Growth inhibition and induction of apoptosis in mesothelioma ... - Nature

Oncogene (2004) 23, 5032–5040

& 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00 www.nature.com/onc

Growth inhibition and induction of apoptosis in mesothelioma cells by selenium and dependence on selenoprotein SEP15 genotype Sinoula Apostolou1, Julian O Klein1, Yasuhiro Mitsuuchi1, Justin N Shetler1, Poulikos I Poulikakos1, Suresh C Jhanwar2, Warren D Kruger2 and Joseph R Testa*,1 1 Human Genetics Program, Fox Chase Cancer Center, 333 Cottman Avenue, Philadelphia, PA 19111-2497, USA; 2Department of Medicine, Memorial Sloan-Kettering Cancer Center, New York, NY, USA

Malignant mesotheliomas (MMs) are aggressive tumors derived from mesothelial cells lining the lungs, pericardium and peritoneum, and are often associated with occupational asbestos exposure. Suppression subtractive hybridization was used to identify genes differentially expressed in MM cells compared to normal mesothelial cells. A gene, SEP15, encoding a 15-kDa seleniumcontaining protein was isolated using this approach and was subsequently shown to be downregulated in B60% of MM cell lines and tumor specimens. A SEP15 polymorphic variant, 1125A, resides in the SECIS recognition element in the 30 -UTR and may influence the efficiency of Sec incorporation into the protein during translation. Since previous studies have implicated a potential role of the trace element selenium as a chemopreventive agent in animal models and in several types of human cancer, we investigated the effect of selenium on MM cells and its dependence on SEP15 genotype. Selenium was shown to inhibit cell growth and induce apoptosis in a dosedependent manner in MM cells but had minimal effect on normal mesothelial cells. However, MM cells with downregulated SEP15 or the 1125A variant were somewhat less responsive to the growth inhibitory and apoptotic effects of selenium than MM cells expressing wild-type protein. RNAi-based knockdown studies demonstrated that SEP15 inhibition makes sensitive MM cells more resistant to selenium. These data imply that selenium may be useful as a chemopreventive agent in individuals at high risk of MM due to asbestos exposure, although those with the 1125A polymorphism may be less responsive to the protective benefits of dietary selenium supplementation. Oncogene (2004) 23, 5032–5040. doi:10.1038/sj.onc.1207683 Published online 26 April 2004 Keywords: mesothelioma; selenium; chemoprevention; apoptosis; SEP15 genotype

Introduction Selenium is an essential trace element in redox pathways and has been utilized as a chemopreventive agent *Correspondence: JR Testa; E-mail: [email protected] Received 22 July 2003; revised 1 March 2004; accepted 1 March 2004; published online 26 April 2004

(Zhang et al., 1997; Ip, 1998; van Lieshout et al., 1998). Chemoprevention studies in animal model systems have shown that selenium is protective against a variety of carcinogens (El-Bayoumy, 1991). An inverse association has also been shown between selenium intake or tissue levels and several human cancers (Hunter et al., 1990; van den Brandt et al., 1993). The manner by which the development of tumors is reduced by selenium is not clear. However, some biological actions of selenium are known to be characteristic of its role as a component of selenoproteins. Selenium exists as the 21st amino-acid selenocysteine (Sec) in all known mammalian selenium-containing proteins. This amino acid is inserted into the growing polypeptide chain at the UGA codon within the ORF of the mRNA (Stadtman, 1996). In mammals, insertion of selenocysteine into the selenoprotein during translation, rather than termination of translation, requires the selenocysteine insertion sequence (SECIS) recognition element within the 30 UTR (Berry et al., 1993; Clark et al., 1996). SECIS elements are present in the mRNAs of all of the selenoproteins and possess a stem-loop structure containing an internal and apical loop of unpaired nucleotides and a conserved 4-bp domain within the stem. The functions of several selenium-containing proteins are known, one of which (the 15-kDa selenoprotein, Sep15) has recently been linked to certain murine cancers (Kumaraswamy et al., 2000). Malignant mesothelioma (MM) is a highly aggressive neoplasm of mesodermal origin often associated with asbestos exposure and is characterized by a long latency. The length of the latent period suggests that multiple genetic alterations may be required for tumorigenic conversion of a mesothelial cell. We used a modified form of suppression subtractive hybridization (SSH) (Diatchenko et al., 1996) to identify genes differentially expressed in MM cells and their normal mesothelial cell counterparts. One gene identified by this approach encodes a human 15-kDa selenoprotein (SEP15). This gene maps to 1p31, a region of frequent (55%) loss of heterozygosity (LOH) (Apostolou et al., 1999) in MM, which may implicate SEP15 as a candidate tumor suppressor gene (TSG). Previous studies have revealed two polymorphisms in the 30 -UTR SECIS of SEP15, at nucleotides 811 (C/T) and 1125 (G/A) (Gladyshev et al., 1998). Nucleotide 1125 resides within the apical loop of the SEP15 SECIS element and may potentially influence

Effects of selenium on mesothelioma cells and SEP15 status S Apostolou et al

5033

Figure 1 SEP15 expression in MM cell lines and tumor (T) samples. Real-time PCR analysis was performed on cDNA from three different normal mesothelial cell cultures and individual MM cell lines. The LOH and nucleotide 1125 status of each MM cell line is indicated. Expression of SEP15 is similar in the three mesothelial cell samples. SEP15 expression in five frozen MM tumor specimens (T) is also shown. Data are normalized to b-actin and are presented as fold-changes in SEP15 mRNA levels, relative to normal mesothelial cells

the efficiency of Sec incorporation (Gladyshev et al., 1998). Since LOH at 1p31 is common in MM and the trace element selenium is a candidate chemopreventive agent in several cancers, we investigated the possible role of SEP15 in MM. We report that expression of this gene is frequently downregulated in human MM cell lines and tumor specimens when compared with normal mesothelial cells. We demonstrate differential effects of selenium on the growth of normal versus malignant mesothelial cells and between MM cell lines possessing downregulated SEP15, wild-type SEP15 or the 1125A variant. These data indicate that SEP15 genotype and expression alterations affect SECIS function. The findings of this study suggest that SEP15 may be involved in the development of MM and could influence response to the chemopreventive properties of selenium.

regulated in 14 of 23 (60%) of the cell lines by two-fold or greater, when normalized to b-actin and compared with the average of three normal mesothelial cell samples (Figure 1). The range of SEP15 expression was found to be similar in the three individual mesothelial cell cultures. In total, 12 of the 20 MM cell lines with LOH at 1p31 showed downregulation of SEP15. SEP15 expression was not detectable in either normal mesothelial cells or MM cells by Northern blot analysis. This was most likely due to the levels of SEP15 being low in MM. However, SEP15 transcripts were detectable in cell lines derived from prostate, breast and colon tissues that were present on the same Northern blot. Expression of SEP15 was also downregulated in three of five MM tumor specimens when compared to that in normal mesothelial cells (Figure 1). The LOH status of these tumors could not be determined since their normal counterparts were not available for analysis.

Results Expression of SEP15 is frequently downregulated in a subset of MM cells We performed SSH to identify genes downregulated in MM cells compared to normal mesothelial cells. One of the genes identified as being downregulated in MM was SEP15. Further analysis of the expression of SEP15 in five MM tumor specimens and a panel of 23 MM cell lines was performed by real-time quantitative PCR. In total, 20 of these cell lines displayed LOH at 1p31, where SEP15 is located. Expression of SEP15 was down-

Differential effects of selenium on MM cell growth are associated with genotype and expression of SEP15 We first tested the effects of selenium on the growth of seven MM cell lines with varying SEP15 status and one normal mesothelial cell culture (7086A1). Three MM cell lines (Meso 8, 14 and 45) had downregulated SEP15 expression, two (Meso 22 and 34) possessed the 1125A allele and two (Meso 6 and 59) expressed only the wildtype 1125G allele. Cell growth was evaluated with increasing concentrations of selenium in the tissue culture media. The cells were incubated in selenium Oncogene


5034

(SeMet, Sec or Na2SeO3) supplemented media at concentrations between 0 and 500 mM for 48 and 72 h, and cell proliferation was assessed. Na2SeO3 and Sec were found to be consistently more potent growth inhibitors than SeMet (data not shown), consistent with reports by other groups (Menter et al., 2000). Figure 2 illustrates the concentration-dependent effects of Sec observed at 48 h. Sec-induced growth inhibition was observed in all MM cell lines but had little or no effect on normal mesothelial cells. Growth inhibition was most pronounced in Meso 6 and 59 cells, which possess wild-type SEP15. MM cell lines with the 1125A allele were less responsive to added selenium than were the wild-type 1125G cells. MM cell lines with reduced expression of SEP15 were least affected by increasing concentrations of Sec. Similar, more marked, effects were observed at the 72-h time point. Collectively, these findings suggest a selective effect of selenium on malignant mesothelial cells compared to their normal counterparts, and that the degree of drug response in MM cells is also influenced by SEP15 status. Induction of apoptosis by selenium varies depending on SEP15 status We also examined the effects of selenium on apoptosis in these same MM cell lines. MM cells were treated with 5 and 25 mM of Na2SeO3 or Sec for 48 and 72 h, and the early event of apoptosis involving the translocation of phospholipid phosphatidylserine from the inner part of the plasma membrane to the cell surface was assayed. At both concentrations of Na2SeO3, induction of apoptosis was much higher in MM cells with wild-type SEP15 protein (Meso 59) than in cells with either downregulated SEP15 (Meso 8) or the 1125A variant (Meso 34). Importantly, selenium induced only low levels of

apoptosis in normal mesothelial cells. Similar results were obtained with Sec-treated cells, further demonstrating a differential effect of selenium in normal versus malignant mesothelial cells and between MM cells with varying SEP15 status. The proportion of apoptotic MM cells treated with Na2SeO3 and Sec is summarized in Figure 3. Apoptosis is increased in cells transfected with wild-type SEP15, but not the 1125A variant, following treatment with selenium We next examined the effect of selenium on an MM cell line transiently transfected with two different SEP15 constructs, to test experimentally whether the differential effects of selenium described above are mediated through SEP15. Meso 8 cells, which express low levels of SEP15 and are resistant to the effects of selenium, were co-transfected with vector containing GFP and either wild-type SEP15 or 1125A variant SEP15 constructs. Cells were treated with 0, 5 or 25 mM of Sec for 72 h, and apoptotic levels were assayed by immunofluorescence using Annexin V-Cy3 and 200–250 cells per concentration of Sec were counted. Transfected cells were green, apoptotic cells were red, and yellow cells indicated apoptotic, transfected cells. Untreated cells and cells treated with 5 mM Sec displayed similar levels of apoptosis (3–4%). However, at a concentration of 25 mM Sec, the proportion of apoptotic cells was higher in cells transfected with the wild-type SEP15 construct (B16%) than in cells transfected with the 1125A SEP15 construct (4–6%) (Figure 4). Thus, the 1125A variant was demonstrated to be less responsive to added selenium than the wild-type protein. Expression of the SEP15 protein in transfected cells from individual constructs was shown to be similar by Western blot analysis (Figure 4).

Figure 2 Growth of MM cells treated with Sec at a concentration range of 1–500 mM for 48 h. Dose-dependent growth suppression by Sec of MM cell lines possessing wild type (Meso 6 and 59), 1125A polymorphic (Meso 22 and 34) or downregulated (Meso 8, 14 and 45) SEP15 is presented. MM cells with the polymorphic variant of SEP15 or downregulated SEP15 did not respond to Sec as efficiently as MM cells with wild-type SEP15. Growth of normal mesothelial cells (7086A1) was not affected by selenium. Data are presented as the mean of triplicate cultures following normalization to untreated cells. Bars, s.d. Oncogene


5035

Figure 3 Sensitivity of MM cells to Sec and Na2SeO3. The proportion of apoptotic cells following treatment with 0, 5 and 25 mM Sec for 72 h was quantified by calculating the percentage of annexin V-positive cells. MM cells with downregulated SEP15 displayed levels of apoptosis comparable to that of cells with the 1125A polymorphism. The percentage of apoptotic cells was highest in MM cells with wild-type SEP15. Apoptosis induction by selenium was minimal in normal mesothelial cells

specific siRNA prior to exposure to selenium. To monitor expression changes following siRNA transfection, we used real-time PCR to determine the extent of SEP15 inhibition. As shown in Figure 5a, SEP15 expression was markedly diminished by SEP15 siRNA. Cells transfected with an SEP15-specific siRNA exhibited increased cell proliferation following exposure to selenium compared to cells transfected with a nonspecific siRNA pool (Figure 5b). In addition, cells treated with SEP15 siRNA exhibited decreased apoptosis following selenium treatment at a concentration of 25 mM Sec (Figure 5c). Collectively, these data suggest that SEP15 expression affects the response of MM cells to growth inhibition and apoptosis induced by selenium.

Figure 4 Co-transfected Meso 8 cells were examined for the expression of 1125A or wild-type SEP15 following treatment with 0–25 mM Sec for 72 h. The numbers of transfected, apoptotic cells were counted and are presented as percentages. Bars, s.d. Expression of the SEP15 protein in transfected cells from individual constructs was also shown to be similar by Western blot analysis (insert)

SEP15 inhibition by siRNA makes sensitive MM cells more resistant to selenium We previously demonstrated that SEP15 is involved in cell proliferation and apoptosis following exposure to selenium in MM cells. To further verify that SEP15 plays a potentially significant role in cell proliferation and apoptosis subsequent to selenium treatment, sensitive MM cells with wild-type SEP15 (Meso 6) were transiently transfected with a pooled mixture of SEP15-

SEP15 genotype frequencies are similar in MM patients and the general population We determined the frequencies of the SEP15 genetic variations within the general human population, MM tumors and cell lines, and peripheral blood from MM patients. We initially performed SSCP analyses to determine the prevalence of nucleotide 1125 G or A in 60 MM cell lines, 44 of which exhibited LOH at 1p31. A total of 17 samples displayed band shifts indicating nucleotide changes (Figure 6). DNA sequencing demonstrated that each of these variants was due to an 1125 nucleotide alteration. The frequencies of the 1125G, 1125 G/A and 1125A genotypes in MM cell lines were 72, 13 and 15%, respectively. We also ascertained the distribution of the SEP15 allele frequencies in nonmalignant tissue (PBL) from MM patients and in the general human population using SSCP or RFLP analysis. The frequencies of the Oncogene


5036

Figure 5 Meso 6 cells transiently transfected with siRNA against SEP15. (a) SEP15 expression in siRNA transfected Meso 6 cells determined by real-time PCR analysis. SEP15 expression is decreased in cells transfected with SEP15-specific siRNA in comparison to cells transfected with a nonspecific siRNA pool. (b) Growth of Meso 6 cells treated with Sec at a concentration range of 1–25 mM for 48 h. MM cells transfected with SEP15-specific siRNA exhibited increased cell proliferation following exposure to selenium compared to cells transfected with a nonspecific siRNA control pool. Data are presented as the mean of triplicate cultures following normalization to control-treated cells. Bars, s.d. (c) Noticeable inhibition of apoptosis (red cells) was observed in cells transfected with SEP15 siRNA compared to control cells following Sec treatment at a concentration of 25 mM. Cells stained with DAPI are shown in blue

MM cases (61%) is due to LOH in malignant cells from MM patients. LOH analysis was performed on tumor biopsies from 24 of these patients, and similar patterns of allelic loss were observed in tumor tissue and derived MM cell lines (data not shown). Preferential loss of one allele versus the other did not appear to occur in tumor cells from MM patients exhibiting LOH at 1p31. The genotypes for all of the samples analysed are shown in Table 1. Finally, to determine if nucleotide changes occurred in the coding region of SEP15, we used SSCP to analyze cDNA from 20 MM cell lines displaying LOH at 1p31. cDNA from normal mesothelial cells was included as a control. Band shifts were not observed in any of the MM samples, indicating no nucleotide alterations in the coding region of SEP15.

Figure 6 SSCP mutation analysis of SEP15 in MM cell lines and tumors. SSCP was performed on cDNA and genomic DNA. Autoradiograph shows PCR products generated from the SECIS element in the 30 -UTR of SEP15 encompassing nucleotide 1125. Lanes 1 and 6, wild-type 1125G; lane 2, 1125A variant; lanes 3–5, heterozygous 1125G/A

1125G, 1125 G/A and 1125A genotypes in the general population were 46, 50 and 4%, respectively (Table 1). The frequencies in PBL from MM cases were 39, 61 and 0%. w2 analysis revealed that the difference in the distribution of 1125A and G alleles between MM blood samples and the general population was not statistically significant, suggesting that the 1125A SEP15 variant does not predispose to the development of MM. The low frequency of the heterozygous 1125 genotype in MM cell lines (13%) compared to that in PBL from

Discussion An abundance of data implicating a potential role of selenium as a chemopreventive agent in animal models

Table 1 Allelic distribution of 1125G/A alleles in human DNA samples 1125G Normal population (n ¼ 48) Blood from MM cases (n ¼ 36) MM tumors/cell lines (n ¼ 60) MM tumors/cell lines without LOH (n ¼ 16) MM tumors/cell lines possessing LOH (n ¼ 44) Oncogene

22 14 43 8 35

(46%) (39%) (72%) (50%) (80%)

Heterozygous 24 22 8 8 0

(50%) (61%) (13%) (50%) (0%)

1125A 2 0 9 0 9

(4%) (0%) (15%) (0%) (20%)


5037

and humans (Hunter et al., 1990; El-Bayoumy, 1991; van den Brandt et al., 1993; Zhang et al., 1997; Ip, 1998; van Lieshout et al., 1998) prompted our study of the protective effects of selenium in MM cells in connection with a gene encoding a selenium-containing protein, SEP15. In this investigation, SEP15 was found to be frequently underexpressed in MM cells relative to normal mesothelial cells, suggesting a possible role of SEP15 in the pathogenesis of MM. We have demonstrated that selenium has differential effects on cell growth and survival in MM cells compared to normal mesothelial cells. In addition, MM cells with a polymorphic variant of SEP15 or loss of expression of SEP15 did not respond to selenium as efficiently as MM cells with wild-type SEP15. Importantly, our data suggest that SEP15 is involved in inducing apoptosis in MM cells, since its inhibition by siRNA resulted in increased proliferation and decreased apoptosis in Meso 6 cells that possess wild-type SEP15. Since mutations were not detected in the coding region of SEP15, downregulation of this gene in MM cells may be caused by epigenetic mechanisms, such as aberrant promoter hypermethylation associated with inappropriate gene silencing (Jones and Baylin, 2002). Such epigenetic changes are common in cancer and are thought to affect multiple steps in tumor progression. To our knowledge, this is the first report of altered expression of SEP15 in a human tumor. However, aberrant expression of SEP15 has previously been reported in two rodent tumor types. Expression of Sep15 was undetectable in mouse prostate adenocarcinoma cells and liver tumors, whereas normal mouse prostate and liver tissues exhibited high levels of the Sep15 protein (Kumaraswamy et al., 2000). Attempts to confirm SEP15 expression in MM by northern analysis demonstrated that SEP15 was not detectable in either normal mesothelial cells or MM cells by northern blot analysis. This was most likely due to the levels of SEP15 being low in MM. However, SEP15 transcripts were detectable in cell lines derived from prostate, breast and colon tissues that were present on the same Northern blot. Overexpression of SEP15 was also observed in four of 23 MM cell lines. This may be due to the fact that MMs have a great deal of genetic variability. Although recurrent genetic alterations exist in MM, individual cell lines can also show considerable variability. Thus, it was not unexpected that SEP15 was not downregulated in all MM cases. Additionally, the number of primary and tumor samples (Figure 1) are too few to draw firm conclusions regarding SEP15 expression in primary tumors and will require further study in the future. However, the situation from the MM cell lines could be the same as in the MM tumors. Previous work has demonstrated that the SEP15 nucleotide at position 1125, within the apical loop of the SECIS element, can influence the efficiency of translation of the in-frame UGA codon (Kumaraswamy et al., 2000). Our cell proliferation and apoptosis data on selenium-treated MM cell lines possessing either wildtype SEP15 or the 1125A variant are consistent with the

fact that the nucleotide at position 1125 does indeed influence SECIS function (Figures 3 and 4). MM cells with the 1125A variant were less responsive to the antiproliferative and proapoptotic effects of selenium compared to cells with the wild-type allele. Tissue culture media is generally believed to be deficient in selenium when compared with the levels observed in human tissues and plasma (Diplock, 1993). Thus, the effects we detected in our in vitro studies are attributed to selenium supplementation. We also observed that different chemical forms of selenium can vary somewhat with respect to their effect on cell proliferation and apoptosis. Since tumor cells with the 1125A allele were less sensitive to the effects of selenium, this allele may result in a defect in SEP15 function or production of less of the protein. While our data suggest a correlation between selenocysteine sensitivity and SEP15 expression, it is also possible that other unknown genetic alterations differentially present in MM cell lines may influence selenocysteine sensitivity and SEP15 expression. However, RNAi-based knockdown studies with Meso 6 cells, which contain wild-type SEP15 and are sensitive to selenium, revealed that SEP15 downregulation makes sensitive MM cells more resistant to the compound (Figure 5). Moreover, transfection with only the wild-type SEP15 construct induced an increased apoptotic response to selenocysteine, whereas the SEP15 variant construct or vector alone did not (Figure 4). These data imply that at least some effects of selenium are mediated through SEP15 and that this gene has a significant role in selenium sensitivity in MM tumor cells. Importantly, normal mesothelial cells containing the wild-type allele were not affected by selenium at the concentrations used in this study. A similar differential in vitro effect of selenium on growth inhibition and apoptosis has been observed in prostate cancer cells compared with normal prostate cells (Menter et al., 2000). The SEP15 gene maps to human chromosome band 1p31. LOH involving the short arm of chromosome 1 is often observed in MM, with most cases showing allelic losses in the 1p21–31 region (Lee et al., 1996; Apostolou et al., 1999). Recurrent LOH at a discrete chromosomal region is generally considered an indication of the presence of a TSG, whose loss/inactivation contributes to tumorigenesis (Gray and Collins, 2000). Owing to its location at 1p31, we considered the possibility that SEP15 is a TSG. However, several facts argue against this possibility. First, LOH analysis of overlapping 1p deletions from a large series of MMs indicated that the minimally deleted (common) region, that is, 1p22, is proximal to the SEP15 locus (Lee et al., 1996). Second, no SEP15 mutations were observed in any of the MM cell lines we tested. Third, we did not observe a statistically significant correlation between SEP15 downregulation and LOH at 1p31 in our MM cases. These findings imply that SEP15 is not the critical target of 1p allelic loss in MM. However, the frequent downregulated SEP15 expression observed in MM suggests that this change may contribute to the pathogenesis of MM. Oncogene


5038

Genotype analyses indicated that the frequency of the 1125A allele is similar in MM patients and in the general population. While most of our MM cases showed LOH at the SEP15 locus, there did not appear to be a selection for either allele. However, the sample size was small. A difference in 1125G/A allele frequencies was recently reported in tumors from patients with either breast cancer or head and neck tumors when compared with cancer-free African-American controls (Hu et al., 2001). In these tumor types, the 1125A allele was associated with increased cancer risk. In summary, selenium has substantially different effects on normal and malignant mesothelial cells, exhibiting significant antiproliferative and proapoptotic activity against MM cells. In tumor cells, differences in SEP15 genotype or expression are associated with variation in response to selenium. Collectively, if our in vitro findings are confirmed in an in vivo model, these may provide a rationale for the possibility of using selenium as a chemopreventive agent in asbestos workers or other groups occupationally exposed to these carcinogenic mineral fibers, who are at high risk of developing MM. Moreover, SEP15 genotyping may be useful in these high-risk groups, since individuals with the wild-type (GG) genotype might be more likely to benefit from dietary selenium supplementation.

Materials and methods Cell lines and RNA preparation Normal human mesothelial cells (including 7086A1, Coriell Cell Repositories; NM23 and WTU3) and MM cell lines established from surgically explanted primary tumors were grown under the same conditions in RPMI 1640 media (Invitrogen) supplemented with 2 mM glutamine, 100 U of penicillin and streptomycin, and 10% FBS. For RNA extraction, cell lines were grown to 70–80% confluence prior to harvest. Total RNA was isolated from cell lines using TRIzol (Invitrogen).

AGG GCG TGG TGC GGA GGG CGG T-30 ; 50 -GAT CAC CGC CTT CCG-30 ). Suppression subtractive hybridization SSH was used to identify genes downregulated in MM cell lines compared to normal mesothelial cells. Two sets of subtractive hybridizations were performed: 40 ng of tester (containing adapters 1a/1b or 2a/2b) cDNA was combined with 2 or 4 mg of driver cDNA and purified. The ethanol precipitated mixture was then washed with 80% ethanol and air dried before being resuspended in 4 ml of hybridization buffer (50 mM HEPES, pH 8.3/0.5 M NaCl/0.02 mM EDTA, pH 8.0/10% PEG 8000). The mixtures were subsequently denatured and then hybridized for 8 h at 681C. Two of the first hybridization reactions between normal tester/MM cDNA driver were combined and mixed with an excess of each driver. This mixture, comprising the second hybridization, was further incubated for 20 h at 681C. Finally, the cDNAs were PCR-amplified using primers 1 and 2 (50 -GTA TAC GAC TCA CTA TAG GGC-30 ; 50 -TGT AGC GTG AAG ACG ACA GAA-30 ). To increase the amount of final difference product, nested PCR using nested 1 and 2 primers (50 -TCG AGC GGC CGC CCG GGC AGG T-30 ; 50 -AGG GCG TGG TGC GGA GGG CGG T-30 ) was performed. Product analysis Radiolabeled final difference products, representative of the mRNA present in normal mesothelial cells, but missing in the MM cell lines, were generated by PCR-amplification with nested 1 and 2 primers, then separated in 6% polyacrylamide/ 8 M urea gels. Following autoradiography, 40 bands, mostly common to each difference product were excised from the gel. The DNA was eluted then cloned into pGEM-T (Promega). The inserts were PCR-amplified using nested 1 and 2 primers, then purified and sequenced. Reverse northern hybridization (Hufton et al., 1999) was used to examine the differential expression of the identified clones. PCR products were electrophoresed on agarose gels. In addition, actin was loaded as a hybridization control. Replica blots were prepared then probed individually with radiolabeled DpnII cut cDNA from tester and each driver. Verification of the differential expression of a number of these clones in MM cell lines as compared to normal mesothelial cells was performed by RT–PCR.

Tester and driver preparation RNA utilized for the SSH procedure was purified using the Fast Track mRNA isolation kit (Invitrogen). cDNA subtraction was performed using a modification of the protocol described by Roberts et al. (2002). Briefly, double-stranded cDNA (5 mg) synthesized (Copy Kit cDNA kit, Invitrogen) from MM cell lines was digested with DpnII (New England Biolabs). cDNA restriction digests were purified and precipitated prior to being ligated to the R-Bgl-12/24 (50 -GAT CTG CGG TGA-30 ; 50 -AGC ACT CTC CAG CCT CTC ACC GCA-30 ) adapters. Ligation reactions were then diluted and PCR-amplified (951C, 1 min, 721C, 3 min for 35 cycles) using the R-Bgl-24 primer to obtain B30 mg/cell line. PCR products were purified, precipitated and digested with DpnII, prior to separation by electrophoresis. Products corresponding to 200– 2000 bp were gel purified to remove the adapters. Then 1.2 mg of tester cDNA was religated to a new set of adapters (adapter 1a/1b: 50 -GTA TAC GAC TCA CTA TAG GGC TCG AGC GGC CGC CCG GGC AGG T-30 ; 50 -GAT CAC CTG CCC G-30 ; or 2a/2b: 50 -TGT AGC GTG AAG ACG ACA GAA Oncogene

Real-time quantitative PCR of SEP15 First-strand cDNA synthesis was performed with 2 mg of total RNA and 500 ng of oligo-dT primer (Invitrogen), according to the manufacturer’s instructions. Real-time PCR analysis was performed on cDNA from MM samples using primers SEP15F (50 -GCA GCT CTT GTG ATC TTC TCG-30 ) and SEP15-R (50 -CTT GGA CTT GAG GGA ACC TTC-30 ) to amplify the cDNA together with the 50 -FAM-TCA ACC TGC TTC AGC TGG ATC CTG-BHQ1-30 probe (Biosource International). b-actin primers and probes were obtained from Perkin-Elmer/ Applied Biosystems. Amplification reactions were performed with 1 ml of cDNA, 3 ml of each of the specific primers (3 mM) and 4 ml of the probe (2 mM) added to Ready-To-Go beads (Amersham). All reactions were performed with the Smart Cycler System (Cepheid), and the thermal cycling conditions were as follows: 5 min at 941C, followed by 45 cycles of 941C for 15 s, 521C for 30 s and 681C for 30 s. To quantitate the amount of specific mRNA in the samples, a standard curve was generated for each run using 10-fold serially diluted


5039 plasmid containing SEP15 cDNA sequence (dilutions ranging from 1 fg to 1 ng). In addition, a standard curve was generated for b-actin ranging from 1 fg to 1 ng. This enabled standardization of the initial RNA content of a tissue relative to the amount of b-actin. Cell proliferation assay MM cells were plated into 96-well tissue culture plates (Costar) and allowed to establish monolayers (B48 h). Various concentrations of selenomethionine (SeMet), sodium selenite (Na2SeO3) or Sec were added to selected wells and incubated for 24, 48 or 72 h. Following incubation, the medium was removed and fresh medium with CCK-8 solution (Alexis Biochemicals), an indicator of proliferating cells, was added to the cells and incubated for 1 h. Absorbance was measured at 450 nm using a microplate reader. Apoptosis assay Normal mesothelial cells and MM cell lines were plated into four-well poly-D-lysine-coated microscope slide glass chamber slides (Costar) and allowed to establish monolayers. Cells were treated with 5 and 25 mM Na2SeO3 or Sec for 48 and 72 h. Following incubation the medium was removed and fresh medium with Annexin V-Cy3 (Medical and Biological Laboratories), an indicator of the induction of apoptosis by detecting phosphatidylserine on the cell surface, was added to the cells and incubated for 5 min. The cells were subsequently stained with diamidino-2-phenylindole (DAPI) and visualized under a Zeiss Axiophot fluorescence microscope using rhodamine and DAPI filters. Digital images of DAPI staining and fluorescein signals were captured with a cooled CCD camera and merged using Quips software. Generation of SEP15 constructs for functional analysis of the SECIS element Oligonucleotide primers 50 -AGC GAT GGC GGC TGG GCC GAG TGG-30 and 50 -GAT TTT TGA AAC TTT TTA TTT ATA TTT TGG-30 complementary to cDNA of human SEP15 (Accession number NM_004261) were designed with terminal BglII and XhoI sites, respectively. An entire cDNA encoding human wild-type (wt-SEP15cHA) or the 1125A polymorphic variant (poly-SEP15cHA) of SEP15 was amplified by RT– PCR utilizing total RNA from mesothelial cells. The nucleotide sequence encoding the hemagglutinin (HA) epitope (50 TAC CCT TAT GAT GTG CCA GAT TAT GCC TCT-30 ) was inserted into both constructs at the carboxyl terminus, nucleotide 491, using PCR with the primers listed above. Nucleotide sequences of the constructs were validated by sequence analyses. Transient transfection of SEP15 in MM cells Transient transfections were carried out with the MM cell line, Meso 8, using FuGENE6 transfection reagent (Roche Molecular Biochemicals), according to the manufacturer’s instructions. MM cells were plated onto six-well culture plates and incubated for at least 24 h prior to DNA transfection. Cells were then transfected with 1 mg of wt-Sep15cHA or poly-Sep15cHA expression vector. To monitor transfection, cells were co-transfected with a GFP-containing vector at a ratio of 1 : 2 and then incubated for 24 h. To quantitate apoptosis, cells were treated with Sec at concentrations of 0–100 mM for 72 h, followed by the Annexin V-Cy3 apoptosis assay. A total of 200 to 250 cells per concentration of Sec were counted.

siRNA transfection SMARTpools of siRNA specific for SEP15 (M-007249-00-50), or nonspecific control RNA duplexes (D-001206-13-05) were purchased from Dharmacon Inc. (Lafayette, CO, USA). Meso 6 and Meso 59 cells were plated into 10 cm dishes. After 24 h, cells were transfected with Oligofectamine reagent (Invitrogen) and gene-specific or nonspecific control siRNAs at a final concentration of 100 nM. After 12 h, cells were trypsinized and plated into 96-well plates for cell proliferation assays or into four-well chamber slides for apoptosis assays. DNA analysis of SEP15 For single-strand conformation polymorphism (SSCP) analysis, genomic DNA and cDNA from MM cell lines were studied. The entire coding sequence of the SEP15 gene was amplified with the following sets of primers (SEP15-1: F, 50 AGT GGG TGT CTG GTG CCG GCG TTT-30 and R, 50 TTG GAC TTG AGG GAA CCT TCC C-30 ; SEP15-2: F, 50 TAT GCA GGA GCT ATT CTT GAA G-30 and R, 50 -AAG GTA ACA AAA GGA TAG GAC-30 ). In total, 20 ng genomic DNA or 1 ml cDNA, 75 ng of each primer and 0.2 ml a32P-dCTP (800 Ci/mmol) (DuPont NEN) were used in a reaction of 10 ml. After an initial denaturation at 941C for 5 min, 35 cycles were carried out at: 941C for 1 min, 531C for 1 min and 721C for 1 min. For each primer pair utilized, genomic DNA from two peripheral blood lymphocyte (PBL) samples from normal individuals or normal mesothelial cells were routinely included as controls in each gel. Samples were analyzed in both 10% nondenaturing acrylamide gels containing 5% glycerol and 0.5 mutation detection enhancement acrylamide (FMC Products) gels. Band shifts were considered indicative of a potential mutation or polymorphism when they were reproducible on at least two separate occasions. Direct sequencing of PCR products displaying band shifts was performed following purification with the Wizard PCR Product Purification Kit (Promega), according to the manufacturer’s instructions. Restriction fragment length polymorphism (RFLP) analysis was used to study nucleotide 1125 in MM cell lines, tumors and blood samples, as described above. The SECIS element in the 30 UTR of SEP15 was amplified with primers (SECIS-F 50 ATA AGA TAT ACT GAG CCT CAA-30 and SECIS-R 50 TGG TCT TAC AA TGA TCA CTT-30 ). PCR amplification of a 474-bp product from genomic DNA of PBL controls was achieved with the following primers (SEPUTR-F, 50 -CAG ACT TGC GGT TAA TTA TGC-30 and SEPUTR-R, 50 -TGG TCT TAC AAA TGA TCA-30 ). PCR was performed at 951C for 3 min, followed by 35 cycles of 951C for 1 min, 521C for 1 min and 721C for 1 min 30 s. Amplified DNA was digested with BfaI (New England Biolabs), recognition sequence (50 -CTAG-30 ), to identify the nucleotide at position 1125. Nucleotide G at position 1125 results in the sequence 50 TTTCTAGCCTAA-30 , which is cleaved by BfaI, whereas an A results in a sequence not recognized by this enzyme. DNA digestion was evaluated by gel electrophoresis in 2.5% agarose. Acknowledgements Supported by the Mesothelioma Applied Research Foundation, National Institutes of Health Grants CA-45745 and CA06927, a gift from the Local #14 Mesothelioma Fund of the International Association of Heat and Frost Insulators & Asbestos Workers in memory of Hank Vaughan and Alice Haas, and by an appropriation from the Commonwealth of Pennsylvania. Oncogene


5040 References Apostolou S, De Rienzo A, Murthy SS, Jhanwar SC and Testa JR. (1999). Cell, 97, 684–686. Berry MJ, Banu L, Harney JW and Larsen PR. (1993). EMBO J., 12, 3315–3322. Clark LC, Combs Jr GF, Turnbull BW, Slate EH, Chalker DK, Chow J, Davis LS, Glover RA, Graham GF, Gross EG, Krongrad A, Lesher JL, Park HK, Sanders BB, Smith CL and Taylor JR. (1996). J. Am. Med. Assoc., 276, 1957–1963. Diatchenko L, Lau YF, Campbell AP, Chenchik A, Moqadam F, Huang B, Lukyanov S, Lukyanov K, Gurskaya N, Sverdlov ED and Siebert PD. (1996). Proc. Natl. Acad. Sci. USA, 93, 6025–6030. Diplock AT. (1993). Am. J. Clin. Nutr., 57, 256S–258S. El-Bayoumy K. (1991). Cancer Prevention, De Vita VT, Hellman S and Rosenberg SA (eds). J.B. Lippincott Co.: Philadelphia, PA, pp. 1–15. Gladyshev VN, Jeang KT, Wootton JC and Hatfield DL. (1998). J. Biol. Chem., 273, 8910–8915. Gray JW and Collins C. (2000). Carcinogenesis, 21, 443–452. Hu YJ, Korotkov KV, Mehta R, Hatfield DL, Rotimi CN, Luke A, Prewitt TE, Cooper RS, Stock W, Vokes, Dolan ME, Gladyshev VN and Diamond AM. (2001). Cancer Res., 61, 2307–2310.

Oncogene

Hufton SE, Moerkerk PT, Brandwijk R, de Bruine AP, Arends JW and Hoogenboom HR. (1999). FEBS Lett., 10, 77–82. Hunter DJ, Morris JS, Stampfer MJ, Colditz GA, Speizer FE and Willett WC. (1990). JAMA, 264, 1128–1131. Ip C. (1998). J. Nutr., 128, 1845–1854. Jones PA and Baylin SB. (2002). Nat. Rev. Genet., 3, 415–428. Kumaraswamy E, Malykh A, Korotkov KV, Kozyavkin S, Yajun Hu Y, Kwon S, Moustafa ME, Carlson BA, Berry MJ, Lee BJ, Hatfield DL, Diamond AM and Gladyshev VN. (2000). J. Biol. Chem., 275, 35540–35547. Lee WC, Balsara B, Liu Z, Jhanwar SC and Testa JR. (1996). Cancer Res., 56, 4297–4301. Menter DG, Sabichi AL and Lippman SM. (2000). Cancer Epidemiol. Biomarkers Prev., 9, 1171–1182. Roberts D, Williams SJ, Cvetkovic D, Weinstein JK, Godwin AK, Johnson SW and Hamilton TC. (2002). DNA Cell Biol., 21, 11–19. Stadtman TC. (1996). Annu. Rev. Biochem., 65, 83–100. van den Brandt PA, Goldbohm RA, van’t Veer P, Bode P, Dorant E, Hermus RJ and Sturmans F. (1993). Cancer Res., 53, 4860–4865. van Lieshout EMM, Ekkel MPC, Bedaf MMG, Nijhoff WA and Peters WHM. (1998). Oncol. Rep., 5, 959–963. Zhang Z, Kimura M and Itokawa Y. (1997). Biol. Trace Element Res., 57, 147–155.