Oncogene (2000) 19, 289 ± 297 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc Characterization of the encoding pinin/DRS/memA and evidence for its potential tumor suppressor function

Yujiang Shi1, Pin Ouyang2 and Stephen P Sugrue*,1

1Department of Anatomy and Cell Biology, University of Florida College of Medicine, Gainesville, Florida, FL 32610-0235, USA; 2Chang Gung University Medical College, Tau-Yuan, Taiwan, Republic of China

Several cell adhesion-related have been shown to It has been shown that 14q harbors at act as tumor-suppressors (TS) in the neoplastic progres- least two potential TS loci, one of which is located in sion of epithelial-derived tumors. Pinin/DRS/memA was the distal 14q31-tel region, and the other in the 14q13 ®rst identi®ed in our laboratory and it was shown to be a region (Bandera et al., 1997; Chang et al., 1995). cell adhesion-related molecule. Our previous study Deletions and translocations of 14q13 have been demonstrated that restoration of pinin expression in correlated with kidney adenocarcinomas (Mitelman et transformed cells not only positively in¯uenced cellular al., 1997). In addition, LOH of the 14q TS adhesive properties but also reversed the transformed (D14S75-D14S288) has been correlated with advanced phenotype to more epithelial-like. Here, we show by stage transitional cell carcinoma (TCC) (Chang et al., FISH analysis that the gene locus for pinin is within 1995), nasopharyngeal carcinoma (Mutirangura et al., 14q13. The alignment of the pinin gene with STS 1998), ovarian epithelial tumor (Bandera et al., 1997), markers localized the gene to the previously identi®ed TS bronchial epithelial tumor (Weaver et al., 1997), and locus D14S75-D14S288. Northern analyses revealed renal cell carcinoma (Beroud et al., 1996). The speci®c diminished pinin mRNA in renal cell carcinomas gene(s) within the 14q13 D segment responsible for the (RCC) and certain cancer cell lines. Immunohistochem- tumor-suppressive activity have not yet been deter- ical examination of tumor samples demonstrated absent mined. or greatly reduced pinin in transitional cell carcinoma Alterations in cell ± cell and cell ± matrix adhesive (TCC) and RCC tumors. TCC-derived J82 cells as well qualities underlie many of the phenotypic changes as EcR-293 cells transfected with full-length pinin cDNA associated with tumor progression including changes in demonstrated inhibition of anchorage-independent growth cell morphology and tissue invasiveness (Ben-Ze'ev, of cells in soft agar. Furthermore, methylation analyses 1997). Down-regulation of adhesion-related molecules revealed that aberrant methylation of pinin CpG islands in tumors have indicated that these proteins may was correlated with decreased/absent pinin expression in function as TSs (Birchmeier, 1995; Bullions et al., 1997; a subset of tumor tissues. These data lend signi®cant Hoover et al., 1998; Hsieh et al., 1995; Simcha et al., support to the hypothesis that pinin/DRS/memA may act 1996; Taverna et al., 1998). The loss of E-cadherin- as a tumor suppressor in certain types of cancers. mediated cell adhesion was shown to be an important Oncogene (2000) 19, 289 ± 297. rate-limiting step in the progression of several somatic and hereditary carcinomas (Berx et al., 1998; Keywords: pinin; tumor suppressor; 14q13; desmosome; Birchmeier et al., 1996; Guilford et al., 1998; DNA methylation; TCC Hirohashi, 1998; Perl et al., 1998). Alterations in desmosomal structure and/or composition have been correlated with the invasiveness and metastatic potential of carcinomas (Garrod et al., 1996; Garrod, Introduction 1995). Furthermore, transfections of nonadhesive L929 cells with a combination of desmosomal components The gain-of-function oncogenes and the loss-of- inhibit the invasiveness of the cells (Tselepis et al., function tumor suppressor (TSG) are two 1998). Recently, Degen et al. (1999) have described an essential elements of the multistep progression of alteration in mRNA levels of pinin/DRS/memA in tumorigenesis. Tumor suppressors are often a€ected invasive melanomas. Thus, it appears that many cell ± by alterations of the TS locus. These alterations include cell adhesion related proteins including desmosomal chromosomal deletions, intragenic mutations, or components are involved in epithelial-derived tumor inactivation of genes by epigenetic mechanisms such progression, and alterations in them may contribute to as methylation (Levine et al., 1991; Clurman and tumor cell invasiveness and metastasis. Groudine, 1997; Haber and Harlow, 1997). The Pinin was ®rst identi®ed as a desmosome-asso- approximate chromosomal locations of TSGs have ciated (Ouyang and Sugrue, 1992, 1996). been frequently determined by loss of heterozygocity Cell ± cell adhesion assembly-assay revealed that pinin (LOH) analyses (Cavenee, 1993; Fearon, 1997). assembled to pre-formed desmosomes and its localization of desmosomes was correlated with reinforcement of intercellular adhesion and increased IF organization (Ouyang and Sugrue, 1992). While *Correspondence: SP Sugrue Received 20 July 1999; revised 21 October 1999; accepted 27 October pinin is associated with mature desmosomes, it is 1999 absent from nascent desmosomes, as well as Pinin as a tumor suppressor YShiet al 290 desmosomes of actively migrating epithelial cells Results (unpublished observations), indicating that pinin is dynamically associated with the desmosome-IF Isolation and characterization of the human pinin gene complex. Transfecting full-length pinin cDNA into transformed 293 cells resulted in enhanced cell ± cell Detailed organization of the pinin gene was deduced adhesion and increased expression of junctional from human genomic clones HG03 and HG07. The complex-associated proteins (Ouyang and Sugrue, gene contains nine exons ranging in size from 72 bp to 1996). It has become evident that pinin/DRS is a 2.5 kb, with intron sizes ranging from 70 bp to 1.3 kb, cell adhesion-related and nuclear protein (Brandner et that contain typical TG-AG splice donor/acceptor sites al., 1997; Ouyang, 1999; Shi et al., 1997), yet (Figure 1). The NCBI UniGene program revealed two functions of pinin's dual localization has yet to be UniGene clusters, Hs.44499 (cluster 1) and Hs.83389 resolved. Nevertheless, it is tempting to speculate that (cluster 2), within the 3' portion of the human pinin pinin, like other cell adhesion/nuclear proteins such gene (Figure 1). Interestingly, three poly(A)+ addition as plakoglobin, and b-catenin (Ben-Ze'ev and Geiger, sites were found in Unigene cluster 1, and a fourth 1998; Kirkpatrick and Peifer, 1995), may play a poly(A)+ addition site was located in Unigene cluster 2. central role in coordinating cell adhesive and nuclear We speculate that the clusters Hs.44499 and Hs.83389 events that are essential in development, tissue present within the pinin gene most likely correlate with remodeling and tumor progression. the alternative usages of poly(A)+ addition sites. There Here, we report a detailed analysis of the human is neither a TATA box nor a CAAT cis-element in the pinin gene which is localized to a previously promoter region of the pinin gene, however, a identi®ed 14q TS locus D14S75-D14S288. We also consensus sequence of transcriptional initiation, which provide evidence supporting the tumor-suppressive is usually present in TATA-less promoters, was found activity of pinin in tumorigenesis of various primary 69 bp upstream of the pinin ATG (translation tumors. Our data collectively suggest that pinin may initiation) (Figure 1). The 5' region of the pinin gene, play a TS role in certain types of cancer. which includes exon 1 and intron 1, contains typical

Figure 1 The organization of the human gene for pinin. The structure of the human pinin gene was deduced from a 12 kb sequence obtained from genomic library screenings. The gene contains nine exons, E1 to E9 (solid black boxes), separated by eight introns (solid line). The sizes of exons and introns, the splice donor/acceptor sites are listed (table). Six typical Alu elements (solid circles) are found in non-coding regions. A transcription initiator sequence (Inr) was identi®ed in the ®rst exon and the position of nucleotide A within the Inr is designated as +1. The translational start codon ATG is located at +69. The region in which exon 1 is located displays typical features of CpG islands. CpG islands are enlarged with multiple vertical lines revealing the density and distribution of CpG doublets. Alignment of the CpG152f, a deposited CpG islands sequence in GenBank, is underlined. The stop codon TAA of the ORF is located within exon 9 (nt+6429 to nt+6431). Two NCBI Unigene clusters Hs.44499 (nt *+6300 to *+7000) and Hs.83389 (nt *+7800 to *+8500) were also found within exon 9. The gene sequence is deposited in GenBank under accession number AF195139

Oncogene Pinin as a tumor suppressor YShi et al 291 CpG islands spanning nucleotide positions *7400 to tissue controls (Figure 3a, individuals 1 and 4). *+600. The CpG islands are comprised of 71 CpG Examination of various cancer cell lines also revealed doublets (Figure 1, vertical lines) and the CG content depressed pinin mRNA levels in two of eight cell lines, was greater than 68%. including a B cell lymphoma cell line (Raji, Figure 3b, lane 5) and a melanoma cell line (G361, Figure 3b, lane 8). The depressed pinin mRNA level in both primary Pinin genetic locus and cancer cell lines suggests that expression of the Fluorescence in situ hybridization (FISH) revealed the pinin gene may be dis-regulated in tumorigenesis of human pinin gene to reside at 14q13 (Figure 2, certain cancers. bottom). The sequences of four human STS mar- kers ± STS A005O20, STS SGC31797, STS stSG4334, The expression and subcellular localization of pinin were and STS stSG407 ± were aligned to the human pinin altered in various primary tumor tissues sequence (Figure 2, top). The pinin gene locus was adjacent to genetic marker D14S306 and within the D We conducted immuno¯uorescence studies on a panel segment between markers D14S75 and D14S1053. This of 16 primary tumor samples of various origins. Down- D segment has an approximate site of 0.6 cM (HGSI regulated pinin expression was observed in transitional database). Thus, the pinin gene locus was within a cell carcinomas (TCC), renal cell carcinomas (RCC) 3 cM interval (D14S75-D14S288) that has been and some metastasic ovarian tumors. Normal (unin- previously identi®ed to be a TS locus by a number of volved) tissue samples as well as neck region squamous studies (Bandera et al., 1997; Beroud et al., 1996; cell carcinoma, colonic adenocarcinoma, and well- Chang et al., 1995; Mutirangura et al., 1998; Weaver et di€erentiated ovarian carcinomas revealed typical al., 1997). desmosomal staining of pinin (Table 1). The most dramatic decrease of pinin expression was found in a TCC resected from the renal pelvis. While typical pinin Pinin mRNA levels were depressed in cancer-derived staining was seen in tissue at the tumor margin (Figure tissues and cancer cell lines 4a), no pinin immunoreactivity was detectable in cells To investigate whether pinin gene expression is a€ected comprising most of the tumor mass (Figure 4c,e). in primary tumors and tumor cell lines, we examined Strikingly, the expression levels of three other major pinin mRNA levels in epithelial-derived RCC tumors, epithelial cell markers, desmoplakin (Figure 4b), as well as in a panel of cancer cell lines. Northern blot plakoglobin (Figure 4b,f), and keratin (not shown), analyses of four control and four tumor tissues remained similar to normal expression levels through- revealed depressed levels of pinin mRNA in two renal out the tissue sample. Interestingly, some RCCs also cell tumors as compared to their corresponding normal demonstrated decreased pinin expression levels. The

Figure 2 Genetic locus of the human pinin gene. The human pinin gene was initially localized to human at position 14q13 by FISH (bottom) using a 7 kb HG07 genomic fragment of human pinin gene as a probe. The scale of the 14q13 tumor suppressor locus is enlarged (middle). The position and distance of ®ve genetic markers in this region are indicated by black diamonds. The pinin gene (arrowhead) is located adjacent to D14S306, and within the region between the genetic markers D14S75 and D14S288. Four of these genetic markers D14S75, D14S278, D14S306, and D14S288 ¯anking both sides of the pinin gene locus are targets of LOH in di€erent tumors. Within the simpli®ed map of human pinin gene (top), four STS markers (black boxes), the sequences of which were identical to the pinin gene, were used to localize the pinin gene to the 14q13 tumor suppressor locus

Oncogene Pinin as a tumor suppressor YShiet al 292 immunostaining of a representative RCC sample for pinin is shown in Figure 5b,d. Remarkably, in a RCC tumor sample, while epithelial cells within collecting ducts exhibited normal pinin staining, the undi€erentiated tumor cells exhibited aberrant over-expression of pinin with large cytosolic accumulation of pinin immunoreactive material (Figures 6a,b). However, the intensity and distribution pattern of desmoplakin staining in the pinin over-expressing cells remained largely una€ected (Figures 6b,d).

Exogenous expression of pinin inhibited anchorage-independent growth of J82 and 293 cells To gain further insight into the tumor suppressor role of pinin, we have conducted soft agar assays to determine the capability of pinin to inhibit ancho- rage-independent cell growth. TCC-derived J82 cells, which exhibited very low endogenous expression of pinin, were transfected with wt-pinin cDNA (Ouyang and Sugrue, 1996). Parental cells and the vector-only transfectants grew eciently in soft agar, forming numerous colonies of greater than 100 cells. The pinin- transfected J82 cells grew poorly and exhibited very limited colony formation in soft agar (data not shown). To con®rm that the inhibition of anchorage-indepen- dent growth was due to the expression of pinin, we generated two cell lines using the ecdysone-inducible mammalian expression system, inducible GFP-only Figure 3 Human pinin mRNA levels in various human tumors and inducible wt-pinin with C-terminal-tagged GFP and tumor cell lines. (a) Tumor (T) and corresponding normal (EcR-293-HPGFP). While non-induced ECR-293-GFP (N) tissues were excised from the same operational site. Twenty and non-induced ECR-293-HPGFP as well as induced mg each of total RNA isolated from human kidney tumor tissues ECR-293-GFP cells were able to eciently grow in soft and corresponding normal tissues of four di€erent donors (1 ± 4) were analysed by Northern blot. Expression levels of pinin were agar (Figure 7a ± c), the anchorage-independent growth normalized to GAPDH expression levels in the same lane. of induced ECR-293-HPGFP cells was suppressed Comparing to corresponding normal tissues, the pinin mRNA (Figure 7d). level was 47-fold lower in the tumor sample of individual 1, and 1.8-fold lower in the tumor sample of individual 4. (b) Northern blot of mRNA from multiple human tumor cell lines. Two mgof The detection of aberrant methylation status of pinin + poly(A) RNA was loaded in each lane. The blot was probed and CpG islands in several primary tumors analysed as described in a. Decreased pinin mRNA levels were observed in lane 5 (Raji), and lane 8 (G361). Lanes 1 to 8 are: Aberrant methylation of CpG islands preceding tumor promyelocytic leukemia HL-60, Hela cell S3, chronic myelogen- ous leukemia K-562, lymphoblastic leukemia MOLT-4, Burkitt's suppressor genes has been demonstrated as one of the lymphoma Raji, colorectal adenocarcinoma SW480, lung mechanisms for down-regulating expression of TS carcinoma A549, and melanoma G361 cell lines genes (Baylin et al., 1998; Herman et al., 1994;

Table 1 Pinin expression and pinin CpG islands methylation in human tumor tissues Vial No. Tissue type Diagnosis Staining pattern Methylation pattern R502D07 Neck region Squa. sell ca. + N/A R410B06 Colon Adenoca. + no R410B09 Colon Uninvolved colon tissue + no R504D06 Ovary Metastatic adenoca. +/7 yes R503J09 Ovary Endometroid adenoca. (stage IV) +/7 yes R506B04 Ovary Cystadenoca. +/7 no R507B04 Ovary Well-di€. Adenoca. + no R412H09 Ovary Endometroid adenoca. + no R506I07 Ovary Granulosa Cell Tumor +/7 no R10604 Renal Pelvis Transitional cell Ca. 7 yes R412J01 Kidney Renal Cell Ca. 7 yes R504I02 Kidney Renal Cell Ca. ++++* no R104F03 Kidney Renal Cell Ca. +/7 no R106D03 Kidney Renal Cell Ca. +/7 no R507A07 Kidney Renal Cell Ca. +/7 no R408H01 Kidney Renal Cell Ca. +/7 yes R106C02 Kidney Renal Cell Ca. 7 no 7no staining; +typical boundary staining; ++++*aberrant and heavy cytosol staining; N/A not available

Oncogene Pinin as a tumor suppressor YShi et al 293

Figure 6 Increased pinin expression levels and its altered distribution in another RCC tumor sample. Triple staining of RCC sample with anti-pinin (a and d), anti-desmoplakin (b and e), and DAPI (c and f). The left panels (a, b and c) are shown at Figure 4 Immuno¯uorescence staining of TCC tumor sample. low magni®cation (106). Detailed triple staining patterns of Cryostat sections of a human TCC tumor sample taken from tumor nodules are illustrated in the right panels (d, e and f) renal pelvis were double-immunostained with anti-pinin (a, c and (shown at 7506 magni®cation). Desmoplakin exhibited typical e), anti-plakoglobin (b and f), or anti-desmoplakin (d) mAbs. intercellular boundary staining, even within the tumor nodule Normal patterns of punctate lateral pinin staining (a), and more area (T in B). Pinin staining within tumor nodules was linear lateral plakoglobin staining (b) were shown in sections abnormally elevated and ®lled the cytosol (T in A), while pinin obtained from uninvolved area of the tumor margin. In the vast staining in duct-like structures appeared normal (arrows in a and majority of cells within the body mass of the tumor, pinin staining b). The magni®cation of a, b and c is 756; b, d and f is 7506 was absent (c and e), yet punctate desmosomal staining of desmoplakin and linear lateral staining of plakoglobin remained unchanged (d and f). The magni®cation of a, b, c and d is 7506; e and f is 1806

Figure 7 Induction of pinin-expression in EcR-293-HPGFP cells inhibits its anchorage-independent capability in soft agar. The uninduced EcR-293-GFP (a), induced EcR-293-GFP (b), unin- duced EcR-293-HPGFP (c) and induced EcR-293-HPGFP (d) cells were assayed for anchorage-independent growth in soft agar. Figure 5 Renal cell carcinoma (RCC) tumor tissue exhibited The appearance of large colonies can be readily seen in the diminished immunostaining for pinin. Cryostat sections from a uninduced and induced EcR-293-GFP as well as in uninduced RCC tumor sample were double immunostained with anti-pinin EcR-293-HPGFP cells, whereas there is almost no countable (b and d) and anti-desmoplakin (a and c) mAbs. In one area of colony forming in pinin induced EcR-293-HPGFP. The the tumor, desmoplakin staining exhibited a bright and punctate magni®cation of a, b, c and d is 506 lateral pattern (arrows in a), whereas pinin staining was low/ absent (arrowheads in b). In other areas of the tumor sample, in addition to pinin absence (d), desmoplakin staining was reduced and exhibited aberrant distribution (c). The magni®cation of a, b, c and d is 7506 and the down-regulation of pinin, we conducted methylation analysis for the pinin CpG islands. Methylated pinin CpG islands were detected in ®ve Kashiwabara et al., 1998; Merlo et al., 1995). In order out of the previously mentioned 16 primary tumor to explore the possibility that there is a correlation samples (Table 1 and Figure 8b). Interestingly, between aberrant methylation of pinin CpG islands correlation between decreased pinin expression and a

Oncogene Pinin as a tumor suppressor YShiet al 294

Figure 8 Methylation analysis of pinin CpG islands region. A partial map of 5' pinin CpG islands was shown in a. The positions of the HpaII/MspI sites and primers used in this analysis for the detection of methylation are indicated as vertical lines and arrows, respectively. Control PCR primers P1 and P2 are designated to amplify the region (P1 ± P2) of pinin CpG islands not containing HpaII/MspI sites. Primers P3 and P4 are designated to amplify DNA region (P3 ± P4) spanning for HpaII/MspI sites. The presence of a visible PCR product in the U lanes indicates that the primers and PCR condition used to amplify the P1 ± P2 region and the P3- P4 region is proper. The presence of product in the H lanes indicate that the P3 ± P4 region from the tumor sample is insensitive to HpaII digestion, therefore, the sites were methylated. The absence of product in the M lanes indicate that the P3 ± P4 region is sensitive to MspI digestion, therefore, the HpaII/MspI site is present. The presence of product in each lane of lower b served as internal control for each DNA sample and indicates the DNA quality of each sample after digestion is remaining unchanged

methylation of the pinin CpG islands was found in that the loss of expression of major components of the TCC #R10604 and RCC #R413J01 samples. These desmosome correlated with invasiveness of TCC. results suggest that the aberrant methylation may be However, we show in this study that the absence of one mechanism responsible for silencing of the pinin pinin at desmosomes in a subset of TCC tumor cells gene. occurs without the depression of other desmosomal components. We suggest that pinin may regulate the stability of cell ± cell adhesion and its inactivation may Discussion be requisite in certain tumor progressions. The evidence that the pinin cDNA-recipient J82 as well The tumor-suppressive role of the pinin/DRS/memA as induced EcR-293-HPGFP cells displayed a suppres- protein, which was initially based on its genetic locus, sion of tumorigenic characteristics in soft agar further has been supported by the demonstration of pinin support that pinin may possess tumor suppressive down-regulation in various primary tumors, the qualities. identi®cation of the methylated CpG islands of the Multiple mechanisms, both genetic and epigenetic, pinin gene in the tumor tissues, and the suppression of account for the down-regulation of tumor suppres- anchorage-independent growth of tumor cells by sors. In our initial attempt to de®ne the mechanism transfection of pinin's cDNA. of decreased pinin levels, we employed Southern The postulated role of pinin in the reinforcement of analyses and PCR-based genomic sequencing to epithelial cell ± cell adhesion is consistent with its examine DNA from multiple tumor samples for identi®cation as a potential tumor suppressor. The evidence of intragenic alterations. However, we have requisite of down regulation of epithelial cell ± cell not yet identi®ed any genetic alterations that would adhesion in tumor progression has received much account for the loss of pinin expression (unpublished attention over the years (Shiozaki et al., 1996; Tlsty, data). The detection of aberrant methylation in pinin 1998). Numerous studies have implicated cell ± cell CpG islands that correlated with lack of pinin adhesion as a key regulatory point in the transition expression in TCC and RCC presents one possible from benign to highly invasive tumors. Focus has mechanism for silencing pinin gene expression. been directed toward the tumor suppressor role of However, the absence of pinin expression in some adhesion molecules such as cadherins, C-CAM, a- tumors (e.g. #R106C02) correlated with neither catenin, and plakoglobin in tumor progression (Ben- aberrant methylation nor intragenic alterations. Ze'ev, 1997; Davies et al., 1998; Ewing et al., 1995; Therefore, we will entertain other mechanisms of Simcha et al., 1996). Conn et al. (1990) demonstrated pinin gene or protein inactivation.

Oncogene Pinin as a tumor suppressor YShi et al 295 The observation that expression of pinin could be Northern analyses depressed or elevated in tumor samples may be a more Human cancer cell line (Clontech) and human kidney tumor general phenomenon in tumors. Degen et al. (1999) (Invitrogen) Northern blots were probed with32P-labeled full- provided convincing data of the overexpression of length human pinin cDNA, and then probed with 32P-labeled pinin/DRS/memA in a subset of melanomas. It should GAPDH cDNA according to manufacturer's protocols. The be noted that eight of 11 melanoma metastatic lesions blots were then subjected to autoradiography using Kodak and 13 of 17 tumor cell lines included in this study, ®lm (Kodak). The quantitative analysis of pinin and GAPDH actually exhibited depressed or absent expression of expression was conducted using the Quantity One1 program pinin/DRS/memA. Here we observed elevated protein (BIO-RAD). expression and uncharacteristic localization of pinin protein within cells comprising a RCC tumor mass. Fluorescence in situ hybridization (FISH) Although it is generally assumed that tumor suppres- Fluorescence in situ hybridization was done according to sors are negative regulators of tumor formation and standard protocol (Lichter and Cremer, 1992). The probe are often down-regulated in cancer development, used for the hybridization was generated from the genomic observations of over-expression of TS proteins such DNA fragment HG07 containing coding region of pinin gene as pRb, p53, p73, p27, p16 in di€erent tumors are not by nick translation with digoxigenin-11-2'-deoxy-uridine- uncommon (Cheng et al., 1999; Doki et al., 1997; Kaur 5'triphosphate (Boehringer Mannheim, Germany). et al., 1998; Shapiro et al., 1995; Warneford et al., 1991; Yokomizo et al., 1999). The over-expression of a Preparation of frozen human tissue sections and tumor suppressor may imply a compensatory mechan- immuno¯uorescence microscopy ism which may function to circumvent the disrupted A total of 17 samples representing di€erent stages and grades regulatory pathway of wild-type TS protein (Benedict of TCC, RCC, ovarian, colon, and neck region tumors as et al., 1999; Emig et al., 1998). well as corresponding uninvolved tissues were obtained from Our initial study of pinin expression has revealed the Tumor Tissue Bank at the University of Florida Cancer that pinin is present in numerous cell types, including Center (Gainesville, FL, USA). Tumor and corresponding those lacking desmosomes (Ouyang and Sugrue, 1996). uninvolved tissues were embedded in OCT compound and Recently, our laboratory and other groups have snap frozen in liquid nitrogen. Frozen cryostat sections of reported that pinin/DRS/memA is also a nuclear 6 mm thickness were ®xed with acetone for 2 min at 720 8C protein (Brandner et al., 1997). Since pinin expression and extensively washed in PBS. Prepared sections were can be detected in lymphocytes, leukocytes, and in incubated with anti-pinin mAb 08L and/or cell adhesion related antibodies against desmoplakin, plakoglobin, and some T cell and B cell leukemias (Ouyang and Sugrue, desmoglein (ARP). Double immuno¯uorescence staining was 1996), the marked depression of pinin mRNA in Raji conducted as described previously (Ouyang and Sugrue, cells may indicate involvement of pinin in tumorigen- 1992), using primary Ab mixtures of anti-pinin/anti- esis of non-epithelial tissue-derived cancers as well. We desmoplakin, or anti-pinin/anti-plakoglobin. Epi¯uorescence are currently investigating the possibility that pinin microscopy was conducted with a Zeiss Axiophot. Images tumor suppressor activity may extend beyond the were digitally captured using a SPOT2 camera (Diagnostic realm of stabilization of desmosomal adhesions. Instruments, Inc.). Here we presented evidence regarding the involve- ment of pinin in tumorigenesis and its potential role as Methylation analysis of pinin CpG islands a tumor suppressor. We believe this study sheds Methylation analyses were conducted as described by Kane et signi®cant light on the current understanding of the al. (1997) with minor modi®cations. One mg each of genomic alterations in cell ± cell adhesion which allow for DNA extracted from 16 tumor samples and one uninvolved epithelial cells to become invasive. Furthermore, more tissue were endonuclease digested. Reactions containing no detailed characterizations of pinin's function in a enzyme, 100 units of HpaII, or 200 units of MspI were tumor suppressor pathway may facilitate clinical carried out overnight in 50 ml volumes at 378C. Twenty-®ve diagnosis as well as the future design of rational ng of column-puri®ed DNA from each digest was analysed therapies for cancer. by PCR. Each 30 ml reaction contained 2 mM MgCl2,10mM Tris-HCl (pH 8.3), 50 nM KCl, 200 mM dNTPs, 1 unit of Taq polymerase (Boehringer Mannheim) and 3 pmol of primers P3 (5'-CGCCCCGACTCTGCGCTAGCAGC) and P4 (5'- Materials and methods CCGGGCTATGCCGCGCCGCCG). Ampli®cation was conducted as follows: (958C, 4 min)61; [(948C, 45 s), Identi®cation and isolation of human genomic clones for pinin (608C, 45 s), (728C, 45 s)]633; (728C, 7 min)61. PCR A human genomic lambda library (Lambda Fix II, products were analysed on 2% agarose gels. Blind assays Stratagene) was probed with 32P-labeled 2.1 kb coding region were conducted as all DNA samples were numbered and of human pinin cDNA. The isolation and puri®cation of coded prior to setting up reactions. PCR reactions and positive clones were done according to manufacturer's template DNA qualities were controlled by carrying out PCR protocol (Stratagene). with the primers P1 (5'-GCATGCGCACCAACAGAATAG) and P2 (5'-GTTAGCTGCGGGTGCCTG). Control PCR reactions were performed in parallel with previous P3-P4 Determination of the human pinin gene structure PCR reactions. Two lambda clones, HPG03 and HPG07, were puri®ed and digested with NotI. NotI DNA fragments containing the Plasmids and constructs pinin gene were veri®ed and characterized by Southern blot analysis and subcloned into pBluescript KS (Stratagene). To The construct of pinin expression vector pcDNA3-S208L have obtain the entire pinin gene sequence, eight small overlapping been previously described (Ouyang and Sugrue, 1996). The restriction fragments were generated and subcloned into pIND vector is an ecdysone-inducible mammalian expression pBluescript KS and sequenced. vector (Invitrogen), and was used to generate pIND-GFP and

Oncogene Pinin as a tumor suppressor YShiet al 296 pIND-HPGFP. The 0.7 kb coding region of humanized GFP the green ¯uorescent proteins by ¯uorescence microscopy (gift from Dr Nick Muzyczka) was ligated to the 3' end of were selected for further analysis. 2.2 kb coding region of human pinin cDNA and this 2.9 kb chimeric cDNA fragment was then cloned into the pIND Growth in soft agar vector thereby constructing 9 kb pIND-HPGFP. To con- struct the control pIND-GFP vector, the 0.7 kb coding An assay for tumor cell growth in soft agar was conducted as region of humanized GFP cDNA was cloned into pIND described by Marshall et al. (1977). Con¯uent monolayers of vector. cultured cells were trypsinized, and single cell suspensions were prepared. Cell suspensions were mixed with DMEM containing 10% FBS and 0.3% agar. In the case for Cell culture and transfection induction of GFP or HPGFP expression, 2 mg/ml of The human TCC-derived J82 cell line was purchased from muristeron A was added to the mixture of complete medium ATCC. Cells were cultured in Dulbecco's modi®ed Eagle and 0.3% soft agar for those induced clones. 16105 cells medium (BioWhittaker) containing 10% FBS (Mediatech were plated onto a pre-cast 3 ml base of DMEM-10% FBS- Cellgro), 2 mM glutamine, and 200 U/mL each of strepto- 0.5% agar in 60 mm petri dishes. Cultures were incubated in

mycin and penicillin G. The methodology for calcium a humidi®ed CO2 incubator at 378C. Cultures were fed at 4- phosphate transfection and cloning stable transfectants have day intervals with DMEM ± 10% FBS ± 0.3% agar or the been previously described (Ouyang and Sugrue, 1996). EcR- medium containing muristeron A. Twenty-one days later, 293 cells stably expressing the ecdysone receptor were colonies greater than 0.1 mm in diameter were counted under purchased from Invitrogen and maintained in aforemen- low-magnifying microscope. The representative phase con- tioned culture medium plus 400 mg/ml of Zeocin (Invitrogen). trast photographs were taken from uninduced or induced The EcR-293 cells were transfected with 10 mg of pIND-GFP EcR-293-GFP controls and EcR-293-HPGFP cells. All or pIND-HPGFP by calcium phosphate transfection. Stable experiments were repeated three times. transfectants were selected for 4 weeks in the medium containing 400 mg/ml of G418 (Life Technologies) and Acknowledgments 400 mg/ml of Zeocin (Invitrogen). Surviving colonies were We thank Dr Nick Muzyczka for the pindGFP construct, individually cloned from the stable pool and expanded. For Mohammad Tabesh, Summer Carter and Michael Ruten- induction of GFP or HPGFP expression, 2 mg/ml of berg for technical assistance, and Todd Barnash for muristerone A (Invitrogen) was added to the cell cultures, graphic assistance. We also thank Matthew N. Simmons and the GFP expression level was reached to maximum after for assistance in the manuscript revision. Work supported 24 h induction. Clones that demonstrate 100% expression of by National Institutes of Health grant EY07883.

References

Bandera CA, Takahashi H, Behbakht K, Liu PC, LiVolsi Doki Y, Imoto M, Han EK, Sgambato A and Weinstein IB. VA, Benjamin I, Morgan MA, King SA, Rubin SC and (1997). Carcinogenesis, 18, 1139 ± 1148. Boyd J. (1997). Cancer Res., 57, 513 ± 515. Emig R, Magener A, Ehemann V, Meyer A, Stilgenbauer F, Baylin SB, Herman JG, Gra€ JR, Vertino PM and Issa JP. Volkmann M, Wallwiener D and Sinn HP. (1998). Br. J. (1998). Adv. Cancer Res., 72, 141 ± 196. Cancer, 78, 1661 ± 1668. Ben-Ze'ev A. (1997). Curr. Opin. Cell Biol., 9, 99 ± 108. Ewing CM, Ru N, Morton RA, Robinson JC, Wheelock MJ, Ben-Ze'ev A and Geiger B. (1998). Curr. Opin. Cell Biol., 10, Johnson KR, Barrett JC and Isaacs WB. (1995). Cancer 629 ± 639. Res., 55, 4813 ± 4817. Benedict WF, Lerner SP, Zhou J, Shen X, Tokunaga H and Fearon ER. (1997). Science, 278, 1043 ± 1050. Czerniak B. (1999). Oncogene, 18, 1197 ± 1203. Garrod D, Chidgey M and North A. (1996). Curr. Opin Cell Beroud C, Fournet JC, Jeanpierre C, Droz D, Bouvier R, Biol., 8, 670 ± 678. Froger D, Chretien Y, Marechal JM, Weissenbach J and Garrod DR. (1995). Cancer Surv., 24, 97 ± 111. Junien C. (1996). Genes Cancer, 17, 215 ± Guilford P, Hopkins J, Harraway J, McLeod M, McLeod N, 224. Harawira P, Taite H, Scoular R, Miller A and Reeve AE. Berx G, Becker KF, Ho¯er H and Van Roy F. (1998). Hum. (1998). Nature, 392, 402 ± 405. Mutat., 12, 226 ± 237. Haber D and Harlow E. (1997). Nat. Genet., 16, 320 ± 322. Birchmeier W. (1995). Bioessays, 17, 97 ± 99. Herman JG, Latif F, Weng Y, Lerman MI, Zbar B, Liu S, Birchmeier W, Behrens J, Weidner KM, Hulskem J and SamidD,DuanDS,GnarraJR,LinehanWM,et al. Birchmeier C. (1996). Curr. Top Microbiol. Immunol., 213, (1994). Proc. Natl. Acad. Sci. USA, 91, 9700 ± 9704. 117 ± 135. Hirohashi S. (1998). Am. J. Pathol., 153, 333 ± 339. Brandner JM, Reidenbach S and Franke WW. (1997). Hoover KB, Liao SY and Bryant PJ. (1998). Am.J.Pathol., Di€erentiation, 62, 119 ± 127. 153, 1767 ± 1773. Bullions LC, Notterman DA, Chung LS and Levine AJ. Hsieh JT, Luo W, Song W, Wang Y, Kleinerman DI, Van (1997). Mol. Cell Biol., 17, 4501 ± 4508. NT and Lin SH. (1995). Cancer Res., 55, 190 ± 197. Cavenee WK. (1993). J. Clin. Invest., 91, 3. KaneMF,LodaM,GaidaGM,LipmanJ,MishraR, Chang WY, Cairns P, Schoenberg MP, Polascik TJ and Goldman H, Jessup JM and Kolodner R. (1997). Cancer Sidransky D. (1995). Cancer Res., 55, 3246 ± 3249. Res., 57, 808 ± 811. ChengM,OlivierP,DiehlJA,FeroM,RousselMF,Roberts Kashiwabara K, Oyama T, Sano T, Fukuda T and Nakajima JM and Sherr CJ. (1999). EMBO J., 18, 1571 ± 1583. T. (1998). Int. J. Cancer, 79, 215 ± 220. Clurman B and Groudine M. (1997). Nature, 389, 122 ± 123. Kaur J, Srivastava A and Ralhan R. (1998). Int. J. Cancer, Conn IG, Vilela MJ, Garrod DR, Crocker J and Wallace 79, 370 ± 375. DM. (1990). Br. J. Urol., 65, 176 ± 180. Kirkpatrick C and Peifer M. (1995). Curr.Opin.Genet.Dev., Davies G, Jiang WG and Mason MD. (1998). Oncol. Rep., 5, 5, 56 ± 65. 1567 ± 1576. Levine AJ, Momand J and Finlay CA. (1991). Nature, 351, Degen WG, Agterbos MA, Muyrers JP, Bloemers HP and 453 ± 456. Swart GW. (1999). Biochim. Biophys. Acta, 1444, 384 ± 394.

Oncogene Pinin as a tumor suppressor YShi et al 297 Lichter P and Cremer T. (1992). Human Cytogenetics: a Shi Y, Tabesh M, Shi J and Sugrue SP. (1997). Mol. Biol. practical approach. Oxford University Press. Cell, 8, 418a. Marshall CJ, Franks LM and Carbonell AW. (1977). J. Natl. Shiozaki H, Oka H, Inoue M, Tamura S and Monden M. Cancer Inst., 58, 1743 ± 1751. (1996). Cancer, 77, 1605 ± 1613. MerloA,HermanJG,MaoL,LeeDJ,GabrielsonE,Burger Simcha I, Geiger B, Yehuda-Levenberg S, Salomon D and PC, Baylin SB and Sidransky D. (1995). Nat. Med., 1, Ben-Ze'ev A. (1996). J. Cell Biol., 133, 199 ± 209. 686 ± 692. Taverna D, Ullman-Cullere M, Rayburn H, Bronson RT and Mitelman F, Mertens F and Johansson B. (1997). Nat. Hynes RO. (1998). Cancer Res., 58, 848 ± 853. Genet., 15, 417 ± 474. Tlsty TD. (1998). Curr. Opin. Cell Biol., 10, 647 ± 653. Mutirangura A, Pornthanakasem W, Sriuranpong V, Tselepis C, Chidgey M, North A and Garrod D. (1998). Proc. Supiyaphun P and Voravud N. (1998). Int. J. Cancer, 78, Natl. Acad. Sci. USA, 95, 8064 ± 8069. 153 ± 156. Warneford S, Townsend M, Rowe PB, Dalla-Pozza L and Ouyang P. (1999). Biochem. Biophys. Res. Commun., 263, Symonds G. (1991). Cell Growth Di€er., 2, 439 ± 445. 192 ± 200. Weaver DA, Hei TK, Hukku B, McRaven JA and Willey JC. Ouyang P and Sugrue SP. (1992). J. Cell Biol., 18, 1477 ± (1997). Carcinogenesis, 18, 1251 ± 1257. 1488. Yokomizo A, Mai M, Tindall DJ, Cheng L, Bostwick DG, Ouyang P and Sugrue SP. (1996). J. Cell Biol., 135, 1027 ± Naito S, Smith DI and Liu W. (1999). Oncogene, 18, 1042. 1629 ± 1633. PerlAK,WilgenbusP,DahlU,SembHandChristoforiG. (1998). Nature, 392, 190 ± 193. Shapiro GI, Edwards CD, Kobzik L, Godleski J, Richards W, Sugarbaker DJ and Rollins BJ. (1995). Cancer Res., 55, 505 ± 509.

Oncogene