Gene amplification and overexpression of protein ... - Nature

Oncogene (2006) 25, 5517–5526

& 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc

ORIGINAL ARTICLE

Gene amplification and overexpression of protein phosphatase 1a in oral squamous cell carcinoma cell lines L-C Hsu1,2, X Huang2,3,5, S Seasholtz3, DM Potter2,4 and SM Gollin2,3 1 Department of Obstetrics, Gynecology and Reproductive Sciences, School of Medicine, University of Pittsburgh, Magee-Womens Research Institute, Pittsburgh, PA, USA; 2University of Pittsburgh Cancer Institute, Pittsburgh, PA, USA; 3Department of Human Genetics, University of Pittsburgh Graduate School of Public Health, Pittsburgh, PA, USA and 4Department of Biostatistics, University of Pittsburgh Graduate School of Public Health, Pittsburgh, PA, USA

Gene amplification of chromosomal band 11q13 is observed frequently in oral squamous cell carcinomas (OSCC). Several genes have been identified in the 11q13 amplicon, including FGF3, FGF4, CCND1, EMS1 and TAOS1. Some of these genes show good correlation between gene copy number and gene expression, and are thought to play a role in driving 11q13 amplification. The PPP1CA gene, which encodes the catalytic subunit of serine/threonine protein phosphatase protein phosphatase 1a (PP1a), is also located in 11q13. Protein phosphatase 1a, one of the isoforms of PP1, regulates critical cellular events, such as cell cycle progression, and apoptosis. We sought to explore the possibility that PPP1CA was amplified and overexpressed in OSCC cells. Indeed, some OSCC cell lines had PPP1CA gene amplification, as analysed by fluorescence in situ hybridization. We have also demonstrated that PPP1CA gene copy number is increased in 21% of the OSCC cell lines determined by quantitative microsatellite analysis. PP1a RNA expression determined by quantitative reverse transcription–polymerase chain reaction was significantly higher in OSCC cell lines with 11q13 amplification compared to those without 11q13 amplification (P ¼ 0.011). The difference was even more significant between cell lines with at least three copies of the PPP1CA gene and those with less than three copies of the gene (P ¼ 0.00045). Relative PP1a protein levels were also significantly associated with PPP1CA gene copy number (P ¼ 0.014). Furthermore, knockdown of PP1a and/or cyclin D1 by small interfering RNA suppressed OSCC cell growth, at least in part by modulating pRB phosphorylation, resulting in G0 growth arrest. These data suggest that like the cyclin D1 gene, CCND1, amplification and overexpression of the PP1a gene, PPP1CA, may be involved in OSCC tumorigenesis and/or progression. Oncogene (2006) 25, 5517–5526. doi:10.1038/sj.onc.1209563; published online 17 April 2006 Correspondence: Dr L-C Hsu, Department of Obstetrics, Gynecology and Reproductive Sciences, School of Medicine, Magee-Womens Research Institute, 204 Craft Avenue, Pittsburgh, PA 15213, USA. E-mail: [email protected] 5 Current address: Stanford University School of Medicine, Center for Clinical Science Research, S1250 Stanford University, 269 Campus Drive, Stanford, CA 94305, USA. Received 3 October 2005; revised 3 March 2006; accepted 3 March 2006; published online 17 April 2006

Keywords: PP1a; cyclin D1; gene amplification; OSCC tumorigenesis

Introduction Amplification of chromosomal band 11q13 is frequently observed in human cancers, including breast (15%), lung (13%), bladder (21%) and esophageal cancers (50%), and is associated with a poor prognosis (Gaudray et al., 1992; A˚kervall et al., 1995; Dickson et al., 1995; Schuuring, 1995; Schwab, 1998; Gollin, 2001; Savelyeva and Schwab, 2001). Amplification of 11q13 occurs in the form of a homogeneously staining region in about 45% of oral squamous cell carcinomas (OSCC) (Lese et al., 1995; Jin et al., 1998; Gollin, 2001). Several genes have been identified in the 11q13 amplicon, including FGF3, FGF4, CCND1, EMS1 and TAOS1 (Gaudray et al., 1992; Dickson et al., 1995; Schuuring, 1995; Gollin, 2001; Huang et al., 2002). Some of them show good correlation between gene copy number and gene expression, and are thought to play a role in driving 11q13 amplification (Huang et al., 2002). It is believed that CCND1, which encodes cyclin D1, is one of the key oncogenes in the 11q13 amplicon. The human cyclin D1 gene was originally cloned as an oncogene involved in chromosomal breakpoint rearrangement in parathyroid adenomas (Motokura et al., 1991). To date, it is well known that amplification and overexpression of cyclin D1 are associated with the development of a subset of human cancers (reviewed by Fu et al., 2004). Cyclin D1 is the regulatory subunit of a CDK holoenzyme that phosphorylates and inactivates the retinoblastoma protein (pRB), and promotes the G1/S cell cycle progression. Cyclin D1 also has CDKindependent functions, such as regulation of cell differentiation, gene transcription, chromatin remodeling and cellular metabolism (reviewed by Fu et al., 2004). It has been demonstrated that antisense cyclin D1 inhibits proliferation of head and neck squamous cell carcinoma cells both in vitro and in vivo (Wang et al., 1998, 2001), providing evidence to support the hypothesis that gene amplification and overexpression of cyclin

Overexpression of PP1a in OSCC cells L-C Hsu et al

5518

D1 may promote OSCC cell growth and contribute to tumorigenesis. The PPP1CA gene, which encodes the catalytic subunit of serine/threonine protein phosphatase 1a (PP1a), one of the PP1 isoforms, also maps to chromosomal band 11q13 (Barker et al., 1990). The PP1 catalytic subunit can form complexes with many regulatory subunits (also called targeting subunits) in a mutually exclusive manner. The regulatory subunits then direct the PP1 catalytic subunit to specific subcellular compartments, target the catalytic subunit to its substrates, and regulate diverse cellular activities including cell cycle progression, apoptosis, growth factor signal transduction, centrosome separation and cellular metabolism (reviewed by Mumby and Walter, 1993; Shenolikar, 1994; Cohen, 2002; Ceulemans and Bollen, 2004). Protein phosphatase 1 is involved in mitotic progression including mitotic entry and mitotic exit (reviewed by Mumby and Walter, 1993; Cohen, 2002; Ceulemans and Bollen, 2004). It has been demonstrated that PP1a can dephosphorylate the tumor suppressors pRB (Alberts et al., 1993; Durfee et al., 1993) and BRCA1 (Liu et al., 2002), suggesting its involvement in regulating cell cycle checkpoint control function of these tumor suppressors. Protein phosphatase 1a also interacts with Bcl-2, which targets PP1a to dephosphorylate Bad, and may play a role in regulating apoptosis (Ayllon et al., 2001). It has been reported that PP1a is upregulated and PP1 activity is significantly higher in preneoplastic lesions of the liver as well as in hepatomas, suggesting that PP1a may be involved in hepatocarcinogenesis (Saadat et al., 1995; Imai et al., 1999). Another isoform of PP1, PP1g1 is overexpressed and may be involved in accelerated growth of breast

cancer cells (Sogawa et al., 1997). The contribution of PP1a to the pathogenesis of other cancer types has not been well documented yet. Protein phosphorylation and dephosphorylation control many cellular functions. Overexpression or mutation of protein kinases has been found frequently in various types of cancer. In contrast, deregulation of protein phosphatases has not been observed or studied as frequently. Here, we sought to explore the possibility that the PPP1CA gene is amplified and overexpressed in OSCC cells, and that overexpression of PP1a, together with cyclin D1, may contribute to OSCC tumorigenesis.

Results PPP1CA gene amplification in oral squamous cell carcinoma cells Approximately 50% of OSCC cell lines are characterized by chromosomal band 11q13 amplification (Lese et al., 1995; Jin et al., 1998; Gollin, 2001; Huang et al., 2002). The 11q13 amplicon core contains 13 genes, including CCND1, TAOS1, FGF4, FGF3 and EMS1. The PPP1CA gene (adjacent to the D11S1889 microsatellite marker) is proximal to the CCND1 gene and is located just outside the 11q13 amplicon core, as illustrated in Figure 1a. To test whether the PPP1CA gene is amplified in OSCC cell lines, the RPCI-11 BAC clone 678D20, which contains PPP1CA (confirmed by PCR using primers specific for the PPP1CA gene), was labeled with SpectrumOranges and used as a probe for fluorescence in situ hybridization (FISH), along with a SpectrumGreens-labeled chromosome 11 centromere probe (CEP11, Vysis Inc.) as a control for chromosome

Figure 1 PPP1CA gene amplification in oral squamous cell carcinoma (OSCC) cells. (a) The PPP1CA gene is located proximal to the CCND1, TAOS1, FGF4, FGF3 and EMS1 genes, which map to the 11q13 amplicon core. (b) The PPP1CA gene is amplified in some OSCC cell lines. Fluorescence in situ hybridization (FISH) was performed using the RPCI-11 BAC clone 678D20, which contains the PPP1CA gene. Oncogene


5519

copy number. The PPP1CA gene was considered amplified when the ratio of SpectrumOrange/SpectrumGreen (R/G) signals was X3 in at least 10% of the cells examined. As shown in Figure 1b and Table 1, the majority of UPCI:SCC032 cells contained equal copy numbers of the PPP1CA gene and CEP 11 (71% of cells with PPP1CA/CEP11 ¼ 1). In contrast, UPCI:SCC084 and UPCI:SCC103 cells showed multiple copies of the PPP1CA gene in metaphase spreads and interphase nuclei (71% of UPCI:SCC084 and 48% of UPCI:SCC103 cells with PPP1CA/CEP11 X3, highlighted in bold in Table 1). Of the 10 UPCI:SCC cell lines analysed by PPP1CA FISH, seven cell lines had 11q13 amplification as determined previously by quantitative microsatellite analysis (QuMA) and FISH (Huang et al., 2002), and two of the seven cell lines (UPCI:SCC084 and UPCI:SCC103) also had PPP1CA amplification. Table 1 UPCI:SCC

029B 032 072 077 084 099 103 131 136 142

Correlation between PPP1CA gene copy number and RNA expression We then performed QuMA (Ginzinger et al., 2000) using a microsatellite marker D11S1889, which is approximately 144 kb distal to the PPP1CA gene, to determine PPP1CA gene copy number in 28 OSCC cell lines. Fourteen of these cell lines are 11q13 Amp þ and 14 are Amp as determined by QuMA for genes in the 11q13 amplicon core including CCND1 (Huang et al., 2002). The results are illustrated in Figure 2. We define those cell lines with more than three copies of the PPP1CA gene as having copy number increase, which can be due to gene amplification and/or gain of chromosome number. Based on this criterion, six of the 28 OSCC cell lines (21%) showed an increase in PPP1CA gene copy number (X3). All six cell lines also had 11q13 amplification. Quantitative reverse transcription–polymerase chain reaction (RT–PCR) was performed to

Distribution of PPP1CA FISH in OSCC cell lines

% Cells with relative loss (R/Go1)

% Cells with equal copy number (R/G ¼ 1)

% Cells with relative gain (1oR/Go3)

% Cells with amplification (R/GX3)

11q13 amplification by QuMA and FISH

17 20 8 26 2 6 0 1 8 5

74 71 76 73 5 75 8 41 89 70

9 9 16 1 22 18 43 58 3 22

0 0 0 0 71 1 48 0 0 3

+ + + + + + +

Abbreviations: (R) SpectrumOrange, PPP1CA; (G) SpectrumGreen, CEP11.

Figure 2 Copy number and relative RNA expression of the PPP1CA gene in OSCC cell lines determined by quantitative microsatellite analysis (QuMA) and quantitative reverse transcription–polymerase chain reaction (RT–PCR). 11q13 amplification status was previously determined by QuMA for genes in the 11q13 amplicon. Correlation coefficient (r) of gene copy number and relative RNA expression is indicated in the graph. Oncogene


5520

Figure 3 Protein phosphatase 1a (PP1a) protein levels in OSCC cell lines determined by Western analysis. PP1a expression levels were first normalized with g-tubulin protein levels and relative PP1a protein levels were calculated by dividing the normalized level with that in UPCI:SCC104, which expressed the least PP1a.

determine RNA expression in these OSCC cell lines. Protein phosphatase 1a RNA levels in the OSCC cell lines were normalized with the average expression level in normal human oral keratinocytes (n ¼ 4). UPCI: SCC084, which had the highest gene copy number (6.84) also expressed the highest level of PP1a RNA (10.5). Indeed, five out of six OSCC cell lines with an increase in PPP1CA gene copy number expressed more than threefold PP1a RNA compared to the average PP1a RNA in normal human oral keratinocytes. Thus, copy number of the PPP1CA gene correlated very well with relative PP1a RNA expression in OSCC cell lines as illustrated in the right panel of Figure 2 (Pearson’s correlation coefficient, r ¼ 0.884, 95% confidence interval 0.780–0.988). Correlation between PPP1CA gene copy number and protein phosphatase 1a protein level To further confirm the correlation between the PPP1CA gene copy number and PP1a expression in OSCC cell lines, we determined PP1a protein levels by Western blot analysis in four OSCC cell lines with more than three copies and four OSCC cell lines with less than three copies of the PPP1CA gene. Protein phosphatase 1a expression levels were normalized with g-tubulin protein levels. As shown in Figure 3, all four OSCC cells with higher PPP1CA gene copy number expressed more PP1a protein than those with less than three copies of the gene. Overexpression of protein phosphatase 1a in oral squamous cell carcinoma cell lines is significantly associated with 11q13 amplification and PPP1CA gene copy number The relative PP1a RNA expression was 3.5472.81 in 11q13 Amp þ OSCC cell lines (n ¼ 14) compared to 1.6270.66 in 11q13 Amp OSCC cells (n ¼ 14), indicating an association between 11q13 amplification and PP1a RNA expression (P ¼ 0.011). As expected, the difference in PP1a RNA expression was even more significant between cell lines with PPP1CA gene copy number X3 (5.6673.24, n ¼ 6) and PPP1CA gene copy number o3 (1.7470.64, n ¼ 22) (P ¼ 0.00045). LikeOncogene

Table 2 Overexpression of PP1a in OSCC cell lines is significantly associated with 11q13 amplification and PPP1CA gene copy number n

P-value

Relative PP1a RNA expression 11q13 Amp+ 3.5472.81 11q13 Amp 1.6270.66

14 14

0.011

Relative PP1a RNA expression PPP1CA X3 copies 5.6673.24 PPP1CA o3 copies 1.7470.64

6 22

0.00045

Relative PP1a protein levels PPP1CA X3 copies 4.4370.67 PPP1CA o3 copies 1.4570.54

4 4

0.014

wise, relative PP1a protein levels were also significantly associated with PPP1CA gene copy number (4.4370.67 in cells with PPP1CA gene copy number X3, n ¼ 4, vs 1.4570.54 in cells with PPP1CA gene copy number o3, n ¼ 4, P ¼ 0.014) (Table 2). PPP1CA gene amplification also occurred in one of 15 primary OSCC tumors analysed (copy number 3.74), indicating that the findings in the OSCC cell lines were not due to a cell culture artifact. Protein phosphatase 1a regulates oral squamous cell carcinoma cell growth To determine whether gene amplification/copy number increase of PPP1CA and overexpression of PP1a contributed to OSCC cell growth, we compared the growth of four OSCC cell lines, two that express low levels of PP1a (UPCI:SCC036 and UPCI:SCC104) and two that express high levels of PP1a (UPCI:SCC103 and UPCI:SCC131). Cells were seeded in 24-well plates and cell growth was determined by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay for 4 consecutive days. The growth curves illustrated in Figure 4 revealed that UPCI:SCC103 and UPCI:SCC131 cells grew faster than UPCI:SCC036 and UPCI:SCC104, suggesting that overexpression of PP1a may provide a growth advantage to OSCC cells. Many genetic changes might have occurred in these tumor cell lines that influence cell growth. To directly address the


5521

effect of PP1a expression on cell growth, we transiently introduced PP1a small interfering RNA (siRNA) into two OSCC cell lines that overexpressed PP1a, UPCI:SCC103 and UPCI:SCC131. Protein phosphatase 1a protein levels were then monitored by Western analysis. Knockdown of PP1a protein by siRNA transfection was very efficient, occurring within the first 3 days following transfection. Protein levels were o30% of those in control siRNA-transfected cells (Figure 5a). The PP1a knockdown effect was slightly diminished 4 days after transfection and PP1a levels were 30–50% of those in control siRNA-transfected cells (Figure 5a). When transfected with a green fluorescent protein (GFP) expression vector, pEGFP-N1, under the same transfec-

Figure 4 Oral squamous cell carcinoma (OSCC) cell growth. UPCI:SCC036 and UPCI:SCC104 expressed low levels of PP1a and UPCI:SCC103 and UPCI:SCC131 expressed high levels of PP1a protein.

tion conditions, more than half of transfected OSCC cells expressed GFP, indicating that the transfection efficiency in OSCC cells was at least 50% (data not shown). Cell growth was determined by the MTT assay and by cell counting two to four days after transfection as described in Materials and methods. As shown in Figure 5b, PP1a siRNA-transfected cells consistently grew slower than control siRNA-transfected cells, either by the MTT assay (an example of the UPCI:SCC131 experiment is illustrated in the left panel) or by cell counting (an example of the UPCI:SCC103 experiment is illustrated in the right panel). The differences in cell growth between control and PP1a siRNA-transfected cells were statistically significant in both cell lines (Po0.0001). Thus, knockdown of PP1a suppressed OSCC cell growth. Protein phosphatase 1a cooperates with cyclin D1 to promote oral squamous cell carcinoma cell growth Cyclin D1, encoded by the CCND1 gene, has been considered the driving force behind 11q13 amplification (Dickson et al., 1995; Schuuring, 1995). Overexpression of cyclin D1 is associated with human tumorigenesis and cancer metastasis (Fu et al., 2004). Antisense cyclin D1 inhibits proliferation of head and neck squamous cell carcinoma cells (Wang et al., 1998, 2001). To determine whether there is a combinatorial effect of PP1a and cyclin D1 on OSCC cell growth, UPCI:SCC131 cells were transfected with cyclin D1 and/or PP1a siRNA, subcultured 1 day after transfection, and harvested for Western analysis 1 day later (2 days after transfection) or subjected to cell growth assays 2 days later (3 days after transfection). Protein phosphatase 1a was knocked down to B30% of control levels when transfected with

Figure 5 Knockdown of Protein phosphatase 1a (PP1a) suppresses OSCC cell growth. (a) Western analysis of PP1a in small interfering RNA (siRNA)-transfected UPCI:SCC131 and UPCI:SCC103 cells 2–4 days after transfection. C: control siRNA, PP1a:PP1a siRNA. g-tubulin was used as a loading control. (b) Cell growth determined by the 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyl-2H-tetrazolium bromide (MTT) assay (left panel) or cell counting (right panel). The differences in cell growth curves between control and PP1a siRNA-transfected cells were statistically significant (Po0.0001). Oncogene


5522

80 pmol of PP1a siRNA alone and to B45% of control levels when transfected with 40 pmol of PP1a siRNA together with 40 pmol of cyclin D1 siRNA (Figure 6c). Cyclin D1 siRNA transfection did not affect PP1a protein levels (Figure 6c). Cyclin D1 was reduced to 37% of control levels after transfection with 80 pmol of cyclin D1 siRNA, and reduced to B45% of control levels when transfected with 40 pmol of cyclin D1 siRNA together with 40 pmol of PP1a siRNA (Figure 6c). Protein phosphatase 1a siRNA transfection did not cause a substantial change in cyclin D1 protein levels (Figure 6c). Cyclin D1 siRNA transfection

suppressed UPCI:SCC131 cell growth compared to control siRNA-transfected cells by as determined by either the MTT assay (Figure 6a) or cell counting (Figure 6b). Protein phosphatase 1a siRNA transfection also suppressed UPCI:SCC131 cell growth. Interestingly, when cyclin D1 siRNA was co-transfected with PP1a siRNA, the growth suppression effect was more dramatic (Figure 6a and b). Thus, knockdown of both PP1a and cyclin D1 cooperatively inhibits OSCC cell growth, suggesting that PP1a may have a combinatorial effect with cyclin D1 and promote cell proliferation in OSCC cells. Protein phosphatase 1a regulates oral squamous cell carcinoma cell growth in part by modulating pRB phosphorylation and cell cycle progression The cyclin D1/CDK holoenzyme can phosphorylate and inactivate pRB (reviewed by Fu et al., 2004), and consequently promotes the G1/S transition. In contrast, it has been reported that PP1a can dephosphorylate and activate pRB (Alberts et al., 1993; Durfee et al., 1993). We next determined the pRB phosphorylation status in siRNA-transfected UPCI:SCC131 cells. Surprisingly, PP1a or cyclin D1 siRNA alone or in combination induced hypophosphorylation of pRB as shown in Figure 6d, a process that may result in G0 growth arrest. Consistent with this finding, PP1a and/or cyclin D1 siRNA-transfected cells had more resting G0 cells as assessed by immunofluorescence staining with Ki-67 antibody, which labels proliferating cells (Scholzen and Gerdes, 2000). G0 cells were defined by negative staining with Ki-67 (Figure 7a). Cell cycle analysis by flow cytometry also showed an increase in G0/G1 peaks in PP1a and/or cyclin D1 siRNA-transfected cells (Figure 7b). Thus, like cyclin D1, PP1a may regulate OSCC cell growth by modulating pRB phosphorylation.

Discussion

Figure 6 Knockdown of both protein phosphatase 1a (PP1a) and cyclin D1 (CCND1) cooperatively inhibits OSCC cell growth in part by modulating pRB phosphorylation. (a) Growth assay by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT). (b) Growth assay by cell counting. Significant differences in cell growth between C (control) and PP1a or cyclin D1 siRNAtransfected cells are marked with ‘*’(Po0.01). (c) Western analysis of PP1a and cyclin D1. (d) Western analysis of hypophosphotylated pRB (HypoRB) and pRB. g-tubulin was used as a loading control. Oncogene

We provide here the first evidence that the PPP1CA gene is amplified and overexpressed in OSCC cells. An increase in PPP1CA gene copy number (X3) was observed in six of 28 (21%) cell lines analysed by QuMA, a real-time PCR-based method that is suitable for accurately and efficiently screening a large number of samples (Ginzinger et al., 2000). Four OSCC cell lines that showed an increase in PPP1CA gene copy number by QuMA were also analysed by FISH. Two of them (UPCI:SCC084 and UPCI:SCC103) displayed PPP1CA gene amplification, while two other cell lines (UPCI:SCC072 and UPCI:SCC131) did not show PPP1CA amplification according to the FISH results (see Table 1). This discrepancy between results from QuMA and FISH is most likely due to the fact that UPCI:SCC072 and UPCI:SCC131 had many copies of chromosome 11, which technically is not considered gene amplification, although it clearly represents extra copies of the genes. The frequency and magnitude of PPP1CA gene amplification is much lower than genes mapped within


5523

Figure 7 Knockdown of protein phosphatase 1a (PP1a) and cyclin D1 induces growth arrest in G0 phase. (a) PP1a and/or cyclin D1 siRNA transfection induced an increase in Ki-67() G0 cells compared to C (control) siRNA-transfected cells. Significant differences in G0 cells between C (control) and PP1a or cyclin D1 siRNA-transfected cells are marked with ‘*’(Po0.0001). (b) Cell cycle distribution of C (control), PP1a and/or cyclin D1 siRNA-transfected UPCI:SCC131 cells.

the 11q13 amplicon core, such as CCND1 (Huang et al., 2002). One may argue that PPP1CA gene amplification occurred at low frequency simply because of its proximity to the 11q13 amplicon. PPP1CA copy numbers derived from FISH using a PPP1CA probe and from QuMA using a microsatellite marker D11S1889, which is approximately 144 kb distal to PPP1CA (cen-PPP1CA-D11S1889-CCND1-Tel), were very close in OSCC cells, with the exception of UPCI:SCC103. The average PPP1CA gene copy number in UPCI:SCC103 cells was 13.46 determined by FISH, and was 3.78 determined by QuMA. This discrepancy in gene copy numbers derived from FISH and QuMA suggested a possibility of a separate amplification peak at the PPP1CA locus, which was not detected by QuMA using D11S1889, suggesting that PPP1CA could be a driving force for 11q13 amplification as well. Alternatively, it simply may reflect the heterogeneity in PPP1CA copy number in the cell line, since FISH is carried out on a cell-by-cell basis and QuMA averages copy number across the population, which is often quite heterogeneous. Quantitative RT–PCR and Western analysis revealed a very good correlation between PPP1CA gene copy number, PP1a RNA, and protein expression. Overexpression of genes is likely a consequence of the increased gene copy number, which in turn could be due to an increase in chromosome copy number and/or gene amplification. Tumor cells are usually aneuploid. The majority of OSCC cells are near triploid (Gollin, 2001). Even though some OSCC cell lines do not have PPP1CA amplification, they may have multiple copies of chromosome 11. This may explain why PP1a RNA expression levels in most of the OSCC cell lines were higher than in normal keratinocytes (see Figure 2). Thus, an increase in PPP1CA gene copy number, which

is partly due to PPP1CA gene amplification and partly due to an increase in chromosome 11 copy number, is significantly associated with PP1a overexpression in OSCC cells. We have also found PPP1CA gene amplification in one of 15 primary OSCC tissues, which supports our findings in the OSCC cell lines. The frequency was lower than in OSCC cell lines, possibly due to the fact that gene amplification and overexpression of PP1a leads to a growth advantage in cell culture and/or contamination of normal tissues in primary tumor samples, which reduces the apparent copy number detected by QuMA. Several protein phosphatases function as tumor suppressor proteins, such as PTEN (reviewed by Chu and Tarnawski, 2004) and PP2A (Janssens et al., 2005). Protein phosphatase 1 plays diverse biological functions in different cell types. Alterations of the PPP1R3 gene, which encodes a PP1 regulatory subunit, have been found in a variety of human cancers, suggesting that it may be a tumor suppressor gene (Kohno et al., 1999). Protein phosphatase 1a can dephosphorylate and activate the tumor suppressor protein pRB, and therefore may function as a tumor suppressor. In contrast, PP1a also dephosphorylates the breast and ovarian tumor suppressor protein BRCA1 and may have an oncogenic effect since phosphorylated BRCA1 is considered to be the active form. It has also been reported that overexpression and/or increase in PP1 activity is associated with accelerated growth of malignant cells (Saadat et al., 1995; Imai et al., 1999). Whether PP1a is a tumor suppressor gene or an oncogene remains unclear. In this report, we tested whether PP1a regulate OSCC cell growth using an siRNA approach, and have demonstrated that similar to cyclin D1, PP1a promotes cell growth in OSCC cells. Furthermore, a double knockdown experiment using cyclin D1 and PP1a Oncogene


5524

siRNAs resulted in combinatorial growth inhibition in OSCC cells. The mechanism(s) by which cyclin D1 and PP1a cooperatively promote cell growth is not clear and will be the focus of future studies. The cyclin D1/CDK holoenzyme can phosphorylate and inactivate pRB (reviewed by Fu et al., 2004), whereas, PP1a can dephosphorylate and activate pRB (Alberts et al., 1993; Durfee et al., 1993). It seems unlikely that both cyclin D1 and PP1a regulate OSCC cell growth through modifying pRB phosphorylation. However, we found that both knockdown of PP1a and cyclin D1 induced hypophosphorylation of pRB and an increase in the G0 cell population (Figures 6d and 7). Thus, modulation of pRB phosphorylation may at least in part account for growth regulation by PP1a. It has been reported that PP1a regulates both mitotic entry and exit (reviewed by Mumby and Walter, 1993; Cohen, 2002; Ceulemans and Bollen, 2004). We have demonstrated that PP1a can dephosphorylate BRCA1 (Liu et al., 2002) and may compromise the G2 checkpoint function of BRCA1, thereby promoting mitotic entry. It is likely that in addition to the effect on pRB, overexpression of PP1a may also result in premature mitotic entry and/or untimely exit from mitosis, and consequently may promote cell growth under the circumstances that some checkpoint control proteins, such as p53, are inactivated, and thereby contribute to tumorigenesis. Taken together, our data suggest that the PPP1CA gene may also be a target for gene amplification and overexpression, and may be involved in OSCC tumorigenesis and/or progression in cooperation with other co-amplified genes in 11q13, such as CCND1. We are currently investigating whether overexpression of PP1a also occurs in other cancer types. Protein kinases and phosphatases are good targets for drug development. Thus, our finding may provide a basis for a new cancer therapeutic strategy for oral cancer and may be other cancers as well.

Materials and methods Cell culture Oral squamous cell carcinoma cell lines were cultured using methods described previously (Huang et al., 2002). Normal human oral keratinocytes were cultured from tissues obtained from patients undergoing uvulopalatopharyngoplasty at the University of Pittsburgh Medical Center (consent obtained through the Head and Neck SPORE at the University of Pittsburgh under their Institutional Review Board Guidelines) as described (Park et al., 1991). Fluorescence in situ hybridization A RPCI-11 BAC clone 678D20 containing the PPP1CA gene was purchased from Invitrogen (Carlsbad, CA, USA) and confirmed by PCR using primers derived from the CA repeat D11S2072 (NCBI locus HSU10996), a marker within 100 kb of the PPP1CA gene (forward primer: 50 -CAAACCATGGCTT CTGTTAA-30 , reverse primer: 50 -CTCAGTGTTACATTATA AGT-30 , product size 184 bp) and primers derived from the Oncogene

PPP1CA gene (forward primer: 50 -CACACCACCCTGTGC CCCAGATGAT-30 , reverse primer: 50 -AAGGTCCATGTTC CCCGTGACAGGT-30 , product size 109 bp) (UniSTS:23991). The BAC DNA was isolated and labeled with SpectrumOranges-deoxyuridinetriphosphate using the Vysis nick translation kit (Vysis, Inc., Downers Grove, IL, USA). A CEP11 probe labeled with SpectrumGreens from Vysis was used for enumerating the chromosome 11 centromere. The labeled BAC clone and CEP11 were then used as probes for FISH using standard procedures described previously (Shuster et al., 2000; Reshmi et al., 2004). Approximately 200 interphase nuclei were analysed for each sample. DNA and RNA extractions Cell line DNA was isolated by using the DNeasy tissue kit (Qiagen, Chatsworth, CA, USA) and quantified by spectrophotometry. RNA was extracted using the Trizols reagent according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA, USA) and purified using the RNeasy mini kit (Qiagen). RNA was then treated with DNase using the DNAFree kit (Ambion, Austin, TX, USA) and quantified by spectrophotometry. Quantitative microsatellite analysis Analysis of DNA copy number of the PPP1CA gene at microsatellite locus D11S1889 was carried out according to the procedure described earlier (Ginzinger et al., 2000; Huang et al., 2002). The PCR primer sequences for D11S1889 are: forward 50 -AGCTGGACTCTCACAGAATG-30 , reverse 50 -CAAGAGGCTGGTAGAAGGTG-30 . The TaqMan CA-repeat fluorogenic probe used for all loci consisted of the following sequence: 50 FAM (6-carboxy fluorescein)TGTGTGTGTGTGTGTGTGTGT-TAMRA (6-carboxy tetramethyl rhodamine) 30 . All of the probes and primers for QuMA were purchased from Integrated DNA Technology (Coralville, IA, USA). Copy number X3 was considered gene copy number increase. Quantitative reverse transcription–polymerase chain reaction TaqMan primers and probe were designed with the PRIMER EXPRESS V.2.0.0 program (Applied Biosystems, Foster City, CA, USA). Polymerase chain reaction primer sequences for the PPP1CA are: forward 50 -TGTGGCGAGTTTGACA ATGC-30 ; reverse 50 - GGGCTTGAGGATCTGGAAAG-30 . The probe sequence is 50 FAM-CGCCATGATGAGTGTGG ACGAGACC-TAMRA. Ribosomal 18S RNA was used as an endogenous control. 18S RNA primer and probe sequences were the same as published (Huang et al., 2002). The reverse transcriptions were carried out as described earlier (Huang et al., 2002). No-reverse transcriptase controls were carried out for the highest RNA input each time. Quantitative PCR (QPCR) was performed on the cDNA using the ABI 7700 Sequence Detection Instrument (Applied Biosystems, Foster City, CA, USA) and analysed using the relative quantitation method (Perkin-Elmer Applied Biosystems User Bulletin No. 2, 1997). For the QPCR, the final concentrations of the reaction components were as follows: 1 PCR buffer A (Applied Biosystems), 300 nM each dNTP, 3.5 mM MgCl2, 0.06 U/ml Amplitaq Gold (Applied Biosystems), 500 nM (100 nM for 18S RNA) primers and 200 nM probe (100 nM 18S RNA). The thermocycler conditions were 951C Taq activation for 12 min and 40 cycles (30 cycles for 18S RNA) of 951C denaturation for 15 s followed by 601C anneal/extend for 60 s. Expression levels were normalized with the average expression level in normal keratinocytes (n ¼ 4).


5525 Western analysis Exponentially growing OSCC cells were harvested by trypsinization, lysed in RIPA buffer, and subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis, followed by Western blotting to nitrocellulose membrane as described previously (Hsu and White, 1998). Protein phosphatase 1a (B37 kDa) was detected by a goat polyclonal anti-PP1a C-19 (Santa Cruz Biotech, Santa Cruz, CA, USA), cyclin D1 (B36 kDa) was detected by a mouse monoclonal antibody HD11 or DCS-6 (Santa Cruz Biotech), hypophoshorylated pRB was detected by a mouse monoclonal antibody G99-549 (BD Biosciences, San Jose, CA, USA), total pRB regardless of phosphorylation was detected by a mouse monoclonal antibody G3-245 (BD Biosciences), and g-tubulin was detected by a mouse monoclonal antibody GTU-88 (Sigma, St Louis, MO, USA). Quantitative analysis was performed using Quantity One software (BioRad, Hercules, CA, USA). Small interfering RNA transfection and growth assays Target-specific 20–25 nucleotide siRNAs for PP1a and cyclin D1, as well as control siRNA, which is a non-targeting 20–25 nucleotide siRNA designed as a negative control, were purchased from Santa Cruz Biotech. Oral squamous cell carcinoma cells were plated in 60 mm culture dishes and subjected to siRNA transfection the following day. Eighty pmol of PP1a, cyclin D1, or control siRNA, or 40 pmol of PP1a siRNA together with 40 pmol of cyclin D1 siRNA were used for each transfection using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. At 1 day after transfection (day 1), transfected cells were trypsinized, plated in 24-well plates (B2.5 104 per well) and 60 mm dishes, and subjected to growth assays and Western analysis respectively, on days 2–4 after transfection. The MTT assay is a measure of cell proliferation based on the cleavage of the yellow tetrazolium salt MTT to purple formazan crystals by metabolically active cells. Growth assessment was performed by the MTT assay (Roche, Indianapolis, IN, USA) according to the manufacturer’s instructions, measuring optical density 550–650 nM (OD550–650), which has a linear correlation with cell number on a log scale; or by cell counting using a hemacytometer. Cells for growth assays were plated in duplicate (data in Figure 4) or in triplicate (data in Figure 5). Each experiment was performed at least two to three times, and consistent results were obtained.

Immunofluorescence staining of Ki-67 was used to assess the G0 cell population and flow cytometry of propidium iodidestained cells was used to determine cell cycle distribution. UPCI:SCC131 cells were trypsinized 1 day after siRNA transfection and subjected to cytospin and immunostaining, or flow cytometric analysis as described previously (Hsu and White, 1998). Primary antibody for immunofluorescence staining was anti-Ki-67 (DakoCytomation, Carpinteria, CA, USA) at 1:500 dilution. G0 cells were negative for Ki-67 staining. Statistical analysis Data were presented as mean7s.d. One-sided Mann–Whitney rank-sum tests were used to test association of overexpression of PP1a with 11q13 amplification and PPP1CA gene copy number. Two-sided t-tests on log-transformed data were used to assess the relationship between knockdown of both PP1a and cyclin D1, and OSCC cell growth. Cell growth curves were compared using an analysis of variance (ANOVA) of the logtransformed data, and treating time as a three-level factor. Fisher’s exact tests were used to assess the significance of PP1a and cyclin D1 knockdown on growth arrest in G0. P-values o0.05 were considered statistically significant.

Abbreviations FISH, fluorescence in situ hybridization; GFP, green fluorescent protein; OSCC, oral squamous cell carcinoma; PP1a, protein phosphatase 1a; QPCR, quantitative PCR; QuMA, quantitative microsatellite analysis; siRNA, small interfering RNA. Acknowledgements We thank Dale Lewis, Suman Barua, Linda Klei, Dr Rahul Parikh and Dr Zhisheng Yu for technical support. This work was supported by funds from the Jennie K Scaife Ovarian Cancer Center (to LCH), NIH Grants R01 CA111436 (to LCH) and R01 DE14729 (to SMG). The molecular cytogenetic analyses were carried out in the UPCI Cytogenetics Facility supported in part by NIH Grant P30 CA47904 to Dr RB Herberman.

References A˚kervall JA, Jin Y, Wennerberg JP, Zatterstrom UK, Kjellen E, Mertens F et al. (1995). Cancer 76: 853–859. Alberts AS, Thorburn AM, Shenolikar S, Mumby MC, Feramisco JR. (1993). Proc Natl Acad Sci USA 90: 388–392. Ayllon V, Cayla X, Garcia A, Roncal F, Fernandez R, Albar JP et al. (2001). J Immunol 166: 7345–7352. Barker HM, Jones TA, da Cruz e Silva EF, Spurr NK, Sheer D, Cohen PT. (1990). Genomics 7: 159–166. Ceulemans H, Bollen M. (2004). Physiol Rev 84: 1–39. Chu EC, Tarnawski AS. (2004). Med Sci Monit 10: RA235–RA241. Cohen PTW. (2002). J Cell Sci 115: 241–256. Dickson C, Fantl V, Gillett C, Brookes S, Bartek J, Smith R et al. (1995). Cancer Lett 90: 43–50. Durfee T, Becherer K, Chen PL, Yeh SH, Yang Y, Kilburn AE et al. (1993). Genes Dev 7: 555–569. Fu M, Wang C, Li Z, Sakamaki T, Pestell RG. (2004). Endocrinology 145: 5439–5447.

Gaudray P, Szepetowski P, Escot C, Birnbaum D, Theillet C. (1992). Mutat Res 276: 317–328. Ginzinger DG, Godfrey TE, Nigro J, Moore II DH, Suzuki S, Pallavicini MG et al. (2000). Cancer Res 60: 5405–5409. Gollin SM. (2001). Head Neck 23: 238–253. Hsu L-C, White RL. (1998). Proc Natl Acad Sci USA 95: 12983–12988. Huang X, Gollin SM, Raja S, Godfrey TE. (2002). Proc Natl Acad Sci USA 99: 11369–11374. Imai Y, Kakinoki Y, Takizawa N, Nakamura K, Shima H, Kikuchi K. (1999). Int J Oncol 14: 121–126. Janssens V, Goris J, Van Hoof C. (2005). Curr Opin Genet Dev 15: 34–41. Jin Y, Ho¨glund M, Jin C, Martins C, Wennerberg J, kervall J et al. (1998). Genes Chromosomes Cancer 22: 312–320. Kohno T, Takakura S, Yamada T, Okamoto A, Tanaka T, Yokota J. (1999). Cancer Res 59: 4170–4174. Lese CM, Rossie KM, Appel BN, Reddy JK, Johnson JT, Myers EN et al. (1995). Genes Chromosomes Cancer 12: 288–295. Oncogene


5526 Liu Y, Virshup DM, White RL, Hsu L-C. (2002). Cancer Res 62: 6357–6361. Motokura T, Bloom T, Kim HG, Jupner H, Ruderman JV, Kronenberg HM et al. (1991). Nature 350: 512–515. Mumby MC, Walter G. (1993). Physiol Rev 73: 673–699. Park NH, Min BM, Li SL, Huang MZ, Cherick HM, Doniger J. (1991). Carcinogenesis 12: 1627–1631. Reshmi SC, Saunders WS, Kudla DM, Ragin CR, Gollin SM. (2004). Genes Chromosomes Cancer 41: 38–46. Saadat M, Mizuno Y, Takizawa N, Kakinoki Y, Rikuchi R. (1995). Cancer Lett 94: 165–170. Savelyeva L, Schwab M. (2001). Cancer Lett 167: 115–123. Scholzen T, Gerdes J. (2000). J Cell Physiol 182: 311–322.

Oncogene

Schuuring E. (1995). Gene 159: 83–96. Schwab M. (1998). BioEssays 20: 473–479. Shenolikar S. (1994). Annu Rev 10: 55–86. Shuster MI, Han L, Le Beau MM, Davis E, Sawicki M, Lese CM et al. (2000). Genes Chromosomes Cancer 28: 153–163. Sogawa K, Masaki T, Miyauchi A, Sugita A, Kito K, Ueda N et al. (1997). Cancer Lett 112: 263–268. Wang MB, Billings KR, Venkatesan N, Hall FL, Srivatsan ES. (1998). Otolaryngol Head Neck Surg 119: 593–599. Wang SJ, Mintz LE, Natarajan V, Lee NJ, Srivatsan ES, Wang MB. (2001). Otolaryngol Head Neck Surg 124: 656–662.