Impaired Thymic Selection and Abnormal Antigen-Specific T Cell Responses in Foxn1Δ/Δ Mutant Mice

Background Foxn1Δ/Δ mutant mice have a specific defect in thymic development, characterized by a block in TEC differentiation at an intermediate progenitor stage, and blocks in thymocyte development at both the DN1 and DP cell stages, resulting in the production of abnormally functioning T cells that develop from an atypical progenitor population. In the current study, we tested the effects of these defects on thymic selection. Methodology/Principal Findings We used Foxn1Δ/Δ; DO11 Tg and Foxn1Δ/Δ; OT1 Tg mice as positive selection and Foxn1Δ/Δ; MHCII I-E mice as negative selection models. We also used an in vivo system of antigen-specific reactivity to test the function of peripheral T cells. Our data show that the capacity for positive and negative selection of both CD4 and CD8 SP thymocytes was reduced in Foxn1Δ/Δ mutants compared to Foxn1+/Δ control mice. These defects were associated with reduction of both MHC Class I and Class II expression, although the resulting peripheral T cells have a broad TCR Vβ repertoire. In this deficient thymic environment, immature CD4 and CD8 SP thymocytes emigrate from the thymus into the periphery. These T cells had an incompletely activated profile under stimulation of the TCR signal in vitro, and were either hypersensitive or hyporesponsive to antigen-specific stimulation in vivo. These cell-autonomous defects were compounded by the hypocellular peripheral environment caused by low thymic output. Conclusions/Significance These data show that a primary defect in the thymic microenvironment can cause both direct defects in selection which can in turn cause indirect effects on the periphery, exacerbating functional defects in T cells.


Introduction
The thymus is the primary site of T cell development. In the thymus, heterogeneous thymic epithelial cells (TECs) and other stromal cells including fibroblasts, dendritic cells, and macrophages constitute a 3-dimensional space that provides a unique microenvironment for the differentiation, proliferation, and maturation of T cells. During differentiation, T cells undergo VDJ recombination generating a diverse repertoire of TCRs. Thymic positive selection ensures the differentiation and maturation of thymocytes expressing a self-restricted TCR; negative selection then removes those thymoctyes that have dangerously high reactivity to self [1]. The resulting T cells migrate out of the thymus resulting in a diverse repertoire of T cells that are able to respond to foreign peptides while remaining tolerant to selfpeptides. The positive selection of CD4 and CD8 T cells is restricted by MHCII and MHCI respectively. The development of TCRab CD4 T cells can be typically supported by MHCII expressing TECs, while CD8 T cells can be supported by MHCI signals from TECs and other stromal cells in thymus [2,3,4,5,6]. The thymic medulla has been shown to be largely responsible for negative selection [7,8]. In the cortico-medullary junction and medulla, dendritic cells (DCs) play a major role in negative selection; medullary TECs have also been reported to be sufficient to drive this process [9].
The development, differentiation, and function of fetal TECs are controlled by the forkhead-class transcription factor Foxn1 [10,11,12,13,14]. Foxn1 is widely if not ubiquitously expressed in fetal TECs from a very early stage [13,15]. In the absence of Foxn1 (i.e. in nude homozygotes), the thymic rudiment is formed, but lymphoid progenitor cells (LPCs) fail to populate it [12,14,16]. In a severe hypomorphic allele, Foxn1 D , initial organogenesis appears normal, but later thymocyte-dependent TEC differentiation is blocked [2]. Thus, Foxn1 is required for both thymocytedependent and -independent TEC differentiation. In the postnatal thymus, Foxn1 expression is distributed in both cortex and medulla [12,17,18]. We have recently reported that correct expression of Foxn1 is also required to maintain the postnatal thymus, as postnatal reduction of Foxn1 expression leads to a phenotype resembling premature thymic involution [18]. Down regulation of Foxn1 has also been implicated in initiating thymic involution [19]. Thus, Foxn1 plays multiple crucial roles in the development and maintenance of TECs throughout the lifespan of the organism.
Foxn1 D/D mice have a very small thymus with abnormal architecture lacking cortical and medullary domains, in which most TECs display an intermediate progenitor phenotype [10]. Thymocytes are specifically blocked at both the DN1 and DP differentiation stages, express constitutively low TCR, and very few SP thymocytes develop and immigrate into the periphery [10]. The resulting peripheral T cells have an atypical activated/ memory phenotype with low homeostatic potential and impaired TCR signaling [20]. T regulatory (T reg ) cells are generated but have reduced function, and T cells are hyper-responsive in both allo-and auto-MRL assays in vitro [21]. These abnormal T cells develop from CD117 2 atypical progenitors without passing through the DN2 or DN3 stages, but directly develop into DN4 from DN1 cells in the postnatal thymus [20]. As these atypical progenitors are normally present in the wild-type thymus, these may represent a normal sub-population of T cells that may serve a specific as yet unknown function in the normal T cell repertoire. As these T cells also have constitutively low TCR expression, the question of whether this abnormal microenvironment is promoting normal positive and negative selection is also important to understanding the mechanism by which these T cells are generated.
In the current study, we investigated the processes of positive and negative selection in Foxn1 D/D mutants. We show that TEC expression of both MHC Class I and II are reduced significantly, with MHC Class II expression affected more severely. We generated Foxn1 D/D BALB/c (MHCII I-E molecule positive) mice, and crossed Foxn1 D/D to the DO11.10 CD4 and OT1 CD8 restricted TCR transgenes. By focusing on the TCR Vb usage and the phenotypes of CD4 and CD8 Tg cells in the thymus or periphery of Foxn1 D/D mice, we analyzed the thymic negative and positive selection and assayed SP thymocyte post-selection maturation in the thymus. We also analyzed T cell responses to specific antigens in vivo. Our results show that the Foxn1 D/D thymus did not generally affect the diversity of the TCR Vb repertoire, although the frequencies of specific Vb usage were highly variable in the mutants compared to controls. However, thymic negative selection was partially reduced and positive selection was dramatically reduced for CD4 T cells, but relatively normal for CD8 T cells. These T cells generated abnormal responses to antigen-specific stimulation in vivo. At least part of this abnormal response was secondary to the previously reported memory-like phenotype of T cells after lymphopenia-induced proliferation and reduced T reg cell function in the Foxn1 D/D periphery [20]. Thus, the peripheral T cell phenotype in the Foxn1 D/D mutants is the result of both direct effects of abnormal selection within the thymus, compounded by secondary effects of low T cell output and a hypocellular peripheral environment.

Peripheral T cells have broad but variable TCR Vb usage in Foxn1 D/D mutant mice
Our previous data reported that T cells in Foxn1 D/D thymus appear to develop from an atypical CD117 low/2 progenitor that develops directly from DN1-DN4 stages, but did undergo TCR VDJ recombination [20]. This result raises the question of whether this atypical pathway generates a diverse TCR repertoire. Because the number of SP thymocytes in the Foxn1 D/D thymus was too low to be analyzed directly for TCR Vb usage, we assayed TCR Vb expression on peripheral T cells (Fig. 1). We found that all TCR Vb alleles were detected on both CD4 and CD8 Foxn1 D/D T cells. Compared to TCRab + T cells in control mice, there was no consistent difference in TCR Vb expression between the Foxn1 D/D mutant mice and controls. The one exception was TCR Vb5, which was significantly over-represented in the Foxn1 D/D CD4 T cells. However, a high variability of TCR Vb expression was seen in Foxn1 D/D individual mouse, compared to the consistent pattern of TCR Vb expression on T cells from control mice. These results indicated that while all TCR Vb alleles were utilized in the Foxn1 D/D CD117 low/2 progenitor T cells, either the relative frequency was variable, or the relative frequencies were skewed in the periphery as a consequence of low thymic output and differential expansion of specific Vb clones in response to a hypocellular periphery.
Thymic negative selection was partially reduced in the Foxn1 D/D mutant mice Based on the expression of TCR Vb, we designed an experiment to test thymic negative selection. Endogenous superantigen-mediated TCR Vb T cell clonal deletion generally occurs in mice that express an endogenous super-antigen and the MHC-II I-E molecule [22]. The expression of Vb5.1 and Vb11 were clear on T cells in the Foxn1 D/D and Foxn1 +/D mice, both of which are maintained on the C57BL/6 background ( Fig. 2A), because these mice failed to delete them due to lack of MHCII I-E molecules on TECs. In BALB/c mice, which express the MHCII I-E molecule, Vb5.1 and Vb11 bearing T cells are deleted ( Fig. 2A, B, D) [22]. We therefore backcrossed the C57BL/6 I-E 2 Foxn1 D/D mice with BALB/c (I-E + ) mice, generating BALB/c (I-E + ) Foxn1 D/D mice after more than 5 generations. On the BALB/c genetic background, the expression of Vb5.1 and Vb11 was greatly reduced in BALB/c Foxn1 +/D control mice, with around 90% of Vb5.1 and Vb11 bearing cells deleted ( Fig. 2B-E, Table 1). However only around 60% of Vb5.1 and Vb11 bearing cells were deleted in the Foxn1 D/D mutant mice on this genetic background. These data showed that although the thymic environment was completely disorganized with lack of cortico-medullary junction and medulla, thymic negative selection did occur in the Foxn1 D/D thymus, although with reduced efficiency. Because the percentage of MHC II + CD45 + CD11c + DCs was higher in total thymic stromal cells in the Foxn1 D/D than in the Foxn1 +/D mice (data not shown), these DCs may partially compensate for the mTEC deficiency in the Foxn1 D/D thymus.
Positive selection for CD4 T cells was greatly reduced in Foxn1 D/D thymus To measure the positive selection of CD4 + SP cells, we crossed the Foxn1 D/D ;I-E + mice with the DO11.10 CD4 restricted TCR transgenic mice (BALB/c background) to generate Foxn1 D/D ;I-E + ;DO11.10 Tg mice (called Foxn1 D/D ;DO11.10 below) [23]. Compared to Foxn1 +/D mice without the DO11.10 transgene, the percentage of CD4 SP thymocytes was increased 4-fold in Foxn1 +/ D ;DO11.10 mice (from 7.5% to 30.6%) (Fig. 3A, B), with around 93% of CD4 + SP cells being Kj1-26 + transgene-expressing T cells (Fig. 3C). The number of total thymocytes was also increased (Fig. 3D). In contrast, in Foxn1 D/D ;DO11.10 mice the percentage of CD4 + SP thymocytes was not significantly increased compared to Foxn1 D/D alone. Only 78% of CD4 cells were Kj1-26 + , and the number of total thymocytes was not significantly changed ( Fig. 3A, B, C, D). In peripheral T cells of Foxn1 +/D ;DO11.10 control mice, the percentage of splenic CD4 + T cells was increased 1.7 fold compared to the non-transgenic mice, with 95% of them Kj1-26 + . However, in the Foxn1 D/D ;I-E + ;DO11.10 mice, the percentage of splenic CD4 T cell dropped dramatically down over 6 fold, and only 47% were Kj1-26 + cells (Fig. 3E, F, G, H). These results indicated that the capability for positive selection of the DO11.10 transgenic TCR was reduced in the Foxn1 D/D mice. This phenotype showed high variability based on the wide range in percentage of Kj1-26 + CD4 T cells in the Foxn1 D/D ;DO11.10 mice. In addition, we found a low percentage of Kj1-26 + CD4 T cells in the periphery compared to that seen in the thymus of Foxn1 D/D ;DO11.10 mice, suggesting that the transgenic CD4 + cells might not survive well in the periphery of the Foxn1 D/D ;DO11.10 mice.
Positive selection for CD8 T cells was partially reduced in Foxn1 D/D thymus To measure the positive selection of CD8 + T cells, we crossed the Foxn1 D/D mice with OT1 TCR transgenic mice, in which thymocytes are restricted to development of CD8 + T cells. Tg CD8 T cells were identified by expression of TCR Vb5 and Va2. In the thymus of Foxn1 +/D ;OT1 mice, the percentage of CD8 + SP cells was increased 5.4 fold (from 3% to 16.2%) compared to Foxn1 +/D mice without the transgene, and 94% of CD8 + SP thymocytes were transgene positive. In Foxn1 D/D ;OT1 mice, the percentage of CD8 SP thymocyte was increased 3 fold (from 2.1% to 6.2%), and 81.9% CD8 SP thymocytes were CD8 Tg cells (Fig. 4A, B, C). Compared to non-transgenic mice, the total thymic cell number was reduced in the Foxn1 +/D ;OT1 control mice, but was unchanged in Foxn1 D/D ;OT1 mice ( Fig. 4D). For peripheral T cell analysis, the results were similar to thymus, except there was no significant difference in total cell number between mice of different genotypes (Fig. 4E, F, G, H). These results indicated that Foxn1 D/D ;OT1 mice had a reduced capability for producing CD8 + OT1 transgenic T cells compared to the Foxn1 +/D ;OT1 control mice, although to a lesser extent than was seen for positive selection of CD4+ T cells.

Reduced MHC expression in Foxn1 D/D TECs
The evidence for decreased positive and negative selection in the Foxn1 D/D mutant thymus combined with our previous data showing serious deficits in TEC differentiation and corticomedullary organization in these mutants suggested that MHC expression could well be affected. MHC II hi stromal cells were almost completely absent from the Foxn1 D/D mutant thymus, and the percentage of stromal cells expressing intermediate levels of MHC II was variable in the mutants, although not significantly different (Fig. 5). MHC I hi cells were also reduced significantly, although not as dramatically, while MHC I lo were increased proportionately. The difference in magnitude of the effects on Class I and Class II MHC was consistent with the more severe effect on the positive selection of CD4 + SP cells in the Foxn1 D/D mutants.
Maturation of SP thymocytes was delayed in Foxn1 D/D mice and immature SP thymocytes migrate into the periphery In the postnatal thymus, SP thymocytes stay in the thymic medulla after positive selection for 3 to 5 days for further maturation [24,25]. We previously reported that the thymic architecture was completely disorganized in the Foxn1 D/D mice, with lack of cortico-medullary junction and medullary organiza- tion, as well as decreased TEC differentiation [10], although SP thymocytes do leave the thymus and colonize the periphery [20]. This raises the question of whether and how SP thymocytes undergo maturation in this deficient environment. To address this issue, we assayed expression of the maturation markers HSA and Qa2 on adult peripheral T cells. As we have previously shown that SP thymocytes in the Foxn1 D/D thymus are contaminated with recirculated peripheral T cells reentering the very hypocellular Foxn1 D/D thymus [10,20], we tracked the expression of these markers on thymocytes from newborn through 28 days postnatal, and compared them to Foxn1 +/D controls.   In control mice at the newborn stage, most CD4 and CD8 SP thymocytes are HSA hi , while in the spleen the vast majority of T cells have down regulated HSA, and there is essentially no Qa2 expression in either population (Fig. 6A). By Day 14, over half of SP thymocytes have down regulated HSA, and a few also have low levels of Qa2. In contrast, over 50% of CD4 and over 80% of CD8 splenic T cells are Qa2 + , and virtually all are HSA negative (Fig. 6B). This trend continues in the thymus such that 4 weeks of age most thymic SP cells are HSA lo Qa2 2 , while most splenic T cells are HSA 2 Qa2 + (Fig. 6C).
In the Foxn1 D/D mutants at all ages examined, both SP thymocytes and splenic T cells showed differences in their HSA and Qa2 profiles compared to controls. At the newborn stage, over 90% of thymocytes were HSA hi , while over 50% of CD4 and 35% of CD8 splenic T cells also showed this phenotype (Fig. 6A). At Day 14, about twice as many SP thymocytes remained HSA hi in the mutants compared to controls, although the frequency of HSA lo Qa2 + thymocytes was similar. However, in the spleen the effect was much greater, with 40% of CD4 and 30% of CD8 T cells remaining HSA hi , and the percentage of Qa2 + cells was roughly half that of controls (Fig. 6B). Even at Day 28, roughly 15% of SP thymocytes and 5-7% of splenic T cells retained high HSA levels, ten-fold that of control mice, and fewer splenic T cells were Qa2 + (Fig. 6C). Thus, down-regulation of HSA and upregulation of Qa2 on both CD4 and CD8 SP thymocytes and on peripheral T cells were delayed in Foxn1 D/D mice.
One explanation for the presence of HSA hi peripheral T cells is that some immature Foxn1 D/D SP thymocytes might migrate out of the thymus. Consisted with this possibility, we found that CD4 peripheral T cells from the Foxn1 D/D mice could not be completely activated. Around 20% of CD4 cells from Foxn1 D/D mice could not up-regulate the expression of CD25 or CD69 after stimulation with anti-CD3 within 24 hours in vitro (Fig. 6D). Almost all Foxn1 +/D CD4 cells were activated with the same stimulation. In contrast, Foxn1 D/D CD8 SP cells were more similar to the profile of the Foxn1 +/D CD8 cells, with almost complete activation. Consistent with our previous results, Foxn1 D/D T cells had constitutively high expression of CD44, so this was not a reliable marker for activation status [21] (Fig. 6D).

Foxn1 D/D peripheral CD4, CD8 T cells have different responses to SEA super-antigen in vivo
We next assayed how these T cells responded to specific antigen in the periphery. We first tested the response of TCR Vb T cell clones to the super-antigen SEA in vivo. Before and after exposure to SEA, the expression of Vb3 and Vb11 were measured in individual mice. Because the frequency of Vb3 and Vb11 bearing cells in individual Foxn1 D/D mice was quite variable, the percentage change in Vb3/11 bearing cells in individual mice was shown by the ratio of before and after injection. As expected, the percentage of Vb3/11 bearing CD4 and CD8 T cells increased 2-4 fold after exposure to SEA at day 2 post-injection in the Foxn1 +/D mice, and then went down rapidly to the same or even below the level seen before exposure (Fig. 7), similar to a previous report [20]. However in the Foxn1 D/D mice, the Vb3bearing CD4 T cells showed a delayed but a stronger and longer lasting response, with a 5-fold increase in percentage at day 6 after SEA exposure, and then a slow decline. CD8 T cells also had a delayed response although it was weaker than CD4 T cells, and even weaker than Foxn1 +/D CD8 T cells (Fig. 7A). Both Foxn1 D/D Vb11 bearing CD4 and CD8 cells had a delayed and weaker response to SEA compared to the same cells in the Foxn1 +/D mice (Fig. 7B). These data indicated that different Foxn1 D/D T cell clones have different responses to the same antigen in vivo. Thus, some clones might increase and some clones might be reduced in response to the same antigen.
The thymus defects in the Foxn1 D/D mice result in very low SP cell production and decreased T cell output from the thymus, causing a hypocellular peripheral environment [10,20]. The hypocellular periphery contributes to the peripheral T cell phenotypes by promoting an activated/memory phenotype in most T cells [20]. To distinguish whether the abnormal response of Foxn1 D/D T cells was directly due to the developmental defect in the thymus or the environmental defect in periphery, 4610 6 sorted Ly5.1 wild-type peripheral T cells per mouse were retroorbitally transferred into Foxn1 D/D adult mice. One day later, 1 mg SEA per mouse was injected. The percentages of Vb3 bearing T cells of both host and donator were measured before and after injection of SEA and shown as the ratio of percentage (Fig. 7B). After 2 days of exposure to SEA, the percentage of transferred Ly5.1 Vb3 bearing wild-type CD4 cells was increased more than 2 fold, which reached a peak of 8 fold at day 4 and then reduced. A similar result was showed in Ly5.1 wild-type CD8 cells. The percentages of transferred Vb3 bearing CD4 and CD8 clones did not increase in the absence of SEA injection, indicating that the increase in transferred Vb3 bearing CD4 and CD8 clones was not due to homeostatic proliferation. These results suggest that the hypocellular peripheral environment shapes the response of T cells to specific antigen. The host Vb3 bearing CD4 cells had a weaker response to SEA than transferred Ly5.1 wild-type cells, although the magnitude of their response was similar to that seen in panels A and B. This suggests that in direct comparison to wild-type cells, the Foxn1 D/D peripheral T cells have a weaker response to specific antigen. It is also possible that the low expression of surface TCR on Foxn1 D/D T cells resulted in a reduced ability to compete for limited antigen stimulation in the same mouse.  of Tg CD8 population in percentage, and 2 mg OVA gave a 1.3 fold increase. Doses of OVA below 2 mg caused a decrease in the percentage of CD8 + ;OT1 Tg cells (Fig. 8A), as previously reported [26,27]. In the Foxn1 D/D ;OT1 Tg mice, a hypersensitive response was seen in the CD8 + ;OT1 Tg population. Doses of 0.5 mg were enough to drive a dramatic expansion of CD8 + ;OT1 Tg cells at day 2 to a level higher than that seen with the 20 mg dose in Foxn1 +/D ;OT1 Tg control mice. This response lasted 2 days longer than the response in Foxn1 +/D ;OT1 Tg control mice, then fell to the level of day 0 at day 6. At no dose did the CD8 + ;OT1 Tg population significantly decrease to below the level of day 0 in Foxn1 D/D mice (Fig. 8B). Thus, the Foxn1 D/D ;OT1 Tg T cells had a hypersensitive proliferative response to low doses of antigen, and never responded to antigen by with clonal deletion. We have previously shown that peripheral T cells in the Foxn1 D/D mice have an activated/memory phenotype due to lymphopenia-induced proliferation (LIP) [20]. To test whether the hypersensitive response of Foxn1 D/D ;OT1 Tg CD8 + cells was due in part to exposure to the lymphopenic environment of the Foxn1 D/D mice, rather than to differences in intrathymic development, we transferred CD8 + ;OT1 Tg cells derived from Foxn1 +/D ;OT1 Tg mice into Foxn1 D/D mice that did not carry the OT1 transgene ('host' in Fig. 8C-E). Two days later (Day 0) we injected a low dose of OVA at 0.25 mg per mouse. As above, the percentage of CD8 + ;OT1 Tg cells was measured before and after injection of OVA, and the change of percentage of CD8 + ;OT1 Tg cells was shown as a ratio. As expected, host T cells showed no change either in the presence or absence of OVA over a 12-day period (Fig. 8C). The percentage of transferred CD8 + ;OT1 Tg cells increased at day 2 after injection of OVA, and then reached a peak to 8 fold higher than the level of day 0, consistent with an LIP response in the hypocellular Foxn1 D/D peripheral environment [20]. In recipient mice that were given injection of OVA, the percentage of CD8 + ;OT1 Tg cells increased more slowly and reached a lower peak (3 fold increase) than with cell transfer alone. Our interpretation of this result is that the expansion of CD8 + ;OT1 Tg cells was driven by LIP, and that the low dose of 0.25 mg OVA induced some level of clonal deletion in the CD8 + ;OT1 Tg, as it did in the Foxn1 +/D ;OT1 Tg mice (Fig. 8A). If the OVA had further stimulated the LIP effect, the percentage of CD8 + ;OT1 Tg cell should have been higher than the percentage in the cell transfer control group (Fig. 8C). Thus, with OVA challenge immediately after cell transfer, the transferred CD8 + ;OT1 Tg cells underwent clonal deletion in response to the i.v. injection of OVA.
The possibility that we were inducing simultaneous LIP and clonal deletion in the above experiment raised the issue of whether we could separate the two responses. Another experiment was performed which OVA was injected 12 days after CD8 + ;OT1 Tg cells were transferred. At this time, all the transferred CD8 + ;OT1 Tg cells had already undergone LIP, and displayed a proliferating status and were CD44 hi , but the percentage had returned to a level similar to day 0 ( Fig. 8C and data not shown). This situation was therefore more similar to that of the host T cells in the Foxn1 D/D mice [20]. In controls that received cells only, the percentage of CD8 + ;OT1 Tg cells did not change over the additional 12-day period without OVA injection ( Figure 8D). In contrast, after injection of OVA the percentage of CD8 + ;OT1 Tg cells increased 2 days later, with a overall profile similar to that seen with injection of OVA soon after cell transfer ( Figure 8C), rather than undergoing deletion as seen in the Foxn1 +/D ;OT1 Tg mice. This result indicated that he response to OVA was dictated not by the thymic environment in which they developed, but by the peripheral environment in which they were exposed to OVA. We interpret this result to mean that exposure to the lymphopenic peripheral environment of the Foxn1 D/D mice and subsequent LIPinduced development of a CD44 hi memory like phenotype conferred to the transferred cells a hypersensitive response to specific antigen. CD4 + CD25 + regulatory T cells can partially suppress the response of the Foxn1 D/D ;OT1 Tg CD8 + cells to OVA in vivo One possibility for the high response of CD8 + ;OT1 Tg cells in the Foxn1 D/D peripheral environment to OVA was the lack of normal regulatory T cell (T reg ) function in Foxn1 D/D mice [20]. To address this issue, sorted CD4 + CD25 + Treg cells from the Foxn1 +/D control mice were transferred into the Foxn1 D/D ;OT1 Tg mice together with i.v. injection of OVA at 0.4 mg per mouse, an compared to injection of OVA or PBS alone. In the OVA only group, the percentage of CD8 + ;OT1 Tg cells increased as expected (compare to Figure 8E with 8B). In the OVA plus Treg group, the percentages of Tg CD8 cells were decreased comparing to the percentage in OVA non-Treg cells group (Fig. 8E), indicating that the co-injected Treg cells could suppress the proliferative response to antigen. These results indicated that the reduced function of CD4 + CD25 + Treg cells in Foxn1 D/D might contribute to the high response of CD8 + ;OT1 Tg cells to OVA in the Foxn1 D/D peripheral environment.

Discussion
Our previous work has shown that the thymic microenvironmental defects in the Foxn1 D/D mouse result in a variety of secondary effects on thymocyte development and peripheral T cell function [10,20,21]. The primary result from these previous studies was that T cells in the Foxn1 D/D thymus developed from kit negative progenitor cells via an atypical differentiation pathway. Our current data further explore the mechanisms by which this defective microenvironment influences thymocyte development, specifically positive and negative selection, and how these defects impact T cell function. We show that these atypical T cells have surprisingly normal Vb usage, but significant defects in positive and negative selection, especially for CD4 + SP cells. Furthermore, immature SP cells were able to exit the thymus prematurely. These defects were exacerbated by peripheral lymphopenia caused by the low thymic output, and the poor function of T reg cells generated in the Foxn1 D/D mice. Thus, peripheral T cell phenotypes were due to both primary defects in the thymic Canonical TCRab progenitors perform VDJ recombination randomly after commitment, but positive selection for different Vb may need continuous development of a variety of TECs [5]. In spite of the severe defects in Foxn1 D/D TEC differentiation and the lack of organized thymic compartments, we found that Vb usage on peripheral T cells was largely similar to control mice. This result suggested that these atypical progenitors had relatively normal VDJ recombination of TCR Vb, that the severe TEC defect in the Foxn1 D/D thymus did not significantly affect the ability to support the process of TCRab commitment, and that there was  6 sorted Foxn1 +/D ;OT1 Tg CD8 T cells were transferred into Foxn1 D/D mice. 1 day (C) and 10 days (D) later, the percentage of OT1 Tg CD8 cell was measured as day 0 and then OVA 250 mg per mouse was injected. The percentage of OT1 Tg CD8 T cells was continuously measured at the indicated day after OVA injection. Results shown were the average volume from two or three individual experiments. E, the hypersensitive response of Foxn1 D/D ;OT1 Tg CD8 T cells was partially suppressed by the transferred CD4 + CD25 + regulatory T cells. Sorted CD4 + CD25 + T cells were retro-orbitally transferred into Foxn1 D/D ;OT1 mice combining with OVA 400 mg per mouse. The results shown were the average volume from two individual experiments. doi:10.1371/journal.pone.0015396.g008 no selective loss of specific Vb bearing clones during positive selection. However, the peripheral Vb repertoire was highly variable between mice. This variability could reflect variability in the differentiation and/or organization of the thymic environment between different mutant mice. Alternatively, T cells with higher affinity to MHC may undergo selective expansion via homeostatic induced proliferation in the Foxn1 D/D lymphopenic periphery, where MHCII expression should be normal. In the case of TCR Vb5, this allele was reported to advantageously expand in inflammatory bowl disease (IBD) [28,29]. As our previous data also showed that some Foxn1 D/D mice showed autoimmune disease symptoms including prolapse [20], the increased TCR Vb5 expression in the Foxn1 D/D periphery might be driven by an autoimmune response.
While DCs are the major stromal cells that perform negative selection at the cortical-medullary junction (CMJ) and in the medulla, TECs can also contribute to this process [9]. In this study, although the percentage of bone marrow derived DCs in the Foxn1 D/D thymus was relatively increased, the capability of negative selection was still lower in the mice compared to in the Foxn1 +/D mice. This reduction of negative selection may be due to the abnormal thymic architecture, with lack of both CMJ and medulla in the Foxn1 D/D thymus, making it difficult for the SP thymocytes to interact properly with DCs. In addition, the reduction in negative selection could be secondary to premature exit of immature T cells from the thymus. In the normal thymus, SP thymocytes undergo maturation in the medulla for 10-14 days, during which time the expression level of HSA is down-regulated, and Qa2 is gradually up-regulated [24,30,31]. We found that the SP thymocytes and splenocytes expressed a higher percentage of HSA hi and lower percentage of Qa2 + in newborn and day 14 Foxn1 D/D mice. Consistent with the immature phenotype of Foxn1 D/D peripheral T cells, some CD4 + T cells couldn't completely be activated under anti CD3 stimulation in vitro, although the CD8 + T cells appeared more normal than CD4 T cells (Fig. 5D). Other work in our lab has shown that the thymic vasculature in Foxn1 D/D mice does not form properly, including inability to retain FITC-dextran (JL Bryson, et al., in preparation). This functional deficit combined with the lack of an organized medulla could combine to result in HSA hi immature cells migrating out of the thymus prematurely in Foxn1 D/D mice. Thus, the Foxn1 D/D thymus is both inefficient at generating mature SP cells and unable to retain the resulting immature cells in the thymus, resulting in immature and autoreactive T cells in the periphery.
Positive selection of CD4 TCRab thymocytes is completely dependent on MHC II expressing TEC, while the development of CD8 thymocytes can be supported by MHC I on both TEC and other thymic stromal cells in thymus [32,33,34]. In Foxn1 D/D thymus, the positive selection for CD4 transgenic T cells was dramatically damaged, while the selection for CD8 transgenic T cells was less affected (Fig. 3, 4). This difference between CD4 and CD8 is consistent with the greatly reduced expression of MHC II on TECs, while MHC I was almost normal in the Foxn1 D/D thymus [10], (Fig. 5). This result was also consistent with our data from another Foxn1 mutant, Foxn lacz , in which Foxn1 expression is gradually down-regulated after birth. In these mice, MHC II expression on TECs and production of CD4 T cells was reduced, but CD8 T cells appeared normal [18]. The CD1 molecule was reported to support CD4 positive selection in MHC II 2/2 mice [35]. CD1 expression was normal in Foxn1 D/D thymus (data not shown), suggesting that it might support some positive selection for CD4 T cells in Foxn1 D/D mice. Alternatively, the reduced positive selection could be due in part to an inability of the thymocytes themselves to respond to selection signals, either due to prior developmental deficits, or to intrinsic qualities based on their differentiation from an atypical progenitor cell [20].
These intrathymic defects in T cell differentiation and selection gave rise to peripheral T cells with defects in in vivo antigen-specific response. The antigen-specific response in vivo is a complicated process. Multiple facts may affect the process such as the dose of antigen, preparation of antigen, usage of adjuvant, the way of giving antigen, animal age and immune status [26,27,36,37,38,39]. For example, oral feeding of antigen may result in tolerance through anergy [39]; intravenous injection of antigen may delete antigen specific T cells rapidly, with remaining cells hyporesponsive; while the subcutaneous injection of antigen may delete antigen specific T cells, with the remain cells hypersensitive [36]. Both low and high dose antigen can result in tolerance, but low dose of antigen results in anergy or clone deletion, while high dose of antigen results in active suppression of antigen specific T cells [26,27]. During antigen specific tolerance, the CD25 + CD4 + T regulatory cells that develop in the thymus play an important role in the process of tolerance and immune balance [38,40]. In Foxn1 D/D mice, the peripheral T cells had an abnormal response to specific antigens; CD4 T cells showed a delayed and strong response to SEA, while CD8 T cells had a delayed but weaker response compared to the same cells in the Foxn1 +/D mice (Fig. 7), or to that reported in wild-type mice [41].
Both intrathymic and peripheral defects could contribute to the abnormal antigen response of Foxn1 D/D T cells: 1) the deficient thymic environment, including the down-regulation of MHC II on TEC, might preferentially select the higher avidity T cells, and these cells could not be deleted by negative selection; 2) the abnormally lymphopenic peripheral environment causes lymphopenia induced proliferation (LIP) [20]; this could cause some clones with a higher avidity to be advantageously expanded in the Foxn1 D/D periphery; 3) since CD4 + CD25 + regulatory T cells played a key role in antigen specific tolerance in vivo [38,40,42], some phenotypes could result from the reduced function of CD4 + CD25 + regulatory T cells in the Foxn1 D/D mice [20]; and 4) constitutively low TCR expression on Foxn1 D/D T cells may reduce the response to antigen [21]. The mutant-like response to SEA of wild-type or OT-1 transgenic SP T cells transferred in Foxn1 D/D mice (Fig. 7) implicates the peripheral environment, rather than the development in an abnormal thymic microenvironment, in this abnormal response. Furthermore, we showed that wild-type CD4 + CD25 + regulatory T cells transferred into Foxn1 D/D mice could partially inhibit the strong response (Fig. 8E). Taken together, our results suggest that that the Foxn1 D/D lymphopenic environment and the lack of normal function of CD4 + CD25 + regulatory T cells were the major reasons for the hypersensitive response in the Foxn1 D/D mice.
Taken together, our results show that both deficient development of T cells in the thymus and an abnormal antigen specific response in the periphery contribute to the phenotype of peripheral T cells in Foxn1 D/D mice. Further analysis on this experimental system may be useful for the studies of clinical immune tolerance or vaccine immune response in humans with immune deficient disease.

Mice and treatment
Foxn1 D/D mutant mice generated in our laboratory were described previously [10]. The mice were of a mixed C57BL/6J and 129SvJ background. BL/6-Ly5.1 mice, BALB/c mice, DO11.10 and OT1 TCR-transgenic mice were obtained from Jackson Labs. All transgenic strains were routinely genotyped by PCR. Foxn1 D/D ;BALB/c mice were generated by backcrossing Foxn1 D/D mice (H-2 b , IE -) with BALB/c (H-2 d , IE + ) for more than 5 generations. To control for the H-2 d background, a pair of primers that are linkage marker D17MIT22 for H-2 IA and IE genes were used for genotypic PCR analysis. This primer pair produces a 157 bp fragment for the C57BL/6 allele, and a 183 bp-sized fragment from BALB/c allele. Foxn1 D/D ;DO11.10 Tg and Foxn1 D/D ;OT1 Tg mice were generated by mating Foxn1 D/ D ;BALB/c mice with DO11.10 mice and Foxn1 D/D mice with OT1 mice, respectively. For testing the response of Vb T cell clones to superantigen, mice were retro-orbitally injected with a single dose of 1 mg SEA (Staphylococcal Enterotoxin A, Sigma-Aldrich) per mouse, in 200 ml PBS. For response of OT1 Tg CD8 cells to OVA protein (Sigma, Steinheim, Germany, MW:44.287 Da), mice received single retro-orbital injections of a serial dose range from 0.06 to 20 mg in 200 ml PBS. All the mice were maintained in a specific pathogen-free (SPF) facility at University of Georgia; experiments were approved by the University of Georgia's Institutional Animal Care and Use Committee. The internal IACUC approval number currently is A2008-10181.

Activation of peripheral T cells by anti-CD3 antibody in vitro
26106 peripheral lymphocytes prepared from spleen and LNs were plated in 24-well plates previously coated with 10 mg/ml of anti-CD3e Ab and cultured for 24 h. As a control, cells were cultured in media alone. Cells were harvested after 9 and 24 hour culture and stained with anti-CD4, anti-CD8, and anti-CD44, anti-25 and anti-CD69 markers, then placed in FACS store solution for analysis by flow cytometry.

Adoptive transfer
Peripheral lymphocyte suspensions were prepared from freshly isolated spleen and lymph nodes from C57BL6/J Ly5.1 or OT1 donor mice and further isolated by density gradient centrifugation and lysis of RBC in ACK lysis buffer (Cambrex Bio Science, Walkersville, MD). Around 4610 6 cells were transferred into Foxn1 D/D mice. For isolation of T regulatory cells, peripheral lymphocyte suspension from Foxn1 +/D mice was stained by anti-CD4 PE and anti-CD25 allophycocyanin mAbs, then sorted by MoFlo FACS (DakoCytomation). 2610 5 T regulatory cells per mice were transferred into Foxn1 D/D mice.