Tbata modulates thymic stromal cell proliferation and thymus function

While investigating stromal cell function in aged C57BL/6 thymi, we observed a 10-fold increase in steady-state mRNA levels of Tbata in C57BL/6 thymus from 4-mo-old mice versus 1-mo-old mice (Fig. 1 A). The increase in thymic Tbata expression at 4 mo of age correlates with the onset of decreased TEC proliferation and numbers observed in the C57BL/6 thymus during aging (Gray et al., 2006). In C57BL/6 mice aged 4–24 mo, the expression of Tbata remains high during a period when thymic stromal cells do not proliferate and their numbers steadily decline (Gray et al., 2006). Modest increases (two- to threefold) in Tbata expression were observed in cerebellum and testes during aging (Fig. S1, A and B). Although the pattern of Tbata expression during aging is similar in thymus, cerebellum, and testes, the magnitude of the age-dependent increase in the thymus is distinct. To extend our findings in the thymus, we examined DBA/2 mice that exhibit a more rapid and severe age-dependent decrease in thymus function compared with C57BL/6 mice (Hsu et al., 2003, 2005). Although Tbata levels in DBA/2 thymus were similar to C57BL/6 mice at 1 mo of age, we noted an 8–10-fold increase in Tbata mRNA levels in DBA/2 mice by 2 mo. By 3 mo, the level of Tbata mRNA had doubled again and remained high as the mice aged up to 9 mo (Fig. 1 A). Therefore, we observed high-level Tbata expression during the period when thymus function is decreasing in these two strains of mice.

Given the changes in cellular composition of the thymus during aging, it was important to confirm that the increased Tbata expression observed was intrinsic to TEC. We examined Tbata levels in stromal cells sorted from 1- or 10-mo-old C57BL/6 mice (Fig. S1 C). Tbata RNA levels were increased about threefold in Ly51⁺ cells and ∼27-fold in UEA⁺ cells sorted from the aged thymi (Fig. 1 B). Therefore, TEC isolated from aged thymi express higher levels of Tbata than those sorted from young thymi.

These data suggested that increased Tbata expression might play a role in decreasing thymus size. To explore whether Tbata expression might be dynamically regulated, we examined Tbata expression in response to a treatment that increases thymus size. An increase in thymus size coincident with an increase in the proportion and numbers of proliferating TEC occurs within 1 wk after castration in C57BL/6 mice (Gray et al., 2006; Williams et al., 2008). We measured Tbata expression in thymi from 4-mo-old castrates and sham-castrates over a 3-wk period after surgery. Within 1 wk, thymus weight increased ∼2-3-fold (not depicted), whereas Tbata expression dropped by ∼50% in castrated mice compared with sham-castrated controls (Fig. 1 C). Tbata expression continued to be lower in total thymi of castrated mice compared with controls over a 3-wk period. To confirm that the change in Tbata expression was TEC-specific, we sorted UEA⁺ and Ly51⁺ TEC from the thymi of castrated mice and sham-castrated mice one week after surgery. We observed a threefold decrease in both UEA⁺ and Ly51⁺ TEC (Fig. 1 D). The reduction of Tbata expression closely coincides with the onset of TEC proliferation and increased thymus size in C57BL/6 mice after castration that we, and others, have observed (Gray et al., 2006; Williams et al., 2008).

To explain these results, we hypothesized that high-level Tbata expression negatively regulates thymus function by regulating TEC proliferation. To further explore the connection between Tbata expression and TEC proliferation, we measured Tbata expression during medullary TEC (mTEC) development. During mTEC development, immature mTECs proliferate. After terminal differentiation, mature mTEC are MHC class II^hi, express AIRE, and exit the cell cycle (Gray et al., 2007). We sorted mTECs (CD45⁻, Ly51⁻) based on low (MEC^lo) or high (MEC^hi) MHC class II expression levels from 5-wk-old C57BL/6 mice (unpublished data). To evaluate our sort, we examined AIRE expression in both samples and found that expression was greater in the mature MEC^hi subset, as expected (Fig. 1 E). Tbata was expressed in each subset but was increased by 5–10-fold in the MEC^hi subset containing mature nonproliferating mTECs compared with the immature MEC^lo subset (Fig. 1 E). Therefore, these data show that mTEC up-regulate Tbata expression as they mature and leave the cell cycle and provide further data supporting a relationship between Tbata expression and TEC proliferative status.

To directly test whether Tbata regulates the cell cycle, we performed growth assays after Tbata transfection. In these studies, we tested two Tbata isoforms, Tbata transcript variant 3 (Ttv3) and Tbata transcript variant 5 (Ttv5), created by alternative splice acceptor sites that differ by 34 aa (Flomerfelt et al., 2000). These two isoforms account for most Tbata expression in the thymus (unpublished data). In accord with our model, we observed that transient transfection of either Ttv3-EGFP or Ttv5-EGFP fusion proteins, but not control EGFP, blocked cell growth in transfected HEK 293T cells (Fig. 2 A). To examine division of Ttv3-EGFP–transfected cells we used DiI, a red fluorescent lipophylic dye that is equally portioned to each daughter during cell division (Huang et al., 1999). EGFP-transfected cells divided over a 4-d period, as indicated by a reduction in dye intensity (Fig. 2 B, top). In contrast, Ttv3-EGFP effectively blocked the majority of transfected cells from dividing (Fig. 2 B, bottom middle). The cell growth block was specific to cells expressing Ttv3-EGFP because untransfected cells (Fig. 2 B, bottom right) in the same culture divided normally. We also confirmed that Tbata mediates growth arrest in NIH3T3 cell lines that contain stably transfected mifepristone (mif)-inducible Ttv3-EGFP or EGFP expression cassettes (unpublished data). Using these inducible cell lines, we found that Tbata-dependent growth arrest is reversible because sorted Ttv3-EGFP– and Ttv5-EGFP–expressing cells resumed growth when placed in media lacking mif (unpublished data). Therefore, these data show that Tbata expression results in growth arrest.

To gain insight into Tbata function in regulating cell proliferation at a molecular level, we used the yeast 2-hybrid assay to screen an 18-d-old mouse embryo cDNA library for Tbata-interacting proteins and identified a cDNA (designated Uba3-346) encoding 235 aa of the carboxyl terminus of Uba3. To confirm the interaction, the Uba3-346 prey plasmid was transformed into yeast strains expressing Ttv3, Ttv5, or Lamin C (Lam). Strains expressing a Tbata isoform and Uba3-346 grew on selective media, whereas those expressing the LaminC-Uba3-346 combination did not (Fig. 3 A), providing genetic evidence that Tbata and Uba3 interact.

We used GST-pulldown and coimmunoprecipitation assays to confirm the interaction. Full-length radiolabeled Uba3 interacted with GST-Ttv3 and GST-Ttv5, but not with GST–Jun fusion proteins (Fig. 3 B, top). Tbata and Uba3 also interact when transfected and expressed in 293T cells (Fig. 3 C). In this experiment, Myc-Uba3-346 was immunoprecipitated by anti-HA antibody in samples that contained HA-Tbata (Fig. 3 C, lane 2), but not in untransfected cells (Fig. 3 C, lane 3) or in cells transfected with Uba3-346 alone (Fig. 3 C, lane 1). In addition, Tbata-Uba3 binding was observed in protein extracts obtained from thymus when immunoprecipitated using an anti-Tbata rabbit antisera (Fig. 3 D). To map the interaction site, a series of Uba3 deletion mutants was produced and tested for specific interaction with the Tbata isoforms Ttv3 and Ttv5 (Fig. 3 E). GST-Ttv3 and GST-Ttv5, but not GST, interacted with Uba3 mutants that lack the amino and carboxyl terminus of the protein. Removal of the central portion of Uba3 abolishes the interaction, whereas amino acids 195–321 of Uba3 retain the ability to interact with Tbata.

Because Tbata has been previously reported to also physically associate with KIF17 (Irla et al., 2007; Saade et al., 2007), we examined whether Tbata preferentially binds with either protein. The results of a series of Tbata binding assays performed with differing amounts of either radiolabeled protein showed that Tbata can bind both Uba3 and KIF17 simultaneously. Tbata interacts with both KIF17 and Uba3 even when these proteins were present at a 1:19 ratio (Fig. 3 F). Therefore, these proteins do not appear to compete for binding to Tbata in the conditions used in these in vitro binding assays.

Given the evidence that Uba3 plays a central role in the Nedd8 pathway, we examined the effect of the Tbata–Uba3 interaction on the assembly of the Nedd8 E1. GST, Ttv3-GST, or Ttv5-GST was combined with radiolabeled Uba3 (Myc-tagged) and AppBp1. Immunoprecipitation of Myc-tagged Uba3 efficiently coprecipitated AppBp1 as expected (Fig. 4 A, IP lane). In contrast, when Ttv3-GST or Ttv5-GST interacted with Uba3, AppBp1 was not appreciably coprecipitated using glutathione–Sepharose (Fig. 4 A, lanes Ttv3 and Ttv5). The slight amount of AppBp1 observed may be caused by the ability of Uba3 to form dimers (Walden et al., 2003). The neddylation of Cullin1 was examined in NIH3T3 cell lines that contain mifepristone (mif)-inducible Ttv3-EGFP or EGFP expression cassettes. A Nedd8 antibody was used to immunoprecipitate protein from extracts obtained from EGFP or Ttv3-EGFP–expressing cells. The immunoprecipitated proteins were analyzed by immunoblotting using an anti-Cullin1 antibody. There was a significant reduction in the level of neddylated Cullin1 in mif-treated cells expressing Ttv3-EGFP compared with mif-treated cells expressing EGFP (Fig. 4 B). Therefore, Tbata binding to Uba3 interferes with the Uba3–AppBp1 interaction and greatly reduces the level of neddylated Cullin1.

Given the role of the Nedd8 pathway in cell cycle control, it seemed likely that Tbata-mediated growth arrest was caused by its ability to bind Uba3 and block its function. To test this prediction, we examined whether overexpression of Uba3 would prevent Tbata-dependent growth arrest in cell lines expressing transfected Tbata. Increasing amounts of either Uba3 or control d-uba3 (Fig. 3 E, deletion 121–304) expression plasmids were cotransfected in HEK 293 cells with a fixed amount of Ttv3-EGFP. Ttv3-EGFP–transfected cells grew more over a 3-d period when Uba3 was overexpressed (Fig. 4 C). The growth was dose dependent (compare 1.25 µg to 2.5 µg Uba3 transfections). Tbata-mediated growth arrest was unaffected in control cells overexpressing D-uba3 that does not bind Tbata. Finally, labeling with DiI confirmed that coexpression of Tbata-EGFP and Uba3 partially restored cell division, whereas cotransfection with D-uba3 did not (Fig. 4 D). The Ttv3-EGFP–dependent growth inhibition was also confirmed by experiments using both BrdU and DNA content analysis (unpublished data). Given that Tbata can simultaneously bind both Uba3 and the kinesin motor protein KIF17, we wanted to determine whether KIF17 coexpression affects Tbata-dependent growth suppression. We determined that KIF17 is not expressed in the thymus, in purified TEC, or in our inducible NIH3T3 cells as measured by RT-PCR (Fig. S1 d). Tbata mediates potent growth suppression in HEK 293 cells (Fig. 2 and Fig. 4) that express KIF17 mRNA (Fig. S1 d). Tbata-mediated growth suppression in HEK 293 cells is reversed by Uba3 overexpression (Fig. 4 D). These data indicate that KIF17 does not affect the ability of Tbata to block the Nedd8 pathway or affect Tbata-mediated growth inhibition in HEK 293T cells. Tbata-mediated growth inhibition was unaffected by overexpression of transiently transfected KIF17 fused to mCherry in our mifepristone-inducible Tbata NIH3T3 cell line Ttv3-C (Fig. S1 e). Although Tbata interacts with KIF17 in other organs (Irla et al., 2007; Saade et al., 2007), it is likely that KIF17 does not have a role in TEC function and does not affect Tbata-mediated growth suppression in the experimental models used in this study. Collectively, these data suggest that Tbata inhibits cell division by blocking the Nedd8 pathway.

The in vitro data concerning the molecular pathways by which Tbata inhibited the Nedd8 pathway and cell proliferation supported the previously generated data linking Tbata expression and thymus function. To more directly examine the in vivo role of Tbata on thymus function, a Tbata KO mouse model was developed (Fig. S2, A–D). Tbata^−/− mice were viable, and both females and males were normally fertile throughout life. Because Tbata is also expressed in the testis and cerebellum, we also examined the structure and proliferative status of these organs in Tbata^−/− mice. Male Tbata^−/− mice exhibited normal testes morphology (Fig. S2 E) and function. Furthermore, there were no obvious neurological problems (unpublished data) or any obvious defects in gross brain anatomy. We did, however, note abundant, abnormal nests of retained external germinal layer (EGL) in the cerebellum. The EGL is normally depopulated by postnatal day 18 in WT mice as progenitors migrate to the internal granule cell layer and differentiate (Hatten and Heintz, 1995). In Tbata^−/− mice, clumps of EGL cells remained evident at 1 mo of age and persisted in aged mice (Fig. S2 F). Therefore, the Tbata deficiency appears to have a minimal impact on testes function but did effect some changes in cerebellum development.

In contrast, we did observe changes in the thymus of Tbata^−/− mice. In 1–2-mo-old Tbata^−/− mice, there were no significant differences in the thymus between Tbata^−/− and WT littermates (Fig. 5 A and not depicted). However, in 5–9-mo-old Tbata^−/− mice, the thymus contained about twice as many thymocytes as WT littermates (Fig. 5 A). In 1-yr-old mice there was a fourfold greater number of thymocytes (Fig. 5 A) and about twice as many stromal cells (Fig. 5 B) in Tbata^−/− mice compared with WT littermates. The rate of age-dependent thymus size decrease was slowed in the Tbata^−/− and thymus size was consistently greater in aged Tbata^−/− mice compared with WT littermates (Fig. 5 A). In spite of the differences noted in thymic cellularity, there were no differences in the numbers or ratios of either their B or T cells or in the CD4/CD8 ratios in spleen or lymph nodes (unpublished data) when Tbata^−/− mice were compared with WT controls. We did observe a 1.5-fold increase in CD44^lo, CD62L⁺ splenic T cells in aged mice consistent with their larger thymus (unpublished data). Thus, the Tbata deficiency resulted in a larger thymus in aged mice.

We next addressed whether T cell development was affected in aged Tbata^−/− mice. The absolute numbers of all thymocyte subsets were consistently greater in aged Tbata^−/− mice compared with 9–12-mo-old WT mice (Fig. 5, C–E). There were 4–5 times more CD4 and CD8 double-negative DN cells in Tbata^−/− mice than in WT littermates (Fig. 5, C and E). CD4 and CD8 double-positive, CD4, and CD8 populations were increased by three- to fourfold in the Tbata^−/− mice (Fig. 5, D and E). We noted a slight increase in the percentage of DN cells in the Tbata^−/− thymus, whereas the percentages of cells in the other subsets were similar to WT controls (Fig. 5, C–E). No defects in positive or negative selection were observed when Tbata^−/− were crossed to the HY or 5CC7 TCR transgenics, although thymus size was larger compared with Tbata⁺ control mice (unpublished data). We also observed normal thymic architecture in 5-mo-old TKO, using anti-keratin 8 (red) to reveal cortical TEC and anti-keratin 5 (green) to visualize medullary TEC in frozen sections (Fig. S3 A). Mature T cells from Tbata^−/− lymph nodes proliferated normally to immunization with a defined antigen (Fig. S3 B). Therefore, while aged Tbata^−/− mice have larger thymi, the Tbata deficiency did not disrupt normal thymocyte development, thymus morphology, or T cell function.

To further examine the effect of Tbata-deficiency on thymus function, Tbata^−/− mice were crossed to Rag2^−/− mice (Shinkai et al., 1992) to create Tbata^−/−Pαγ2^−/− double KO mice. The proportions of DN1-3 were normal in thymi from Tbata^−/−Rag2^−/− mice and, as expected, thymocyte development was arrested at the DN3 stage (Fig. 6 A). We observed an increase in total thymocyte numbers in Tbata^−/−Rag2^−/− mice 8 wk after birth (not depicted) and by 3 mo the thymi of Tbata^−/−Rag2^−/− mice contained ∼3-4 times more thymocytes (Fig. 6 B). There were about twice as many early thymocyte progenitors (ETPs; CD44^hi, c-kit^hi, CD25⁻, and Lin⁻) in Tbata^−/−Rag2^−/− thymi (Fig. 6 C) than in Rag2^−/− mice. Therefore, the Tbata deficiency causes an increase in thymocyte numbers in Rag2^−/− mice, as well as in aged mice.

To confirm that the phenotype observed in Tbata^−/− thymus was caused by alterations in stromal cells, we examined thymus recovery after BM transplantation (BMT) in the Tbata^−/−Rag2^−/− versus control littermates. T cell depleted Ly5.1⁺ BM cells from WT mice were injected into adult Tbata^−/−Rag2^−/− mice. Age-matched littermates (Tbata^+/+, Rag2^−/−) were used as controls. Ly5.1⁺, B220⁺ cell numbers were used to show that the engraftment of donor cells in different mice was comparable (Fig. S4). Interestingly, thymic reconstitution in the Tbata^−/−Rag2^−/− thymus occurred faster than was observed in a Rag2^−/− mouse (Fig. 6 F). Within 3 wk, all thymocyte developmental stages, derived from donor cells were present in transplanted Tbata^−/−Rag2^−/− mice (Fig. 6, D and E). In Rag2^−/− mice, donor cells were present but a comparable state of reconstitution was not reached for another 10–14 d (unpublished data). At 5 wk after transplant, both the Tbata^−/−Rag2^−/− and the Rag2^−/− mice had full reconstitution of all thymic subsets, but the total number of donor thymocytes in the Tbata^−/−Rag2^−/− was about sevenfold greater than in the Rag2^−/− mice (Fig. 6 F). By 9 wk after BMT; thymic reconstitution in both Rag2^−/− mice and Tbata^−/−Rag2^−/− mice was similar (Fig. 6 F). The rapid thymic reconstitution in Tbata^−/−Rag2^−/− mice was accompanied by a more rapid repopulation of the periphery by donor T cells (Fig. 6 G). Therefore, Tbata^−/−Rag2^−/− mice have improved thymus reconstitution relative to control mice as measured by reconstitution of thymocyte subpopulations and production of mature T cells providing additional evidence that Tbata affects thymus function.

We hypothesized that the increase in thymus size observed in aged Tbata^−/− mice was caused by enhanced TEC proliferation. Therefore, we examined TEC proliferation in 6–9-mo-old mice because this is the earliest period when differences in thymus size between WT and Tbata^−/− mice are evident. We analyzed mTEC (UEA⁺, ClassII⁺) and cTEC (Ly51⁺, ClassII⁺) from 6–9-mo-old WT or Tbata^−/− mice using flow cytometry. The percentages of TEC subsets in Tbata^−/− and WT mice at these ages were similar, except for a twofold increase in the percentage of MEC-lo cells as a fraction of total TEC (P = 0.009; unpublished data). There were significantly more TECs in the Tbata^−/− thymi (P = 0.02) compared with WT littermates, and both cTEC and mTEC numbers were increased (Fig. 7 A). However, the cTEC/mTEC ratios (WT, 2.06; Tbata^−/−, 2.07; unpublished data) were the same in both sets of mice. Using Ki67 staining, we observed a significant increase in the numbers of proliferating UEA⁺ mTEC in Tbata^−/− mice both in MEC^lo and MEC^hi fractions, whereas numbers of proliferating cTECs were modestly increased (Fig. 7 B). Similar data were obtained in other experiments using BrdU to assess TEC proliferation (unpublished data). Therefore, Tbata appears to play a role in regulating TEC proliferation in this setting.

T cell development depends on a complex interplay between BM-derived lymphoid progenitor cells and resident stromal cells (Anderson and Jenkinson, 2001). Recent findings highlight the relationship between stromal cell proliferation and numbers with thymus function as observed in a K5-CyclinD1 transgenic, in Nude-Wt chimeras, or in aged thymus (Robles et al., 1996; Klug et al., 2000; Gray et al., 2006; Jenkinson et al., 2008). Therefore, an understanding of the mechanisms that regulate stromal cell numbers might identify control points of thymus function and provide strategies to improve T cell production in immunodeficient patients. To increase our basic understanding of stromal cell function in the thymus, we cloned Tbata (Kim et al., 1998). Initially, we reported that Tbata expression was restricted to the subcapsule of adult mice (Flomerfelt et al., 2000). Using more sensitive techniques, however, we now report that Tbata is also expressed in cortical and medullary TECs in accord with the findings of others (Flomerfelt et al., 2000; Saade et al., 2007). In this study, we provide strong evidence that Tbata exerts a reversible negative regulatory effect on cell proliferation and thymus function.

Our data show that changes in the level of Tbata expression in the thymus correlates with the state of TEC proliferation, as we demonstrated in four different situations. First, Tbata levels are lowest in young thymus when TEC proliferation is robust (Gray et al., 2006) and are increased 10-fold in C57BL/6 mice 4 mo of age or older, coincident with reduced proliferation and a reduction of TEC numbers over time. Second, Tbata levels vary in mouse strains with different rates of age-dependent loss of thymus function. DBA/2 mice rapidly lose thymus function (Hsu et al., 2003) and exhibit an earlier and greater increase in Tbata expression compared with C57BL/6 mice. In accord with our phenotypic data, Tbata^−/− resides on chromosome 10 in a region identified as a genetic locus involved in the early reduction of thymus function in DBA/2 mice (Hsu et al., 2003). Third, in thymi enlarged after castration, we found a threefold reduction in Tbata expression in TECs from 4-mo-old mice 1 wk after surgery compared with sham castrated mice. In this same time frame, TEC proliferation increases and a 2.5-fold increase in TEC numbers is observed (Gray et al., 2006; Williams et al., 2008). These findings suggest that Tbata-mediated cell cycle arrest is reversible in TECs, as we observed in our in vitro experiments. Fourth, we observed a 5–10-fold increase in Tbata expression in mature mTECs (UEA⁺MHC class II^hi) compared with the preceding stage (UEA⁺ MHC Class II^lo) of mTEC development. Mature mTECs cease proliferation during this developmental transition. We also observed increased proliferation of mature mTECs in Tbata^−/− mice, suggesting that Tbata may also play a role in promoting exit from the cell cycle during mTEC development. Together, these data suggest that Tbata is involved in the negative regulation of TEC proliferation.

To further explore how Tbata might regulate TEC proliferation, we characterized the function of the Tbata protein. We showed that Tbata mediates a profound growth arrest in cells by specifically interacting with Uba3. This interaction inhibits the Nedd8 pathway and results in cell cycle arrest, because neddylation of cullins is a prerequisite for targeted degradation of several cell cycle control proteins (Podust et al., 2000; Doronkin et al., 2003). Linkage of Tbata to the Nedd8 pathway was confirmed by showing that Tbata-mediated growth arrest was attenuated by overexpression of Uba3, whereas overexpression of an Uba3 mutant that does not bind Tbata did not affect Tbata-mediated growth arrest. Based on our experimental evidence, it appears that KIF17 does not affect Tbata-mediated growth suppression. Therefore, Tbata exerts a negative regulation on cell cycle progression that is consistent with blocking the Nedd8 pathway. These data support Tbata as a negative regulator of TEC proliferation and provide a molecular mechanism for its effect.

Different steps of the Nedd8 pathway have been used in cells to provide additional levels of cell cycle regulation. Mechanisms are based on selective de-neddylation of cullins by the COP9 signalsome protein complex or other proteins (Gong et al., 2000; Yang et al., 2002; Gan-Erdene et al., 2003; Mendoza et al., 2003; von Arnim, 2003; Wu et al., 2003). NUB1 sequesters and degrades Nedd8 monomers and Nedd8 conjugates (Kamitani et al., 2001; Kito et al., 2001) and TIP120A binds to cullins blocking their modification by Nedd8 (Yogosawa et al., 1999; Hwang et al., 2003). Our data reveal that Tbata mediates a novel inhibition mechanism of the proximal Nedd8 pathway by preventing formation of the Nedd8 E1 via its disruption of the Uba3–AppBp1 complex. The crystal structure of the human Uba3–AppBp1 heterodimer suggests how Tbata might disrupt this complex (Walden et al., 2003). 72 aa of Uba3 contact AppBp1 to form functional adenylation and catalytic cysteine domains. An interface between Uba3 and AppBp1 is mediated by a closely spaced high-affinity interaction of 32 residues of Uba3 within aa 270–334 that directly contact AppBp1. Tbata specifically interacts with a portion of Uba3 that includes aa 214–321, a region that contains 28 residues that contact AppBp1 (Walden et al., 2003). Because of the close proximity of AppBp1 to Uba3 at aa 214–321, it is likely that a high-affinity Tbata interaction with Uba3 in this region would sterically inhibit the binding of AppBp1 to Uba3 and block Nedd8 E1 formation. Therefore, the restricted expression of Tbata in thymus appears to mediate a unique cell cycle control mechanism in TECs to regulate thymus function.

The dominant phenotypic feature of Tbata deficiency is the improved thymus function observed in aged Tbata^−/− mice compared with WT mice. One likely explanation for the Tbata^−/− phenotype is that enhanced proliferation of thymic stromal cells in aged mice creates a larger niche size that can accommodate more thymocytes. This change results in greater thymus function in aged mice compared with the WT control, but does not affect T cell development or thymus architecture. Although thymus function decreases during aging in Tbata^−/− mice, it occurs at a slower rate. In 6–9-mo-old Tbata^−/− mice, we observed statistically significant increases in total TEC number and in the number of proliferating UEA⁺ cells. The phenotype of Tbata^−/− mice appears to depend on changes in thymic stromal cells because Tbata is only expressed in these cells (Flomerfelt et al., 2000) and Tbata^−/−Rag2^−/− mice show a robust and more rapid thymic reconstitution compared with Rag2^−/− littermates after BMT from WT donor mice, thereby mapping the function to the TEC compartment. Rapid reconstitution in the Tbata^−/−Rag2^−/− may be caused by enhanced stromal cell proliferation and a more rapid response to increase niche size when normal thymocytes enter the thymus. A similar model has been proposed to explain the more rapid reconstitution of IL-7R^−/− mice compared with Rag2^−/− mice (Prockop and Petrie, 2004). It is probable that thymus function depends on integration between niche size and precursor import. For example, studies using IGF-1 to enhance thymus function concluded that all parameters of thymus function, including import, appear to be related to available niche size (Chu et al., 2008). In the case of the Tbata^−/−Rag2^−/−, it is possible that the “reception” for recently imported precursor cells favors rapid expansion and onset of differentiation.

The regulation of thymic stromal cell proliferation is likely involved in controlling thymus function in response to environmental signals, aging, or disease states. We hypothesize that Tbata regulates thymus function by a reversible negative regulation of stromal cell proliferation and propose that this creates a greater availability of thymus stromal cell niches accessible for thymocyte development. It is possible that the Tbata–Uba3 interaction will provide a molecular target for developing therapeutic tools to enhance T cell production in adult immunocompromised patients and to augment function of the thymus in aged patients.

Tbata transcript variant 3 (AF257502; available from GenBank/EMBL/DDBJ under accession no. NM_023064), and Tbata transcript variant 5 (AF257503; available from GenBank/EMBL/DDBJ under accssion no. NM_001017407) cDNAs were cloned from mouse thymus (Flomerfelt et al., 2000). Human AppBp1 (BC000480) and mouse Uba3 (BC002002) cDNAs were obtained from the IMAGE consortium. Human Uba3 cDNA was a gift from E. Yeh (University of Texas M.D. Anderson Cancer Center, Houston, TX). GST-Jun plasmid was a gift from Z.-G. Liu (National Cancer Institute, Bethesda, MD). Mouse KIF17 fused to mCherry was purchased from GeneCopoeia Inc. C57BL/6, DBA/2, and Ly5.1 congenic mice were obtained from the National Cancer Insitute (NCI), Frederick, animal facility. C57BL/6 Rag2^−/− mice were obtained from Taconic. Tbata^−/−Rag2^−/− or Rag2^−/− littermates were bred in our mouse facility. All animal care, handling, and procedures were done following protocols detailed in Animal Science Protocols approved by the NCI Animal Care and Use Committee using guidelines approved by the National Institutes of Health Animal Research Advisory Committee.

A P1 clone was used to construct a targeting vector with a 2.3-kb EcoRV-Xma1 fragment upstream of the thymus-specific start site and a 4.5-kb Xho1–BamH1 fragment downstream of the termination codon (Flomerfelt et al., 2000). The linearized targeting vector was transfected into E14.1 ES cells derived from 129 mice. Homologous recombination was confirmed by Southern blot analysis (Fig. S2, A–D). Chimeric mice were bred to C57BL/6 mice. Heterozygous mice were backcrossed onto a C57BL/6 background four times, and heterozygotes were interbred to obtain Tbata^−/− mice and control WT littermates.

FACS data were collected on a FACSCalibur or a LSRII (both from BD) and analyzed using CellQuest (BD) and FlowJo (Tree Star, Inc.) software. 10 µg/ml 2.4G2 was added to 10⁷ cells/ml. Double-negative subsets were studied on gated lineage-negative cells using anti–CD44-PE and anti–CD25-FITC along with biotinylated or Tricolor-conjugated anti-lineage markers (CD3, CD4, CD8, CD11b, Pan NK, B220, and Ter119), followed by Streptavidin Tricolor or Streptavidin allophycocyanin conjugates. TEC analysis was done on CD45- or CD90-negative cells and included HSA⁻ cells to quantitate TEC numbers in the aging studies. Ki67 staining was done on fixed cells using the BrdU staining kit (BD) after surface staining with UEA-FITC, Ly51-PE, Class II-Bio-Pacific blue avidin, and CD45 PE Cy5.5. Cells stained with Ki67 were not DNase-treated.

The thymus was dissected and minced with a Mickle Tissue Homogenizer. The tissue was resuspended in RPMI, and 50 µl of Liberase4 (28 U/ml) and DNase1 (2 mg/ml) were added and incubated for 15 min at 37°C, shaking at 250 RPM. For sorting and proliferation analysis, the cell suspension was thymocyte-depleted using the anti-CD90 MACS system (Miltenyi Biotec).

Frozen thymus sections were incubated in 1% 24g2, 3% normal donkey serum. Keratin staining was done with rat anti-Keratin 5 (2 µg/section; Covance) and rabbit anti-Keratin 8 (2 µg/section; Troma-1; Developmental Studies Hybridoma Bank, The University of Iowa) and visualized with the appropriate FITC- or Texas red–conjugated secondary antibodies. Images were collected on an Olympus IX71 with an UPanFLN20x/0.5 objective using an iCys Research Laser Scanning Cytometer. A series of overlapping 200× magnification images of the whole section were obtained and digitally stitched together using an iCys Research Imaging Cytometer and iCyte Cytometric Analysis Software 3.2. The software produces the horizontal lines when the image is digitally stitched together.

RNA from sorted stromal cells was linearly amplified using the RiboAmp HS Plus (Applied Biosystems) kit. cDNA synthesis was performed using random primers and 3–5 µg of total RNA. Real-time PCR quantitation was performed using the LightCycler (Roche). Standard curves for quantitation were generated from serial dilutions of cDNA from 5 µg of total thymus mRNA from C57BL/6 mice. Undiluted cDNA was assigned an arbitrary value, and the instrument software calculated concentration values. Assays were designed to amplify all Tbata isoforms or only the Ttv3 or Ttv5 isoforms. Primer sequences are available upon request. For the aging studies on sorted TEC, the relative expression was normalized to input RNA. For all other experiments, Tbata levels were normalized using GAPDH expression in the same samples. All experiments were done in triplicate and were repeated 3–5 times with independent samples.

Recombinant GST, GST-Tbata, and GST-Jun fusion protein lysates were purified from bacteria. Radiolabeled proteins were produced using separate coupled transcription-translation reactions (Promega). A sample of the reaction was run on a gel to confirm that equal amounts of radiolabeled proteins were mixed with equal amounts of either GST or GST fusion proteins bound to glutathione-coated Sepharose beads. Samples were incubated for 2–12 h at 4°C, followed by five 1-ml washes using ice-cold TBS with 0.1% Tween-20. Bound proteins were eluted from the beads in PAGE sample buffer at 70°C, and subjected to PAGE.

The Matchmaker Yeast Two-Hybrid System 3 (Takara Bio Inc.) was used following the manufacturer’s recommendations to identify proteins that interact with Tbata. A full-length cDNA of the Ttv3 Tbata isoform was fused to the Gal4 DNA binding domain in the vector pGBKT7. This construct did not activate reporter genes based on growth analysis and was used as the bait. Prey constructs were obtained from a day 18 mouse embryo library cloned into the vector pACT2. The day 18 embryo has a fully functional thymus that expresses Tbata (Flomerfelt et al., 2000).

24 h after transfection, cells were labeled with DiI. 1 h later, DiI label intensity was analyzed by FACS (day 0) and each day after transfection. We noted a slight increase in DiI intensity in nondividing cells over a 4-d period. MODFIT software (Verity Software House) was used to analyze the FACS data using a progressive Gaussian estimation based on the intensity and the SD of the parent population. Profiles were generated to measure cell division by gating the analysis on EGFP⁺, Ttv3-EGFP⁺, or EGFP⁻ cells.

Statistical analysis was performed using StatView 5.0.1 or Microsoft Excel software. Error bars on graphs show the SD. Where indicated, studies were analyzed using Student’s t test, with a value of 0.05 or less considered significant.

Fig. S1 shows Tbata expression in the cerebellum and testes, sort scheme for TEC cells, KIF17 expression analysis in different tissues and cells, and the effect of KIF17 on Tbata-mediated growth suppression. Fig. S2 shows the generation and verification of the Tbata^−/−, and stained sections of WT and Tbata^−/− brains and testes. Fig. S3 shows keratin5 and keratin8 staining of WT and Tbata^−/− thymus and the T cell response to DNP-OVA of WT and Tbata^−/− mice. Fig. S4 shows enumeration of splenic B cells 3 wk after BM transplantation in Rag2^−/− and Tbata^−/− mice.

The authors thank Krishnendu Roy, Allan Weissman, Emanuela Lacana, Yu-Waye Chu, Richard Hodes, Kirsten Williams, and Fran Hakim, for helpful discussions about the manuscript. Vanessa Baxendale kindly provided her expertise in evaluating testis anatomy. The National Institute of Allergy and Infectious Disease Gene Targeting Unit produced the Tbata^−/−.

The authors have no commercial affiliations or competing financial interests to disclose.