Subcellular Expression of Autoimmune Regulator Is Organized in a Spatiotemporal Manner* □ S

Autoimmune regulator (AIRE) is responsible for the development of organ-specific autoimmune disease in a monogenic fashion. Rare and low levels of tissue expression together with the lack of AIRE-expressing cell lines have hampered a detailed analysis of the molecular dynamics of AIRE. Here we have established cell lines stably transfected with AIRE and studied the regulatory mechanisms for its subcellular expression. We found that nuclear body (NB) formation by AIRE was dependent on the cell cycle. Biochemical fractionation revealed that a significant proportion of AIRE is associated with the nuclear matrix, which directs the functional domains of chromatin to provide sites for culturing g/ml G418 selection media, G418-resistant fluorescence from the cells representative fluorescent expanded. For construction retrovirus vector, human cDNAs subcloned into pMX-puro-FLAG vector pro- Medullary thymic epithelial cell (mTEC) line 14 the g/ml puromycin Biosciences). block- ade of proteasome activity, cells 100 gene expression Lipo- fectAMINE transfection reagent (Invitrogen) for COS-1 and RD cells for NB4 cells. pSI-AIRE and CMX-PML expression Western Blot Analysis— Sequential isolation of subcellular components was performed as described previously (15), with a slight modification. Briefly, wild-type HeLa cells and pEGFP-AIRE-transfected HeLa cells were washed once with PBS and resuspended in 400 (cid:1) l of ice-cold lysis buffer containing 20 m M HEPES (pH 7.9), 1 m M EDTA, 1 m M dithiothreitol, 0.2% Nonidet P-40 (Sigma), and a mixture of protease inhibitors. The cells were vortex-mixed vig- orously and kept on ice for 10 min. The mixture was then centrifuged for 1 min at 5,000 (cid:3) g , and the supernatant was harvested for the cyto- plasmic protein. After isolation of the cytoplasmic protein, nuclear pellets were resuspended in 50 (cid:1) l of cytoskeletal (CSK) buffer (10 m M PIPES (pH 6.8), 100 m M NaCl, 300 m M sucrose, 3 m M MgCl 2 , 1 m M EGTA, 0.5% Triton X-100, 1 m M dithiothreitol, and protease inhibitors) containing 100 units of RNase-free DNase I (Roche Applied Science) and incubated for 50 min at 30 °C. Ammonium sulfate was then added to a final concentration of 0.25 M , and the samples were kept on ice for 5 min. After centrifugation, the supernatant was harvested (chromatin fraction), and the pellets were then extracted with 2 M NaCl in CSK buffer. After further centrifugation, the supernatant was harvested (2 M NaCl wash), and the pellets were lysed with 8 M urea to provide the nuclear matrix fraction. Western blot analysis was performed as de- scribed previously (16) with anti-GFP polyclonal Ab (Invitrogen), anti-I (cid:2) B (cid:3) polyclonal Ab, and anti-lamin A/C mAb (Santa Cruz Biotechnol- ogy, Santa Cruz, CA). An

Breakdown of self-tolerance is considered to be the key event for the development of autoimmune diseases (1). Although a propensity to appear in families is one of the common features of autoimmune diseases, only a few genes relevant to the pathogenetic processes that underlie the development of autoimmune diseases per se are actually known (2). One of these genes is the autoimmune regulator (AIRE), 1 whose mutation is responsible for the development of autoimmune-polyendocrinopathy-candidiasis ectodermal dystrophy (APECED: OMIM 240300) with autosomal recessive inheritance (3)(4)(5)(6).
The AIRE gene encodes a predicted 58-kDa protein carrying a conserved nuclear localization signal, two plant homeodomain (PHD)-type zinc fingers, four LXXLL motifs or nuclear receptor interaction domains, and the recently described homogenously staining region and SAND domains (3,4). Based on the fact that PHD resembles the RING finger, which functions as an E3 ubiquitin (Ub) ligase (7), in both sequence and structure (8,9), we have recently found that AIRE acts as an E3 Ub ligase through the N-terminal PHD domain (PHD1) (10). The significance of this finding was underscored by the fact that disease-causing missense mutations in the PHD1 (i.e. C311Y and P326Q) abolished its E3 ligase activity. We speculate that AIRE might exert its function by facilitating polyubiquitination of substrate(s) in yet undetermined processes.
One of the key features of AIRE in the context of autoimmunity is its limited tissue expression in medullary epithelial cells and cells of the monocyte-dendritic cell lineage of the thymus (11,12), and both cell types are considered to play major roles in the establishment of self-tolerance by eliminating autoreactive T cells (1). Another characteristic feature of AIRE is its subcellular localization. Immunocytochemical staining has revealed its presence in the nucleus, appearing as a speckled pattern known as a nuclear body (NB) (11)(12)(13). Despite intense efforts, a coherent explanation of NB function has yet to emerge. * This work was supported in part by Special Coordination Funds of the Ministry of Education, Culture, Sports, Science and Technology, the Japanese Government (MEXT) (to M. M.), a Grant-in-Aid for Scientific Research from the MEXT (to S. H. and M. M.), the Tampere University Hospital Medical Research Fund, the Emil Aaltonen Foundation, the Finnish Medical Foundation, the Pirkanmaa Regional Fund of the Finnish Cultural Foundation (to J. P.), the Tampere University Hospital Medical Research Fund, the Finnish Academy, the Sigrid Juselius Foundation (to P. P.), Grant-in-Aid for Scientific Research (A) and Fund for "Research for the Future" Program from the Japan Society for the Promotion of Science and MEXT (to N. S.), and the CNRS (to V. D.). 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.
Although understanding the relationship between AIRE gene malfunction and the breakdown of self-tolerance promises to help unravel the pathogenesis of not only APECED but also other types of autoimmune disease, rare and low levels of tissue expression have hampered a detailed analysis of the molecular dynamics of AIRE. This situation is compounded by the lack of availability of established cell lines constitutively expressing AIRE. To overcome these difficulties, we have established cell lines stably transfected with AIRE and studied the molecular dynamics of AIRE in detail, especially focusing on the factors that influence its subcellular expression. The establishment of stably transfected cell lines with AIRE has enabled us to investigate many aspects of AIRE that could not be easily addressed by the transient expression systems. Our results suggest the existence of a group of mechanisms that together control AIRE expression within a cell in a spatiotemporal manner.

EXPERIMENTAL PROCEDURES
Construction of the Fusion Gene for AIRE cDNA with Marker Genes-Human AIRE cDNA was amplified by PCR from Marathon-Ready human thymus cDNA (BD Biosciences, San Jose, CA). Briefly, AIRE cDNA was first amplified with adaptor primers according to the manufacturer's instructions, and then nested PCR was performed with AIRE-specific primers in which EcoRI and SalI restriction sites were added at the 5Ј and 3Ј ends, respectively. The following primers were used: AIRE-U1, CGGAATTCATGGCGACGGACGCGGCGCTAC; AIRE-D1, ACGCGTCGACTCAGGAGGGGAAGGGGGCCG. The resulting fragment containing the human AIRE open reading frame was ligated into the cloning site downstream of the enhanced fluorescent gene in pEGFP-C1 (BD Biosciences), FLAG-tag in pCR3 vector (Invitrogen), or 6Myc-tag in pCR3 vector. Site-directed mutagenesis was performed using human AIRE cDNA cloned into pBluescript II SKϩ (Stratagene, San Diego, CA) as a template. pFLAG-AIRE C299A and C434A were generated with QuikChange (Stratagene) (10). The sequence of each constructed vector was confirmed by the dideoxy chain termination method with automated sequencing (Applied Biosystems, Foster City, CA).
Establishment of Stable Transfectants Expressing the AIRE Fusion Genes and Transient Gene Expression Studies-HeLa cells were transfected with pEGFP-AIRE plasmid by use of the FuGENE 6 transfection reagent (Roche Applied Science) according to the manufacturer's instructions. Forty eight hours later, culturing of the cells in selection media containing 400 g/ml G418 (Invitrogen) was started. After 14 days of culture in selection media, G418-resistant clones were isolated. The fluorescence signals from the cells were determined with the aid of fluorescence microscopy, and the representative fluorescent clones were further expanded. For the construction of retrovirus vector, human AIRE cDNAs were subcloned into pMX-puro-FLAG vector (kindly provided by Dr. T. Kitamura). Medullary thymic epithelial cell (mTEC) line 1C6 (kindly provided by Dr. M. Kasai) (14) or NIH3T3 cells were infected with the pMX-puro-FLAG-AIRE vectors and cultured for 14 days in the presence of 2 g/ml puromycin (BD Biosciences). For blockade of proteasome activity, cells were treated with 100 M N-acetyl-Leu-Leu-norleucinal (ALLN; Nacalai Tesque, Kyoto, Japan) or with 5 M benzyloxycarbonyl-Leu-Leu-Leu-H (MG132; Calbiochem) for 16 h. Transient gene expression studies were performed using the Lipo-fectAMINE transfection reagent (Invitrogen) for COS-1 and RD cells and electroporation for NB4 cells. The pSI-AIRE and CMX-PML expression plasmids and anti-AIRE Abs have been described previously (11,13).
Isolation of Cellular Proteins and Western Blot Analysis-Sequential isolation of subcellular components was performed as described previously (15), with a slight modification. Briefly, wild-type HeLa cells and pEGFP-AIRE-transfected HeLa cells were washed once with PBS and resuspended in 400 l of ice-cold lysis buffer containing 20 mM HEPES (pH 7.9), 1 mM EDTA, 1 mM dithiothreitol, 0.2% Nonidet P-40 (Sigma), and a mixture of protease inhibitors. The cells were vortex-mixed vigorously and kept on ice for 10 min. The mixture was then centrifuged for 1 min at 5,000 ϫ g, and the supernatant was harvested for the cytoplasmic protein. After isolation of the cytoplasmic protein, nuclear pellets were resuspended in 50 l of cytoskeletal (CSK) buffer (10 mM PIPES (pH 6.8), 100 mM NaCl, 300 mM sucrose, 3 mM MgCl 2 , 1 mM EGTA, 0.5% Triton X-100, 1 mM dithiothreitol, and protease inhibitors) containing 100 units of RNase-free DNase I (Roche Applied Science) and incubated for 50 min at 30°C. Ammonium sulfate was then added to a final concentration of 0.25 M, and the samples were kept on ice for 5 min. After centrifugation, the supernatant was harvested (chromatin fraction), and the pellets were then extracted with 2 M NaCl in CSK buffer. After further centrifugation, the supernatant was harvested (2 M NaCl wash), and the pellets were lysed with 8 M urea to provide the nuclear matrix fraction. Western blot analysis was performed as described previously (16) with anti-GFP polyclonal Ab (Invitrogen), anti-IB␣ polyclonal Ab, and anti-lamin A/C mAb (Santa Cruz Biotechnology, Santa Cruz, CA). An anti-AIRE polyclonal Ab was raised by immunizing rabbits with peptides corresponding to the C terminus of human AIRE. In situ extraction of the subcellular components from stable transfectants grown on coverslips was performed with the same method as described above.
Immunocytochemical Staining-pEGFP-AIRE-transfected HeLa cells seeded on coverslips were subjected to immunocytochemistry. Briefly, cells on the coverslips were fixed with 4% paraformaldehyde and then permeabilized with PBS containing 0.5% Triton X-100. After washing the cells with PBS, nonspecific binding was blocked with 1% bovine serum albumin in PBS. Cells were then incubated with appropriate Abs (anti-PML and anti-CBP/p300 polyclonal Abs were from Santa Cruz Biotechnology). Alexa 594-conjugated anti-mouse and anti-rabbit Abs (Invitrogen) were used for detection of the primary Ab. After staining with 2 g/ml 4Ј,6-diamidino-2-phenylindole (DAPI; Roche Applied Science) in PBS, coverslips were mounted with ProLong Antifade (Invitrogen), and fluorescence signals were observed with the aid of a fluorescence microscope or confocal microscope equipped with an argon laser.
Assessment of Ubiquitination of AIRE-Proteins were expressed by transfecting expression constructs with the Myc-tagged human AIRE cDNA and HA-tagged ubiquitin cDNA into COS-7 cells (17). Immunoprecipitation with anti-Myc mAb (clone 9E10) and Western blot analysis with anti-HA polyclonal Ab, both from Santa Cruz Biotechnology, were performed as described previously (16,17). For proteomics analysis, human AIRE was expressed in 293T cells with a modified tandem affinity purification system (18). Purified proteins were subjected to SDS-PAGE on a 9% gel, and the gel was stained with Coomassie Brilliant Blue. The bands were excised from the gel and then in-gel digested with trypsin (Promega, Madison, WI). The molecular masses of the tryptic peptides were determined using a matrix-assisted laser desorption ionization time-of-flight mass spectrometer (Bruker Daltonics, Bremen, Germany), and protein identification was performed with the MASCOT search engine (Matrix Science, London, UK).
In Vitro Ubiquitination Assay-The production of full-sized recombinant AIRE and the in vitro ubiquitination assay were performed as described (10,17). In brief, reaction mixtures (20 l) containing 1 g of recombinant AIRE, 0.1 g of recombinant rabbit E1 (Boston Biomedica, Cambridge, MA), 1 l of crude Escherichia coli lysate containing human Ubc4, 0.5 units of phosphocreatine kinase, 1 g of Ub (Sigma), 25 mM Tris-HCl (pH 7.5), 120 mM NaCl, 2 mM ATP, 1 mM MgCl 2 , 0.3 mM dithiothreitol, and 1 mM creatine phosphate were incubated for 2 h at 30°C. The reaction was terminated by the addition of SDS sample buffer containing 4% 2-mercaptoethanol and heating at 95°C for 5 min. Samples were resolved by SDS-PAGE on a 6% gel and then subjected to Western blot analysis with a mouse mAb to Ub (clone 1B3; MBL, Nagoya, Japan) and horseradish peroxidase-conjugated rabbit polyclonal Ab to mouse immunoglobulin (Promega). Signals were detected with ECL (Amersham Biosciences).

Establishment of Stable Transfectants Expressing AIRE-It
has been demonstrated that AIRE is predominantly expressed in the mTEC with the characteristic morphology of NB (11)(12)(13). However, the low levels of expression together with the lack of established cell lines constitutively expressing AIRE have hampered a detailed analysis of the molecular dynamics of AIRE. To approach this issue, we constructed a fusion gene in which the AIRE gene was fused with a green fluorescence protein (GFP) marker, and we obtained a stable transfectant of this fusion gene in HeLa cells of epithelial cell lineage (GFP-AIRE/HeLa). We first examined the expression of GFP-AIRE in HeLa cells with the aid of a fluorescence microscope. As reported for endogenous AIRE as well as for transiently expressed AIRE gene in cultured cells, GFP-AIRE was expressed primarily in the nucleus (Fig. 1A, left). In addition to the moderate nucleoplasmic expression, GFP-AIRE was found concentrated in discrete nuclear speckles of NBs. Unexpectedly, not all of the cells showed typical NB morphology, and approx-imately two-thirds of the cells displayed faint and homogeneous nuclear fluorescence without apparent NBs. The low frequency of occurrence of cells with typical NBs was not due to the mixture of GFP-AIRE-expressing cells with untransfected cells, because limiting dilution culture twice did not significantly change the population of cells with NBs (data not shown and see below). Most of the GFP-AIRE NBs, when observed, were not colocalized with endogenous PML bodies, as revealed by double immunocytochemistry (Fig. 1B).
It has been demonstrated that NB formation by PML, a prototypic NB protein that is thought to play a role in the control of apoptosis (19,20), is cell-cycle related; PML formed NBs at a significantly higher level in the G 1 phase (21). We therefore suspected that the infrequent expression of NBs in the GFP-AIRE/HeLa cells might be due in part to a similar mechanism. With the use of time-lapse analysis, we have observed that NBs largely disappeared before mitosis but are quickly re-formed after cell division (supplemental Movie 1), clearly indicating that NB formation by AIRE is cell cycle-dependent.
Association of AIRE with the Nuclear Matrix-Although the structural characteristics of AIRE protein suggest that it is a transcriptional regulating factor (13,(22)(23)(24), biochemical data supporting this idea are still limited. To obtain further insight into the function of AIRE, we investigated its subcellular localization. Nuclear matrix, the insoluble skeletal framework of the nucleus, is a dynamic structural subcomponent of the nucleus that directs functional domains of chromatin to provide sites for the specific control of nucleic acids (25). To test whether AIRE might exert a transcriptional regulating function in the nuclear matrix fraction, GFP-AIRE/HeLa cells grown on coverslips were sequentially extracted in situ with a detergent (CSK buffer), DNase I, ammonium sulfate, and high salt, leaving behind the nuclear matrix. The efficiency of chromatin digestion and removal was assessed by the disappearance of DNA staining with DAPI, and the presence of nuclear matrix after this treatment was confirmed by staining with a mAb against lamin A/C, a core nuclear matrix-associated protein (26) (Fig. 2A). The speckled pattern of GFP-AIRE NBs was still detectable after these extractions, suggesting that a significant proportion of AIRE protein was associated with the nuclear matrix. This finding was also supported by Western blot analysis, in which nuclear matrix protein was recovered with 8 M urea after high salt extraction and assayed for the presence of GFP-AIRE protein. We found significant amounts of GFP-AIRE protein in this fraction from GFP-AIRE/HeLa cells but not from wild-type HeLa cells (Fig. 2B). Successful fractionation of the cytoplasmic and nuclear matrix proteins was ascertained by Western blot analysis with Abs specific for IB␣ and lamin A/C, respectively. Although we did not observe apparent fluorescence signals from the cytoplasm of GFP-AIRE/HeLa cells, the cytoplasmic fraction contained a certain amount of GFP-AIRE protein. This discrepancy could be due to differences in the sensitivities of the detection systems used. Association of AIRE protein with the nuclear matrix was also observed in other cell types stably transfected with AIRE (i.e. FLAG-AIRE/mTEC cells and Myc-AIRE/NIH3T3 cells; see below) (supplemental Fig. S1).
Ubiquitin-dependent Subcellular Targeting of AIRE-It has been demonstrated that PML changes its subcellular localization in a proteasome-dependent manner; many PML NBs accumulated in nucleoli upon treatment with proteasome inhibitors such as MG132 (27). We used GFP-AIRE/HeLa cells to examine whether the subcellular localization of AIRE is controlled by a similar proteasome-dependent mechanism. Upon proteasome inhibition with ALLN or MG132 (data not shown), we observed enhanced fluorescence signals from the cells (Fig.  1A, right); not only was there an increase in the percentage of cells exhibiting NBs but there was also an increase in the size of each dot (supplemental Movie 2). Of note, NBs accumulated in proximity to the nucleoli, where DAPI staining is faint with this treatment. In contrast to GFP-AIRE/HeLa cells, HeLa cells stably transfected with GFP alone (GFP/HeLa) showed homogenous fluorescence signals from both cytoplasm and nucleus (data not shown). Upon proteasome inhibition, the distribution pattern of GFP in GFP/HeLa cells remained unchanged, although the fluorescence signals were moderately enhanced.
Because AIRE and PML protein are not colocalized in untreated GFP-AIRE/HeLa cells, as described above, but both change their localization within the nucleus upon inhibition of the proteasome pathway (Fig. 1A) (27), we attempted to determine whether AIRE and PML were colocalized after proteasome inhibition. Most of the cells showed no colocalization of AIRE and PML after proteasome inhibition, as observed for untreated cells (data not shown). Thus, AIRE and PML do not largely share their metabolic pathways, although subcellular targeting of these NBs might be controlled by a similar proteasome-dependent mechanism.
We have also investigated proteasome-dependent subcellular targeting of AIRE in the mouse mTEC line 1C6 (14) stably transfected with FLAG-tagged AIRE (FLAG-AIRE/mTEC). In contrast to GFP-AIRE/HeLa cells, FLAG-AIRE/mTEC cells displayed scattered fine dots of AIRE not only in the nucleus but also in the cytoplasm (Fig. 3A, a-c). Upon proteasome inhibition, AIRE in the cytoplasm was decreased, and AIRE predominantly accumulated in the nucleus (compare Fig. 3, A, d-f, with A, a-c) (supplemental Fig. S2A). Enhancement of FLAG-AIRE expression after MG132 treatment, as assessed by Western blot analysis, was much weaker compared with that in GFP-AIRE/HeLa cells. 2 We also established NIH3T3 cells sta- bly transfected with FLAG-tagged AIRE (FLAG-AIRE/ NIH3T3: Fig. 3B) as well as with Myc-tagged AIRE (Myc-AIRE/ NIH3T3: Fig. 3C). Upon treatment of FLAG-AIRE/NIH3T3 cells with MG132, AIRE in the cytoplasm was decreased and AIRE predominantly accumulated in the nucleus (compare Fig.  3, B, j-l with B, g-i) (supplemental Fig. S2B); both the size of dots within the nucleus and the numbers of NB were moderately increased. Although AIRE NBs were concentrated in some area within the nucleus upon proteasome inhibition, it was not in the nucleoli; AIRE and nucleolin were not colocalized in Myc-AIRE/NIH3T3 cells (Fig. 3C, d). Thus, proteasomedependent subcellular targeting of AIRE was not confined to some particular cell types and/or tags for detection (i.e. GFP, FLAG, and Myc). In contrast, the increased amount of AIRE protein after proteasome inhibition was somewhat dependent on cell type; GFP-AIRE/HeLa cells showed a more dramatic increase compared with FLAG-AIRE/mTEC cells or FLAG-AIRE/NIH3T3 cells.
AIRE Is Subject to Modification with Ubiquitin-To confirm that AIRE is under proteasome-dependent regulation for its subcellular targeting, we have tested whether AIRE can be ubiquitinated in vivo. To investigate this, Myc-tagged AIRE and HA-tagged Ub were coexpressed in COS-7 cells (which do not express endogenous AIRE as assessed by reverse transcriptase-PCR; data not shown), and the ubiquitination of AIRE was assessed by immunoprecipitation with anti-Myc mAb followed by detection with anti-HA Ab (Fig. 4A). Although plasmid expressing the Myc tag alone did not show any association with Ub, ubiquitinated AIRE was easily detected when Myc-tagged AIRE was used for the transfection, clearly indicating that AIRE is subject to modification with Ub in vivo. Furthermore, AIRE expressed in 293T cells exhibited various molecular masses identified by peptide mass fingerprinting, consistent with the presence of polyubiquitinated forms of AIRE (supplemental Fig. S3).
We have demonstrated recently that AIRE acts as an E3 Ub ligase through the N-terminal PHD domain (PHD1) (10). Because many E3 ligases exert ubiquitination of their own protein (self-ubiquitination) as well as of specific target protein(s), we tested whether the ubiquitination of AIRE is due to self-ubiquitination by AIRE itself. To this end, an in vitro ubiquitination assay was performed by mixing recombinant full-length AIRE with Ub, recombinant E1, and Ubc4 (E2) in the presence of ATP. We then prepared immunoprecipitates for AIRE from this total reaction mixture and performed Western blot analysis with anti-Ub Ab. Although the total reaction mixture contained polyubiquitinated proteins, no polyubiquitinated proteins were observed from the immunoprecipitates for AIRE (Fig. 4B); total reaction mixture and immunoprecipitates for AIRE contained similar amounts of AIRE as revealed by the Western blot analysis with anti-His Ab. This result suggests that AIRE does not ubiquitinate itself. Rather, we speculate that the E3 ligase activity of AIRE is mostly directed toward ubiquitination of specific target protein(s), and that there should be another E3 ligase responsible for the ubiquitination of AIRE. This hypothesis is supported by the fact that NIH3T3 cells expressing a PHD1 mutant (C299A) that lacks E3 ligase activity (10) showed changes in the subcellular localization of AIRE upon proteasome inhibition similar to that seen in cells expressing wild-type AIRE (Fig. 3B, p-r). The effect of proteasome inhibition was also independent of the PHD2 domain, because a PHD2 mutant (C434A) also showed similar subcellular targeting to that of wild-type AIRE-transfected cells (Fig. 3B, v-x) (supplemental Fig. S2B). Similar results were obtained from FLAG-AIRE/mTEC cells (supplemental Fig. S2A).

Regulation of AIRE Activities Determined by Colocalization with Other Transcriptional Regulators of NB Morphology-By
using coprecipitation experiments, we have demonstrated previously (22) that AIRE is associated with CBP/p300, a common coactivator of transcription. On the other hand, CBP/p300 has been demonstrated to colocalize with PML (28). Because AIRE and PML are not colocalized, as demonstrated in this study, but both NBs are colocalized with CBP/p300, we suspected that AIRE activities might be regulated and/or modulated in part by competition with PML for CBP/p300. To test this hypothesis, we used RD cells, in which both AIRE and endogenous PML are undetectable. Immunofluorescence of untreated RD cells showed no discrete nuclear dot formation by CBP/p300 (Fig.  5A, a). Instead, CBP/p300 was found in a diffuse and/or speckled pattern. When AIRE was introduced into RD cells, endogenous CBP/p300 was recruited to the AIRE NBs, confirming the in vivo association between AIRE and CBP/p300 (Fig. 5B,  a-c). In turn, when PML was expressed in RD cells, similar translocation of endogenous CBP/p300 to the PML bodies was observed (Fig. 5B, d-f), as demonstrated previously (28). When AIRE and PML were introduced simultaneously, AIRE and FIG. 2. Association of AIRE with the nuclear matrix. A, GFP-AIRE-transfected HeLa cells were extracted sequentially in situ with detergent, DNase I plus ammonium sulfate, and high salt, leaving behind the nuclear matrix. NBs were still present in the nuclear matrix fraction prepared using this treatment. The efficiency of chromatin digestion and removal was assessed by the disappearance of DNA staining with DAPI, and the presence of nuclear matrix after the treatment was confirmed by staining with anti-lamin A/C mAb. Original magnification, ϫ40. B, cytoplasmic, chromatin, and nuclear matrix proteins were extracted from wild-type HeLa cells (W) or from GFP-AIRE-transfected HeLa cells (A), and GFP-AIRE protein was detected by anti-GFP Ab with Western blot (IB) analysis. A significant amount of GFP-AIRE protein was detected in the nuclear matrix fraction obtained from GFP-AIREtransfected cells (top). Successful fractionation of the cytoplasmic and nuclear matrix proteins was ascertained by anti-IB␣ Ab (middle) and anti-lamin A/C mAb (bottom), respectively.
PML were mostly observed in distinct NB structures (Fig. 5B,  g-i), as expected. Of note, in cells doubly transfected with AIRE and PML, CBP/p300 predominantly colocalized with PML bodies (Fig. 5B, j-l) rather than with AIRE NBs (Fig. 5B, m-o). These results suggest that the activities of AIRE might be indirectly regulated by PML through competition for CBP/p300. Preferential colocalization of CBP/p300 with PML bodies rather than with AIRE NBs was also observed in NB4 cells. Endogenous PML-retinoic acid receptor (due to the chromosomal translocation) displayed a speckled pattern instead of typical PML bodies in NB4 cells (29); colocalization of CBP/ p300 with PML was still observed in this speckled pattern, as demonstrated in Fig. 5A, b. When both AIRE and wild-type PML were transfected, AIRE and PML were mostly observed in distinct NB structures (Fig. 5C, a), as observed in RD cells. Again, endogenous CBP/p300 showed colocalization with PML (Fig. 5C, b) but not with AIRE (Fig. 5C, c). Thus, although AIRE associates with CBP/p300, CBP/p300 preferentially colocalizes with PML rather than with AIRE when both NB proteins are available, at least in the NB structures. We do not see this, however, as excluding CBP/p300 as an AIRE cofactor. Rather, we speculate that competition for CBP/p300 by AIRE and PML might be a novel and unique mechanism for the control of AIRE function. DISCUSSION We have studied the molecular dynamics of AIRE, a putative transcription-regulating gene responsible for the hereditary type of the human autoimmune disorder APECED. Although AIRE-deficient mice exhibit an autoimmune phenotype similar to that seen in the human disease (30,31), and have provided us with some important insights into how the immune system discriminates between self-and non-self with the help of AIRE (32), the molecular mechanism for these actions by AIRE still remains unknown. Because the lack of cell lines constitutively expressing AIRE in sufficient amounts has been the main reason for these difficulties, we first established cell lines stably transfected with GFP-AIRE that enable us to monitor the molecular dynamics of AIRE within a single cell. As expected, we have observed GFP-AIRE expression primarily in the nucleus, with the characteristic morphology of NB. Unexpectedly, however, not all of the cells showed typical NB morphology, and many cells displayed faint and homogeneous nuclear fluorescence without apparent NBs. With the use of time-lapse analysis, we have found that NB formation is regulated by the cell cycle. These results suggest that not all of AIRE is NB-bound. Instead, we speculate that AIRE forms NB structures under certain cellular conditions.
It has been shown that many transcription factors bind to the nuclear matrix (25). Our studies have demonstrated that a significant amount of AIRE protein, including that in its NB form, exists in the nuclear matrix fraction. This is an important biochemical finding that supports a role for AIRE as a transcriptional regulating factor. However, it remains to be determined whether AIRE NBs themselves are the major FIG. 4. Polyubiquitination of AIRE is not due to self-ubiquitination by its E3 ligase activity. A, Myc-tagged AIRE or Myc tag alone was coexpressed with HA-tagged ubiquitin in COS-7 cells, and the ubiquitination of AIRE was assessed by immunoprecipitation (IP) with anti-Myc mAb followed by detection with anti-HA mAb. The position of Myc-tagged AIRE is indicated by an arrow. B, immunoprecipitates for AIRE were prepared from the total reaction mixture of in vitro ubiquitination, which contains full-sized recombinant His-tagged AIRE as E3 ligase, and then subjected to Western blot (IB) analysis with anti-Ub Ab. AIRE was not polyubiquitinated, although other substrate(s) are polyubiquitinated in the total reaction mixture (top). Immunoprecipitates for AIRE and total reaction mixture contained similar amounts of AIRE as revealed by Western blot analysis with anti-His Ab (bottom).
FIG. 3-continued site of transcriptional regulation by AIRE. Furthermore, it is not yet clear which form of AIRE, its nucleoplasmic form or NB form, is mainly responsible for the physiological function of AIRE. It is equally possible that AIRE in NBs is functionally distinct from that existing homogenously in the nucleoplasm. In this respect, it is interesting to note that PML bodies have not been reliably shown to be the sites of transcription nor do they contain chromatin or RNA, although nascent RNA is present in the immediate neighborhood of PML bodies, and they might be linked to regulation of transcription (20,33). Thus, the functional significance of NB formation by AIRE requires further study.
Given that AIRE is a transcriptional regulator, targeting of AIRE into the nucleus and/or to a particular site within the nucleus should be a relevant process that determines the effect of AIRE. In the present study, we have suggested that ubiquitination is an important mechanism that determines the subcellular targeting of AIRE. We have observed altered localization of NBs within the nucleus from GFP-AIRE/HeLa cells after treatment with proteasome inhibitor. In both FLAG-AIRE/mTEC and Myc-AIRE/NIH3T3 cells, AIRE NBs were increased with concomitant reduced expression in the cytoplasm upon proteasome inhibition (Fig. 3). These results suggest that ubiquitination of AIRE is a critical modification process that dictates subcellular targeting. It is possible to speculate that ubiquitination, facilitating nuclear targeting of AIRE, is a positive regulatory process for AIRE function.
We have suggested previously that AIRE shuttles between the nucleus and the cytoplasm (13) and that this process appears to be under the control of CRM-1, a protein responsible for nuclear export (34). Treatment of the cells transiently AIRE-transfected with leptomycin B (LMB), a CRM-1 inhibitor, resulted in preferential localization of AIRE dots within the nucleus (34). Unexpectedly, however, treatment of GFP-AIRE/ HeLa cells with LMB reduced the numbers of NBs to some degree. 2 We suspect that the difference in experimental conditions (i.e. transient versus stable AIRE expression) led to the formation of a different AIRE protein complex. In the stable expression system, AIRE may form a protein complex with factor(s) whose subcellular localization is under the control of CRM-1 and is responsible for the degradation of the AIRE protein; accumulation of such factor(s) within the nucleus as a result of LMB treatment would result in the loss of AIRE NBs.
The effect of proteasome inhibition on the recruitment of AIRE onto NB structures may support the idea that most of AIRE is not NB-bound; a fraction of AIRE is located on the discrete subnuclear structures, the AIRE NBs, and the rest are in the nucleoplasm with homogeneous distribution. In this respect, it is interesting to note that FANCD2, a gene responsible for Fanconi anemia, is mono-ubiquitinated in response to DNA damage and is targeted to nuclear foci (dots), providing the link between the ubiquitination pathway and the targeting of gene products to a defined nuclear organelle (35). It needs to be determined what kinds of stimuli cause the ubiquitination of AIRE under physiological conditions.
Because both a PHD1 mutant (lacking E3 ligase activity) and a PHD2 mutant (devoid of intrinsic transactivation properties (10)) showed distributions similar to that of wild-type AIRE in untreated FLAG-AIRE/NIH3T3 cells, intracellular targeting is independent of the function of PHD domains. This is consistent with a recent report (36) demonstrating that mutated AIRE lacking PHD domains retained at least nuclear entry. Upon proteasome inhibition, both FLAG-AIRE/NIH3T3 cells and FLAG-AIRE/mTEC cells transfected with the PHD1 mutant showed similar altered subcellular targeting to that of wildtype AIRE-transfected cells (Fig. 3B and supplemental Fig. S2), suggesting that auto-ubiquitination by AIRE itself is not responsible for this action. Supporting this notion, AIRE itself was not polyubiquitinated in the in vitro ubiquitination assay FIG. 5. Preferential colocalization of CBP/p300 with PML bodies rather than with AIRE NBs. In RD cells, CBP/p300 was found in a diffuse and/or speckled pattern (A, a). When either AIRE or PML was introduced into RD cells, endogenous CBP/p300 was recruited to the AIRE NBs (B, a-c) or to the PML bodies (B, d-f), respectively. When AIRE and PML were introduced simultaneously, AIRE and PML were mostly observed in distinct NB structures (B, g-i). In those cells, CBP/ p300 colocalized preferentially with PML bodies (B, j-l) rather than with AIRE NBs (B, m-o). Colocalization of CBP/p300 with PML was observed in a speckled pattern in NB4 cells (A, b). When both AIRE and wild-type PML were transfected, AIRE and PML were mostly observed in distinct NB structures (C, a). In those cells, endogenous CBP/p300 showed colocalization with PML (C, b) but not with AIRE (C, c). Green colors represent the fluorescein isothiocyanate signals, and Texas Red staining is in red. Original magnification, ϫ60. when full-sized recombinant AIRE protein was used as an E3 ligase. Therefore, there must be other E3 ligase(s) responsible for AIRE ubiquitination. Identification of such E3 ligase(s) is required for the elucidation of detailed metabolic and regulatory pathways for the control of AIRE function.
Finally, our results suggest that autoimmunity is controlled in part by multiple regulatory steps affecting the function of transcriptional regulator(s), as exemplified by AIRE. Knowing this, one of the most important questions to be answered is what kind of gene(s) is the target of AIRE. Such putative target gene(s) should maintain self-tolerance by eliminating autoreactive T cells in the thymus. Supporting this hypothesis, AIRE expression in the thymus was enhanced when negative selection, but not positive selection, was induced (37). Future work will clarify how self-tolerance is physiologically established and maintained in the thymus by studying the link between abnormal function of AIRE at the molecular level and the pathogenetic process of the disease.