset) and foamy cytoplasm containing nuclear debris of lymphoid cells (Figure 5A, arrows in inset). These TEC resembled those found in the thymi of the adult Stat3⫺/⫺ mice (Figure 2F). Likewise, Stat3⫺/⫺ mice exhibited a higher susceptibility to a low dose of ␥-irradiation. Numerous TUNELpositive apoptotic cells appeared in the thymi of Stat3⫺/⫺ GFP reporter mice following 1.5 Gy administration (Figure 5B, right), whereas very few TUNEL-positive cells were detected in controls (Figure 5B, left). GFP expression did not overlap with most TUNEL-positive cells, and this finding is consistent with spontaneous apoptosis of the adult mutant thymi (Figure 4D). When mice were ␥-irradiated at 2 Gy, the reduction in thymocytes was more pronounced in Stat3⫺/⫺ mice than in controls (Figure 5C). Upon total body exposure to 2 Gy of ␥-irradia-

tion, their thymi showed severe loss of lymphoid cells (Figure 5D, asterisk), whereas control thymi were only mildly affected, except for scattered clusters of nuclear debris (Figure 5D, arrowheads). To summarize, the thymi of Stat3⫺/⫺ mice, despite their normal appearance at the preadolescent stage, show a higher sensitivity to apoptotic stimuli, resulting in the loss of thymocytes and the destruction of thymic architecture. To determine if the susceptibility of the thymocytes themselves increased in epithelium-specific Stat3⫺/⫺ mice, isolated thymocytes were directly exposed in vitro to DEX and ␥-irradiation. Figure 5E shows that there was no difference in the sensitivity of the thymocytes to DEX or ␥-irradiation between the mutant and control mice. These results suggest that the increased sensitivity of in vivo thymocytes of epithelial-specific Stat3-disrupted mice to either DEX or ␥-irradiation is not directly due to an increased susceptibility of the thymocytes, but rather to a functional impairment of Stat3⫺/⫺ TEC, resulting in failure to protect thymocytes from apoptotic stimuli. Impaired Regeneration of Thymocytes in Irradiated Thymi of Stat3⫺/⫺ Mice High susceptibility of the thymi of K5-specific Stat3⫺/⫺ mice to DEX or to irradiation suggests the possibility that, upon environmental stress, Stat3⫺/⫺ TEC may be defective in supporting de novo T cell development in

Immunity 266

Figure 4. Apoptosis in the Thymus of Adult Epithelium-Specific Stat3⫺/⫺ Mice Thymi of 9-week-old mice carrying K5-Cre:STOP flox-GFP:Stat3 flox/⫹ (⫹/⫺; [A and B]) and K5-Cre:STOP flox-GFP:Stat3 flox/⫺ (⫺/⫺; [C and D]) are subjected to TUNEL analysis probed with rhodamine. The scale bars represent 100 ␮m for (A) and (C) and 20 ␮m for (B) and (D). Signals of GFP and rhodamine are merged.

the thymus. To examine the capability of the irradiated thymi of epithelium-specific Stat3⫺/⫺ mice to induce thymopoiesis, ␥-irradiated neonatal thymi from these mutant mice were grafted subcutaneously in nude mice. Four weeks after grafting, normal thymic colonization was evident in control grafts (Figure 6A, ⫹/⫹ and ⫹/⫺); however, K5-specific Stat3⫺/⫺ thymic grafts were either completely absorbed or remained as minute rudiments (Figure 6A, ⫺/⫺). Immunofluorescence analyses of these thymus sections revealed normal cortical TEC (Figure 6B, arrowheads) and medullary TEC (Figure 6B, arrows) together with thymocytes regenerated from the donor origin (data not shown) in control grafts, whereas Stat3⫺/⫺ grafts contained far fewer thymocytes (data not shown), along with disorganized architecture without discernible TEC (Figure 6C). Furthermore, sublethally irradiated K5-specific Stat3⫺/⫺ mice were reconstituted with MHC-compatible normal spleen cells containing lymphocyte progenitor cells (Sprent et al., 1995). Irradiated mutant mice were not capable of developing thymocytes within 4 weeks after the reconstitution, unlike the efficient thymocyte regeneration seen in control mice (Figure 6D). Thymocyte repopulation even without cell reconstitution in control mice, perhaps as a result of colonization of relatively radio-resistant T cell progenitors, was also impaired in the mutant mice. These re-

sults indicate that irradiated Stat3⫺/⫺ TEC are defective in their support of de novo thymopoiesis and suggest that Stat3 in TEC plays a crucial role in maintaining the thymic microenvironment required for the survival of developing thymocytes. Upregulation of Gene Expression within Stat3⫺/⫺ TEC It is highly possible that the molecule(s) downstream of Stat3 within TEC is (are) involved in thymocyte survival and the maintenance of thymic architecture. We next investigated if there was any difference in gene expression between Stat3⫺/⫺ and control TEC. DNA microarray analysis used to examine the expression of approximately 800 genes revealed that significant upregulation of 14 genes (ratio ⬎ 2.0) was obtained in Stat3⫺/⫺ TEC compared with Stat3⫹/⫺ TEC, whereas there were no significant number of downregulated genes (ratio ⬍ 0.5) in Stat3⫺/⫺ TEC. Five representative upregulated genes (ratio ⬎ 2.5) are shown in Figure 7A. The most intriguing upregulated gene was H2-O␤, which belongs to the nonclassical class II and plays an important role in antigen processing in B cells, although its role in TEC is unknown (Alfonso and Karlsson, 2000). In fact, the thymi from newborn Stat3⫺/⫺ mice expressed the H2-O␤ protein at a high density not only in the medullary, but also in the cortical, TEC (Figure 7B, right); although, H2-O␤ was

Role of Stat3 in Thymic Epithelia 267

Figure 5. Increased Sensitivity of Thymi in Stat3⫺/⫺ Mice to Apoptotic Stimuli (A) Wild-type (⫹/⫹) and Stat3⫺/⫺ mice (⫺/⫺) (17 days old) were treated intraperitoneally with saline only (0) or 3, 10, and 30 ␮g DEX. Their thymi were examined 24 hr later. Compared with controls, Stat3⫺/⫺ thymi exhibit much higher sensitivity to DEX (3–30 ␮g). Enlargement of the medullae is seen even in the subcapsular area (asterisks). Higher magnification of the Stat3⫺/⫺ thymus treated with 10 ␮g DEX reveals frequently occurring epithelial cells bearing ample nuclei (arrowheads in inset) and foamy cytoplasm, containing nuclear debris of lymphoid cells (arrows in inset). Hematoxylin-eosin stain. The scale bar represents 200 ␮m, and it represents 20 ␮m for the inset. (B) TUNEL staining (rhodamine) of thymus sections from 3-day-old Stat3⫹/⫺ and Stat3⫺/⫺ mice carrying a GFP reporter gene 20 hr after ␥-irradiation at 1.5 Gy. The scale bar represents 40 ␮m. Signals of GFP and rhodamine are merged. (C) Differences in sensitivity of thymocytes to total body ␥-irradiation between controls (⫹/⫹, white bars; ⫹/⫺, shaded bars) and Stat3⫺/⫺ mice (black bars). Three-day-old mice were subjected to ␥-irradiation at the indicated doses, and 20 hr after the radiation, thymocyte numbers were counted. The values represent the mean cell numbers per thymic lobe ⫾ SEM. The number of mice examined (⫹/⫹, ⫹/⫺, and ⫺/⫺): 0 Gy, n⫽ 4, 4, and 4, respectively; 1 Gy, n ⫽ 4, 4, and 5, respectively; 2 Gy, n ⫽ 7, 4, and 5, respectively. Two asterisks indicate that p ⬍ 0.01. (D) ␥-irradiation (2 Gy) has a mild effect on wild-type thymi (3 days old), with scattered apoptotic cell debris in the cortex (arrowheads). Stat3⫺/⫺ thymi at the same dose show a severe reduction in cellularity in the cortex (asterisk). Hematoxylin-eosin stain. The scale bar represents 100 ␮m. (E) Isolated thymocytes from both Stat3⫹/⫺ (squares) and Stat3⫺/⫺ mice (triangles) of 7-day-old littermates showed essentially identical sensitivity to DEX (left) and ␥-irradiation (right). After the in vitro treatment (20 hr) with DEX or ␥-irradiation at the indicated doses, cell viability was determined in triplicate by trypan blue dye exclusion. Dead cells (%) are calculated as follows: (% experimental dye positive ⫺ % dye positive in medium)/(100 ⫺ % dye positive in medium). Values represent the mean percentage of dead cells ⫾ SEM. Thymocytes from three mice in each group were analyzed.

predominantly expressed in the medullary TEC in control mice (Figure 7B, arrows in left panel), as previously reported (Karlsson et al., 1991; Surh et al., 1992). The expression of other genes that were not included in the array was screened by means of RT-PCR. No differences were seen between the two kinds of TEC in the level of mRNAs of stromal-derived lymphokines/chemokines, including IL-7, TSLP, TNF-␣, TARC, and TECK (Figure 7C and data not shown). Surprisingly, ErbB2/neu mRNA in Stat3⫺/⫺ TEC showed a significant increase compared

with controls (Figure 7C). Bol et al. reported that the overexpression of the neu oncogene behind the K5 promoter resulted in the disintegration of thymic architecture and thymocyte apoptosis (Bol et al., 1998), both of which are remarkably similar to the findings for the phenotype in Stat3⫺/⫺ mice. Densitometric semiquantification analysis, normalized with HPRT transcripts, revealed that ErbB2/neu mRNA expression had increased 9-fold in Stat3⫺/⫺ TEC compared with control, while no difference was seen in IL-7 mRNA between the two

Immunity 268

Figure 6. Impaired Regeneration of Thymocytes in Irradiated Thymi of Epithelium-Specific Stat3⫺/⫺ Mice (A–C) Thymic graft to nude mice. Neonatal thymi were ␥-irradiated (10 Gy) and implanted in Balb/c nu/nu mice subcutaneously. Graft thymi were harvested 4 weeks later. The scale digit represents 1 mm. (B) Immunofluorescence examination of control graft with ER-TR4 and ERTR5 reveals well-organized cortical and medullary TEC (arrowheads and arrows, respectively) with fibrotic tissues (asterisks). (C) However, no discernible TEC are found in the remainder of the thymic grafts from Stat3⫺/⫺ mice. The scale bar represents 300 ␮m. (D) Thymus reconstitution with MHC-compatible spleen cells. Thymocyte numbers of ␥-irradiated (5 Gy) Stat3⫹/⫺ mice (hatched bars) or Stat3⫺/⫺ mice (black bars). They were left untreated or transplanted with spleen cells (C57BL/6; 2 ⫻ 107/head) 6 hr later, and the number of thymocytes was assessed at the time indicated. The values represent the mean thymocyte numbers per lobe of two mice in each group.

types of TEC (Figure 7D). Stat3⫺/⫺ thymi also demonstrated an upregulated, diffuse expression of the ErbB2/ neu protein both in the cortical and medullary regions, whereas it is expressed weakly and predominantly in medullary TEC in controls, as also previously reported (Figure 7E) (Bol et al., 1998). Collectively, these results suggest that Stat3 in TEC plays a crucial role in either directly or indirectly repressing putatively responsible genes. As a result, the crosstalk between TEC and thymocytes is destroyed in response to environmental stimuli or with aging, leading to thymocyte apoptosis, destruction of architectural integrity, and, finally, thymic atrophy.

Discussion Age-Associated Development of Thymic Alterations in Stat3⫺/⫺ Mice Epithelium-specific Stat3⫺/⫺ mice exhibit normal thymic development at birth and, hence, normal T cell development. However, the thymi of mutant mice aged older than 5 weeks undergo hypoplasia, which becomes more severe with age. The phenotype of the mutant thymus includes a reduction in the number of thymocytes and morphological changes in TEC, the cortical TEC in particular. Normal thymic involution begins from early adolescence (George and Ritter, 1996), which approximates

Role of Stat3 in Thymic Epithelia 269

Figure 7. Upregulated Gene Expression in Stat3⫺/⫺ TEC (A) DNA microarray analysis. A scatter plot of fluorescence intensity to compare gene expression in thymic stroma from newborn Stat3⫹/⫺ (Cy5) and Stat3⫺/⫺ mice (Cy3). Differential expressions of signals between Stat3⫹/⫺ and Stat3⫺/⫺ stroma are attributed to TEC, because Stat3 is ablated exclusively in TEC, not in other stromal cells in Stat3⫺/⫺ thymi. The numbers on the diagonal lines show ratios by which Cy3 signal intensity increases over that of Cy5. (B) H2-O␤ expression in Stat3⫺/⫺ thymi. Thymic sections from Stat3⫹/⫺ and Stat3⫺/⫺ mice (3 days old) are stained with ER-TR4 together with biotinylated anti-H2-O␤ (K535), followed by FITC-anti-rat IgG and avidin-rhodamine, respectively. Different antiserum to H2-O␤ (K536) yielded identical results (data not shown). H2-O␤ expression does not overlap that of cortical TEC in control (arrows), while it is also expressed in cortical TEC in Stat3⫺/⫺ thymi. The scale bar represents 100 ␮m. Signals of FITC and rhodamine are merged. (C) RT-PCR analysis of Stat3⫺/⫺ TEC. Total RNAs from thymic stromal tissue of 3-dayold Stat3⫹/⫺ and Stat3⫺/⫺ mice are reverse transcribed and subjected to PCR using specific primers for IL-7 (top), ErbB2/neu (middle), and HPRT (bottom) at the respective optimal number of cycles: 32, 35, 28. To compare the quantities of mRNAs in the two kinds of stroma, the number of cycles is reduced by 2 (⫺2) and 4 (⫺4) cycles from the optimal cycles (0) for each PCR reaction. Results show representative data of four independent experiments from different batches of siblings with similar results. (D) Semiquantification of mRNA. Signals of PCR products at the optimal cycles are analyzed with a densitometer on the basis of signals of HPRT products. Shaded bar, Stat3⫹/⫺; black bar, Stat3⫺/⫺. Representative data are shown for five independent experiments with similar results. (E) Immunohistochemical staining of ErbB2/ neu in thymi of 28-day-old Stat3⫹/⫺ (left) and Stat3⫺/⫺ (right) mice. The scale bar represents 200 ␮m.

the time of the onset of thymic changes in Stat3⫺/⫺ mice. Hirokawa et al. reported that aging resulted in a selective loss of cortical TEC (Hirokawa and Makinodan, 1975), a selectivity that also characterized the early change in our adult Stat3⫺/⫺ mice. Therefore, it is likely that Stat3 ablation in TEC results in the acceleration of thymic changes associated with aging. Normal thymic involution is synchronized with sexual maturity (Hirokawa and Makinodan, 1975). Many lines of evidence have demonstrated that sex hormones induce thymic involution (Brunelli et al., 1992; Tibbetts et al., 1999). Severe thymic atrophy is, however, observed in Stat3⫺/⫺ mice regardless of sex or pregnancy. Also, it is known that the serum glucocorticoid levels reach adult levels after adolescence in rodents (Henning, 1978) and that the administration of glucocorticoids can induce thymic atrophy (Cohen, 1992). Since Stat3⫺/⫺ mice develop skin abnormalities such as ulcers and alopecia (Sano et al., 1999), it was considered possible that Stat3⫺/⫺ mice exhibited

a higher level of stress-associated glucocorticoids (Sapolsky et al., 1986). However, there was no significant increase in the circulating level of a representative murine glucocorticoid, corticosterone, in Stat3⫺/⫺ mice bearing skin lesions (S. Sano, unpublished data). Rather, ablation of Stat3 in TEC appeared to increase the sensitivity to glucocorticoids. Histologically, the thymi of epithelium-specific Stat3⫺/⫺ mice showed much higher sensitivity to DEX, approximately ten times higher than those of wild-type. Taken together, the observations made in our study suggest that Stat3 ablation in TEC results in intrathymic alterations that accelerate thymic aging. Thymocytes Are Prone to Apoptosis in Epithelium-Specific Stat3⫺/⫺ Mice Stat3 is an antiapoptotic regulator, which activates the transcription of Bcl-xL or Bcl-2 (Fukada et al., 1996; Grad et al., 2000; Takeda et al., 1998). Antisense Stat3

Immunity 270

therapies reduce Bcl-xL and induce apoptosis in carcinoma (Grandis et al., 2000). Interestingly, aged or stressinduced phenotypes of the thymus of epithelial-specific Stat3⫺/⫺ mice were characterized by numerous apoptotic thymocytes while TEC were viable, even though Stat3 was disrupted in the TEC, not in the thymocytes. Our results contrast with the report that apoptotic thymocytes did not increase in the thymi of T cell lineagespecific Stat3-disrupted mice, in which thymocytes were affected but TEC remained intact (Takeda et al., 1997). Combining this finding with our result that isolated thymocytes from epithelial-specific Stat3⫺/⫺ mice exhibited no increased sensitivity to apoptotic stimuli in vitro strongly suggests that TEC determine the fate of thymocytes in vivo. Furthermore, cytokines secreted by thymic stroma are critical for thymocyte survival (Ritter and Boyd, 1993). We examined mRNAs for IL-7 and TSLP, which share certain properties in vitro (Levin et al., 1999), but found no decrease in Stat3⫺/⫺ TEC. It was recently reported that glucocorticoids produced by TEC may induce thymocyte apoptosis, although this remains controversial (Ashwell et al., 2000; Pazirandeh et al., 1999; Purton et al., 2000). As far as we could determine, however, there was no substantial difference between the thymi of Stat3⫺/⫺ and control mice in the expression of P450scc, a rate-limiting enzyme for steroidogenesis (Ashwell et al., 2000) (S. Sano, unpublished data). Abnormalities of Stat3⫺/⫺ TEC Thymic development is a complex process that depends on mutually inductive and stepwise, two-way interactions between thymocytes and TEC (van Ewijk et al., 2000). Since the T cell progenitors of Stat3⫺/⫺ mice remain intact, autonomous changes in TEC appear to be primarily responsible for the phenotype. Our results clearly demonstrated that thymocytes failed to develop in the mutant thymus with a microenvironment damaged by ␥-irradiation. A microarray of approximately 800 genes revealed that 14 genes were upregulated differentially in TEC from newborn Stat3⫺/⫺ mice, whereas no genes were downregulated. It has not been determined whether ␣-fetoprotein and ␥-glutamyl transferase, the two genes showing the highest upregulation in the mutant thymi, are involved in the thymic events. The most notable gene found in the array was H2-O␤, a nonclassical MHC class II gene that is largely restricted to B cells and TEC (Karlsson et al., 1991; Surh et al., 1992). It was reported that H2-O is tightly associated with H2-M in the endosomal system, resulting in alterations of the repertoire of antigens in B cells (Alfonso and Karlsson, 2000). Unlike B cells, however, TEC still expressed H2O␤ in H2-O␣-deficient mice (Liljedahl et al., 1998), suggesting a specific function in the thymus. In fact, the H2-O␤ protein was expressed with higher density in Stat3⫺/⫺ thymi than in control thymi and was distributed both in the cortex and medulla, while it was expressed largely in the medullary TEC in control mice. It remains unknown, however, whether H2-O is involved in thymocyte apoptosis. Further, we found that ErbB2/neu was upregulated in Stat3⫺/⫺ TEC. ErbB2/neu belongs to the ErbB family, one of the receptor tyrosine kinase families. This family consists of the EGF receptor (ErbB1), ErbB2/neu, ErbB3,

and ErbB4 (Bargmann et al., 1986). A specific ligand of ErbB2/neu has not been identified. Rather, it has been suggested that ErbB2/neu functions primarily as a counterpart of heterodimers with other ErbB members to expand the array of downstream signalings upon stimulation by ligand binding (Wada et al., 1990). Interestingly, neu oncogene transgenic mice behind the K5 promoter developed not only skin tumors, but also changes in the thymus, which share remarkable similarities to those found in Stat3⫺/⫺ mice (Bol et al., 1998). We suggest that ErbB2/neu overexpression occurs when Stat3 is deficient. Moreover, the K5 protein was expressed in both cortical and medullary TEC in adult Stat3⫺/⫺ mice, as is also the case in hCD3⑀tg mice (Klug et al., 1998). A low dose of ␥-irradiation (1 Gy) converted K5⫺ TEC into K5⫹ TEC in the cortex of the Stat3⫺/⫺ neonates, unlike in that of wild-type mice (S. Sano, unpublished data). Ablation of Stat3 may result in “dedifferentiation” of cortical TEC by aging or apoptotic stimuli, although K5⫺ cortical TEC would normally have differentiated from K5⫹ TEC stem cells (Klug et al., 2000). It is reported that K5 promoter activity is regulated by AP-1 transcription factors, which are distributed in a differentiationspecific manner (Rossi et al., 1998) and, most importantly, which are the downstream target of ErbB2/neu (Galang et al., 1996). We therefore suggest that aberrant K5 expression in the Stat3⫺/⫺ thymus might be attributed to ErbB2/neu overexpression. To summarize them, the Stat3⫺/⫺ thymus is characterized by aberrant expression of the molecules; K5, ErbB2/neu, and H2-O␤, all of which are otherwise expressed predominantly in medullary TEC and presumably downregulated in the cortex as they differentiate. Although growth factors or cytokines required for Stat3 activation in TEC are undefined, an unbalanced signaling cascade, such as the SHP2-MAPK predominant pathway, may be elicited by a given ligand in the absence of Stat3 (Ohtani et al., 2000). In addition, overexpressed ErbB2/neu could associate with gp130 to augment the MAP kinase pathway as previously reported (Qiu et al., 1998). We conclude that Stat3 of TEC regulates and coordinates the signal crosstalks so that they can promote TEC differentiation, while Stat3 represses the specific genes that induce thymocyte apoptosis and destroy the thymic architecture in response to environmental stimuli, including aging. Experimental Procedures Mice The generation of keratinocyte-specific Stat3-disrupted mice was previously described. Briefly, mice carrying a keratin 5 promoterdriven (K5)-Cre transgene and a Stat3null allele (K5-Cre:Stat3⫹/⫺) were mated with Stat3flox/flox mice. Offspring carrying a floxed Stat3 allele and/or K5-Cre transgene (K5-Cre:Stat3flox/⫹, K5-Cre:Stat3flox/⫺, Stat3flox/⫹, Stat3flox/⫺) were used for subsequent analyses. For each experiment, controls were taken from littermates of Stat3flox/⫹ mice as Stat3⫹/⫹ (wild-type), Stat3flox/⫺, or K5-Cre:Stat3flox/⫹ mice as Stat3⫹/⫺. In some experiments, K5-Cre:Stat3flox/flox mice were used as Stat3⫺/⫺. Mice carrying a GFP reporter gene were described previously (Kawamoto et al., 2000). Briefly, the reporter transgene contains the CAG promoter, the loxP-flanked CAT gene with an SV40 polyadenylation signal, and the GFP gene at the 3⬘ end (STOPflox-GFP). Mice carrying a K5-Cre:STOPflox-GFP:Stat3flox/flox gene were generated to visualize TEC with disrupted Stat3. Care of the mice was in accordance with our institutional guidelines.

Role of Stat3 in Thymic Epithelia 271

Immunohistology The antibodies that were used included: anti-cortical TEC (ER-TR4), anti-medullary TEC (ER-TR5), anti-K5 (Babco), anti-ErbB2/neu (sc284; Santa Cruz Biotechnology), anti-H2-O␤ (K535 and K536, kindly provided by Dr. C. Surh and Dr. L. Karlsson, respectively), and normal rat or rabbit sera (DAKO, Glostrup) as controls for the first antibodies. For immunohistochemical staining, 5 ␮m frozen sections were stained with antibodies, incubated with a rabbit anti-rat horseradish peroxidase conjugate (DAKO), and visualized by diaminobenzidine. For immunofluorescence staining, sections were incubated with the first antibodies and then with FITC- or rhodamine-conjugated second antibodies (Sigma). For some biotinylated antibodies, streptavidin-rhodamine (Sigma) was used for the second reagent. For detection with anti-ErbB/neu, the tissues were fixed in formalin, embedded in paraffin, and cut into 4-␮m sections, which were microwaved before staining. For observation of GFP, tissues were fixed with 4% paraldehyde overnight and then cut with a cryostat. TUNEL staining was performed according to the manufacturer’s protocol (Apoptosis Kit II, MBL) and probed with rhodamine (Sigma). Specimens were examined under a fluorescence microscope or a confocal fluorescence microscope (Olympus). FACS Analysis The antibodies comprised anti-CD4, anti-CD8, anti-CD3, and antiTCR␤ chain (Pharmingen). Thymocytes were incubated with the antibodies for 30 min at 4⬚C, and then analyzed with a FACSsort (Becton Dickinson). Isolation of Thymic Stroma and In Vitro Culture of TEC TEC were isolated according to the method of Imaizumi et al., with a minor modification (Imaizumi et al., 1993). Briefly, thymi from newborn to 5-day-old mice were minced gently, and the fragments were pipetted in a layer on fetal calf serum (FCS) to remove thymocytes. After they had been left standing for 5 min, the thymic capsule fragments were harvested in the sediments, which were used as thymic stromal fractions, and suspended in MCDB medium (Kyokuto) supplemented with 5 ␮g/ml insulin (Sigma), 50 ng/ml EGF (Upstate Biotechnology), 0.1 mM monoethanolamine, 0.1 mM phosphoryl ethanolamine, and 60 ␮g/ml kanamycin monosulfate (Sigma), and cultured in dishes precoated with collagen type I (Iwaki Glass) at 37⬚C in an atmosphere of 5% CO2/95% air for 7–10 days, during which TEC outgrew other stromal cells. Allele-Specific Genomic PCR For detection of the floxed Stat3 allele and the truncated Stat3 allele in the genomic DNA from thymic stroma and thymocytes, PCR reactions were performed with a mixture of three primers as previously reported (Sano et al., 1999). PCR products were electrophoresed on a 2% agarose gel and visualized with ethidium bromide staining. Western Blot Analysis Cultured TEC were lysed in RIPA buffer (10 mM Tris-HCl [pH 7.5], 1% NP40, 0.1% deoxycholate, 0.1% SDS, 0.15 M NaCl, 1 mM EDTA, 10 ng/ml aprotinin, 0.2 mM PMSF, 1 mM Na3VO4), and equal amounts of protein were electrophoresed on 7.5% SDS-polyacrylamide gels, transferred to nitrocellulose membranes, and blotted with anti-Stat3 (C-20; Santa Cruz), anti-phophorylated Stat3 (Tyr705; Cell signaling Technology), or anti-␤ actin (Sigma). After treatment with horseradish peroxidase-conjugated second antibodies, bands were visualized with the ECL detection system (Amersham Pharmacia Biotech). Thymic Graft Donor thymi were collected from neonatal wild-type, Stat3⫹/⫺, and Stat3⫺/⫺ mice and ␥-irradiated at 10 Gy. The thymi were then implanted subcutaneously in 11-week-old Balb/c nu/nu mice. Four weeks later, the grafts were harvested and examined. Cell Reconstitution Assay Four-week-old Stat3⫹/⫺ and Stat3⫺/⫺ mice were ␥-irradiated at 5 Gy, and 6 hr later, spleen cells from C57BL/6 mice were intravenously

transplanted (2 ⫻ 107/head) into the hosts. Thymi of the recipient mice were examined 4 weeks later. T Cell Proliferation Assay Lymph node cells from 9-week-old mice were collected and suspended in RPMI1640 medium supplemented with 10% heat-inactivated FCS, 2 mM L-glutamine, 50 ␮M 2-mercaptoethanol, and antibiotics. Cells (2 ⫻ 105 per well) were seeded in triplicate in a 96-well plate (Corning) in the absence or presence of 50 U/ml IL-2 (Shionogi), 1 ␮g/ml ConA (Sigma), 2 ␮g/ml PHA (Sigma), 0.1 ␮g/ml anti-CD3 antibody (2C11, Pharmingen), or 0.1 ␮g/ml SEB (Toxin Technology), cultured for 48 hr at 37⬚C in an atmosphere of 5% CO2/95% air, and pulsed with 0.5 ␮Ci [3H]thymidine for the last 4 hr. Cells were harvested onto a glass filter, and the radioactivity was counted with a ␤-scintillation counter. RT-PCR Total RNA from thymic stroma was extracted with the RNA isolation kit (Promega), reverse transcribed using random oligonucleotide hexomers, and PCR amplified for mouse hypoxanthine phosphoribosyltransferase (HPRT) (5⬘-CACAGGACTAGAACACCTGC-3⬘ and 5⬘-GCTGGTGAAAAGGACCTCT-3⬘), mouse IL-7 (5⬘-ACCCAG CGCAAAGTAGAAAC-3⬘ and 5⬘-ACAGGCAGCAGAACAAGGAT-3⬘), mouse TSLP (5⬘-TGCAAGTACTAGTACGGATGGGGC-3⬘ and 5⬘GGACTTCTTGTGCCATTTCCTGAG-3⬘), and mouse ErbB2/neu (5⬘-TCTGCCTGACATCCACAGTG-3⬘ and 5⬘-AATAGATTCCAATGC CATCC-3⬘). The optimal annealing temperature was 60⬚C for all the primers, except for ErbB2/neu, for which the temperature was 57⬚C, and reaction cycles were set as indicated in the text. In some experiments, amplified signals were scanned with a densitometer and analyzed on NIH Image software to evaluate the transcripts semiquantitatively. Microarray Hybridization Total RNAs isolated from thymic stroma of Stat3⫹/⫺ and Stat⫺/⫺ mice were pooled (15 mice each, ranging from newborn to 3 weeks old), and 14 ␮g total RNA from each group was reverse transcribed in the presence of Cy5- or Cy3-labeled dUTP, respectively, mixed, and simultaneously hybridized to the IntelliGene Mouse CHIP Set I (Takara Biomedical) with approximately 600 known and 300 EST mouse genes (for detailed information, refer to http://www.takara. co.jp/bio/). The array was rinsed and scanned for the two fluorescent colors independently (GMS 428 Array Scanner, Affymetrix). The signal intensity for each spot was determined with ImaGene software (Ver.4.0; BioDiscovery). Significant signals were obtained for approximately 800 genes in the array. Acknowledgments We would like to thank Mrs. S. Okamoto for her outstanding technical assistance with immunofluorescence studies, Dr. A. Farr for helpful discussions, Drs. S. Akira and K. Takeda for providing the mice with floxed Stat3 and Stat3null alleles, and Drs. C. Surh and L. Karlsson for their generous gifts of anti-H2-O␤ antibodies. Received January 9, 2001; revised June 29, 2001. References Akira, S. (1997). IL-6-regulated transcription factors. Int. J. Biochem. Cell. Biol. 29, 1401–1418. Akira, S. (2000). Roles of STAT3 defined by tissue-specific gene targeting. Oncogene 19, 2607–2611. Alfonso, C., and Karlsson, L. (2000). Nonclassical MHC class II molecules. Annu. Rev. Immunol. 18, 113–142. Anderson, G., Moore, N.C., Owen, J.J., and Jenkinson, E.J. (1996). Cellular interactions in thymocyte development. Annu. Rev. Immunol. 14, 73–99. Ashwell, J.D., Lu, F.W., and Vacchio, M.S. (2000). Glucocorticoids in T cell development and function. Annu. Rev. Immunol. 18, 309–345. Bargmann, C.I., Hung, M.C., and Weinberg, R.A. (1986). The neu

Immunity 272

oncogene encodes an epidermal growth factor receptor-related protein. Nature 319, 226–230. Bol, D., Kiguchi, K., Beltran, L., Rupp, T., Moats, S., Gimenez-Conti, I., Jorcano, J., and DiGiovanni, J. (1998). Severe follicular hyperplasia and spontaneous papilloma formation in transgenic mice expressing the neu oncogene under the control of the bovine keratin 5 promoter. Mol. Carcinog. 21, 2–12. Boyd, R.L., Tucek, C.L., Godfrey, D.I., Izon, D.J., Wilson, T.J., Davidson, N.J., Bean, A.G., Ladyman, H.M., Ritter, M.A., and Hugo, P. (1993). The thymic microenvironment. Immunol. Today 14, 445–459. Brunelli, R., Frasca, D., Spano, M., Zichella, L., and Doria, G. (1992). Gonadectomy in old mice induces thymus regeneration but does not recover mitotic responsiveness. Ann. N Y Acad. Sci. 673, 252–255. Capone, M., Romagnoli, P., Beermann, F., MacDonald, H.R., and van Meerwijk, J.P. (2001). Dissociation of thymic positive and negative selection in transgenic mice expressing major histocompatibility complex class I molecules exclusively on thymic cortical epithelial cells. Blood 97, 1336–1342. Clarke, A.R., Purdie, C.A., Harrison, D.J., Morris, R.G., Bird, C.C., Hooper, M.L., and Wyllie, A.H. (1993). Thymocyte apoptosis induced by p53-dependent and independent pathways. Nature 362, 849–852. Cohen, J.J. (1992). Glucocorticoid-induced apoptosis in the thymus. Semin. Immunol. 4, 363–369. Fuchs, E. (1995). Keratins and the skin. Annu. Rev. Cell Dev. Biol. 11, 123–153. Fukada, T., Hibi, M., Yamanaka, Y., Takahashi-Tezuka, M., Fujitani, Y., Yamaguchi, T., Nakajima, K., and Hirano, T. (1996). Two signals are necessary for cell proliferation induced by a cytokine receptor gp130: involvement of STAT3 in anti-apoptosis. Immunity 5, 449–460. Galang, C.K., Garcia-Ramirez, J., Solski, P.A., Westwick, J.K., Der, C.J., Neznanov, N.N., Oshima, R.G., and Hauser, C.A. (1996). Oncogenic Neu/ErbB-2 increases ets, AP-1, and NF-kappaB-dependent gene expression, and inhibiting ets activation blocks Neu-mediated cellular transformation. J. Biol. Chem. 271, 7992–7998. George, A.J., and Ritter, M.A. (1996). Thymic involution with aging: obsolescence or good housekeeping? Immunol. Today 17, 267–272. Grad, J.M., Zeng, X.R., and Boise, L.H. (2000). Regulation of BclxL: a little bit of this and a little bit of STAT. Curr. Opin. Oncol. 12, 543–549. Grandis, J.R., Drenning, S.D., Zeng, Q., Watkins, S.C., Melhem, M.F., Endo, S., Johnson, D.E., Huang, L., He, Y., and Kim, J.D. (2000). Constitutive activation of Stat3 signaling abrogates apoptosis in squamous cell carcinogenesis in vivo. Proc. Natl. Acad. Sci. USA 97, 4227–4232.

Kawamoto, S., Niwa, H., Tashiro, F., Sano, S., Kondoh, G., Takeda, J., Tabayashi, K., and Miyazaki, J. (2000). A novel reporter mouse strain that expresses enhanced green fluorescent protein upon Cremediated recombination. FEBS Lett. 470, 263–268. Klug, D.B., Carter, C., Crouch, E., Roop, D., Conti, C.J., and Richie, E.R. (1998). Interdependence of cortical thymic epithelial cell differentiation and T-lineage commitment. Proc. Natl. Acad. Sci. USA 95, 11822–11827. Klug, D.B., Crouch, E., Carter, C., Coghlan, L., Conti, C.J., and Richie, E.R. (2000). Transgenic expression of cyclin D1 in thymic epithelial precursors promotes epithelial and T cell development. J. Immunol. 164, 1881–1888. Leonard, W.J., and O’Shea, J.J. (1998). Jaks and STATs: biological implications. Annu. Rev. Immunol. 16, 293–322. Levin, S.D., Koelling, R.M., Friend, S.L., Isaksen, D.E., Ziegler, S.F., Perlmutter, R.M., and Farr, A.G. (1999). Thymic stromal lymphopoietin: a cytokine that promotes the development of IgM⫹ B cells in vitro and signals via a novel mechanism. J. Immunol. 162, 677–683. Liljedahl, M., Winqvist, O., Surh, C.D., Wong, P., Ngo, K., Teyton, L., Peterson, P.A., Brunmark, A., Rudensky, A.Y., Fung-Leung, W.P., and Karlsson, L. (1998). Altered antigen presentation in mice lacking H2-O. Immunity 8, 233–243. Mackall, C.L., and Gress, R.E. (1997). Thymic aging and T-cell regeneration. Immunol. Rev. 160, 91–102. Mackall, C.L., Punt, J.A., Morgan, P., Farr, A.G., and Gress, R.E. (1998). Thymic function in young/old chimeras: substantial thymic T cell regenerative capacity despite irreversible age-associated thymic involution. Eur. J. Immunol. 28, 1886–1893. Ohtani, T., Ishihara, K., Atsumi, T., Nishida, K., Kaneko, Y., Miyata, T., Itoh, S., Narimatsu, M., Maeda, H., Fukada, T., et al. (2000). Dissection of signaling cascades through gp130 in vivo: reciprocal roles for STAT3- and SHP2-mediated signals in immune responses. Immunity 12, 95–105. Pazirandeh, A., Xue, Y., Rafter, I., Sjovall, J., Jondal, M., and Okret, S. (1999). Paracrine glucocorticoid activity produced by mouse thymic epithelial cells. FASEB J. 13, 893–901. Purton, J.F., Boyd, R.L., Cole, T.J., and Godfrey, D.I. (2000). Intrathymic T cell development and selection proceeds normally in the absence of glucocorticoid receptor signaling. Immunity 13, 179–186. Qiu, Y., Ravi, L., and Kung, H.J. (1998). Requirement of ErbB2 for signalling by interleukin-6 in prostate carcinoma cells. Nature 393, 83–85. Ritter, M.A., and Boyd, R.L. (1993). Development in the thymus: it takes two to tango. Immunol. Today 14, 462–469.

Henning, S.J. (1978). Plasma concentrations of total and free corticosterone during development in the rat. Am. J. Physiol. 235, E451– E456.

Ropke, C., Van Soest, P., Platenburg, P.P., and Van Ewijk, W. (1995). A common stem cell for murine cortical and medullary thymic epithelial cells? Dev. Immunol. 4, 149–156.

Hirano, T., Ishihara, K., and Hibi, M. (2000). Roles of STAT3 in mediating the cell growth, differentiation and survival signals relayed through the IL-6 family of cytokine receptors. Oncogene 19, 2548– 2556.

Rossi, A., Jang, S.I., Ceci, R., Steinert, P.M., and Markova, N.G. (1998). Effect of AP1 transcription factors on the regulation of transcription in normal human epidermal keratinocytes. J. Invest. Dermatol. 110, 34–40.

Hirokawa, K., and Makinodan, T. (1975). Thymic involution: effect on T cell differentiation. J. Immunol. 114, 1659–1664.

Sano, S., Itami, S., Takeda, K., Tarutani, M., Yamaguchi, Y., Miura, H., Yoshikawa, K., Akira, S., and Takeda, J. (1999). Keratinocytespecific ablation of Stat3 exhibits impaired skin remodeling, but does not affect skin morphogenesis. EMBO J 18, 4657–4668.

Hollander, G.A., Wang, B., Nichogiannopoulou, A., Platenburg, P.P., van Ewijk, W., Burakoff, S.J., Gutierrez-Ramos, J.C., and Terhorst, C. (1995). Developmental control point in induction of thymic cortex regulated by a subpopulation of prothymocytes. Nature 373, 350–353. Hugo, P., Kappler, J.W., and Marrack, P.C. (1993). Positive selection of TcR alpha beta thymocytes: is cortical thymic epithelium an obligatory participant in the presentation of major histocompatibility complex protein? Immunol. Rev. 135, 133–155.

Sano, S., Kira, M., Takagi, S., Yoshikawa, K., Takeda, J., and Itami, S. (2000). Two distinct signaling pathways in hair cycle induction: Stat3-dependent and -independent pathways. Proc. Natl. Acad. Sci. USA 97, 13824–13829. Sapolsky, R.M., Krey, L.C., and McEwen, B.S. (1986). The neuroendocrinology of stress and aging: the glucocorticoid cascade hypothesis. Endocr. Rev. 7, 284–301.

Imaizumi, A., Goldschneider, I., and Yoshida, T. (1993). Reproducible procedures for establishing mouse thymic stromal cell lines. Cell. Immunol. 150, 81–89.

Schindler, C., and Darnell, J.E., Jr. (1995). Transcriptional responses to polypeptide ligands: the JAK-STAT pathway. Annu. Rev. Biochem. 64, 621–651.

Karlsson, L., Surh, C.D., Sprent, J., and Peterson, P.A. (1991). A novel class II MHC molecule with unusual tissue distribution. Nature 351, 485–488.

Shores, E.W., Van Ewijk, W., and Singer, A. (1991). Disorganization and restoration of thymic medullary epithelial cells in T cell receptornegative scid mice: evidence that receptor-bearing lymphocytes

Role of Stat3 in Thymic Epithelia 273

influence maturation of the thymic microenvironment. Eur. J. Immunol. 21, 1657–1661. Sprent, J., Hurd, M., Schaefer, M., and Heath, W. (1995). Split tolerance in spleen chimeras. J. Immunol. 154, 1198–1206. Surh, C.D., Gao, E.K., Kosaka, H., Lo, D., Ahn, C., Murphy, D.B., Karlsson, L., Peterson, P., and Sprent, J. (1992). Two subsets of epithelial cells in the thymic medulla. J. Exp. Med. 176, 495–505. Takeda, K., Kaisho, T., Yoshida, N., Takeda, J., Kishimoto, T., and Akira, S. (1998). Stat3 activation is responsible for IL-6-dependent T cell proliferation through preventing apoptosis: generation and characterization of T cell-specific Stat3-deficient mice. J. Immunol. 161, 4652–4660. Takeda, K., Noguchi, K., Shi, W., Tanaka, T., Matsumoto, M., Yoshida, N., Kishimoto, T., and Akira, S. (1997). Targeted disruption of the mouse Stat3 gene leads to early embryonic lethality. Proc. Natl. Acad. Sci. USA 94, 3801–3804. Tarutani, M., Itami, S., Okabe, M., Ikawa, M., Tezuka, T., Yoshikawa, K., Kinoshita, T., and Takeda, J. (1997). Tissue-specific knockout of the mouse Pig-a gene reveals important roles for GPI-anchored proteins in skin development. Proc. Natl. Acad. Sci. USA 94, 7400– 7405. Tibbetts, T.A., DeMayo, F., Rich, S., Conneely, O.M., and O’Malley, B.W. (1999). Progesterone receptors in the thymus are required for thymic involution during pregnancy and for normal fertility. Proc. Natl. Acad. Sci. USA 96, 12021–12026. van Ewijk, W. (1991). T-cell differentiation is influenced by thymic microenvironments. Annu. Rev. Immunol. 9, 591–615. van Ewijk, W., Hollander, G., Terhorst, C., and Wang, B. (2000). Stepwise development of thymic microenvironments in vivo is regulated by thymocyte subsets. Development 127, 1583–1591. van Ewijk, W., Shores, E.W., and Singer, A. (1994). Crosstalk in the mouse thymus. Immunol. Today 15, 214–217. Wada, T., Qian, X.L., and Greene, M.I. (1990). Intermolecular association of the p185neu protein and EGF receptor modulates EGF receptor function. Cell 61, 1339–1347. Windmill, K.F., and Lee, V.W. (1999). Influences of surgical castration on the thymus of male rats. J. Reprod. Immunol. 44, 29–39. Zhong, Z., Wen, Z., and Darnell, J.E., Jr. (1994). Stat3: a STAT family member activated by tyrosine phosphorylation in response to epidermal growth factor and interleukin-6. Science 264, 95–98.