A Novel Rho Gtpase-Activating Protein, Is Downregulated in Cancer and Inhibits Tumor Cell Growth

Oncogene (2007) 26, 4580–4589 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE Deleted in liver cancer 3 (DLC-3), a novel Rho GTPase-activating protein, is downregulated in cancer and inhibits tumor cell growth

ME Durkin1, V Ullmannova1, M Guan and NC Popescu

Laboratory of Experimental Carcinogenesis, Center for Cancer Research, National Cancer Institute, Bethesda, MD, USA

Two related Rho GTPase-activating proteins, DLC-1 wide range of cellular processes (Ridley, 2001; Jaffe and (deleted in liver cancer 1) and DLC-2, are emerging as Hall, 2005). Rho activity is frequently increased in bona fide tumor suppressor genes that inhibit cancer cell human tumors, but as activating mutations in Rho growth. In this report, we characterized a gene on protein genes are rare, the regulators of Rho GTPases chromosome Xq13 that encodes DLC-3 (also known as may be targeted during oncogenesis (Gomez del Pulgar KIAA0189 and STARD8), a third member of the DLC et al., 2005). Rho proteins transmit signals in the active, family. The DLC-3 gene has transcripts with alternative GTP-bound state, which is positively controlled by 50 ends, one of which, DLC-3a, encodes an 1103-amino guanine nucleotide exchange factors that catalyse the acid polypeptide highly similar to DLC-1 and DLC-2. A exchange of bound GDP for GTP and negatively second isoform (DLC-3b) would yield a protein lacking controlled by Rho GTPase-activating proteins (GAPs) the N-terminal sterile alpha motif domain. The DLC-3 that stimulate GTP hydrolysis (Van Aelst and D’Souza- gene is widely expressed in normal tissues, but DLC-3 Schorey, 1997). Decreased GAP activity could result in mRNA levels were low or absent in a significant number of sustained Rho signaling that facilitates the growth and breast, ovarian, liver and prostate cancer cell lines. Using metastasis of tumor cells. a cancer profiling array to compare matched tumor and The human RhoGAPs are a diverse family of proteins normal human tissues, downregulation of DLC-3 mRNA that share a conserved, 150–200 amino acid GAP was observed in kidney, lung, ovarian, uterine and breast domain that contains the catalytic activity (Bernards, cancer samples. By quantitative reverse transcriptase– 2003; Moon and Zheng, 2003). One subgroup of the polymerase chain reaction, DLC-3 expression was re- human RhoGAPs includes DLC-1 (deleted in liver duced in primary prostate carcinomas relative to normal cancer 1), the human homologue of rat p122RhoGAP prostate tissue. Transfection of human breast and prostate (Homma and Emori, 1995; Yuan et al., 1998), and cancer cells with a DLC-3a expression vector inhibited DLC-2, also known as STARD13 (Ching et al., 2003). cell proliferation, colony formation and growth in soft The DLC-1 and DLC-2 genes encode polypeptides of agar. These results indicate that deregulation of DLC-3 approximately 1100 amino acids (aa) with a character- may contribute to breast and prostate tumorigenesis. istic modular architecture, consisting of an N-terminal Oncogene (2007) 26, 4580–4589; doi:10.1038/sj.onc.1210244; sterile a motif (SAM) domain, a serine-rich domain, published online 5 February 2007 a RhoGAP domain and a C-terminal steroidogenic acute regulatory protein-related lipid transfer (START) Keywords: tumor suppressor gene; Rho GTPase-activating domain (Homma and Emori, 1995; Yuan et al., 1998; protein; prostate cancer; breast cancer Ponting and Aravind, 1999; Ching et al., 2003). DLC-1 and DLC-2 were shown to have in vitro GAP activity for RhoA and Cdc42 and to influence cell morphology and cytoskeletal organization (Homma and Emori, 1995; Sekimata et al., 1999; Ching et al., 2003; Wong et al., Introduction 2003, 2005). DLC-1 inhibits the growth, colony-forming ability, tumorigenicity and invasiveness of human liver, The development of neoplasia is associated with breast, ovarian, nasopharyngeal, esophageal and non- alterations in signal transduction pathways that regulate small cell lung cancer cells (Ng et al., 2000; Yuan et al., cell proliferation, survival and invasiveness (Hanahan 2003, 2004; Zhou et al., 2004; Goodison et al., 2005; and Weinberg, 2000). The members of the Rho family of Syed et al., 2005; Wong et al., 2005; Seng et al., 2006), small GTPases act as molecular switches that influence a and DLC-2 has been shown to inhibit the ras-induced transformation of rodent cells and the growth of breast Correspondence: Dr NC Popescu, Laboratory of Experimental and liver cancer cells in culture (Nagaraja and Kandpal, Carcinogenesis, Center for Cancer Research, National Cancer Institute, 2004; Leung et al., 2005). Bethesda, MD 20892, USA. A third member of the deleted in liver cancer gene E-mail: [email protected] 1These two authors contributed equally to this work. family is present in the human genome. The KIAA0189 Received 13 September 2006; revised 17 October 2006; accepted 17 cDNA clone, isolated from a human myeloid cell line October 2006; published online 5 February 2007 library, was noted to encode a protein similar to DLC-3 and cancer ME Durkin et al 4581 p122RhoGAP (Nagase et al., 1996). The gene encoding genomic DNA sequence revealed the exon structure of KIAA0189 (official gene name, STARD8, for ‘START the gene (Figure1b and Table 1) and showed that the domain protein 8’) has been localized to the X heterogeneity at the 50 ends of the DLC-3 transcripts chromosome and is distinct from the DLC-1 and appears to arise through the use of alternative promo- DLC-2 genes on 8p22 and 13q13.1, respectively. In this ters and exon skipping. communication, we have characterized additional tran- A transcript represented by a cDNA from a human scripts of the STARD8 locus that have alternative 50 endometrium carcinoma cell line (GenBank CR749411, ends and analysed the sequence of the predicted protein clone DKFZp686H1668), designated DLC-3a, shares product, which we term DLC-3. We also present exons 2–14with KIAA0189, now termed DLC-3 b. evidence that DLC-3 expression is altered in some Nucleotides 1–160 and nt 161–194of the DLC-3 a human cancers and that DLC-3 can suppress tumor cell sequence are present in exons 1A and 1B, respectively, growth. located upstream of the first exon of DLC-3b (exon 1C, nt 1–176 of D80011). A third transcript, DLC-3g, represented by a cDNA from a uterine leiomyosarcoma Results library (IMAGE 5518429, GenBank BC035587) appears to use the same transcription start site as DLC-3a Analysis of DLC-3 transcripts but lacks exon 3, causing a reading frame shift that The STARD8 gene encoding DLC-3 is closely linked to would result in premature translation termination the androgen receptor (AR), oligophrenin 1 (OPHN1) after 52 aa. Reverse transcriptase–polymerase chain and ephrin-B1 (EFNB1) genes on chromosome Xq13.1 reaction (RT–PCR) analysis of normal human liver, (Figure 1a). Database searches identified cDNAs that prostate and mammary gland RNA using primers were nearly identical to the original KIAA0189 sequence in exon 1A and exon 5 indicated that the DLC-3a throughout most of their lengths but differed at the transcript, and not the DLC-3g isoform, is present in 50 ends. Comparison of the cDNA sequences to the these tissues (Figure 1d).

Figure 1 Structure of the DLC-3/STARD8 gene. (a) Map of the human chromosome Xq13 region, with boxes representing STARD8 and neighboring genes encoding the androgen receptor (AR), oligophrenin-1 (OPHN1) and ephrin-B1 (EFNB1). The numbers above the line refer to the sequence coordinates, in megabases. (b) Exon organization of the DLC-3 gene. Boxes represent exons and are numbered to correspond with exons 2–14of the homologous DLC-1 gene (Durkin et al., 2002). Arrows indicate the potential transcription start sites upstream of exons 1A and 1C. (c) Schematic representation of the exon sequences present at the 50 ends of the three DLC-3 transcripts. The KIAA0189 cDNA sequence is designated as the DLC-3b isoform, and the GenBank accession numbers of the cDNAs corresponding to the DLC-3a and g isoforms are given in parentheses. The locations of the putative ATG translation start codons in the three transcripts and the premature TGA stop codon in the DLC-3g isoform are marked. (d) Ampliﬁcation of cDNA prepared from normal human liver (L), mammary gland (M) and prostate (P) with DLC-3 primers 1F and 5R2. The major RT–PCR product is a 343 bp band containing exon 3 sequences, and not a 261 bp band missing exon 3. The identity of the PCR product was veriﬁed by sequencing.

Oncogene DLC-3 and cancer ME Durkin et al 4582 Table 1 Exon organization of the human DLC-3 (STARD8) gene Numbera Size (bp) cDNAb (nt) Genomic DNAc (nt) DLC-1/DLC-2 exon sizes (bp)

Exon 1A X259 12–270 85290–85548 Exon 1B 34271–304 103120–103153 Exon 1C X176 1–176 17211–17386 Exon 2 72 177–248 36474–36545 72/72 Exon 3 82 249–330 38854–38935 82/82 Exon 464331–39439928–39991 64/64 Exon 5 1418 395–1812 40772–42189 1424/1361 Exon 6 1741813–1986 177/174 42785–42958 Exon 7 160 1987–2146 43824–43983 160/160 Exon 8 199 2147–2145 44484–44682 199/199 Exon 9 211 2146–2556 45096–45306 214/211 Exon 10 115 2557–2671 45607–45721 115/115 Exon 11 225 2672–2896 45982–46206 219/222 Exon 12 218 2897–3114 47186–47403 218/218 Exon 13 177 3115–3291 47505–47681 174/177 Exon 14115 d 3292–4824 47865–49393 118/115d

aExons are numbered to correspond to the numbers of the equivalent exons of the DLC-1 gene. bThe position of the exon in the cDNA sequence was obtained from GenBank BC035587 for exons 1A and 1B and from GenBank D80011 for exons 1C and 2–14. cThe position of the exon in the genomic DNA sequence was obtained from GenBank AL732324for exons 1A and 1B and from GenBank AL360076 for exons 1C and 2–14. dThe sizes refer to the length of the open reading frame in the sequence of the ﬁnal exon.

Features of the DLC-3 protein sequence Linear Motif server (http://elm.eu.org) revealed the The DLC-3a transcript has a 3309 bp open reading presence of a number of putative serine and threonine frame that encodes an 1103-aa polypeptide containing phosphorylation sites, and there are also several proline- SAM, RhoGAP and START domains (Figures 2 and rich stretches that could interact with proline recogni- 3a). The DLC-3a amino acid sequence is 44% identi- tion motifs such as the SH3 (src homology 3) domain cal to DLC-1 and 52% identical to DLC-2, with the present in many signaling proteins (Li, 2005) (not greatest similarity in the RhoGAP domains. The DLC-3 shown). RhoGAP domain contains the conserved ‘arginine finger’ (Arg688) essential for the catalytic activity of Phylogeny of the DLC family genes RhoGAPs (Moon and Zheng, 2003) and also has two In addition to the similarities at the protein level, the residues (Lys725 and Arg729) equivalent to ones found exon organizations of the DLC family genes are highly to be important for the activity of DLC-1 and DLC-2 conserved (Table 1), suggesting that the three genes have (Sekimata et al., 1999; Leung et al., 2005). Translation a common ancestor. The duplications apparently initiation at the first ATG codon of the longest open involved an extended region of genomic DNA, as some reading frame of the DLC-3b transcript (nt 338–340 of genes linked to the DLC-3 gene at Xq13 have D80011) would yield a 1023-residue protein lacking the paralogues linked to the DLC-2 locus on chromosome SAM domain. The two DLC-3 isoforms are also present 13q13 and to the DLC-1 locus on chromosome 8p22 in rodents; a 4.4 kb mouse cDNA (AK133927) encodes a (Figure 3b). Sequence comparisons indicate that the 1019-aa protein that is 87% identical to the DLC-3b genes on 8p (DLC1, SLC7A2 and MTMR7) are more isoform, and there are mouse expressed sequence tags closely related to their paralogues on 13q (DLC2, with 50 ends similar to the a form (e.g., GenBank SLC7A1 and MTMR6) than to those on Xq (DLC3, CJ163700). SLC7A3 and MTMR8). It therefore appears that an The serine-rich region between the SAM and Rho- initial duplication event yielded the Xq block and an 8p/ GAP domains has the least sequence similarity among 13q precursor that subsequently underwent another the DLC family members, but there are several round of duplication. conserved stretches that could correspond to sites with biological functions. One of the conserved sequence elements (LDDILQHV, residues 388–395 in DLC-3a)is Expression of DLC-3 mRNA in human tissues similar to the consensus LD motif (LDXLLXXL) found The expression pattern of the DLC-3 mRNA in human in paxillin and other signaling proteins, which mediates tissues was examined by Northern blot hybridization the binding of paxillin to vinculin and focal adhesion (Figure 4a). The 5 kb DLC-3 transcript had a broad kinase (Brown et al., 1998). Downstream of this motif in tissue distribution, with the highest levels in kidney, lung human DLC-3 is an E80 amino acid insertion enriched and placenta, similar to earlier results for KIAA0189 in proline, alanine and glutamine residues, which has (Nagase et al., 1996). The localization of DLC-3 mRNA relatively low homology with the mouse sequence was compared to that of DLC-1 and DLC-2 by RT– and is missing from chicken, frog, fish and marsupial PCR (Figure 4b). The transcripts for all three DLC DLC-3 sequences (not shown). A search for sequence family members were detected in solid tissues, but the patterns in the serine-rich domain using the Eukaryotic levels were low in peripheral blood lymphocytes.

Oncogene DLC-3 and cancer ME Durkin et al 4583 DLC3 MPLLDVFWSCFRKVKCFPLLQVKKNAEAEAKRACEWLQATGFPQYVQLFEEGSFPLDIGS 60 DLC1 ------MCRKKPDTMILTQIEAKEACDWLRATGFPQYAQLYEDFLFPIDISL 46 was more abundant in DU-145 prostate cancer, Hoc8 DLC2 DQTTRRSPYRMSRILARHQLVTKIQQEIEAKEACDWLRAAGFPQYAQLYEDSQFPINIVA 90 ovarian cancer, and MDA-MB-468 and MDA-MB-436 DLC3 VKKHNGFLDEDSLGALCRRLMTLNNCASMKLEVHFQSKQNEDSEEEEQCTISSHWAFQQE 120 breast carcinoma cells (Figure 5). In liver cancer cell DLC1 VKREHDFLDRDAIEALCRRLNTLNKCAVMKLEISPHRKRSDDSDEDEPCAISGKWTFQRD 106 DLC2 VKNDHDFLEKDLVEPLCRRLNTLNKCASMKLDVNFQRKKGDDSDEEDLC-ISNKWTFQRT 149 lines, no DLC-3 transcript was detected in ﬁve out of 16 DLC3 SKCWSPMGSSDLLAPPSP------GLPATSSCESVLTELSATSL 158 lines assayed by RT–PCR. DLC-3 expression in primary DLC1 SKRWSRLEEFDVFSPKQDLVPGSPDDSHPKDGPSPGGTLMDLSERQEVSSVRSLSSTGSL 166 DLC2 SRRWSRVDDLYTLLPRGDRNG------SPGGT--GMRNTTSSESVLTDLSEPEV 195 tumors from different tissues was analysed using a

DLC3 PVITVSLPPEPADLPLPGRAPSSSDRPLLSPTQGQEGPQDKAKK------RHR 205 Cancer Proﬁling Array (Figure 6a). Relative to the DLC1 PSHAPPSEDAATPRTNSVISVCSSSNLAGNDDSFGSLPSPKELSSFSFSMKGHEKTAKSK 226 DLC2 CSIHSESSGGSDSRSQPGQCCTDNPVMLDAPLVSSSLPQPPRDVLNHPFHPKNEKPTRAR 255 adjacent normal tissue, DLC-3 mRNA levels were

DLC3 NRSFLKHLESLRRKEKSGSQQAEPKHSPATSEKVSKA------242 signiﬁcantly downregulated in a high percentage of DLC1 TRSLLKRMESLKLKSSHHSKHKAPSKLGLIISGPILQEGMDEEKLKQLNCVEISALNGNR 286 kidney, lung, uterine, ovarian and breast cancers. Owing DLC2 AKSFLKRMETLRGKGAHGRHKGSGRTGGLVISGPMLQQEPESFKAMQC----IQIPNGDL 311 to the low number of prostate cancer cases in this set, DLC3 --SSFRSCRGFLSAGFYRAKNWAATSAGGSGANTRKAWEAWPVAS------285 DLC1 INVPMVRKRSVSNSTQTSSSSSQSETSSAVSTPSPVTRTRSLSACNKRVGMYLEGFDPFN 346 DLC-3 expression was examined in larger group of DLC2 QNSPPPACRKGLPCSGKSSGESSPSEHSSSGVSTPCLKERKCHEANKRGGMYLEDLDVLA 371 prostate neoplasms, using real-time RT–PCR to quan- DLC3 ------FRHPQWTHRGDCLVHVPGDHKPGTFPRSLSIESLCPEDGHRLADWQPGRR 335 DLC1 QSTFNNVMEQNFKNRESYPEDTVFYIPEDHKPGTFPKALTNGSFSPSGNNGSVNWRTGSF 406 titate the transcript levels. In 60% of primary prostate DLC2 GTALPDAGDQSRMHEFHSQENLVVHIPKDHKPGTFPKALSIESLSPTDSSNGVNWRTGS- 430 carcinomas, DLC-3 expression was downregulated DLC3 WGCEGRRGSCGST------GSHASTYDNLPELYPAEPVMVGAEAEDEDDEESGGSY 385 DLC1 HGPGHISLRRENSSDSPKELKRRNSSSSMSSRLSIYDNVPGSILYSSSGDLADLENEDIF 466 relative to normal prostate tissue. DLC-3 expression DLC2 ------ISLGREQVPGAREPRLMASCHRASRVSIYDNVPGSHLYASTGDLLDLEKDDLF 483 was not signiﬁcantly altered in benign prostate hyper- DLC3 AHLDDILQHVWGLQQRVELWSRAMYPDLGPGDEEEEEATSSVEIATVEVKCQAEALSQME 445 plasias (Figure 6b). DLC1 PELDDILYHVKGMQRIVNQWSEKFSDEGDSDSALD------501 DLC2 PHLDDILQHVNGLQEVVDDWSKDVLPELQTHDTLVGEPGLST------525

DLC3 VPAHGESPAWAQAEVQPAVLAPAQAPAEAEPVAQEEAEAPAPAPAPAPAQDSEQEAHSGG 505 DLC1 ------Effect of DLC-3a expression on tumor cell growth DLC2 ------The ability of DLC-1 and DLC-2 to inhibit cell DLC3 EPTFASSLSVEEGHSISDTVASSSELDSSGNSMNEAEAAGSLAGLQASMPRERRDSGVGA 565 DLC1 SVSPCPSSPKQIHLDVDNDRTTPSDLDSTGNSLNEPEEP------SEIPERRDSGVGA 553 proliferation led us to examine whether DLC-3 pos- DLC2 FPSPNQITLDFEGNSVSEGRTTPSDVERDVTSLNESEPP------GVRDRRDSGVGA 576 sessed similar activity. Human tumor cell lines with DLC3 SLTRPCR-KLRWHSFQNSHRPSLNSESLEINRQFAGQINLLHKGSLLRLTAFMEKYTVPH 624 low levels of the endogenous DLC-3 transcript were DLC1 SLTRSNRHRLRWHSFQSSHRPSLNSVSLQINCQSVAQMNLLQKYSLLKLTALLEKYTPSN 613 DLC2 SLTRPNR-RLRWNSFQLSHQPRPAPASPHISSQTASQLSLLQRFSLLRLTAIMEKHSMSN 635 transfected with an expression vector encoding the DLC3 KQGWVWSMPKFMRRNKTPDYRGQHVFGVPPLIHVQRTGQPLPQSIQQAMRYLRSQCLDQV 684 complete open reading frame of the DLC-3a isoform. DLC1 KHGFSWAVPKFMKRIKVPDYKDRSVFGVPLTVNVQRTGQPLPQSIQQAMRYLRNHCLDQV 673 DLC2 KHGWTWSVPKFMKRMKVPDYKDKAVFGVPLIVHVQRTGQPLPQSIQQALRYLRSNCLDQV 695 Overexpression of DLC-3a inhibited cell proliferation in

DLC3 GIFRKSGVKSRIQNLRQMNETSPDNVCYEGQSAYDVADLLKQYFRDLPEPIFTSKLTTTF 744 MCF-7 breast cancer cells and in 22Rv1 prostate cancer DLC1 GLFRKSGVKSRIQALRQMNEGAIDCVNYEGQSAYDVADMLKQYFRDLPEPLMTNKLSETF 733 DLC2 GLFRKSGVKSRIHALRQMNENFPENVNYEDQSAYDVADMVKQFFRDLPEPLFTNKLSETF 755 cells when compared to cells transfected with the control

DLC3 LQIYQLLPKDQWLAAAQAATLLLPDENREVLQTLLYFLSDIASA-EENQMTAGNLAVCLA 803 vector (Figure 7). DLC-3a expression also reduced DLC1 LQIYQYVPKDQRLQAIKAAIMLLPDENREVLQTLLYFLSDVTAAVKENQMTPTNLAVCLA 793 colony formation in MCF-7 and 22Rv1 cells by 67 DLC2 LHIYQYVSKEQRLQAVQAAILLLADENREVLQTLLCFLNDVVNLVEENQMTPMNLAVCLA 815 and 65%, respectively. The effect of DLC-3a expression DLC3 PSIFHLNVSKKDSPSPRIKSKRSLIGRPGPRDLSDNMAATQGLSHMISDCKKLFQVPQDM 863 DLC1 PSLFHLNTLKRENSSPRVMQRKQSLGKPDQKDLNENLAATQGLAHMIAECKKLFQVPEEM 853 on anchorage-independent growth was tested in 22Rv1 DLC2 PSLFHLNLLKKE-SSPRVIQKKYATGKPDQKDLNENLAAAQGLAHMIMECDRLFEVPHEL 874 cells and in the MDA-MB-361 breast cancer cell line, in DLC3 VLQLCSSYSAAELSPPGPALAELRQAQAAGVSLSLYMEENIQDLLRDAAERFKGWMSVPG 923 DLC1 -SRCRNSYTEQELKPL-TLEALGHLGNDDSADYQHFLQDCVDGLFKEVKEKFKGWVSYST 911 which DLC-3 mRNA levels are also low (not shown). DLC2 VAQSRNSYVEAEIHVP-TLEELGTQLEESGATFHTYLNHLIQGLQKEAKEKFKGWVTCSS 933 DLC-3a expression decreased the number of colonies in DLC3 PQHTELACRKAPDGHPLRLWKASTEVAAPPAVVLHRVLRERALWDEDLLRAQVLEALMPG 983 DLC1 SEQAELSYKKVSEGPPLRLWRSVIEVPAVPEEILKRLLKEQHLWDVDLLDSKVIEILDSQ 971 soft agar by 49% in 22Rv1 cells and by 71% in MDA- DLC2 TDNTDLAFKKVGDGNPLKLWKASVEVEAPPSVVLNRVLRERHLWDEDFVQWKVVETLDRQ 993 MB-361 cells. DLC3 VELYHYVTDSMAPHPCRDFVVLRMWRSDLPRGGCLLVSQSLDPEQPVPESGVRALMLTSQ 1043 DLC1 TEIYQYVQNSMAPHPARDYVVLRTWRTNLPKGACALLLTSVDHDRAPVV-GVRVNVLLSR 1030 DLC2 TEIYQYVLNSMAPHPSRDFVVLRTWKTDLPKGMCTLVSLSVEHEEAQLLGGVRAVVMDSQ 1053

DLC3 YLMEPCGLGRSRLTHICRADLRGRSPDWYNKVFGHLCAMEVAKIRDSFPTLQAAGPETKL 1103 DLC1 YLIEPCGPGKSKLTYMCRVDLRGHMPEWYTKSFGHLCAAEVVKIRDSFSNQNTETKDTKSR 1091 Discussion DLC2 YLIEPCGSGKSRLTHICRIDLKGHSPEWYSKGFGHLCAAEVARIRNSFQPLIAEGPETKI 1113 Figure 2 Alignment of the amino acid sequences of the three DLC-1 and DLC-2 are regulators of Rho GTPase human deleted in liver cancer family members. The amino acid activity that inﬂuence the proliferation, cytoskeletal sequences of DLC-3a (residues 1–1103 of UniProt entry Q68DG7, the translation of nt 116–3427 of CR749411), DLC-1 (residues organization and motility of cultured cells. The DLC 1–1091 of UniProt Q96QB1) and DLC-2 (residues 30–1113 of the a family proteins also have important roles in normal isoform, UniProt Q9Y3M8) were aligned using the ClustalW 1.81 development, as null alleles of the mouse DLC-1 gene program, and the alignment in the serine-rich domain was then (Durkin et al., 2005) and the single Drosophila DLC-1- edited manually. The amino acid numbering is on the right, and et al dashes indicated gaps introduced to maximize the sequence like gene (Denholm ., 2005) result in defective identity. Amino acids that are identical in at least two of the three morphogenesis and embryonic lethality. In this report, sequences are highlighted in gray. The three amino acids in the we present evidence that DLC-3, a third member of RhoGAP domains important for biological activity are highlighted the deleted in liver cancer family, may be involved in in black. regulating cell growth. The three DLC genes have an overlapping expression pattern in adult tissues but do not appear to be redundant, as the severe phenotype of DLC-1À/À mice suggests that DLC-2 and -3 were unable Expression of DLC-3 in human cancers to compensate for the loss of DLC-1 during embryo- Northern blot hybridization showed that DLC-3 genesis (Durkin et al., 2005). There may be variations in mRNA levels were low in most of the human breast, the temporal or cell-type expression of the DLC-1, -2 prostate and ovarian cancer lines tested, but the mRNA and -3 genes, or the amino acid sequence differences

Oncogene DLC-3 and cancer ME Durkin et al 4584

Figure 3 DLC-3 protein structure and comparison with other DLC family members. (a) On top is a diagram of the domain organization of the DLC-3a polypeptide, showing the SAM, serine-rich (SR), RhoGAP and START domains. The amino acids comprising the domains are indicated. Below are comparisons of the mouse DLC-3b (UniProt Q8K031), human DLC-1 and DLC-2a proteins, giving the percent identity between their individual domains and those of human DLC-3a. The sequences were aligned, and the percent identities were calculated by GAP program of Wisconsin Package Version 10.2 software suite (Accelrys Inc., San Diego, CA, USA), using the default settings. (b) Schematic map of the paralogous chromosomal regions containing the DLC1, STARD13 (DLC2) and STARD8 (DLC3) loci. The gene order is based on the DNA sequence maps of the three chromosomes; the distances are not drawn to scale. The paralogous blocks on Xq and 13q were previously identiﬁed (McLysaght et al., 2002; available online at http:// wolfe.gen.tcd.ie/dup). Paralogous genes on 8p were identiﬁed by visual inspection of the human genome map at the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/mapview/). To the left is a phylogenetic tree of the Xq, 13q and 8p chromosomal regions, based on the degree of similarity between the DLC, MTMR (myotubularin) and SLC7 (amino acid transporter) family members, as calculated by the Hovergen database of homologous vertebrate genes at the ExPASY Proteomics server (ca.expasy.org).

between the three proteins may be sufficient to alter The high degree of similarity between the GAP their interaction repertoires and lead to functional domain of DLC-3 and those of DLC-1 and DLC-2 specialization. suggests that DLC-3 may also act as a GAP with The DLC-3 gene appears to have two promoters that comparable substrate specificity. The RhoGAP domains yield transcripts with alternative 50 ends. The sequences of DLC-1 and DLC-2 were found to be essential for flanking both of the putative transcription starts sites their effects on cytoskeletal organization and cell upstream of exons 1A and 1C have a high GC content, proliferation (Sekimata et al., 1999; Ching et al., 2003; typical of the CpG islands associated with promoter Leung et al., 2005; Wong et al., 2005). The other regions. The protein products of the DLC-3 a and b domains of the DLC family polypeptides may regulate transcripts are predicted to differ by the presence or the activity of the GAP domain through intra- or absence, respectively, of the N-terminal SAM domain. intermolecular interactions or mediate the subcellular Similar variants have been reported for the human localization of the protein. The START domain is a DLC-2 gene, the DLC-2a (plus SAM) and DLC-2g potential lipid-binding domain (Ponting and Aravind, (minus SAM) forms (Leung et al., 2005). SAM domains 1999; Alpy and Tomasetto, 2005) and could be involved are approximately 70-aa motifs that mediate protein– in membrane targeting of the protein or in conferring protein interactions (Qiao and Bowie, 2005), and the lipid-regulated conformational changes, like the lipid- two DLC-3 isoforms could have different biological binding domains of other RhoGAPs (Bernards and properties, as was shown for variants of diacylglycerol Settleman, 2004). kinase Z with or without a C-terminal SAM domain The largest domain in the DLC-3 protein is the 530-aa (Murakami et al., 2003). serine-rich domain, encoded primarily by a single large

Oncogene DLC-3 and cancer ME Durkin et al 4585

Figure 4 Expression of DLC-3 mRNA in human tissues. (a) Autoradiogram of a multiple tissue Northern blot of poly(A þ ) RNA from the indicated tissues hybridized to a 32P-labeled DLC-3 cDNA probe (5F/5R1). The same blot was re-hybridized to a ribosomal protein L19 probe (RpL19) as a control for loading. (b) Aliquots of cDNA from human brain, heart, kidney, liver, peripheral blood lymphocytes (PBL) and spleen RNA were amplified with primers specific for human DLC-1, DLC-2 and DLC-3 (7F/9R). Control amplifications were performed with primers for human GAPDH.

exon, as in the DLC-1 and -2 genes. This region is predicted to contain a number of potential serine/ threonine phosphorylation sites, and recently Ser322 of Figure 5 Expression of DLC-3 in human cancer cell lines. rat DLC-1/p122RhoGAP was found to be phosphory- Autoradiograms of Northern blots of total RNA (15 mg per lane) lated in adipocytes after insulin treatment (Hers et al., from human breast (a), ovarian (b) and prostate cancer (c) cell lines hybridized to a 32P-labeled DLC-3 cDNA probe (7F/11R). The 2006). Both serine/threonine and tyrosine phosphoryla- blots were re-hybridized to a human b-actin cDNA probe as a tion have been found to regulate the activity of other control for loading. (d) RT–PCR analysis of DLC-3 mRNA in RhoGAP proteins (Moon and Zheng, 2003; Bernards human hepatocellular carcinoma cell lines. cDNAs were amplified and Settleman, 2004). The serine-rich region of DLC-3 with primers specific for human DLC-3 (7F/11R) and GAPDH. also has potential motifs for binding to proline- recognition domains, and thus this domain may possess multiple sites for interacting with other proteins to form additional clinical features resulting from mosaic signaling complexes. DLC-3 deficiency. The gene-encoding DLC-3 is flanked by several well- DLC-3 mRNA levels were low or undetectable in characterized genes that are mutated in human genetic most of the human breast, ovarian and prostate cancer disorders, including androgen insensitivity (AR, Avila cell lines tested and were also decreased in the majority et al., 2001), one form of X-linked mental retardation of primary tumors from these tissues. Our results are (OPHN1, Billuart et al., 1998), and the skeletal disorder consistent with microarray gene expression profiling craniofrontonasal syndrome (EFNB1, Wieland et al., data showing downregulation of DLC-3 expression in 2004). A female with craniofrontonasal syndrome was human prostate tumors (cited in Alpy and Tomasetto, recently found to have a deletion of EFNB1 that also 2005) and in transgene-induced mouse mammary included part of STARD8/DLC3 (Twigg et al., 2006). As tumors (Thakur et al., 2005). Transcription of the STARD8 is subject to X-inactivation (Carrel and DLC-3 gene could be repressed by epigenetic mechan- Willard, 2005), further phenotypic analyses of patients isms such as promoter DNA hypermethylation and with deletions extending to STARD8 could reveal histone hypoacetylation, as has been found for DLC-1

Oncogene DLC-3 and cancer ME Durkin et al 4586

Figure 6 DLC-3 expression in human tumors. (a) The human DLC-3 cDNA (7F/11R) probe was hybridized to the Cancer Proﬁling Array I containing cDNAs from matched normal (N) and tumor (T) tissues of 241 patients. On top is a section of the autoradiogram showing the results for kidney and lung tumor samples, and below is a table summarizing the data for 10 tissues, indicating the percentage of cases in which DLC-3 expression was increased, decreased or unchanged in tumors relative to the adjacent normal tissue. (b) Analysis of DLC-3 expression in prostate neoplasms by quantitative real-time RT–PCR. DLC-3 mRNA levels were measured in three normal prostate tissues, 10 cases of benign prostate hyperplasia (BPH), and 10 prostate carcinomas (PCa), and the expression was normalized to that of GAPDH. In parentheses is the number of samples in which the relative DLC-3 expression was less than one- half of the average normal level, indicated by the horizontal line.

in prostate carcinomas (Guan et al., 2006). Reduced evidence that alterations in DLC-3 may affect cellular DLC-3 mRNA levels could also result from genomic functions. DNA losses. Loss of heterozygosity studies have suggested the presence of one or more tumor suppressor genes on the X chromosome (reviewed in Spatz et al., Materials and methods 2004), and the DLC-3 locus is close to an Xq region commonly lost in ovarian epithelial tumors (Edelson RNA preparation and analysis et al., 1998). Generation of DLC-3-speciﬁc antibodies Human cancer cell lines were obtained and cultured as will be needed to determine whether DLC-3 expression described previously (Guan et al., 2006; Ullmannova and in tumors is reduced at the protein level and whether this Popescu, 2006), and total RNA was isolated using the TRIzol is correlated with changes in mRNA levels. reagent (Invitrogen, Carlsbad, CA, USA) or the RNeasy Mini Consistent with a potential role for DLC-3 as a tumor Kit (Qiagen, Valencia, CA, USA). Total RNA from normal suppressor, we found that overexpression of DLC-3a in human liver, mammary gland and prostate were purchased from Stratagene (La Jolla, CA, USA) or Clontech (Mountain human breast and prostate cancer cell lines inhibited View, CA, USA). First strand cDNA was transcribed from colony formation and cell proliferation. DLC-3 expres- total RNA using the Thermoscript RT–PCR kit (Invitrogen) or sion also reduced anchorage-independent growth in soft the cDNA Archive kit (Applied Biosystems, Foster City, CA, agar, an in vitro measure of tumorigenicity. Over- USA). The six tissue Multiple Choice Human First Strand expression of DLC-3 could decrease the global levels cDNA panel (CH1101) was purchased from OriGene Tech- of active, GTP-bound Rho proteins or inhibit the GDP– nologies (Rockville, MD, USA). The sequences of the primers GTP cycling process, interfering with signals necessary used for PCR ampliﬁcation of cDNAs are listed in Table 2. For for promoting cell growth and survival. A better Northern-blot analysis, aliquots of total RNA from human understanding of the role of DLC-3 in signal transduc- cancer cell lines were denatured, fractionated by agarose gel tion will require additional characterization of the electrophoresis and transferred to nylon membranes as described (Yuan et al., 2003). A 12-tissue human poly (A þ ) polypeptide, to identify its interaction partners and the RNA Northern blot was purchased from OriGene Technolo- upstream factors that regulate its activity. In a recent gies. DLC-3 cDNA probes were prepared by RT–PCR using large-scale mutational analysis of human cancers, primers 5F/5R1 or 7F/11R and labeled with [32P]dCTP by missense mutations in STARD8 were found in human oligolabeling. Northern blots were hybridized to the probes and breast tumors (Sjo¨ blom et al., 2006), providing further washed as described previously (Yuan et al., 2003).

Oncogene DLC-3 and cancer ME Durkin et al 4587

Figure 7 Effect of DLC-3a overexpression on human breast and prostate cancer cells. (a) DLC-3a inhibits cell proliferation. MCF-7 breast cancer cells and 22Rv1 prostate cancer cells were transfected with the DLC-3a/pcDNA3.1 expression vector (DLC-3) or with the pcDNA3.1 vector with no insert (vector), and cell growth was measured by reduction of MTT. (b) DLC-3a inhibits colony formation. Crystal violet stained colonies of MCF-7 and 22Rv1 cells transfected with the DLC-3a expression vector or the control vector. (c) DLC-3a expression inhibits anchorage-independent growth. Transfection of 22Rv1 cells and MDA-MB-361 breast cancer cells (MDA 361) with the DLC-3a vector reduced colony formation in soft agar compared with cells transfected with the control vector.

Table 2 PCR primers Expression of DLC-3 in human tumors

a The Cancer Profiling Array I, which contains normalized Name Sequence Location cDNAs from matched pairs of tumor and normal tissues from DLC-3b 241 individuals with 19 tumor types, was purchased from BD 1F 50-accatgcctctgctggacgtt nt 223–243 Biosciences (San Jose, CA, USA). Hybridization of the filter to 5F 50-gtaagtgctggtctcctatggg nt 459–480 32P-labeled probes and analysis of the data were performed as 5R1 50-gcttggcatttgacctcaactg nt 1389–1410 described previously (Ullmannova and Popescu, 2006). 5R2 50-ggctactgatggtacactgctc nt 416–437 5R3 50-gagtattggcaccactgccacc nt 899–920 DLC-3 expression in prostate tumor samples 7F 50-ctggaccaagtaggcatcttcc nt 2138–2159 0 Tissue specimens from three normal prostates, 10 prostate 9R 5 -aggcacactgccaggttgcctg nt 2481–2502 carcinomas and 10 cases of benign prostate hyperplasia were 11R 50-ctcttccatgtagaggctcagg nt 2782–2803 obtained from the University Hospital of Heraklion, Crete, DLC-1 (AF035119) Greece, and total RNA was extracted from the tissue samples 7F 50-cacaggacaaccgttgcctcag nt 2271–2292 and transcribed into cDNA as described earlier (Guan et al., 9R 50-ctcttcagggtgttgagatgga nt 2714–2735 2006). Gene expression was assessed by quantitative real-time RT–PCR using the ABI PRISM 7900 Sequence Detection DLC-2 (AY082589) System instrument and software (Applied Biosystems) with 7F 50-cagcaactgcctcgatcaggtg nt 2185–2206 QuantiTect SYBR Green PCR reagents (Qiagen). Primers for 0 10R 5 -ccaccaactcgtgtggaacctc nt 2726–2747 human DLC-3/STARD8 (catalogue # PPH10566A) were purchased from SuperArray (Frederick, MD, USA). Glycer- GAPDH (NM_002046) 8F 50-cgaccactttgtcaagctca nt 1014–1033 aldehyde-3-phosphate dehydrogenase (GAPDH) was used as 9R 50-aggggagattcagtgtggtg nt 1197–1216 an endogenous reference gene for normalizing variance in the quality of RNA and the amount of input cDNA, using primers aNumbers refer to the exons in which the sequences are located. described by Song et al. (2006). After an initial denaturation Forward (F) and reverse (R) primers are indicated. bPrimer 1F step at 951C for 15 min, 40 cycles of amplification (941C, 15 s; sequence is from BC035587, and the other sequences are from D80011. 551C, 30 s; 551C, 30 s) were performed. Amplification was

Oncogene DLC-3 and cancer ME Durkin et al 4588 immediately followed by a melt program to validate the insert, using GeneJuice Transfection Reagent (Novagen, specificity of the PCR products. The 2ÀDDCT method was used Madison, WI, USA) for 2 days. To determine the effect to calculate the relative fold difference in DLC-3 expression of DLC-3a expression on cell growth, 2 Â 102,2Â 103 and compared to the average ratio of normal tissues. A twofold 2 Â 104 of MCF-7 and 22Rv1 cells were seeded in triplicate reduction was considered significant. on a 96-well plate and incubated for 10 days. Cell proliferation was measured every other day by the colorimetric DLC-3a expression vector construction 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide First strand cDNA synthesized from Hoc8 ovarian carcinoma (MTT) reduction assay (Promega, Madison, WI, USA). RNA was amplified by PCR using the DLC-3 1F/5R3 primers, For colony formation assays, cells were grown for 2 days and the product was digested with XhoI. The 0.55 kb fragment in medium containing 250 mg/ml G418 sulfate (Geneticin, was gel-purified, cloned into the EcoRV- and XhoI-cut Invitrogen), and then 1 Â 106 cells were seeded in a 100 mm pBluescript SK vector (Stratagene), and sequenced to verify dish with G418-supplemented medium for 2 weeks. Colonies its identity. The plasmid DNA was digested with XhoI and were stained with 0.5% crystal violet after fixation with SalI and ligated to a 3.1 kb XhoI/SalI fragment (nt 683–3810 of 70% ethanol. To test for anchorage-independent growth, GenBank BC035587) from the DLC-3g cDNA clone (IMAGE transfected cells cultured in G418 for 2 days were seeded on 5518429, purchased from the American Type Culture Collec- soft agar plates according to the manufacturer’s protocol tion). The resulting plasmid contained a 3.6 kb insert with the (Chemicon, Temecula, CA, USA) and cultured in the complete open reading frame of the DLC-3a cDNA. The insert appropriate medium for 4weeks. Colonies were then stained was excised as an EcoI/SalI fragment and cloned into EcoRI- overnight with Cell Staining Solution (Chemicon) and and XhoI-cut pcDNA3.1 (Invitrogen), to generate a plasmid counted. that expressed the DLC-3a isoform under the control of the cytomegalovirus promoter. Acknowledgements Analysis of DLC-3-transfected cells Cells at 70% confluency were transfected with the DLC3a/ This work was supported by the Intramural Research Program pcDNA3.1 plasmid or with the pcDNA3.1 vector with no of the National Cancer Institute, NIH.

References

Alpy F, Tomasetto C. (2005). Give lipids a START: the StAR- Edelson MI, Lau CC, Colitti CV, Welch WR, Bell DA, related lipid transfer (START) domain in mammals. J Cell Berkowitz RS et al. (1998). A one centimorgan deletion unit Sci 118: 2791–2801. on chromosome Xq12 is commonly lost in borderline and Avila DM, Zoppi S, McPhaul MJ. (2001). The androgen invasive epithelial ovarian tumors. Oncogene 16: 197–202. receptor (AR) in syndromes of androgen insensitivity and in Gomez del Pulgar T, Benitah SA, Valeron PF, Espina C, prostate cancer. J Steroid Biochem Mol Biol 76: 135–142. Lacal JC. (2005). Rho GTPase expression in tumourigenesis: Bernards A. (2003). GAPs galore! A survey of putative Ras evidence for a significant link. Bioessays 27: 602–613. superfamily GTPase activating proteins in man and Droso- Goodison S, Yuan J, Sloan D, Kim R, Li C, Popescu NC et al. phila. Biochim Biophys Acta 1603: 47–82. (2005). The RhoGAP protein DLC-1 functions as a Bernards A, Settleman J. (2004). GAP control: regulating the metastasis suppressor in breast cancer cells. Cancer Res 65: regulators of small GTPases. Trends Cell Biol 14: 377–385. 6042–6053. Billuart P, Bienvenu T, Ronce N, des Portes V, Vinet MC, Guan M, Zhou X, Soulitzis N, Spandidos DA, Popescu NC. Zemni R et al. (1998). Oligophrenin-1 encodes a rhoGAP (2006). Aberrant methylation and deacetylation of deleted in protein involved in X-linked mental retardation. Nature 392: liver cancer-1 gene in prostate cancer: potential clinical 923–926. applications. Clin Cancer Res 12: 1412–1419. Brown MC, Curtis MS, Turner CE. (1998). Paxillin LD motifs Hanahan D, Weinberg RA. (2000). The hallmarks of cancer. may define a new family of protein recognition domains. Nat Cell 100: 57–70. Struct Biol 5: 677–678. Hers I, Wherlock M, Homma Y, Yagisawa H, Tavare JM. Carrel L, Willard HF. (2005). X-inactivation profile reveals (2006). Identification of p122RhoGAP (deleted in liver extensive variability in X-linked gene expression in females. cancer-1) Serine 322 as a substrate for protein kinase B Nature 434: 400–404. and ribosomal S6 kinase in insulin-stimulated cells. J Biol Ching YP, Wong CM, Chan SF, Leung TH, Ng DC, Jin DY Chem 281: 4762–4770. et al. (2003). Deleted in liver cancer (DLC) 2 encodes a Homma Y, Emori Y. (1995). A dual functional signal RhoGAP protein with growth suppressor function and is mediator showing RhoGAP and phospholipase C-delta underexpressed in hepatocellular carcinoma. J Biol Chem stimulating activities. EMBO J 14: 286–291. 278: 10824–10830. Jaffe AB, Hall A. (2005). Rho GTPases: biochemistry and Denholm B, Brown S, Ray RP, Ruiz-Gomez M, Skaer H, biology. Annu Rev Cell Dev Biol 21: 247–269. Hombria JC. (2005). crossveinless-c is a RhoGAP required Leung TH, Ching YP, Yam JW, Wong CM, Yau TO, Jin DY for actin reorganisation during morphogenesis. Development et al. (2005). Deleted in liver cancer 2 (DLC2) suppresses cell 132: 2389–2400. transformation by means of inhibition of RhoA activity. Durkin ME, Avner MR, Huh CG, Yuan BZ, Thorgeirsson SS, Proc Natl Acad Sci USA 102: 15207–15212. Popescu NC. (2005). DLC-1, a Rho GTPase-activating Li SS. (2005). Specificity and versatility of SH3 and other protein with tumor suppressor function, is essential for proline-recognition domains: structural basis and implications embryonic development. FEBS Lett 579: 1191–1196. for cellular signal transduction. Biochem J 390: 641–653. Durkin ME, Yuan BZ, Thorgeirsson SS, Popescu NC. (2002). McLysaght A, Hokamp K, Wolfe KH. (2002). Extensive Gene structure, tissue expression, and linkage mapping of genomic duplication during early chordate evolution. Nat the mouse DLC-1 gene (Arhgap7). Gene 288: 119–127. Genet 31: 200–204.

Oncogene DLC-3 and cancer ME Durkin et al 4589 Moon SY, Zheng Y. (2003). Rho GTPase-activating proteins Syed V, Mukherjee K, Lyons-Weiler J, Lau KM, Mashima T, in cell regulation. Trends Cell Biol 13: 13–22. Tsuruo T et al. (2005). Identification of ATF-3, caveolin-1, Murakami T, Sakane F, Imai S, Houkin K, Kanoh H. (2003). DLC-1, and NM23-H2 as putative antitumorigenic, proges- Identification and characterization of two splice variants terone-regulated genes for ovarian cancer cells by gene of human diacylglycerol kinase Z. J Biol Chem 278: profiling. Oncogene 24: 1774–1787. 34364–34372. Thakur A, Xu H, Wang Y, Bollig A, Biliran H, Liao JD. Nagaraja GM, Kandpal RP. (2004). Chromosome 13q12 (2005). The role of X-linked genes in breast cancer. Breast encoded Rho GTPase activating protein suppresses growth Cancer Res Treat 93: 135–143. of breast carcinoma cells, and yeast two-hybrid screen shows Twigg SR, Mastumoto K, Kidd AM, Goriely A, Taylor IB, its interaction with several proteins. Biochem Biophys Res Fisher RB et al. (2006). The origin of EFNB1 mutations in Commun 313: 654–665. craniofrontonasal syndrome: frequent somatic mosaicism Nagase T, Seki N, Ishikawa K, Tanaka A, Nomura N. (1996). and explanation of the paucity of carrier males. Am J Hum Prediction of the coding sequences of unidentified human Genet 78: 999–1010. genes. V. The coding sequences of 40 new genes (KIAA0161- Ullmannova V, Popescu NC. (2006). Expression profile of the KIAA0200) deduced by analysis of cDNA clones from tumor suppressor genes DLC-1 and DLC-2 in solid tumors. human cell line KG-1. DNA Res 3: 17–24. Int J Oncol 29: 1127–1132. Ng IO, Liang ZD, Cao L, Lee TK. (2000). DLC-1 is deleted in Van Aelst L, D’Souza-Schorey C. (1997). Rho GTPases and primary hepatocellular carcinoma and exerts inhibitory signaling networks. Genes Dev 11: 2295–2322. effects on the proliferation of hepatoma cell lines with Wieland I, Jakubiczka S, Muschke P, Cohen M, Thiele H, deleted DLC-1. Cancer Res 60: 6581–6584. Gerlach KL et al. (2004). Mutations of the ephrin-B1 gene Ponting CP, Aravind L. (1999). START: a lipid-binding cause craniofrontonasal syndrome. Am J Hum Genet 74: domain in StAR, HD-ZIP and signalling proteins. Trends 1209–1215. Biochem Sci 24: 130–132. Wong CM, Lee JM, Ching YP, Jin DY, Ng IO. (2003). Qiao F, Bowie JU. (2005). The many faces of SAM. Sci STKE Genetic and epigenetic alterations of DLC-1 gene in 2005: re7. hepatocellular carcinoma. Cancer Res 63: 7646–7651. Ridley AJ. (2001). Rho family proteins: coordinating cell Wong CM, Yam JW, Ching YP, Yau TO, Leung TH, Jin DY responses. Trends Cell Biol 11: 471–477. et al. (2005). Rho GTPase-activating protein deleted in liver Sekimata M, Kabuyama Y, Emori Y, Homma Y. (1999). cancer suppresses cell proliferation and invasion in hepato- Morphological changes and detachment of adherent cells cellular carcinoma. Cancer Res 65: 8861–8868. induced by p122, a GTPase-activating protein for Rho. Yuan BZ, Jefferson AM, Baldwin KT, Thorgeirsson SS, J Biol Chem 274: 17757–17762. Popescu NC, Reynolds SH. (2004). DLC-1 operates as a Seng TJ, Low JS, Li H, Cui Y, Goh HK, Wong ML et al. tumor suppressor gene in human non-small cell lung (2006). The major 8p22 tumor suppressor DLC1 is carcinomas. Oncogene 23: 1405–1411. frequently silenced by methylation in both endemic and Yuan BZ, Miller MJ, Keck CL, Zimonjic DB, Thorgeirsson SS, sporadic nasopharyngeal, esophageal, and cervical carcino- Popescu NC. (1998). Cloning, characterization, and chromo- mas, and inhibits tumor cell colony formation. Oncogene somal localization of a gene frequently deleted in human liver [Epub ahead of print]. cancer (DLC-1) homologous to rat RhoGAP. Cancer Res 58: Sjo¨ blom T, Jones S, Wood LD, Parsons DW, Lin J, Barber TD 2196–2199. et al. (2006). The consensus coding sequences of human breast Yuan BZ, Zhou X, Durkin ME, Zimonjic DB, Gumundsdottir K, and colorectal cancers. Science 314: 268–274. Eyfjord JE et al. (2003). DLC-1 gene inhibits human breast Song YF, Xu R, Zhang XH, Chen BB, Chen Q, Chen YM cancer cell growth and in vivo tumorigenicity. Oncogene 22: et al. (2006). High-frequency promoter hypermethylation of 445–450. the deleted in liver cancer-1 gene in multiple myeloma. J Clin Zhou X, Thorgeirsson SS, Popescu NC. (2004). Restoration of Pathol 59: 947–951. DLC-1 gene expression induces apoptosis and inhibits both Spatz A, Borg C, Feunteun J. (2004). X-chromosome genetics cell growth and tumorigenicity in human hepatocellular and human cancer. Nat Rev Cancer 4: 617–629. carcinoma cells. Oncogene 23: 1308–1313.

Oncogene