Resistance to Diverse Drugs and Ultraviolet Light Conferred by Overexpression of a Novel Human 26 S Proteasome Subunit*

We have investigated the usefulness of the fission yeast Schizosaccharomyces pombe as a model organism for the discovery of novel modes of drug resistance in human cells. In fission yeast, overexpression of the essentialpad1 + gene confers pleiotropic drug resistance through a pathway involving an AP-1 transcription factor encoded bypap1 +. We have identified POH1, a human pad1 homologue that can substitute fully forpad1 + and induce AP-1-dependent drug resistance in fission yeast. POH1 also confers P-glycoprotein-independent resistance to taxol (paclitaxel), doxorubicin, 7-hydroxystaurosporine, and ultraviolet light when transiently overexpressed in mammalian cells. Poh1 is a previously unidentified component of the human 26 S proteasome, a multiprotein complex that degrades proteins targeted for destruction by the ubiquitin pathway. Hence, Poh1 is part of a conserved mechanism that determines cellular susceptibility to cytotoxic agents, perhaps by influencing the ubiquitin-dependent proteolysis of transcription factors.

Exposure of eukaryotic cells to any of a wide variety of cytotoxic agents can result in the appearance of multiply resistant subpopulations, a phenomenon that frequently limits the effectiveness of anticancer therapies. Much effort has been directed toward improved understanding of the molecular mechanisms that underlie such pleiotropic resistance. Many of these studies have used mammalian cell lines selected in vitro for their resistance to high concentrations of cytotoxic drugs. Under these circumstances the mechanism of resistance typically involves overexpression of P-glycoprotein (P-gp), 1 a drugexcluding pump with broad specificity (1,2). Unfortunately, P-gp overexpression fails to account for the majority of clinical drug resistances especially in solid tumors (3), and the key determinants of this phenomenon remain largely unknown.
We have taken an alternative route to the identification of human proteins potentially involved in pleiotropic drug resistance. The fission yeast Schizosaccharomyces pombe is established as a model organism in which powerful genetical approaches can be used to elucidate fundamental but complex eukaryotic processes such as mitosis (4). Overexpression of the pad1 ϩ gene in S. pombe was recently found to confer moderate resistance to the protein kinase inhibitor staurosporine and a variety of other drugs including caffeine and the spindle poison thiabendazole (5)(6)(7). We have identified a human functional pad1 homologue (Poh1) that confers moderate resistance to chemotherapeutic drugs and also to ultraviolet light when transiently overproduced in mammalian cells. Poh1 is a previously unidentified component of the 26 S proteasome and genetical data suggest that Poh1-induced drug resistance is mediated through AP-1 transcription factors.

EXPERIMENTAL PROCEDURES
Molecular Cloning of POH1-Polymerase chain reaction amplification was performed using an HT29 human colon carcinoma cDNA library (kindly provided by H. Okayama, University of Tokyo) as template and the degenerate oligonucleotides 1) 5Ј-CTSGCHCTSCT-SAARATG and 2) 5Ј-ACDSWCTGRATDGGRTC, corresponding to the amino acid sequences LALLKM and DPIQSV found both in Pad1 and in the predicted product of the Caenorhabditis elegans cosmid clone F37A4.5 (5). The resulting products were used as template for a second round of amplification using the same primer 2 in combination with a 3rd primer, 5Ј-GARGTBATGGGNCTSATGCT, corresponding to the internal block of conserved amino acids EVMGLML. The resulting 300base pair fragment was radiolabeled and used to isolate full-length POH1 cDNAs from a Basinger human fibroblast cDNA library (also provided by H. Okayama) by standard procedures (8). Multiple clones with inserts of approximately 1.7 kilobase pairs were identified, which correspond to the POH1 open reading frame preceded by approximately 200 base pairs of 5Ј-untranslated sequence and followed by approximately 600 base pairs of 3Ј-untranslated sequence. One clone corresponded to a POH1 mRNA species with the same 5Ј end but polyadenylated immediately 3Ј to the translation stop codon. The complete nucleotide sequence of the open reading frame and 5Ј-untranslated region (GenBank TM accession number U86782) was determined on both strands using a Sequenase kit (United States Biochemical) according to the manufacturer's instructions. A version of the POH1 open reading frame flanked by 5Ј SalI and 3Ј NotI sites and tagged with the hemagglutinin epitope (YPYDVPDYA) was generated by polymerase chain reaction and cloned into the fission yeast vector pREP3X (kindly provided by S. Forsburg, The Salk Institute, La Jolla, CA), which is essentially identical to pREP1 (9), containing 5Ј XhoI and 3Ј NotI sites in the polylinker.
Antibodies and Immunochemistry-A polyclonal rabbit antiserum against a GST-Poh1 fusion protein was raised using standard procedures (10) after cloning the POH1 open reading frame into pGEX4T-2 (Pharmacia Biotech Inc.). Immunoblotting was performed using the ECL method (Amersham Corp.) according to the manufacturer's instructions.
Fission Yeast Methods-The strain HM123 (leu1-32, h Ϫ ) was used unless otherwise stated. Standard procedures for S. pombe genetics * This work was supported in part by the Imperial Cancer Research Fund, the Human Frontier Science Program Organization, the European Society for Medical Oncology, and Fondazione Ticinese per la Ricerca sul Cancro. 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U86782.
§ Recipient of Fondazione Ticinese per la Ricerca sul Cancro Fellowship.
Transfections, Cytotoxicity Assays, and Analysis of Doxorubicin Uptake-A hemagglutinin epitope-tagged version of POH1 cDNA with the Kozak sequence 5Ј-CACC-3Ј immediately 5Ј to the initiator ATG was generated by polymerase chain reaction and inserted into the plasmid vector pcDNA3 containing an SV40 origin (Invitrogen) to give pcDNA3.POH1. COS-1 cells were transiently cotransfected by the calcium phosphate method (12) with either pcDNA3 (empty vector) or pcDNA3.POH1 in a ratio of 4:1 with the plasmid pOPRSV.CD2, which drives expression of a truncated version of rat CD2 (13). 24 h after transfection, cells were selected with anti-CD2 immunomagnetic beads (Dynal) according to manufacturer's guidelines. Cells transfected with pcDNA3.POH1 or pcDNA3 were used to prepare whole cell extracts for immunoblot analyses and to determine clonogenic survival curves. For clonogenic assays, 10 4 CD2 ϩ cells were seeded per plate; after 24 h, cells were exposed to the appropriate cytotoxic agent. Stock solutions of paclitaxel (Sigma), doxorubicin (Farmitalia), UCN-01 (kindly provided by E. Sausville, NCI, Bethesda), and hydrogen peroxide (Sigma) were made in complete medium and added to the cells at the final concentrations indicated. After drug exposure, the medium was removed and plates were washed with phosphate-buffered saline before the addition of fresh medium. For assessment of UV sensitivity, cells in phosphatebuffered saline were exposed to UVC (254 nm) by means of a Stratalinker TM 1800 (Stratagene) followed by the addition of fresh medium. After 10 -14 days of growth, colonies were stained with crystal violet, and colonies with more than 30 cells were counted. For the assessment of cellular uptake of doxorubicin, 48 h after transfection, plates containing COS1/pcDNA3 or COS1/pcDNA3.POH1 cells were exposed to doxorubicin for 6 h, stained with fluorescein isothiocyanate-conjugated anti-CD2 antibodies and assessed by flow cytometry (FACScan, Becton Dickinson). The median fluorescence channel for the doxorubicin-derived signal (red) was measured in CD2 ϩ cells (green) containing pcDNA3 or pcDNA3.POH1.
Proteasome Purification-Lysates of logarithmically growing H1299 cells were prepared exactly as described (14) and were incubated with CNBr-activated Sepharose beads (Sigma) coated with the monoclonal antibody MCP-21 (14). The same procedure was carried out using mock-coated beads as a negative control. After extensive washing, the beads were boiled in SDS-polyacrylamide gel electrophoresis sample buffer and subjected to immunoblotting analysis using the anti-Poh1 rabbit antiserum and ECL detection. Biochemical purification of 26 S proteasomes from human erythrocytes was performed as described (15). Separation of 26 S and 20 S proteasomes by nondenaturing gel electrophoresis was essentially as described by Hendil et al. (14). HeLa cells were lysed by sonication for 5 s at 0°C in 5 volumes of 50 mM Tris pH 7.4, 17% glycerol, and the lysate was centrifuged for 5 min at 14,000 ϫ g at 4°C. 20-l samples of the supernatant were resolved by electrophoresis in gels containing 4.5% acrylamide and Tris borate buffer at 4°C before transfer to nitrocellulose membranes and immunoblotting as described above.

A Human Functional Homologue of Fission Yeast Pad1-
The sequence similarity between fission yeast Pad1 and a C. elegans cosmid-encoded sequence (5) suggested that this determinant of drug resistance might be conserved in human cells. We exploited this sequence similarity to generate a hybridization probe by amplification using degenerate primers and hence clone a pad1 homologue (termed POH1 for pad one homologue) from a human cDNA library (Fig. 1A). The amino FIG. 1. Poh1 is a highly conserved human homologue of fission yeast Pad1 and is related to the human Jab1 and S12/p40 proteins. A, alignment of the predicted Poh1 protein sequence with those of S. pombe Pad1, human Jab1, and S12/p40. Residues present at identical positions in two or more of the proteins are highlighted in black, and conservative substitutions are in gray. acid sequence encoded by the human cDNA shares 68% identity with fission yeast Pad1 and has therefore been highly conserved through eukaryotic evolution. Poh1 is also clearly related, albeit more distantly, to two other human proteins: the c-Jun-binding protein Jab1 (16) and the S12/p40 component of the 26 S proteasome regulatory subunit (17,18). 28 amino acid residues are found at conserved positions in all four proteins (but not in any other sequences in the SwissProt data base) and may therefore contribute to a common structure or function. POH1 mRNA is expressed in a wide variety of human tissues, with highest levels in heart and skeletal muscle (Fig. 1B). A similar pattern of mRNA expression has been described for the human proteasome components S12/p40, X and Y (18).
The high degree of conservation of Pad1 and Poh1 suggested that the two proteins might perform analogous functions. This was confirmed by expression of POH1 in a haploid fission yeast strain bearing a deletion of the essential pad1 ϩ gene (⌬pad1). This strain was able to grow normally when transformed with a plasmid (pREP-POH) in which POH1 expression is driven by the thiamine-repressible nmt1 promoter ( Fig. 2A-C). Growth of the ⌬pad1 strain containing pREP-POH was only seen when the nmt1 promoter was derepressed (thiamine absent from medium; Fig. 2A).
Immunoblotting analysis of the yeast transformant cell extracts was performed using a polyclonal antibody raised against Pad1 (5). This confirmed that Pad1 protein was absent from the ⌬pad1 strain (Fig. 2D, lane 4) and showed that this antiserum cross-reacts with Poh1 protein (Fig. 2D, lanes 3, 4,  and 6). Repression of POH1 expression in the ⌬pad1 strain by the addition of thiamine to the medium resulted in the rapid loss of cell viability, accompanied by a variety of mitotic defects (data not shown). Thus Poh1 is a fully functional human homologue of Pad1. This result contrasts with that reported for Jab1, which although clearly related to Pad1 is unable to com-plement the ⌬pad1 strain (16).
POH1 Confers Pleiotropic Drug Resistance-High copy-number plasmids bearing pad1 ϩ confer pleiotropic drug resistance in fission yeast, in a manner that depends upon the presence of an otherwise inessential AP-1 transcription factor encoded by pap1 ϩ (5,19). As shown in Fig. 3, wild-type cells containing pREP-POH became resistant to staurosporine, a protein kinase inhibitor, in the absence of thiamine (Fig. 3A, promoter derepressed) but not in the presence of thiamine (Fig. 3B, promoter  repressed). Furthermore, as in the case of pad1 ϩ , the drug resistance phenotype induced by pREP-POH was no longer observed in the pap1 deletion (⌬pap1) background (Fig. 3, A  and B, lower halves), though the ⌬pap1 strain expressed high levels of Poh1 (Fig. 2D, lane 6) and was fully viable in the absence of the drug (Fig. 3C, lower half).
As overexpression of POH1 in fission yeast induced a drug resistance phenotype, we examined the possibility that the same might be true in mammalian cells. Our attempts to generate stable mammalian transfectants constitutively overexpressing Poh1 were not successful, so we chose instead to adopt a transient transfection approach. COS-1 cells were cotransfected with a plasmid encoding C-terminally truncated rat CD2 as a marker of transfection and either pcDNA3-POH1, an SV40 origin-containing plasmid in which POH1 expression is driven by the CMV promoter or the empty vector pcDNA3. Cells transiently overexpressing Poh1 or the vector control cells were isolated by means of immunomagnetic beads coated with an anti-CD2 monoclonal antibody and tested in clonogenicity assays for their sensitivity to a variety of cytotoxic treatments. Western blots probed with a rabbit polyclonal antiserum raised against Poh1 demonstrated that the transiently transfected cells expressed approximately two-fold more Poh1 than the control cells (Fig. 5A). Comparisons of the sensitivities of the two populations to 7-hydroxystaurosporine (Fig. 4A), taxol (Fig.  4B), and doxorubicin (not shown) indicated that the Poh1overexpressing cells were significantly resistant to each of these drugs when compared with the vector transfectants. The greatest effect was seen with 7-hydroxystaurosporine, where the Poh1-induced increase in IC 50 was approximately 4-fold. In contrast, no significant differences in sensitivity to hydrogen peroxide (Fig. 4D) or ionizing radiation (not shown) were de-  4 and 6). The extracts were analyzed by immunoblotting using a polyclonal antiserum raised against Pad1 (5) (indicated by bracket on left), which cross-reacts with Poh1 (indicated by arrow on right). Note that the slightly smaller Pad1 protein (33 kDa) observed in lane 1 is truncated at the C terminus, whereas Poh1 migrated at 38 kDa due to the presence of an N-terminal HA epitope tag. An unidentified cross-reactive band of approximately 30 kDa is seen in all the extracts. Acquisition of multidrug resistance in mammalian cell lines is frequently accompanied by overexpression of the P-gp product of the MDR1 gene (1,2,20). To investigate the possibility that POH1 overexpression might lead to elevated MDR1 expression, P-gp levels were examined by Western blotting but were found to be similar in extracts of both populations of transiently transfected cells used for clonogenic assays (Fig.  5B). Thus POH1 induces multidrug resistance by a mechanism that does not involve P-gp overexpression. Moreover, POH1 overexpression also resulted in moderate but significant resistance to ultraviolet light (Fig. 4C), indicating that POH1-induced resistance is likely to be mediated at least in part through a mechanism unrelated to P-gp function.
To investigate further the nature of POH1-induced drug resistance the accumulation of doxorubicin in COS-1 cells transiently overexpressing POH1 was measured by flow cytometry after 6 h exposure to the drug (Fig. 6). The results indicate that intracellular accumulation of doxorubicin was not significantly influenced by POH1 overexpression, adding further weight to the argument that P-gp is not likely to be involved in the Poh1-responsive drug resistance pathway.
Poh1 is a Novel 26 S Proteasome Component-Given the similarity between Poh1 and S12/p40, we investigated the possibility that Poh1 might also be associated with the proteasome. The monoclonal antibody MCP-21, which recognizes a subunit of the core (20 S) human proteasome (14), was used to immunopurify proteasomes from lysates of human H1299 lung cancer cells. Immunoblotting analysis of these proteasome preparations showed that Poh1 was substantially enriched in the immunopurified material (Fig. 7A). The significance of this result was underlined by the finding that a biochemically pu-rified 26 S proteasome preparation from human erythrocytes also contained Poh1 (Fig. 7B, lane 3). Approximately equivalent amounts of Poh1 were present in 50 g of a crude fraction and 1 g of the purified 26 S preparation (Fig. 7B, compare  lanes 2 and 3), suggesting copurification of Poh1 alongside other 26 S components. A Coomassie-stained gel lane of the purified 26 S fraction indicated only a relatively low intensity of staining in the Poh1-containing region. This could indicate that Poh1 is not a major stoichiometric component of the proteasome, but it is also possible that Poh1 is less efficiently stained than other proteasome components under these conditions. To determine what fraction of total cellular Poh1 is associated with the proteasome, crude cellular lysates were fractionated by nondenaturing gel electrophoresis under conditions previously shown to separate intact 26 S and 20 S proteasome particles (14), the positions of which were revealed by immunoblotting with MCP-21 (Fig. 7C, lane 1). Blotting with affinity-purified anti-Poh1 antibodies indicated that all the detectable Poh1 is associated with the 26 S particles (Fig.  7C, lane 2). Thus very little Poh1 is present as free monomers or other relatively low molecular weight forms; this is also the case for the majority of the previously-characterized components of the 20 S proteasome (14). Taking these results together, we conclude that Poh1 is a previously uncharacterized component of the 19 S regulatory cap of the 26 S proteasome. DISCUSSION The validity of fission yeast as a model organism with which to investigate drug resistance in mammals is confirmed by our demonstration that Pad1, a key determinant of pleiotropic resistance in S. pombe, is functionally conserved as Poh1 in human cells. In COS cells the response to a specific subset of cytotoxic insults is modified by overexpression of POH1 in such a way that cell survival is favored. Alterations in P-gp expression or intracellular drug accumulation do not appear to be involved, but it is not clear at this stage if the POH1-induced survival advantage reflects a decreased propensity for cell death or an alteration in the processing of potentially lethal damage. POH1 can also induce drug resistance in fission yeast, which lacks the apoptotic cell death program, suggesting that modulation of apoptosis is unlikely to be solely responsible for POH1-induced resistance in mammalian cells.
Poh1 is a novel component of the 26 S proteasome, indicating that proteolysis is involved in determining the sensitivity of human cells to cytotoxic treatments. A number of recent reports lend weight to this idea. Mutations in the fission yeast proteasome components encoded by mts2 ϩ and mts3 ϩ (21,22) are temperature-sensitive for growth and confer resistance to the spindle poison methyl benzimidazole-2-yl carbamate at the permissive temperature. Interestingly, overexpression of POH1 in fission yeast also induces resistance to this drug. 2 Furthermore tumor necrosis factor-induced killing of human fibrosarcoma cells was modulated by proteasome inhibitors (23). We therefore consider it likely that Poh1 induces drug resistance by modulation of a mammalian proteasome activity.
Poh1 and AP-1 Transcription Factors-The requirement for Pap1 in Pad1/Poh1-induced drug resistance in fission yeast and the similarity between Poh1 and Jab1 suggest that modulation of transcription could underlie the induction of drug resistance by the Poh1 proteasome component. Indeed, overexpression of Pap1 in fission yeast results in a pattern of drug resistance indistinguishable from that seen on overexpression of Pad1/ Poh1 (5,19). What then might be the significant targets of AP-1 that induce resistance to cytotoxic agents? Mammalian AP-1responsive genes reported include those encoding glutathione S-transferase Ya and , elevated levels of which could contribute to cellular resistance (24,25). Similarly in budding yeast the promoter of the GSH1 gene has been shown to be AP-1 responsive (26). This gene encodes ␥-glutamylcysteine synthetase, which performs the rate-limiting step in glutathione biosynthesis and can influence cellular drug resistance. However, the observation in budding yeast of AP-1-induced resistance to nitrosoguanidine suggests that glutathione-independent mechanisms are also significant (27). In line with our observation of unaltered P-gp expression on POH1 overexpression, AP-1mediated drug resistance in budding and fission yeast was found to be independent of endogenous P-gp-related drug transporters (6,28).
Both c-Jun and c-Fos are known to be degraded by the ubiquitin/proteasome pathway (29 -32), and elevated expression of one or other of these AP-1 factors has been correlated with drug resistance in a number of instances (24,33,34). Ubiquitin-dependent degradation of c-Jun appears to be regulated in response to signals transduced by mitogen-activated protein kinase (35). In addition, murine cells lacking c-Fos were found to be sensitized to UV radiation (36), although DNA repair was not defective, suggesting a role for AP-1 factors in a pathway that modulates cell survival after exposure to a given level of cytotoxic insult. Our results suggest that Poh1, a component of the 26 S proteasome, also participates in such a pathway in mammalian cells. This pathway is potentially involved in clinical resistance to anticancer drugs, and our preliminary data indicate that Poh1 is widely but variably expressed in human cancer cell lines. 3 The extent of the involvement of Poh1 in clinical drug resistance awaits further investigation.
A further direct connection between ubiquitin-dependent proteolysis and transcription factors arose with the recent description of budding yeast Sug1 and its mammalian counterpart (mSug1/Trip1/FZA-B) both as a transcriptional co-activator and as a component of the 26 S proteasome (37)(38)(39). Mammalian mSug1 associates with c-Fos via its leucine zipper (40) and also with a number of transcription factors of the nuclear receptor family (41). It is not clear at this stage whether Sug1 (and, by extension, Pad1/Poh1) acts as a proteasomal receptor for transcription factors that have been targeted for ubiquitin-dependent degradation, or if the 26 S proteasome component and transcription factors interact for other reasons. It is nonetheless intriguing that the c-Jun binding protein Jab1, which was identified as a transcriptional co-activator (16) is clearly a close relative of Pad1/Poh1, identified here as a component of the 26 S proteasome. It seems possible that coordination of proteolysis with transcriptional activation could be significant in the modulation of transcriptional responses to a wide variety of stimuli.
Acknowledgments-We thank Susheela Dhut for excellent technical 3 V. Spataro and C. Norbury, unpublished data.  Fig. 3E. Poh1 (arrow) migrates slightly ahead of the HA-tagged derivative but is slightly retarded in lane 2 due to the 50-fold higher protein content. A sample of the purified 26 S proteasome preparation stained with Coomassie Blue is shown for comparison (lane 4). C, crude HeLa cell lysates prepared by sonication were separated by nondenaturing gel electrophoresis and analyzed by immunoblotting using MCP-21 (lane 1), which detects both 26 S and 20 S proteasomes under these conditions (14). A parallel lane was analyzed using affinity-purified anti-Poh1 antibodies (lane 2). assistance, Drs. C. Gordon and M. Karin for communicating results prior to publication, H. Okayama for human cDNA libraries, Edward Sausville (NCI) for UCN-01, and Nic Jones, Mark Toone, Derek McCusker, John Trowsdale, Enzo Cerundolo, Ian Hickson, Trivadi Ganesan and Agnes Norbury for advice and comments on the manuscript.