The thymic stromal niche normally directs the production and export of a self-tolerant T cell repertoire. Many models of spontaneous autoimmunity, however, develop thymic architectural abnormalities before disease onset. Although this is suspected to affect central tolerance induction, creating an autoimmune predisposition, in-depth analysis of the microenvironment within these thymi is lacking, such that the mechanisms and likely direct effects on the T cell repertoire are unknown or speculative. Here we show that NZB mice, the first described model for systemic autoimmunity, demonstrate a complex thymic phenotype, including a lack of the autoimmune regulator (Aire), early defects in thymic epithelial cell (TEC) expansion, and evidence for altered NF-κB2 signaling. Analysis of medullary TEC revealed a numerical loss of the Aire-expressing MHC class IIhigh (mTEC-high) subset as well reduced Aire protein and mRNA per cell. RelB expression was also reduced, while chemokines CCL19 and CCL21 were increased. Unexpectedly, the proportion of cortex and medulla in the NZB mice was normal from 36 wk, despite worsening architectural abnormalities. These data show that the NZB defect is more complex than previously appreciated, segregating into early numerical TEC deficiencies that correct with age, late degeneration of the niche architecture that does not affect TEC number, and a persistent reduction in Aire and RelB expression per cell acquired upon mTEC-high differentiation.
The thymic microenvironment consists of heterogeneous, spatially specialized microniches, which allow thymocytes at different stages of maturation to receive the signals required to continue T cell development (1). In turn, signals from thymocytes to the stromal cells are required for their normal function. Thymic stromal dysfunction can predispose or even directly cause autoimmunity by allowing autoreactive thymocytes to continue development, and by failing to create a regulatory T cell repertoire to keep low-affinity, autoreactive T cells in check. The development of severe thymic architectural abnormalities before spontaneous autoimmunity is highly conserved in mice and has also been reported in humans (2, 3, 4, 5, 6), but the direct cause and effects of these thymic defects remain largely unknown.
Nonhemopoietic thymic stromal cells include cortical and medullary epithelium (cTEC, mTEC),3 reticular fibroblasts, and endothelium (1). During thymocyte development, cTEC impose MHC restriction through positive selection, while dendritic cells and mTEC are required to rid the developing T cell repertoire of high-affinity self-reactivity through negative selection. A proportion of mTEC, termed mTEC-high, express high levels of MHC class II, CD80, and CD86 (7, 8, 9). mTEC-high express the autoimmune regulator, Aire, which is a transcription factor that provides a pivotal contribution to negative selection by driving expression of particular tissue-restricted self-Ags (TRAs), presenting an “antigenic snapshot” of the peripheral environment to developing thymocytes (8, 10, 11). A loss of Aire results in organ-specific autoimmunity in mice and humans (10, 12), and even the loss of a single allele is sufficient to decrease negative selection and increase autoimmunity in susceptible animals (11).
Many murine autoimmune models, including NZB, NOD, (NZB × NZW)F1 (B/W F1), MRL/MP-Faslpr, C3H/HeJ-Fasgld, and BXSB/MpJ-Yaa mice, develop progressive thymic stromal disruption, leading to the hypothesis that abnormal thymic niches can alter T cell selection events, predisposing to autoimmunity (2, 3, 4, 5, 13, 14). NZB mice develop a systemic autoimmune disease, including antierythrocyte, antinuclear, anti-single-stranded DNA, and antithymocyte autoantibodies. Pathogenesis is complex and polygenically controlled, with an array of immune abnormalities (15, 16, 17, 18, 19). Similarly to human systemic lupus erythematosus, disease onset occurs well into adulthood (mean onset, ∼6 mo in mice; ∼30 years in humans) (20, 21). By 1 year of age, NZB mice show 100% incidence of anti-erythrocyte autoantibodies, and death from anemia occurs at around 62 wk (13, 22, 23). Backcrossing studies have shown that autoantibody titers correlate with thymic disruption in B/W F1 mice, and that NZB mice contribute the genes responsible for the development of the thymic defect (24). No mice with normal thymi develop anti-dsDNA Abs, although a small percentage develop proteinuria (24).
Prior studies of NZB thymi utilized histological techniques and reported, variously, premature thymic involution, progressive reduction in mTEC and cTEC during the abnormal expansion of perivascular spaces and keratin-negative areas (KNA), reduced expression of the epithelial cell adhesion molecule (EpCAM), and the ectopic development of isolated mTEC in the cortex (3, 13, 14, 25). However, no formal quantification of TEC or other stromal cell types has been made and compared with other mouse strains, or over time. This study compares NZB thymic stroma to nonautoimmune-prone mice using flow cytometry and quantitative PCR (qPCR). We show that TEC expansion lags behind wild-type (WT) strains, resulting in a reduced stroma/T cell ratio in young mice. This is not normalized by sex steroid withdrawal, a process shown to increase TEC (and thymocyte) number in normal mice (26). NZB mice show reduced numbers of tolerance-inducing mTEC-high cells, which in turn express reduced levels of Aire, TRAs, and RelB, suggesting a new tolerance breakpoint in this model for systemic autoimmunity. Furthermore, we report an increase in the expression of lymphotoxin (LT)-dependent chemokines, which very likely contribute to the hallmarks of the NZB thymic defect.
Materials and Methods
NZB mice were bred in-house at the Monash University Mouseworks facility, Melbourne, or obtained from the Walter and Eliza Hall Institute (WEHI) Central Animal House, Melbourne. C57BL/6J (B6) and BALB/c mice were obtained from Central Animal Services, Monash University, Animal Resource Centre (ARC), Perth, or the Baker Heart Research Institute (BHRI), Prahran. B10.Br mice were obtained from the WEHI Central Animal House, and NOD Lt mice were from ARC. All mice were specific pathogen-free and housed in microisolators at BHRI or Monash University Mouseworks. All experiments were performed with the approval of the Monash University or BHRI animal ethics committees.
Analysis of recent thymic immigrants
Unenriched, red cell lysed splenocytes were labeled with CFSE (Molecular Probes) at a cell density of 10 × 106 cells/ml in sterile PBS with 1 μM CFSE at 37°C for 5 min. Staining was stopped by adding 10× volume RPMI 1640 plus 0.2% BSA. Cells were washed three times in RPMI 1640 alone. Recipient mice received 15 × 106 cells in RPMI 1640 i.v. and mice were humanely killed 4 days after transfer. Recent thymic immigrants were defined as CFSE-positive cells, collecting 5–6 million total events. Perfusion experiments showed that CFSE-positive cells were intrathymic and not traveling through thymic blood vessels at the time the mice were killed (data not shown).
Direct antiglobulin (Coombs) test for antierythrocyte autoantibodies
Aged NZB mice (>20 wk) used in this study were tested at autopsy for antierythrocyte autoantibodies using a hemagglutination assay. Briefly, whole heparanized blood was washed three times in cold FACS buffer (5 mM EDTA in PBS with 0.2% BSA and 0.02% NaN3) then diluted to 2% in FACS buffer before being added to an equal volume of serially diluted anti-mouse Ig (Poly1270; BD Pharmingen) in a 96-well round-bottom plate. Plates were incubated for 60 min at 37°C before being read. Thirty percent (3/10) of 20- and 24-mo-old mice were positive by hemagglutination assay, and the incidence reached 100% (6/6) by 12 mo.
A nonenriching thymic stromal preparation was performed for individual thymi as previously described (27). Cells (5 × 106) were added to 96-well round-bottom plates and suspended in 50 μl of an appropriately diluted primary mAb or conjugate for 15 min at 4°C in the dark, followed by a wash step (200 μl of FACS or EDTA/FACS buffer). Secondary mAbs were then added, incubated, and washed twice before resuspension in 200 μl of EDTA/FACS buffer for acquisition. Intracellular staining for Aire and Ki67 was performed using the BD Cytofix/Cytoperm kit (BD Biosciences) according to the manufacturer’s instructions. A FACSCalibur and CellQuest software (BD Biosciences) were used for flow cytometric analysis, using four fluorescent channels (FL1 for FITC, FL2 for PE, FL3 for PECy5/PerCP Cy5.5, and FL4 for allophycocyanin).
TEC were defined as CD45− and MHC class II+ or EpCAM+. The mTEC-high subset expresses high levels of MHC class II and Ulex europeaus agglutinin 1 (UEA-1) and includes a subset of cells expressing Aire (8). The mTEC-low subset was defined as MHC class IIlow, UEA-1+, while cTEC-high and cTEC-low subset was defined as MHC class IIlow, UEA-1+, while cTEC were Ly51+ and UEA-1−.
Statistical analysis was performed with SPSS v.15.0 software using a Mann-Whitney U test for unpaired nonparametric data.
Nonenriching, pooled thymic stromal preparations (8–10 thymi) followed by autoMACS depletion of CD45+ cells were performed as previously described (27). CD45− cells were enriched to 70–90% before staining with the appropriate immunofluorescence markers, and were then sorted on a FACSAria or FACSVantage cell sorter (BD Biosciences) at a pressure not exceeding 30 psi. Samples were collected in 30% (v/v) FCS in RPMI 1640 and then washed in PBS. Pellets were lysed in RLT buffer (Qiagen) and snap-frozen in liquid nitrogen before being stored at −80°C.
cDNA synthesis and qPCR
Immunohistology and confocal microscopy
Following humane killing, organs were removed, embedded in OCT compound (TissueTek; Miles Scientific), frozen in a liquid nitrogen/isopentane slurry, and stored at −80°C. Frozen tissue sections were cut at a thickness of 8–12 μm using a Leica CM1850 cryostat at −25°C and were stained and mounted as previously described (11). Images were acquired on a Bio-Rad MRC 1024 confocal microscope with a three-line Kr/Ar laser (excitation lines 488, 568 and 647 nm) using Bio-Rad LaserSharp acquisition software v.3.2.
Abs and conjugates
Abs were obtained from BD Pharmingen unless otherwise indicated. Primary Abs and conjugates used were UEA1 lectin (Vector Laboratories), anti-Ly51 (clones 6C3 and BP-1), anti-IA/IE (clone M5/114.15.2), anti-CD45 (clone 30-F11), anti-MTS15 (grown in-house), anti-CD31 (clone MEC13.3), anti-Aire (clone 5H12-2 (29
Delayed expansion of the thymic epithelium in NZB mice
Several causes of NZB thymic abnormalities have been proposed and linked to disease, including premature thymic involution (5, 25, 30). To further define specific stromal defects, and to follow their progression, we quantified various stromal subsets compared with nonautoimmune mouse strains. In particular, we sought to establish whether KNA occur as a result of either loss or displacement of TEC, since an overall reduction in TEC is likely to impact more strongly on tolerance induction. When compared with multiple WT strains, 12-wk-old NZB mice showed a significantly increased thymocyte-to-stromal cell ratio (Fig. 1⇓A) and reduced total number of TEC (Fig. 1⇓B). However, when compared across time, the overall thymus size was similar to B6 mice, expanding to normal size by 6 wk before following a similar pattern of thymic atrophy (Fig. 1⇓C). NZB mice had significantly fewer TEC until 12 wk of age; thereafter, they were similar to B6, BALB/c, and B10.Br mice (Fig. 1⇓, B and D, and data not shown).
Smaller medullary islets were evident by histological examination of 12-wk-old NZB mice, and they showed abnormal KNA in both the cortex (data not shown) and medulla (Fig. 1⇑E). While 3-wk-old NZB mice could not be distinguished from age-matched B6 mice in blind studies, aged NZB mice showed a severely disrupted thymus by histology (Fig. 1⇑F), with B cell follicles, evidence of fibrosis, giant perivascular spaces distinguishable from those that develop during normal thymic atrophy, and KNA infiltrating the cortex and poorly organized medulla.
NZB medullary abnormalities evident before development of KNA
All TEC subsets were present in the NZB mouse, including the tolerance-inducing MHC class IIhigh subset of mTEC (mTEC-high), but at 3 wk of age, all were numerically deficient (Fig. 2⇓A). By 12 wk of age, the TEC deficit was only due to reduced numbers of mTEC-high cells compared across multiple strains (Fig. 2⇓B; data also comparing BALB/c and B10.Br mice are not shown). By 24 wk of age the number of mTEC-high cells was normal (Fig. 2⇓C). NZB mice showed altered TEC proportions at all ages to 24 wk (Fig. 2⇓, D–F). Beyond this age, the proportion of cTEC relative to mTEC increased further in NZB mice, but this process also occurs in WT strains as part of normal thymic atrophy (Fig. 2⇓G). The NZB mice demonstrated a surprisingly normal aging epithelial compartment, with normal numbers of TEC and TEC subsets, despite histological evidence for ongoing degeneration of the three-dimensional stromal network.
NZB thymi are Aire deficient
The mTEC-high deficiency in young NZB mice raised the possibility of a reduction in the number of cells expressing Aire, which fall within this stromal subset. In both 3-wk-old (data not shown) and 12-wk-old NZB mice, we found a significant reduction in the overall proportion of TEC expressing Aire protein compared with both B6 and B10.Br mice (Fig. 3⇓A and data not shown). Total numbers of Aire+ mTEC were 5-fold reduced (Fig. 3⇓B), due to a combined reduction in the overall proportion of TEC, and, within this compartment, a preferential, proportional reduction in the mTEC-high subset. These results were supported by histology, which showed fewer Aire+ cells (Fig. 3⇓C). Strikingly, at higher magnification, the level of Aire expression per cell appeared to be lower by histology (Fig. 3⇓D), and the mean fluorescence intensity for Aire staining as shown in Fig. 3⇓A was significantly lower in NZB mice (B6 mice: 68.9 ± 3.9; NZB mice: 60.2 ± 4.7 (mean ± SD); p < 0.05). Subsequent qPCR analysis of sorted mTEC-high showed that the level of Aire transcript was at least double in WT strains B6 and B10.Br compared with NZB mice (Fig. 3⇓E and data not shown). To determine whether this was linked to loss of Aire function, we examined several Aire-dependent TRA and found these similarly reduced in NZB compared with B6 (Fig. 3⇓E) or B10.Br mice (data not shown), while expression of Aire-independent TRA was normal (Fig. 3⇓E). Hence, the overall reduction in NZB thymi has two components: fewer Aire+ mTEC-high cells, and reduced levels of expression at both protein and mRNA transcript levels per cell.
Recent work implicating receptor activator of NF-κB ligand (RANKL) signaling to lymphoid tissue inducer cells in the up-regulation of CD80 on mTEC, and subsequent Aire expression (9), led us to examine RANKL expression levels in the hemopoietic compartment of NZB thymi. We indeed found a 50% reduction in transcript (Fig. 3⇑F) of 6- to 12-wk-old mice compared with B6 and B10.Br WT strains, providing a possible mechanism for the lag in mTEC-high expansion in these mice. NZB mice were not deficient in the receptor activator of NF-κB itself (data not shown).
NZB TEC enter cell cycle at normal proportions
It is well reported that fibroblasts play an important role in TEC development (31, 32). Consistent with an overall reduction in CD45− stromal cells, we found a 2-fold reduction in the number of fibroblasts stained by the Ab MTS15 in 3- to 6-wk-old NZB mice (Fig. 4⇓A). Despite this, we found no significant difference in the proportion of any stromal subset entering cell cycle at either 3 wk (Fig. 4⇓B), when the number of stromal cells were significantly reduced, or at 24 wk (Fig. 4⇓C), when they were not.
Altered fibroblast and endothelial cell populations
The retention of mature T cells is a feature of many autoimmune thymi (33, 34), and abnormally high recirculation to the thymus of mature lymphocytes also occurs in B/W F1 mice (35, 36), perhaps suggesting endothelial abnormality. NZB mice at 3 wk had fewer endothelial cells (Fig. 5⇓A), while fibroblast and endothelial compartments both normalized numerically by 12 wk (data not shown). Beyond 12 wk of age, however, we found an increased proportion of nonhemopoietic, nonepithelial (EpCAM−) cells expressing MHC class II in the NZB mice. Staining with CD31 revealed that a proportion of these were endothelium (Fig. 5⇓B), increasing with age until most NZB endothelial cells were MHC class II+. With age, there was also a large increase in MHC class II expression on a subset of nonepithelial, nonendothelial (CD31−) stromal cells stained with the fibroblast marker MTS15 (Fig. 5⇓C). Young mice showed normal profiles (data not shown).
Inflammation and increased LT signaling can induce abnormal MHC class II expression and drive overproduction of chemokines CCL19, CCL21, and CXCL13 in peripheral tissues and autoimmune lesions (37, 38, 39, 40, 41, 42, 43, 44). These chemokines induce lymphocyte immigration into the perivascular space, causing enlarged KNA, and the eventual neogenesis of a tertiary lymphoid organ, with discrete T and B cell zones (42, 43, 44, 45, 46, 47), therefore sharing many similarities with NZB thymi. The possibility that abnormal intrathymic LT signals were driving the degeneration of the medullary epithelial network was thus investigated.
Altered expression of LT-dependent genes in NZB mice
LTβR ligands exist as either a membrane-bound heterotrimer (LTα1β2) or a secreted LTα analog, LIGHT (48). In NZB mice, mTEC subsets showed a 2-fold increased LTβ expression, while LTα expression was reduced (Fig. 6⇓A). However, expression of IL-6 and of and LT-dependent chemokines CCL19, CCL21, and CXCL13 was abnormal and differed between mTEC subsets (Fig. 6⇓A). The mTEC-low subset showed increased CCL19 expression, but normal CCL21 and CXCL13, while mTEC-high had normal CCL19, increased CCL21, and decreased CXCL13 expression, pointing to a dysregulation in factors controlling the differential expression of thymic chemokines in these animals. RelB, which controls chemokine transcription as a heterodimer with p52, was reduced in the mTEC-high subset alone. Expression of genes found to differ between B6 and NZB mice were further examined in B10.Br TEC subsets, and in all cases NZB was the abnormal strain (data not shown).
LTα and LTβ were not increased in the hemopoietic compartment (Fig. 6⇑B), which is normally the major source of LT expression in the thymus. Moreover, expression levels of neither subunit increased with age, suggesting no progressive, ongoing inflammatory process in the thymus.
Ordinarily, mTEC produce the LT-dependent chemokines CCL19 and CCL21 to attract developing CCR7-expressing thymocytes to the medulla (49), while CCL19 also induces thymocyte egress to the periphery (50). Although these chemokines attract mature lymphocytes into tissues (51), lymphocyte immigration was not increased in the NZB thymus (Fig. 6⇑C), unlike NOD mice, which also develop autoimmune-associated thymic architectural abnormalities, and showed significantly higher thymic immigration compared with B6 mice.
Castration does not normalize the TEC defects in NZB mice
The thymus is sensitive to sex steroids and, postpuberty, undergoes progressive thymic atrophy, which can be reversed by castration, resulting in significant parallel expansion of both thymocytes and TEC (26, 52, 53, 54). It has been hypothesized that NZB thymic defects occur as a result of premature thymic involution (55, 56). Although we show that TEC defects are present before puberty, since low levels of sex steroids circulate early in life (57), mice were castrated to determine whether hypersensitivity to sex steroids causes the defect, and whether sex steroid ablation could be beneficial in normalizing the NZB thymic defect. NZB thymi underwent significant hypertrophy postcastration (Fig. 7⇓A) compared with age-matched sham castrated controls. This hypertrophy, however, was limited to hemopoietic cells (Fig. 7⇓B). No TEC subset was significantly altered by castration (Fig. 7⇓, C and D).
Taken together, these data depict a complex thymic phenotype in the NZB mouse, consisting of an early numerical defect in TEC expansion that normalizes over time, chronically reduced levels of Aire expression, and a progressive increase in KNA coupled with increased chemokines known to cause thymocyte arrest in the medulla.
NZB mice have been studied for more than four decades as a prototypic model for systemic autoimmunity, both in their own right and as an F1 cross, commonly with the nonautoimmune prone NZW mouse, which generates a model for severe systemic lupus erythematosus (13, 14, 22, 23). The onset of autoimmune disease in NZB mice, well into adulthood, mimics human autoimmunity and makes it ideal among systemic autoimmune models for studying the role of congenital thymic defects before disease onset. Thymic architectural abnormalities in models of spontaneous autoimmunity precede disease, yet they have not yet been linked to a susceptibility gene. Studies in B/W F1 mice have shown that they are not autoimmune in origin (24), since disease-resistant mice can still possess thymic defects. However, the stromal defects were necessary for disease to develop and are therefore suspected to impose susceptibility.
Despite normal numbers of thymocytes from 6 wk of age in young NZB mice, TEC failed to undergo sufficient expansion parallel to thymocytes, despite no obvious loss in the proportion of cells in cycle. However, the numerical and proportional reduction in mTEC-high cells, and reduced Aire expression therein, demonstrated defects in mTEC differentiation. Although Aire−/− mice show altered mTEC differentiation (8, 58), it seems unlikely that NZB mTEC abnormalities occur secondary to the loss of Aire, since Aire−/− mice show an increased proportion of mTEC-high cells (8), which is entirely opposite to young NZB mice.
The reduction in the level of Aire expressed per cell was striking and unexpected, and to our knowledge this is the first demonstration of such a phenotype in a model of systemic autoimmunity. Many studies have reported an overall loss of Aire commensurate with a loss of mTEC (59, 60, 61), but not a loss of Aire transcription per mTEC-high cell, apart from mice heterozygous for the Aire gene (11). In the context of this study, it translated to reduced transcriptional ability shown through reduced expression of Aire-dependent TRA, which is noteworthy given that organ-specific autoimmunity is not generally held to be a feature of disease in this strain. It has been suggested that Aire may have an as yet poorly described role in Ag presentation (62), which could be linked to autoimmune susceptibility on this permissive background. However, if reduced Aire creates or contributes to reduced negative selection, it is probable that various polygenically controlled traits cooperate to break peripheral tolerance and instigate the various symptoms characterizing this complex disease. Indeed, not only are autoreactive T cells present in healthy humans and mice, but, depending on the background strain, Aire-deficient mice are not always susceptible to overt autoimmunity (H. Scott, unpublished observations). An early study also suggested that people heterozygous for an AIRE mutation are also likely to possess several striking yet subclinical immune abnormalities, including increased T cell activation (63), again showing the necessary contribution of multiple tolerance breakpoints to the eventual activation of an autoreactive cell.
Aire expression has also been linked to the alternative NF-κB2 signaling pathway (34, 59, 64, 65, 66). Accordingly, we found a reduction in RelB production in mTEC-high cells. Although gene deletions and mutations from signaling pathways culminating in RelB translocation show reduced or absent Aire expression using whole thymus or whole stroma PCR (59, 61, 67), or even sorted CD80+ TEC (66), the phenotype always included a loss of mature mTEC. The requisite experiments focusing on sorted UEA-1+ MHC class IIhigh (or CD80-high) cells (if any remain) have not been performed to determine whether this pathway directly affects Aire expression itself or merely the development of mTEC-high cells. Additionally, although the NZB thymic transcriptome was shown to be similar to that of the relb−/− mouse (68), this study also performed PCR on whole thymus preparations and, therefore, given the reduced proportion of mTEC in young NZB thymi, not surprisingly found a reduction in RelB and other medullary signaling molecules. Herein we show that the reduction in RelB is restricted to the mTEC-high subset, which is interesting given that it is normal in mTEC-low cells, and suggests that the abnormality is acquired upon differentiation.
The lack of mTEC was not due to simple attrition, shown by Ki67 staining, but reduced availability of RANKL is likely to explain the reduction in the proportion of mTEC-high cells in the young NZB mouse (9). The lack of medullary expansion is commensurate with a lack of relevant mesenchymal growth factors due to reduced numbers of the relevant mesenchymal fibroblast subset (MTS15+) (31). Alternatively, fibroblast number may be reduced secondary to, or even because of, an independent reduction in TEC, although the presence and expansion of fibroblasts in TEC-deficient, castrated rag−/− mice (26) would seem to preclude this.
Despite the early defects in TEC expansion and differentiation, numerically, the NZB epithelial composition was surprisingly similar to the WT in aged mice, suggesting a divergence in thymic defects. First, while lagging TEC expansion and skewed cortex/medulla proportions normalize with age, the level of Aire expression in sorted mTEC-high cells was always lower than WT strains, persisting beyond 36 wk when the numbers of mTEC-high cells in NZB mice no longer significantly differed (data not shown). Furthermore, histology showed progressive degeneration of the epithelial network with age. These results suggest that KNA occur due to TEC displacement rather than to their erosion, and that although TEC are present in normal numbers in aged NZB nice, their physical arrangement becomes progressively more abnormal.
As architectural abnormalities worsened, MHC class II was expressed by all stromal subsets. Endothelial cells in rodents, unlike humans, are generally negative for MHC class II, as are thymic fibroblasts. Whether these cells are either reprogrammed to express it in the aging NZB mouse or acquire molecules from neighboring cells cannot be determined from this study. However, expression of MHC class II on atypical cell types, including mouse endothelium, has been reported in response to inflammation and linked to IFN-γ, TNF-α, and LT exposure (37, 38, 39, 40). Endothelial cells do have the capacity to present Ag in an autoimmune setting, dependent on increased levels of the LT-dependent chemokine CCL21, as seen in NZB mice, to enable lymphocyte adhesion (69).
LT signaling culminates in nuclear translocation of the RelB/p52 dimer and results in transcription of genes critical for maintenance of secondary lymphoid organs and the mTEC network, including CCL19, CCL21, and CXCL13 (34, 45, 70). In this model, the increase in CCL19 and CCL21 was most likely due to dysregulated transcription rather than to increased LT signaling, since these chemokines were not increased uniformly across mTEC subsets. Although LTβ expression was increased in mTEC subsets, LTα was decreased, and there was no alteration in hemopoietic cells, which express the majority of LTαβ, nor was LTβR expression increased by mTEC. Taken together, these data show that the perturbation is likely to occur through signal strength, or downstream. Mediators of increased LT biological activity such as IL-6, also associated with reduced thymic epithelium in aged and autoimmune settings (71, 72, 73) and found here to be increased in the NZB thymus, may play a role, and a lack of negative feedback for chemokine expression is also possible. Regardless of their driving force, increased expression of CCL19 and CCL21 is a plausible mechanism for intrathymic retention of mature T and B cells (43, 50). Since increased thymic immigration does not occur, an arrest in cellular trafficking (likely hindering the exit of thymocytes and newly entered cells) is probably responsible for the increased proportion of mature T and B cells in these thymi (3, 20).
The lack of response from NZB TEC after castration when compared with age-matched sham-castrated NZB mice is interesting, given that thymocytes were capable of significant expansion and that TEC ordinarily increase in number by 10 days postcastration of WT mice (26). Androgen receptors expressed by the nonhemopoietic stromal compartment primarily drive thymic involution (74); however, results from the same study showed that although atrophy is initiated by stromal cells, regeneration requires some direct effect of androgen withdrawal on hemopoietic cells. Ordinarily, thymocyte numbers increase first (by days 5–7) after castration, followed by significant increases in TEC (∼days 7–10), supporting a model where stromal cells initiate thymocyte proliferation, after which thymocytes drive TEC expansion. Here we see expansion of the hemopoietic compartment, showing that NZB stromal cells are able to initiate castration-mediated thymocyte expansion, but that they themselves are unable to respond normally, or that they experience prolonged delay in doing so. These results do not support a sex steroid-related or premature thymic involution-based etiology for the lack of TEC.
Extensive defects in tolerance-inducing mTEC support the hypothesis that the NZB thymic defect contributes to loss of tolerance. The NZB mouse strain is a unique tool for studying Aire expression, since, unlike so many knockout and mutant models described to date, factors controlling the level of Aire expression per cell can be studied independently to a loss of mTEC-high cells.
Findings from this study will be important when considering immunomodulatory therapies for autoimmunity, which should include consideration of the thymic stromal compartment. Since immunity can be restored faster following chemotherapy and hemopoietic stem cell transplantation, using reversible sex steroid blockade (52, 75, 76, 77, 78), hemopoietic stem cell transplantation may be rendered safe enough to reinstate tolerance in some autoimmunity patients given a similar though less myeloablative regime. However, any underlying stromal defects must be considered, as their effects could be magnified following thymic hypertrophy, increasing the risk of relapse, particularly if stromal cells are incapable of expansion.
Finally, note that NZB mice—indeed, most patients with autoimmunity—are disease-free for a large proportion of a normal lifespan before accumulating the defects and activation events required to cross the threshold into pathogenesis. Despite strong genetic predisposition (100%) and the autoimmunity-associated defects reported herein, the fact that backcrossed mice can possess stromal defects in the absence of disease (24) suggests that the polygenic contribution to autoimmunity affords opportunity for protection in the face of the abnormal thymus.
The authors thank Jade Homann for valuable technical assistance, Lisa Bryant for highly expert animal husbandry, and Dr. Tracy Heng for helpful discussions.
R.L.B. is the Chief Scientific officer of Norwood Immunology, which has licensed some of the intellectual property from this project from Monash University.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by funding from Norwood Immunology and the Australian Stem Cell Centre. H.S.S. was supported by National Health and Medical Research Council Fellowships 171601 and 461204, National Health and Medical Research Council Program Grants 257501 and 264573, and Eurothymaide, Sixth Framework Programme of the European Union.
↵2 Address correspondence and reprint requests to Dr. Richard Boyd, Monash Immunology and Stem Cell Laboratories, Monash University, Wellington Road, Clayton, Victoria 3800, Australia. E-mail address:
↵3 Abbreviations used in this paper: cTEC, cortical TEC; Aire, autoimmune regulator; EpCAM: epithelial cell adhesion molecule; KNA, keratin-negative area; LT, lymphotoxin; RANKL, receptor activator of NF-κB ligand; TEC, thymic epithelial cell; mTEC, medullary TEC; mTEC-high, medullary TEC expressing high levels of MHC class II; mTEC-low, mTEC expressing low levels of MHC class II; qPCR, quantitative PCR; TRA, tissue-restricted Ag; WT, wild type; UEA-1, Ulex europeaus agglutinin 1.
- Received June 2, 2008.
- Accepted December 21, 2008.
- Copyright © 2009 by The American Association of Immunologists, Inc.