Differential expression of tapasin and immunoproteasome subunits in adenovirus type 5- versus type 12-transformed cells.

Adenovirus type 12 (Ad12)-transformed baby rat kidney (BRK) cells are oncogenic in syngeneic immunocompetent rats in contrast to adenovirus type 5 (Ad5)-transformed BRK cells, which are not oncogenic in these animals. A significant factor contributing to the difference in oncogenicity may be the low levels of major histocompatibility complex (MHC) class I membrane expression in Ad12-transformed BRK cells as compared with those in Ad5-transformed BRK cells, which presumably results in escape from killing by cytotoxic T lymphocytes. Here we show that, in addition to the decreased levels of expression of the MHC class I heavy chain and the peptide transporter Tap-2, the expression levels of the chaperone Tapasin and the immunoproteasome components MECL-1, PA28-alpha, and PA28-beta also are much lower in Ad12- than in Ad5-transformed BRK cells. The low expression levels of these proteins may contribute to the escape from killing by cytotoxic T lymphocytes, because the generation of optimal peptides and loading of these peptides on MHC class I require these components. Increased levels of phosphorylated signal transducer and activator of transcription-1 protein and expression of IFN regulatory factor-7 were found in Ad5- versus Ad12-transformed BRK cells. Therefore, the critical alteration leading to the plethora of differences may be an interferon (-related) effect.

Oncogenic transformation of cells involves multiple events, including activation of oncogenes, inactivation of tumor suppressor genes, and extension of lifespan or immortalization. The lifespan of primary cells is normally restricted to a defined number of cell divisions, but after transformation the cells usually become immortal. Another frequently observed parameter of transformation is the ability to grow in an unrestricted fashion in a living organism, either in the absence of the T-cell immune defense (nude mice) or in the presence of immune defense (immunocompetent animals). Because transformed cells usually express neoantigens, oncogenicity in immunocompetent organisms requires escape from the T-cell immune surveillance by cytotoxic T lymphocytes.
A suitable model to study differences in oncogenicity is the adenovirus transformation system (1). The transforming activ-ity of adenoviruses (Ad) 1 is a function of the early region 1 (E1) of the viral genome, which encodes the E1A and E1B proteins (2). Primary baby rat kidney (BRK) cell cultures can be transformed by most human Ad serotypes, but only cells transformed by the oncogenic subgroup A adenoviruses are oncogenic in syngeneic animals. Studies on the differences between cells transformed by the non-oncogenic Ad5 and oncogenic Ad12 have led to the identification of a number of differentially expressed cellular genes that could explain the differences in oncogenicity of these cells. Among these are the MHC class I genes, which are down-regulated in Ad12-transformed cells (3). Absence of MHC class I antigens might be particularly important with respect to the escape from the T-cell immune surveillance (4). As a result of the lack of MHC class I membrane expression, viral peptides cannot be presented to the immune system, which causes these cells to escape from elimination by cytotoxic T cells.
Down-regulation of membrane expression of MHC class I protein is caused in part by decreased levels of transcription of the MHC class I heavy chain genes (5-7) but is also because of the low expression level of the Tap protein (8,9). Low Tap protein levels severely limit the amount of peptides available for presentation by MHC class I molecules (10). Because only trimeric complexes consisting of MHC class I heavy chain, light chain (␤2-microglobulin), and peptide are stable, limited availability of peptides severely affects the amount of membraneexpressed MHC class I antigens (10 -14). The availability of optimal peptides might be limited further by the low expression levels of two ␥-interferon-inducible components of the proteasome, LMP-2 and LMP-7, in Ad12-transformed cells (15,9).
In an attempt to restore MHC class I membrane expression in Ad12-transformed cells, we have generated in the present study Ad12-transformed BRK cells that are stably transfected with a plasmid encoding RT1-A u , the MHC class I heavy chain of the rat strain used for these studies, and a plasmid encoding Tap-2. Despite the fact that the genes were expressed as RNA, these cells did not show an increase in MHC class I membrane expression. Additional characterization of Ad5-and Ad12transformed BRK cells revealed that the chaperone protein Tapasin, as well as MECL-1 and PA28-␣ and -␤, three proteins involved in the generation of peptides that can be presented on MHC class I (16), are expressed at higher levels in Ad5-than in Ad12-transformed BRK cells. We present evidence that differ-ential expression of this set of genes can be explained by differences in the activity of STAT-1, a transcription factor involved in the control of expression of these genes (17)(18)(19)(20). STAT-1 has been identified as a downstream component of the interferon signal transduction pathway (21,22), and its functional activation was reflected in the expression of IRF-7 in Ad5-transformed BRK cells but not in Ad12-transformed BRK cells (23).

EXPERIMENTAL PROCEDURES
Plasmids and Antibodies-A partial RT1-A u cDNA, RT16, has been described (24). Cloning of the 5Ј part was based on the homology of RT1-A. RT16 lacks only bp 1040 -1042 of the RT1.A l , whereas 17 of 18 bp around the RT1.A l translational start codon are identical with the corresponding region of RT1.A a . Therefore, an upstream oligo, U1, corresponding to the RT1.A l sequence, and a downstream oligo, D1, derived from RT16 (24), were used for RT-PCR of total RNA from the Ad5-transformed BRK cell line BXc22, which expresses high levels of RT1-A u (see Fig. 1B) (25). The PCR product of 473 bp was subcloned in the AT vector pCR2.1 (Invitrogen), resulting in PCR2.1-RT1-A u 5Ј. The sequence of two independently isolated clones was identical, and this sequence was submitted to GenBank TM (accession number AF400159). The 489-bp EcoRI fragment of plasmid pCR2.1-RT1-A u 5Ј was inserted in the EcoRI site of pBluescript (Stratagene, La Jolla, CA), resulting in pBluescript-RT1-A u 5Ј. The full-length RT1-A u cDNA was constructed in a two-step procedure, inserting the 504-bp StyI/PstI fragment of RT16 in the StyI/PstI sites of pBluescript-RT1-A u 5Ј and subsequently the 783-bp PstI fragment of RT16. The pcDNA3.1-RT1-A u expression vector was made by inserting the HindIII/XbaI fragment of pBluescript-RT1-A u , which encodes full-length RT1-A u , in the HindIII and XbaI site of pcDNA3.1 (Invitrogen). The sequence of pBluescript-RT1-A u was reconfirmed.
The rat Tap-2 expression plasmid was constructed by cloning the BamHI-XhoI fragment from pBluescript-rTap-2 (a kind gift from Dr. F. Momburg, German Cancer Research Center, Heidelberg, Germany) into the BamHI and XhoI sites of plasmid pcDNA3.1 (Invitrogen). The mouse Tapasin expression construct pREP8-mTapasin (26) was a kind gift from Dr. A. Grandea III (Vanderbilt University, Nashville, TN).
Cell Culture, DNA Transfection, and Generation of Stably Transfected Cell Lines-BRK cells transformed with the early region 1 of Ad5 or Ad12 were described elsewhere (25). Monoclonal cell lines 33RT60 and 33RT61 were established by transfection of Ad12E1-transformed BRK cell line RICc33 with plasmid pcDNA3.1-RT1-A u and subsequent selection for G418 resistance. Polyclonal cell lines expressing Tapasin, Tap-2, or both proteins and control lines were established by transfection of 33RT60 with the empty expression vector pcDNA3.1, the plasmid pREP8-mTapasin, or the plasmid pcDNA3.1-Tap-2 or co-transfected with both expression plasmids. The pECV5 plasmid, which encodes the Hygromycin resistance gene, was included in all these transfections, and subsequently cells were selected for hygromycin resistance. AdE1-transformed cells were cultured in minimum Eagle's medium, supplemented with 10% newborn calf serum and antibiotics. U2OS cells were cultured in Dulbecco's modified Eagle's medium, supplemented with 10% fetal-calf serum and antibiotics. Tissue culture media and sera were purchased from Invitrogen. All tissue culture plastics were obtained from Greiner. Treatment with 100 units/ml IFN-␥ (Invitrogen) or 1000 units/ml TNF-␣ (Sigma) was performed for the indicated periods of time. Transfections were performed using the TFX-50 transfection protocol according to the instructions of the manufacturer (Promega) or using the calcium-phosphate protocol as described previously (32).
Western Blot Analysis-Total cell extracts were fractionated by SDS/ PAGE on 10% or 12.6% gels. Proteins were transferred onto Immobilon-P membranes (Millipore) and incubated with specific primary antibodies as indicated. As secondary antibodies horseradish-peroxidase-coupled goat anti-rabbit and rabbit anti-mouse IgG (Jackson) were used. The bound antibodies were visualized with the ECL detection system according to the manufacturer's protocol (Amersham Biosciences).
Metabolic Labeling and Immunoprecipitations-Exponentially growing cells were labeled for 1-6 h with [ 35 S]methionine and lysed in IPB.14 as described previously (33). MHC class I proteins were immunoprecipitated with the mouse monoclonal antibody U9F4 (27), and Tapasin was immunoprecipitated with a rabbit polyclonal antibody raised against mouse Tapasin (29). Immunoprecipitated proteins were fractionated by SDS-PAGE on 10% gels (Tapasin) or 12.6% gels (MHC class I). Proteins were visualized by autoradiography.
Northern Blot Analysis and RT-PCR-Total RNA was isolated from exponentially growing cells as described previously (34). poly(A) ϩ enrichment was performed using the mRNA isolation kit according to the instructions of the manufacturer (Roche Molecular Biochemicals). 10 g of total RNA or 5 g of poly(A) ϩ enriched RNA was size fractionated on a 1% agarose/2.2 M formaldehyde gel by electrophoresis and transferred to Hybond filters (Amersham Biosciences). Filters were hybridized for RT1-A u with the 0.9-kb PvuII/EcoRI fragment of the pcDNA3.1-RT1-A u expression vector, with the 0.8-kb KpnI/SacI fragment of pBluescript-rTap-2 for Tap-2, with the 1.3-kb HindIII/EcoRI fragment of pREP8-mTapasin for Tapasin, and with the PstI fragment of the rat GAPDH cDNA for GAPDH. For RT-PCR RNA isolation and cDNA preparation were as described previously by Martens et al. (35). The primers used were as follows: as control rat elongation factor 1, AAGCTGAGCGTG-AGCGTG and CGGGTGACTTTCCATCCC (464 bp); rat IRF-7, GCAG-CAGTGGTTCTGAAC and GGCGACAAGGATCACCAC (293 bp); rat IFN-␤, GTGACGGGTGCATCACCTCC and CCACTGCCCTCTCCATC-GAC (153 bp); IFN-␣ (consensus primers annealing with all mouse IFN␣ subtypes), AGGGCTCTCCAGAYTTCTGCTCTG and ATGGCTA-GRCTCTGTGCTTTCCT (524 bp) (36).
Cell Surface Protein Expression-Exponentially growing cells were harvested by mild trypsinization. Approximately 2 ϫ 10 5 cells were incubated for 30 min on ice with 30 l of U9F4 hybridoma tissue culture supernatant 1:1 diluted with PBA (phosphate-buffered saline containing 0.5% bovine serum albumin). The cells were washed twice with PBA and incubated with fluorescein isothiocyanate-conjugated goat-antimouse antibody (Jackson) for 30 min on ice. Controls were stained with fluorescein isothiocyanate-conjugated goat-anti-mouse antibody only. Subsequently, the cells were washed twice with PBA, and the fluorescence intensity was analyzed by the FACSCalibur (BD Biosciences).

Overexpression of RT1-A u Heavy Chain mRNA in Ad12transformed BRK Cells Is Not Sufficient to Restore Expression-
The rat MHC class I protein RT1-A u is strongly downregulated in Ad12-transformed BRK cells, which is assumed to contribute to the oncogenicity of these cells in syngeneic Wag-Rij rats (3). In an attempt to functionally restore RT1 membrane expression, we have cloned the cDNA encoding the major RT1 allele expressed in this rat strain, RT1-A u . The RT1-A u expression construct was sequence-verified, and the protein could be detected after transient transfection in U2OS cells (Fig. 1A). Contrary to the result in U2OS cells, stable overexpression of RT1-A u mRNA in Ad12-transformed BRK cells (Fig.  1B) does not lead to high protein levels (Fig. 1C). In 23 independent monoclonal cell lines, no increase in RT1-A u protein levels was observed. This result suggests that in these Ad12transformed cells additional alterations impair the formation of stable MHC class I molecules presenting peptides at the cell surface.
Tap-2 and Tapasin Are Expressed Differentially in Ad5-and Ad12-transformed Cells-Stable MHC class I protein expression requires the availability of small peptides that can be presented by the heavy chain molecule (12,13,37). Peptide loading is a function of the Tap proteins, which are known to be hardly expressed in Ad12-transformed mouse cells (8). Restoring Tap-2, but not Tap-1, partially restored MHC class I membrane expression in the published mouse model. Fig. 2A shows that the Tap-2 protein levels in Ad5-transformed BRK cells are also much higher than in Ad12-transformed BRK cells. As a control, we identified the levels of NF-B1-p105 ( Fig. 2A), which, as we showed previously, are identical in Ad5-and Ad12-transformed BRK cells (25). The difference in Tap-2 level is in line with the difference in Tap-2 mRNA expression (Fig. 2C).
Another protein involved in MHC class I membrane expression (38) is Tapasin, a chaperone protein in the endoplasmic reticulum (39). Therefore, this protein is a candidate that might be involved in differential MHC class I membrane expression in Ad-transformed cells. Hence, we determined the Tapasin protein levels in Ad5-and Ad12-transformed BRK cells. As shown in Fig. 2B, Tapasin is expressed to a higher level in Ad5-than in Ad12-transformed BRK cells. This is probably because of the difference in Tapasin mRNA expression (Fig. 2C).
To test whether Tap-2 and Tapasin might be sufficient to restore MHC class I membrane expression, polyclonal cell lines that stably overexpress Tap-2, Tapasin, or both proteins were established using 33RT60, the Ad12-transformed BRK cell line stably overexpressing the RT1-A u mRNA, as the parental cell line. Tapasin protein levels in these stable cell lines (Fig. 3A, lanes 4 and 5) were somewhat higher than the endogenous Tapasin protein levels in Ad5-transformed cells (lane 1). Tap-2 protein levels (lanes 3 and 5) were not as high as in Ad5transformed cells but were strongly increased compared with the level in the parental cell line transfected with the empty expression construct (lane 2). However, no increase in MHC class I membrane expression was observed in any of these polyclonal cell lines, compared with that in the parental control line (Fig. 3B). Similar results were found in a second set of polyclonal cell lines (data not shown). Therefore, RT1-A u , Tap-2, and Tapasin are not sufficient to restore MHC class I membrane expression in Ad12-transformed cells. The capacity of these cells to express RT1-A u was demonstrated by the stimulation of the expression by IFN-␥ (Fig. 3B, B4).
Induction of MHC Class I in Ad12-transformed Cells-It is well established that many genes encoding antigen presentation factors are targets for IFN-␥. As expected, IFN-␥ treatment could restore MHC class I membrane expression in Ad12transformed cells (Fig. 4, A1), to a level comparable with the MHC class I expression in Ad5-transformed cells (Fig. 4, A2). A possible explanation for the difference between Ad5-and Ad12transformed cells could be that Ad5-transformed cells produce interferon. Therefore we tested whether the conditioned medium from Ad5-transformed BRK cells could induce MHC class I membrane expression in Ad12-transformed cells. As shown in Fig. 4B induction was always observed although the magnitude of the effect depends on the conditioned medium used. Some conditioned medium led to an induction comparable with the effect of IFN-␥. Analysis of media from Ad5-transformed mouse cells showed that these cells do not produce mouse IFN-␥ (data not shown). Previously, it has been shown that Ad5-transformed cloned rat embryo fibroblasts, in contrast to Ad12transformed fibroblasts, express IFN-␤ mRNA (40,41). Therefore, we tested our panel of Ad-transformed BRK clones for expression of IFN-␤ by quantitative RT-PCR (Fig. 4C). Expression of IFN-␤ could be demonstrated in two of the three Ad5- transformed BRK cells but not in the Ad12-transformed BRK cells. Analysis of the expression of IFN-␣ is hampered as for rat IFN-␣ only a single cDNA has been published (42) whereas IFN-␣ consists of a large family of closely related members in mice (23). A comparable intensity was found for all cell lines tested with the strongest signal with a degenerate primer set for mice IFN-␣ (36) (Fig. 4C). These results suggest that IFN-␤ could be responsible for the conditioned medium effect. To determine whether the amounts of IFN-␤ present in the conditioned medium of Ad5-transformed BRK cells are indeed responsible for the induction of MHC class I in Ad12-transformed BRK cells, an IFN-␤ neutralizing antibody (31) was added to the conditioned medium. The neutralizing antibody reduced the conditioned medium effect to 57% (data not shown). Interestingly, this was found for the conditioned medium from the Ad5-transformed BRK clone 22. This indicates that although we fail to detect IFN-␤ expression by quantitative RT-PCR, the amount of IFN-␤ produced by this cell line is still of significant biological relevance. The inability to observe an effective block by anti IFN-␤ of MHC class I membrane expression in Ad5 cells (data not shown) might imply that Ad5-transformed cells produce other MHC class I inducing cytokines.
Differential Expression of IFN-␥ Target Genes-The ability of interferons to induce MHC class I antigen suggests that IFNtarget genes might be expressed differentially in Ad5-and Ad12-transformed BRK cells. Known IFN-␥ target genes involved in antigen presentation are the immunoproteasome subunit MECL-1 (43) and the proteasome activator subunits PA28-␣ and -␤ (44 -48). These proteins were found to be expressed to a much higher level in Ad5-than in Ad12-transformed BRK cells (Fig. 5). Previous studies already demonstrated that LMP-2 and -7, two IFN-␥-inducible proteasome subunits, are also expressed at much higher levels in Ad5-than in Ad12-transformed cells (15). Differential regulation of LMP-2 and -7, MECL-1, and PA28-␣ and -␤ could imply that Ad5-transformed BRK cells have higher proteasome levels than Ad12-transformed BRK cells. However, this is not the case, because the constitutive proteasome subunit MC-3 is

FIG. 4. IFN-␥ and conditioned media of Ad5-transformed BRK cells can induce MHC class I cell surface expression of Ad12transformed cells. A, FACS analysis of cell surface expression of RT1-A u of (A1) Ad12-transformed BRK cells (clone RICc35) and its induction after growth for 3 days with 100 units/ml IFN-␥ and for comparison (A2) analysis of the expression of Ad5-transformed cells (clone BXc23
). Control staining with secondary antibody only is shown as filled black graphs. B, cell surface expression of RT1-A u of Ad12transformed BRK cells (clone RICc35) (B1, B2) and its induction after growth for 3 days on 50% conditioned medium from Ad5-transformed BRK cells clone BXc22 (B1) or clone BXc23 (B2). Filled black graphs are control staining with secondary antibody only. C, RT-PCR for the mRNA expression of elongation factor 1 (EF1), IFN-␤, and IFN-␣ in Ad5-and Ad12-transformed BRK cells. expressed somewhat higher in Ad12-than in Ad5-transformed BRK cells (Fig. 5).
Previously we and others have shown that Ad5-and Ad12transformed BRK cells differ in their basal NF-B activity, which could partially explain the differences in MHC class I heavy chain expression between these cells (25,40,49). In addition, we have shown that TNF-␣ is a strong NF-B inducer in Ad5-and Ad12-transformed BRK cells (50), confirmed in Ad12-transformed cells by the induction of the degradation of IB␣ (Fig. 6B). Interestingly, several of the differentially expressed genes have established (51,52) or putative NF-B binding sites. Therefore, we studied the effect of TNF-␣ on the expression of antigen presentation machinery components but did not find any alteration (Fig. 6A) (data not shown).
The differences in the expression levels of antigen presentation machinery components between Ad5-and Ad12-transformed BRK cells and the expression of IFN-␤ by Ad5-transformed BRK cells indicate a difference in activity of the IFN signal transduction pathway between these cells. The transcription factor STAT-1 mediates the induction of MHC class I in response to interferons (21). We have examined the total protein levels of this transcription factor and found that Ad5transformed cells contain more STAT-1 than Ad12-transformed cells (Fig. 7A). Activation of STAT-1 occurs via phosphorylation of tyrosine 701 (53). Consistent with the hypothesis of active interferon signaling in Ad5-transformed cells, STAT-1 is phosphorylated on Tyr-701 in Ad5-transformed cells but not in Ad12-transformed cells. STAT-1 is part of the heterotrimeric transcription factor ISGF3, which regulates IRF-7 expression as part of the multistage IFN response (23). Using quantitative RT-PCR, we found that IRF-7 was expressed in Ad5-transformed BRK cells but not in Ad12-transformed BRK cells (Fig.  7B). Ad12-transformed cells are still able to activate the interferon signal transduction pathway, because IFN-␥ treatment caused phosphorylation of STAT-1 (Fig. 7, lanes 7 and 8) In conclusion, the increased expression of genes involved in antigen presentation in Ad5-transformed cells can be induced in Ad12-transformed cells by IFN-␥ and also by the conditioned media of Ad5-transformed cells, as shown for Tap-2 and PA28␣ in Fig. 6C. The transcription factor that mediates the interferon response, STAT-1, is active in Ad5-transformed cells but  1-6) and Ad12-transformed BRK cells (lanes 7-12) were treated for the indicated periods of time with IFN-␥, TNF-␣, or with both cytokines. Western blots of total cell lysates were analyzed for Tap-2, MECL-1, and PA28-␣ and as a loading control for NF-B1-p105. B, Western blots for TNF-␣-mediated (30 min) IB␣ processing in Ad12-transformed BRK cells (clone RICc33). NF-B1-p105 was determined as a control. C, up-regulation of PA28-␣ and Tap-2 protein levels in Ad12-transformed BRK cells by conditioned medium from Ad5-transformed BRK cells. Exponentially growing RICc35 cells (lanes 1-3) and RICc33 cells (lanes 4 -6) were grown for 3 days on fresh cell culture medium (lanes 1 and 4) or on fresh cell culture medium diluted 1:1 with cell culture medium conditioned by BXc22 cells (lanes 2 and 5) or BXc23 cells (lanes 3 and 6). Western analysis blots of total lysates were stained for Tap-2, PA28-␣, and NF-B1-p105. As a control, Ad5-transformed BRK cell lines BXc22 (lane 7) and BXc23 (lane 8) were included. not in Ad12-transformed cells, indicating that the difference in antigen presentation between Ad5-and Ad12-transformed cells is partly determined by the production of an IFN (-like) activity in Ad5-transformed cells, most likely IFN-␤. DISCUSSION Differential membrane expression of MHC class I antigens in Ad5-and Ad12-transformed cells most likely contributes to the observed differences in oncogenicity between these cells in immunocompetent animals. Here we show that the Tapasin protein, a chaperone that is part of the MHC class I peptide loading complex, and several components of the peptide generating machinery, MECL-1 and PA28-␣ and -␤, are also expressed at much lower levels in Ad12-than in Ad5-transformed BRK cells. Furthermore, we confirm a very striking difference in expression of the peptide transporter Tap-2 at the protein level between these cells.
The down-regulation of several gene products involved in the generation and presentation of peptides in the context of MHC class I antigens implies that several genes may have to be re-expressed in Ad12-transformed cells to attain higher levels of surface MHC class I antigens. Indeed, we found that overexpression of the RT1-A u class I MHC heavy chain cDNA in Ad12-transformed BRK cells did not lead to increased protein expression. This is possibly because of the fact that the U9F4 antibody recognizes only properly folded trimeric MHC class I complexes consisting of one heavy chain, one light chain (␤2microglobulin), and a peptide (12) and not free heavy chain protein (13,37). Formation of the trimeric complex is impaired because of limited availability of suitable peptides in the endoplasmic reticulum as a result of the low levels of Tap-1 and -2, Tapasin, LMP-2 and -7, MECL-1, and PA28-␣ and -␤ in the Ad12-transformed cells (8,9,15).
For proper loading of peptides on the MHC class I chains the Tap and Tapasin proteins are required. The heterodimeric Tap complex functions by transporting peptides from the cytosol to the endoplasmic reticulum (reviewed in Refs. 54 -56). Mice lacking Tap-1 (10) and mutant cell lines lacking one or both Tap subunits (14, 57-59) fail to express MHC class I complexes on their cell membranes, thus demonstrating that Tap is an essential component of the antigen presentation machinery. In agreement with a previous report (8), we found a strikingly decreased level of Tap-2 in Ad12-transformed cells.
The role of Tapasin in the formation of MHC class I antigens is illustrated by the fact that Tapasin-deficient mice show a 10-fold decreased MHC class I cell surface expression (38,39). Tapasin functions in the MHC class I peptide loading complex by binding independently to Tap and the MHC class I heavy chain, thus enhancing the loading of peptides (29,55,60,61). Tapasin binding stabilizes Tap, resulting in increased Tap protein levels (62), which is also apparent in our experiments (Fig. 3). In view of the significant contribution of Tapasin to antigen presentation, it seems likely that reduced Tapasin expression in Ad12-transformed BRK cells also plays a role in the low MHC class I cell surface expression. However, restoring Tapasin expression, even in combination with RT1-A u and Tap-2, is not sufficient to restore MHC class I cell surface expression (Fig. 3). Possibly, in contrast to results of a previous report (8), Tap-1 overexpression is required in these Ad12transformed BRK cells. Alternatively, insufficient availability of optimal peptides could be responsible for the effect, although low expression levels of the immunoproteasome subunits do not necessarily result in generation of low levels of peptides (14,59,63). However, the peptides generated in Ad12-transformed cells compared with those in Ad5-transformed cells may be less optimal for presentation as a result of low MECL-1 and PA28-␣ and -␤ levels (Fig. 5) and the low LMP-2 and -7 levels reported previously (15). MECL-1, LMP-2, and LMP-7 are IFN-␥-responsive genes and functionally replace the constitutive 20 S catalytic proteasome ␤-subunits X, Y, and Z (16, 44, 64 -66). These replacements enhance cleavage after hydrophobic and basic residues, which improves the generation of some but not all peptides presented by MHC class I molecules. In addition, the proteasome activator PA28 complex, which is also induced by IFN-␥, enhances optimal peptide formation by keeping the proteasome exit open, thus reducing the processivity that supposedly results in increased average length of the generated peptides (44,48,64).
Differential expression in Ad5-and Ad12-transformed cells of several components of the antigen presenting machinery suggests a common regulatory mechanism. We have shown previously that transcriptional down-regulation of MHC class I genes can be attributed to the H2TF1 element in the mhcI promoter (25,49,67), which is regulated by NF-B family members (68,69). Consistently, basal NF-B activity is very low in Ad12-transformed cells, and ectopic expression of NF-B1 partially restores MHC class I protein expression (25,40,49). The precise mechanism of NF-B down-regulation is not clear. Previous experiments indicated that the levels of NF-B1-p50 are low in Ad12-transformed cells compared with those in Ad5-transformed cells (25), whereas expression levels of the precursor protein NF-B1-p105 were comparable. Surprisingly, we now also find equal expression levels of NF-B1-p50 in Ad5-and Ad12-transformed cells using a more potent antibody. 2 The specificity of the antibody used in the latter study has been verified by Pereira et al. (30) using cells from NF-B1-deficient mice. Hence, the reason for the low activity of NF-B in Ad12-transformed cells is still unclear. It has recently been shown that NF-B1-p50 is hypophosphorylated in Ad12-compared with Ad5-transformed cells, and this has been proposed to be the reason for the difference in NF-B activity in these cells (70). It is tempting to speculate that the large difference in NF-B DNA binding activity between Ad5-and Ad12-transformed cells could possibly be responsible for the various differences in gene expression. The promoter of the PA28-␤ gene (51) and the promoter of the human Tapasin gene (71) indeed contain an NF-B binding site (52). However TNF-␣, which is a very strong activator of NF-B in Ad12-transformed BRK cells (50), does not directly induce PA28-␤ and Tapasin, nor Tap-2, MECL-1, and PA28-␣ in Ad-transformed cells (Fig. 6A) (data not shown). Nevertheless, the TNF-␣ pathway is clearly functional in these Ad12-transformed cells as an NF-B-dependent luciferase construct is induced 10-fold upon treatment with TNF-␣ (data not shown). However, these experiments do not rule out a delayed NF-B effect, because NF-B directly upregulates IFN-␤ (72)(73)(74), which in turn lead to induction of MHC class I cell surface expression (75). To study the long term effect of NF-B on MHC class I cell surface expression, Ad12transformed BRK cells were treated in the present study with TNF-␣ for 3 days and subjected to FACS analysis. No increase in MHC class I cell surface expression of two independent Ad12-transformed BRK cell lines was found (data not shown), making it unlikely that the differences in NF-B activity in Ad5-and Ad12-transformed BRK cells explain the difference in MHC class I cell surface expression.
In the present study, we present evidence that IFN-inducible genes may be involved in the differential gene expression. We found that STAT-1 expression is higher and is phosphorylated on the activating Tyr in Ad5-transformed cells but not in Ad12transformed cells. It has been demonstrated that even the presence of non-phosphorylated STAT-1 can maintain basal MHC class I, ␤2-microglobulin, and LMP2 expression relative to cells with no STAT-1 expression (20). Therefore, the constitutively active STAT-1 in Ad5-transformed cells can explain why these cells have high MHC class I cell surface expression whereas lower expression and lack of activation of STAT-1 in Ad12-transformed cells may explain down-modulation of class I in Ad12-transformed cells. Activation of STAT-1 might imply active ISGF3, and indeed one of its targets, IRF-7, was demonstrated to be expressed in Ad5-and not in Ad12-transformed cells. The interferon response is a multistage process, which requires active IRF-7 to reach the final stage in which the members of the IFN-␣ family are switched on (23,36). Our results indicate that both Ad5-and Ad12-transformed cells express low levels of IFN-␣, as found frequently in non-infected cells (76). This indicates that IRF-7 expressed in Ad5-transformed BRK cells is not active.
Despite extensive studies to determine the molecular basis for the difference in oncogenicity between Ad5-and Ad12transformed cells no definitive solution has been found. It is clear that reduced MHC class I expression is not the only factor responsible for the effect, because nearly every protein component of the antigen presentation machinery is down-regulated in Ad12-transformed cells compared with Ad5-transformed cells, including Tap-1 and -2 (8,9) and LMP-2 and -7 (9, 15), as well as Tapasin and the PA28 complex (this study). Attempts to restore the expression levels of all these proteins in Ad12transformed cells by ectopic overexpression are hampered by the large number of factors involved. It is conceivable that an upstream transcription factor that co-regulates the expression level of all these genes involved in antigen presentation is the actual trigger. Although previous data suggested that reduced NF-B activity might be the factor responsible for the low MHC class I heavy chain expression (25), our present study indicates that an interferon-related pathway is a more likely candidate. We present data indicating that the transcription factor STAT-1, a key regulator of MHC class I cell surface expression (19 -21), is constitutively activated in Ad5-but not in Ad12transformed cells possible because of the differential expression of IFN-␤. Future work should focus on the mechanism of differential expression of STAT1 and IFN-␤ by Ad5 and Ad12. So far the evidence indicates that the Ad5-and Ad12-E1A proteins behave similarly in their association with cellular proteins, suggesting that subtle structural differences in the resulting protein complexes may be responsible for the observed effects.