SKY1 Is Involved in Cisplatin-induced Cell Kill in Saccharomyces cerevisiae, and Inactivation of Its Human Homologue, SRPK1, Induces Cisplatin Resistance in a Human Ovarian Carcinoma Cell Line

Introduction

Cisplatin is one of the most widely used anticancer drugs. Platinum-based chemotherapy is active against cancers of the lung, ovary, bladder, head and neck, esophagus, cervix, and endometrium and is curative for the vast majority of patients with testicular cancer (1) . Unfortunately, cellular resistance to cisplatin, either intrinsic or acquired, is encountered regularly and severely limits the therapeutic potential of the drug (2) . A better understanding of the cellular mechanisms of cisplatin sensitivity and resistance could lead to the development of effective specific biological and pharmacological intervention and thus to better treatment results with regard to long-term survival. Molecular pathways leading to cisplatin resistance need to be defined in detail, which will be largely sustained by the identification and characterization of genes regulating cisplatin sensitivity. Multiple mechanisms by which cells may overcome the cytotoxic action of cisplatin have been identified in vitro. These include decreased intracellular drug accumulation, inactivation by glutathione or metallothioneins, aberrations in repair, enhanced tolerance, and defects in pathways modulating cell death (2, 3, 4) . Despite the many potentially important resistance mechanisms that have been identified in vitro, hardly any data demonstrate correlations between presently known in vitro mechanisms of cisplatin resistance and clinical outcome. Apparently, these processes do not fully account for the observed in vivo unresponsiveness of particular tumors to platinum-based chemotherapy. Therefore, additional cisplatin resistance mechanisms for which the genes involved have yet to be identified may exist.

We used the budding yeast Saccharomyces cerevisiae as a model system to study drug sensitivity and resistance (5) . In this study, we performed a genome-wide screen to identify and characterize novel cisplatin sensitivity genes, and we found that disruption of the yeast YMR216C locus corresponding to the SKY1 (SR 3 protein-specific kinase from budding yeast) gene conferred cellular resistance to cisplatin. We demonstrate that down-regulation of the human Sky1p homologue SRPK1 (SR protein-specific kinase) by antisense ODNs in human ovarian carcinoma cells also confers cisplatin resistance. This is the first report in which SKY1 and SRPK1 are associated with alterations in cisplatin sensitivity, and our findings indicate that SRPK1 might potentially be of importance for studying clinical drug resistance.

Materials and Methods

Generation of Cisplatin-resistant Yeast Mutants.

Yeast cells were cultured on complete YNB medium (Difco Laboratories, Detroit, MI) at 30°C as described previously (5) . Cisplatin-sensitive S. cerevisiae strain MGSC131 (6) was transformed with a yeast genomic mini-Tn3::lacZ::LEU2 transposon insertion library (7) , referred to as Δ35 (kindly provided by Dr. P. B. Ross-Macdonald, Yale University, New Haven, CT), by means of a high-efficiency protocol. Transformants were selected on complete YNB medium plates lacking leucine. Leucine-proficient S. cerevisiae cells were replated at a density of 10⁴ cells/94-mm dish on selective YNB medium plates containing 4 μg/ml cisplatin [Platosin; cis-diamminedichloroplatinum(II); Pharmachemie, Haarlem, the Netherlands]. Colonies surviving this one-step drug selection were picked and retested for cisplatin resistance in semiquantitative spot assays and quantitative clonogenic survival assays as described previously (5) . Inverse PCR (8) was used to obtain sequences flanking the transposon elements. PCR products were purified and sequenced using transposon-specific primers (7) . A gene-specific SKY1 disruption was generated in strain MGSC283 (an isogenic ura3-1 derivative of MGSC131) by one-step gene replacement performed as described by Rothstein (9) , which yielded MGSC-sky1Δ cells. The construct used in the one-step gene replacement procedure was made using primers SKY1-5A (5′-GTA-AGA-AAG-CTG-GGA-TGG-GGC-CAC-TTC-TCA-TCT-TTG-ACA-GCT-TAT-CAT-C-3′) and SKY1-3 (5′-CGG-AAC-CAC-TCC-CGT-ACA-ACT-CTC-TAT-CAG-GTA-CCC-ACT-CGT-GCA-CCC-3′). A probe hybridizing to the 0.5-kb 5′ SpeI-HindIII SKY1 fragment was used to confirm successful disruption by Northern blotting. Sequences were analyzed using the Saccharomyces Genome Database. Database entries used for amino acid comparisons between Sky1p and its human counterpart, SRPK1, were RefSeq accession number NP_013943 and Protein Information Resource accession number S45337, respectively.

Construction of Expression Plasmids.

A 2.3-kb SKY1 PCR product was ligated into the low-copy S. cerevisiae expression vector pYCTEF, which consists of the 4.5-kb PvuII backbone fragment from centromeric plasmid YCplac22 (10) and the 1.0-kb PvuII fragment from p424TEF (11) , containing the constitutive translation elongation factor 1α promoter. The PCR product was generated using primers SKYex-5 (5′-ATA-GTG-GAT-CCT-GGT-ATA-AAT-AGA-CAC-CCC-C-3′) and SKYex-3 (5′-CTA-ACC-TCG-AGA-GGG-CAA-AAT-AAA-GGT-ATA-AAG-G-3′). The full-length human SRPK1 coding region, homologous to SKY1, was kindly provided by Drs. X-D. Fu and H-Y. Wang (University of California at San Diego, La Jolla, CA), and dnSRPK1 (12) was made available by Drs. X-D. Fu and J-H. Ding (University of California at San Diego). Both constructs were also cloned into the pYCTEF expression vector, enabling heterologous expression in yeast.

Treatment of Cells with Antisense ODNs.

The human ovarian carcinoma A2780 cell line was obtained from the American Type Culture Collection (Manassas, VA), and maintained in HEPES-buffered RPMI 1640 supplemented with 10% fetal bovine serum (Life Technologies, Inc., Rockville, MD) in a humidified incubator at 37°C with 8.5% CO₂. Morpholino phosphorodiamidate antisense ODNs directed against the translation initiation site of SRPK1 (referred to henceforth as AS-SRPK1; sequence, 5′-CAT-GGT-GAG-ACC-CAA-CAA-AAG-CAG-G-3′), control random antisense ODNs (referred to henceforth as NS; sequence, 5′-CCT-CTT-ACC-TCA-GTT-ACA-ATT-TAT-A-3′), and control antisense ODNs with four mispairs distributed along the sequence (referred to henceforth as MIS; sequence, 5′-CAT-CGT-GTG-ACC-CAA-CAA-TAG-CTG-G-3′; mismatches are shown in bold) were purchased from Gene Tools (Corvallis, OR). Incubation of A2780 cells with ODNs was performed using the manufacturer’s Special Delivery protocol. Cisplatin sensitivity of ODN-treated A2780 cells was determined by MTT assay (13) . Briefly, the cells were seeded in 96-well tissue culture plates. The next day, the culture medium was replaced by medium containing ODNs and 0.6 μm ethoxylated polyethylenimine delivery agent but lacking serum, and the cells were incubated at 37°C with 8.5% CO₂ for 3 h. Subsequently, the medium was exchanged with standard culture medium containing serum, as described above. After 16 h, cisplatin was added at a dose range of 0–24 μm, and the cells were allowed to grow for another 72 h. Growth inhibition was determined by the use of MTT (Sigma-Aldrich, Zwijndrecht, the Netherlands) as described by Perez et al. (13) .

Statistical Analysis.

Relationships between cisplatin concentration and cisplatin-induced cytotoxicity were evaluated using Siphar version 4.0 software (InnaPhase, Philadelphia, PA) and the Number Cruncher Statistical System version 5.X package (Dr. J. L. Hintze, University of Utah, East Kaysville, UT). Data were fitted to a sigmoidal maximum effect (E_max) model based on the modified Hill equation, as follows: E = E₀ + E_max × [(C^γ)/(C^γ + EC₅₀^γ)]. In this equation, E₀ is the “no drug” effect, E_max is the maximum drug effect, C is the drug concentration to which cytotoxicity is related, EC₅₀ is the drug concentration predicted to result in half-maximal cytotoxicity, and γ is the Hill constant describing the sigmoidicity of the curve. The difference in EC₅₀ between two cytotoxicity tests was evaluated statistically using a two-tailed (unpaired) Student’s t test.

Immunofluorescence Detection and Western Blot Analysis of SRPK1.

After 3 h of incubation with 5.6 μm ODNs as described above, followed by recovery in standard culture medium, a monolayer of A2780 cells was deposited on slides in a Cytospin 2 Cytocentrifuge (Shandon Inc., Pittsburgh, PA) at 750 rpm for 10 min. Cells were fixed in methanol:acetone (1:1) at −20°C for 10 min, rehydrated in PBS at room temperature for 15 min, and then blocked in blocking buffer (10% rabbit serum, 1% goat serum, and 1% normal human serum in PBS) at room temperature for 15 min. Next, they were incubated at room temperature for 45 min with primary monoclonal anti-SRPK1 (mouse IgG1; 1:50 dilution of the 250 μg/ml stock purchased from BD Transduction Laboratories, Lexington, KY) or nonspecific mouse monoclonal IgG1 (ICN/Cappel, Costa Mesa, CA) at the same final concentration, washed three times with PBS, and developed at room temperature for 30 min with secondary FITC-conjugated goat antimouse IgG (1:50 dilution of the 1 mg/ml stock obtained from Nordic, Tilburg, the Netherlands). Finally, the slides were washed three times in PBS and mounted in VECTASHIELD medium containing 1 μg/ml 4′,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA) for examination under a Leica DMR fluorescence microscope with the CGH photoimaging system (Leica Microsystems, Wetzlar, Germany). For Western blotting, A2780 cells were incubated similarly, and total protein was isolated by adding SDS-loading buffer to cell pellets and boiling the samples for 3 min. Western blots were blocked in PBS containing 0.2% Tween 20 and 5% Protifar (Nutricia, Zoetermeer, the Netherlands) at 37°C for 1 h and then incubated with a 1:2000 dilution of mouse monoclonal anti-SRPK1 at 4°C for 16 h. SRPK1 was visualized using horseradish peroxidase-linked goat antimouse IgG (1:4000 dilution of the 400 μg/ml stock purchased from Santa Cruz Biotechnology, Santa Cruz, CA) and SuperSignal West Pico chemiluminescent substrate (Pierce, Rockford, IL). Blots were then stripped in 63 mm Tris (pH 6.8), 0.002% SDS, and 114 mm β-mercaptoethanol at 56°C for 45 min and reincubated as described above with a 1:5000 dilution of mouse monoclonal anti-β-actin (Sigma-Aldrich) as first antibody.

Results and Discussion

Disruption of the SKY1 Gene in Yeast Induces Cisplatin Resistance.

In this study, we used the budding yeast S. cerevisiae as a model system to search for novel genes that, upon disruption, confer cellular resistance to the commonly applied anticancer drug cisplatin. A yeast genomic mini-Tn3::lacZ::LEU2 transposon insertion library (7) , referred to as Δ35, was introduced into S. cerevisiae cells. Transformation originates from homologous recombination between yeast DNA sequences flanking the transposon and endogenous genomic sequences, leading to replacement of the original genomic copy with the mutagenized version (9) . A cisplatin-sensitive yeast strain referred to as MGSC131, which displays a steep dose-response curve to the drug (5 , 6) , was used as recipient. Untransformed MGSC131 cells seeded on cisplatin-containing YNB medium plates did not yield any colonies, even after prolonged incubation. In contrast, when 3 × 10⁵ transformants of library Δ35 were plated under identical conditions, 31 colonies surviving the one-step drug selection grew within 3–6 days. Upon verification of the resistance phenotypes of the colonies derived from this genome-wide yeast screening approach, we selected nine strains that showed a similar cisplatin resistance level for further characterization. Southern blotting using a transposon-specific probe revealed that each strain contained one single transposon insertion (data not shown). The sites flanking the transposons (i.e., the loci that had been disrupted) could thus be identified directly by means of inverse PCR (8) , followed by sequencing of the respective products. Sequences of 50–60 bp on either side of the transposon were obtained, and the insertion sites were defined by comparison with public S. cerevisiae databases.

Most prominently, the YMR216C locus corresponding to the SKY1 gene (14) turned out to be disrupted in at least five transposon-containing cisplatin-resistant yeast strains, referred to as Δ35-2, Δ35P-1, Δ35P-3, Δ35P-4, and Δ35P-5. At least three strains (Δ35-2 or Δ35P-1, Δ35P-3 or Δ35P-4, and Δ35P-5) appeared to contain different insertions in the same gene, as judged by the sizes of the PCR products obtained (Fig. 1a) ⇓ . In the remaining transposon-containing cisplatin-resistant yeast strains, other loci had been disrupted, as will be described elsewhere. 4 In a quantitative clonogenic survival assay, strains Δ35-2, Δ35P-1, Δ35P-3, Δ35P-4, and Δ35P-5 were 4-fold cisplatin resistant as compared with parental MGSC131-SKY1⁺ cells. Fig. 1b ⇓ shows a typical cisplatin sensitivity profile of the transposon-derived sky1Δ transformant Δ35-2 as compared with untransformed isogenic SKY1⁺ cells.

The observed cisplatin-resistant phenotype of the transposon-containing transformants could, in principle, have arisen from unrelated mutations acquired during the screening procedure (i.e., cisplatin-induced mutations leading to resistance) rather than library-derived gene disruption. To address this issue, PCR-generated one-step gene replacement constructs (9) were used to disrupt the SKY1 gene from the S. cerevisiae genome independently. Successful disruption was confirmed by PCR (data not shown) and Northern blotting (Fig. 1b ⇓ , inset), and possible cisplatin resistance of the replacement-derived sky1Δ mutant (MGSC-sky1Δ) was monitored. Indeed, the cisplatin sensitivity profile appeared to be similar to that of the original transposon-containing sky1Δ strains (Fig. 1b) ⇓ , i.e., MGSC-sky1Δ cells were 4-fold cisplatin-resistant.

Subsequently, we reintroduced the SKY1 gene into the sky1Δ deletion mutant cells. For this purpose, the SKY1 open reading frame was cloned into yeast expression vector pYCTEF, and the resulting plasmid was transformed into MGSC-sky1Δ cells. The transformants became as sensitive to cisplatin as the original MGSC131-SKY1⁺ strain, whereas transformation with the empty vector alone left the resistance of the MGSC-sky1Δ cells completely unaffected (Fig. 1b) ⇓ . Our findings clearly demonstrate that the cisplatin resistance, which was observed originally, was linked to disruption of the SKY1 gene. Moreover, the data suggest that Sky1p is involved in the cytotoxicity of cisplatin, i.e., that SKY1 might be regarded as a cisplatin sensitivity gene.

Heterologous Expression of the Human Homologue SRPK1 Restores Cisplatin Sensitivity in sky1Δ Mutant Yeast.

Database analysis revealed that SKY1 has a human homologue, which encodes the SR protein-specific kinase SRPK1. SR protein-specific kinases and the SR proteins that they phosphorylate are thought to be key regulators of RNA processing and, in mammalian cells, alternative splicing through multiple mechanisms (14 , 15) . The predicted polypeptide sequence encoded by the human SRPK1 cDNA shares 35% identity over 644 amino acids to Sky1p. Notably, SRPK1 and Sky1p also share structural homology: their kinase domains are interrupted by a unique spacer sequence. Because several data suggest strongly that Sky1p and SRPK1 perform homologous functions in vivo (14 , 16) , we tested SRPK1 for its ability to complement the cisplatin resistance phenotype of S. cerevisiae sky1Δ cells. SRPK1 was cloned into yeast expression vector pYCTEF and transformed into MGSC-sky1Δ mutant cells. Heterologous RNA expression was confirmed by Northern blotting (Fig. 2 ⇓ , inset) using a PCR-generated probe specific for the spacer domain of SRPK1 (14) . In MGSC-sky1Δ yeast cells the cisplatin resistance phenotype appeared to be largely reversed by introduction of the human SRPK1 gene, as was the case when the yeast SKY1 gene was reintroduced. In contrast, MGSC-sky1Δ cells transformed with a construct encoding a well-established dominant-negative kinase-inactive mutant (K109M) of SRPK1 (dnSRPK1; Refs. 12 and 16 ) were still cisplatin resistant (Fig. 2) ⇓ . Apparently, SRPK1 can functionally substitute SKY1 in S. cerevisiae cells and can therefore be considered a cisplatin sensitivity gene as well. To our knowledge, SKY1 and SRPK1 have not been previously associated with cisplatin sensitivity or resistance. The observation that dnSRPK1 (encoding a kinase-inactive protein) could not substitute SKY1 suggests that the kinase function of SR protein-specific kinases is essential for the cytotoxicity of cisplatin in yeast.

Antisense Experiments.

Because SRPK1 was able to act as a cisplatin sensitivity gene in S. cerevisiae cells, we assessed whether it also functions as a gene that is necessary for proper cisplatin-induced cell kill in human cells. Therefore, we designed experiments in which we could monitor the effects of SRPK1 inactivation by antisense ODNs on cisplatin sensitivity in human cells. First, we established the effect of AS-SRPK1 on SRPK1 protein expression levels in ovarian carcinoma A2780 cells by immunofluorescence. In untreated A2780 cells, we found SRPK1 expression in both the cytoplasm and nuclear speckles (Fig. 3, a and b) ⇓ , as demonstrated previously (17) . Treatment of A2780 cells with control antisense ODNs, NS, or MIS had no detectable effect on SRPK1 protein expression levels or subcellular localization (Fig. 3, e–h) ⇓ . In contrast, expression of the SRPK1 protein was largely decreased after treatment of A2780 cells with specific AS-SRPK1 (Fig. 3, c and d) ⇓ . Incubation with AS-SRPK1 almost completely abolished the nuclear staining of SRPK1 protein (Fig. 3, c and d) ⇓ as compared with NS or MIS (Fig. 3, e–h) ⇓ , whereas cytoplasmic staining was reduced to the background level detected using a nonspecific antibody (data not shown). Notably, the anti-SRPK1 monoclonal antibody used in these experiments is highly specific: on Western blots, it recognized a single band of ∼92 kDa which is in agreement with the relative molecular mass of the SRPK1 protein, as published by Gui et al. (15) . Furthermore, Western blotting experiments confirmed the specific down-regulation of SRPK1 expression by AS-SRPK1: there was a marked decrease in the level of SRPK1 protein detected on incubation of A2780 cells with AS-SRPK1 as compared with NS or MIS (Fig. 3i) ⇓ .

Now that we had shown that AS-SRPK1 was able to specifically down-regulate SRPK1 protein expression, we tested it for its effects on cisplatin sensitivity. First, we determined the optimum dose for AS-SRPK1 treatment. A2780 cells were incubated with increasing concentrations of AS-SRPK1 (1.4, 2.8, 5.6, and 11.2 μm) at a fixed dose of cisplatin (6 μm), and survival was estimated by monitoring cell proliferation in MTT assays (Fig. 4 ⇓ , inset). For the ODN dose range tested, the optimum concentration for reduction of cisplatin-induced growth inhibition appeared to be 5.6 μm AS-SRPK1 (Fig. 4 ⇓ , inset). Subsequently, we determined the shift in EC₅₀ value (i.e., the cisplatin concentration predicted to result in half-maximal cytotoxicity) at a fixed dose of 5.6 μm AS-SRPK1, as compared with the same dose of NS or MIS, by MTT assay. Repeatedly, AS-SRPK1 induced a 3- to 4-fold cisplatin resistance in A2780 cells, as evaluated against incubation with NS or MIS. Notably, the growth rate of A2780 cells in the absence of cisplatin was not decreased after treatment with AS-SRPK1, as compared with NS or MIS (data not shown). A typical experiment (of three performed) is shown in Fig. 4 ⇓ ; whereas incubation with NS, MIS, or delivery agent alone resulted in similar EC₅₀ values (1.5 ± 0.19, 2.2 ± 0.63 and 2.2 ± 0.30 μm cisplatin, respectively), AS-SRPK1 treatment led to a 3- to 4-fold increase in EC₅₀ (6.3 ± 0.70 μm cisplatin). Statistical analysis comparing NS, MIS, or delivery agent only versus AS-SRPK1 treatment indicated that the differences in EC₅₀ were significant at P < 0.0001, as estimated by Student’s t test. It can thus be concluded that down-regulation of SRPK1 protein by specific antisense ODNs rendered A2780 cells resistant to cisplatin, i.e., SRPK1 may function as a cisplatin sensitivity gene in this human ovarian carcinoma cell line.

Mechanistic Aspects and Implications for Clinical Studies.

To characterize the specificity and underlying mechanisms of the observed resistance of sky1Δ yeast cells, other DNA-damaging agents, including platinum-based drugs, will be used in future studies. Because, in our hands, sky1Δ strains showed normal growth characteristics, 5 and the growth rate of human A2780 cells was not decreased by incubation with specific ODNs, the cisplatin-resistant phenotypes were obviously not caused simply by distorted growth. Because we did not detect altered levels of yeast SKY1 RNA on cisplatin treatment, 5 the gene is probably not regulated by cisplatin at the transcriptional level. Sky1p (and SRPK1) activity could be regulated instead by autophosphorylation or posttranslational modification by upstream components.

As stated above, our data suggest that the kinase function is essential to the involvement of SR protein-specific kinases in the cytotoxicity of cisplatin. The S. cerevisiae SR protein encoded by NPL3 has been shown (14) to be phosphorylated by Sky1p both in vitro and in vivo and shuttles between the nucleus and the cytoplasm to deliver mRNA and/or proteins. It has been demonstrated (18, 19, 20, 21) that phosphorylation of Npl3p and mammalian SR proteins specifically modulates their protein-protein and protein-RNA interactions and localization. Sky1p and SRPK1 may thus be key players in pathways determining cisplatin sensitivity via mRNA and/or protein delivery by Npl3p or mammalian SR proteins, respectively.

In future studies, we will assess the relative contribution of pathways using SRPK1 to clinical cisplatin sensitivity and resistance. Low levels of cisplatin resistance are generally believed to be sufficient to cause lack of clinical responsiveness. Changes of <2-fold may account for treatment failure in human ovarian carcinoma xenografts (22) . Interestingly, SRPK1 is highly expressed in testis (23) , whereas, at the same time, testicular germ cell tumors are extremely sensitive to cisplatin-containing chemotherapy (1) . Indeed, constitutive expression of SKY1 from a high-copy vector clearly made sky1Δ yeast cells more sensitive to cisplatin, as compared with an appropriate SKY1⁺ control strain, 5 , 6 which suggests that SR protein-specific kinases are cisplatin sensitivity genes in the sense that they are actively involved in the cytotoxicity of cisplatin. In addition, we found that down-regulation of SRPK1 with AS-SRPK1 also made NT2 testicular cancer cells less sensitive to cisplatin, as compared with cells treated with control MIS. 5 Given these observations, we will monitor samples derived from both ovarian and testicular cancers, which are generally treated with cisplatin-containing regimens in the clinic (1) , for SRPK1 expression and correlation with clinical responsiveness.