Melanoma Metastasis Suppression by Chromosome 6: Evidence for a Pathway Regulated by CRSP3 and TXNIP1

[CANCER RESEARCH 63, 432–440, January 15, 2003] Melanoma Metastasis Suppression by Chromosome 6: Evidence for a Pathway Regulated by CRSP3 and TXNIP1

Steven F. Goldberg,2 Mary E. Miele, Naohito Hatta, Minoru Takata,3 Carrie Paquette-Straub, Leonard P. Freedman,4 and Danny R. Welch5 Jake Gittlen Cancer Research Institute, Penn State College of Medicine, Hershey, Pennsylvania 17033 [S. F. G., D. R. W.]; Department of Medical Technology, University of Delaware, Newark, Delaware 19716 [M. E. M., C. P-S.]; The Penn State-National Foundation for Cancer Research Center for Metastasis Research, Hershey, Pennsylvania 17033 [D. R. W.]; Department of Dermatology, Kanazawa University School of Medicine, 13-1 Takaramachi, Kanazawa, 920-8640 Japan [N. H., M. T.]; and Cell Biology Program, Memorial Sloan-Kettering Cancer Center, Sloan-Kettering Division, Joan and Sanford I. Weill Graduate School of Medical Sciences of Cornell University, New York, New York 10021 [L. P. F.]

ABSTRACT We have shown that the MMCT6 of human chr6 into metastatic human melanoma cell lines C8161 or MelJuSo suppressed metastasis Loss of genetic material on chromosome 6 has been associated with without blocking tumor formation (4, 5). Metastasis-suppressed hy- progression of human melanomas. We showed previously that introducing brids of C8161 (neo6/C8161) were blocked at steps in the metastatic chromosome 6 into metastatic human melanoma cell lines suppresses metastasis without affecting the ability of the hybrids to form progres- cascade affecting the survival and proliferation at secondary sites (6). sively growing tumors. By subtractive hybridization comparing nonmeta- We used subtractive hybridization to identify a metastasis suppressor static chromosome 6-containing (neo6/C8161) versus parental (C8161) gene, KISS1, which mapped to 1q32 (7–10). Tumor cells transfected metastatic cells, the KISS1 metastasis suppressor gene was isolated. How- with KISS1 cDNA were nonmetastatic, but remained tumorigenic. ever, KISS1 mapped to chromosome 1q32. To identify upstream regula- MMCT of chr6 with a deletion (ϳ40 Mb) spanning 6q16.3-q23 tor(s) of (and downstream effectors of) KISS1, microarray hybridization (neo6qdel) did not suppresses metastasis (11). Because KISS1 was comparing C8161 and neo6/C8161 variants was performed. TXNIP/ cloned from C8161 cells, the genetic defect must be at the level of VDUP1, a thioredoxin-binding protein, was expressed more highly in KISS1 regulation, not in KISS1 itself. Also, because neo6qdel hybrids neo6/C8161 and in nonmetastatic melanomas. Increased TXNIP expres- did not express KISS1, the upstream regulator(s) of KISS1 was sion inhibited metastasis and up-regulated KISS1. Surprisingly, TXNIP mapped between 6q16.3-q23 (11). Shirasaki et al. verified this hy- also mapped to chromosome 1q. PCR karyotyping that refined the region on chromosome 6 identified CRSP3/DRIP130, a transcriptional coactiva- pothesis in melanoma clinical samples by showing loss of heterozy- tor, as a metastasis suppressor. CRSP3 transfectant cells had up-regulated gosity of markers in that region correlated with decreased KISS1 KISS1 and TXNIP expression and were suppressed for metastasis. Quan- expression (12). titative real-time reverse-transcription PCR of clinical melanoma samples KISS1 is a precursor for secreted neuropeptide ligands [designated showed that loss of CRSP3 expression correlated with decreased KISS1 Metastin (13) or Kisspeptins (14)] for a G-protein coupled [named expression and increased metastasis. Thus, we implicated a specific gene hOT7T175 (13), AXOR12 (15), or hGPR54 (14)]. Although on chromosome 6 in the etiology of melanoma metastasis and identified hOT7T175-transfected B16/BL6 melanoma-injected mice treated potential up-stream regulators of KISS1 and TXNIP. with Metastin developed fewer lung metastases (13), the mechanism of KISS1 suppression is still unknown. Recent publications suggest INTRODUCTION possible mechanisms for KISS1 metastasis suppression. Metastin induces Ca2ϩ in receptor-transfected CHO cells (13), as well as Metastasis is the culmination of neoplastic progression toward phosphorylation of ERK1/2 and weak phosphorylation of p38/MAPK autonomy from host regulation. Acquiring metastatic potential is but not of SAPK/JNK (14). Metastin inhibits motility, chemotaxis, apparently “driven” by genomic instability with coincident or super- and invasion in vitro (13, 16), possibly by repressing the transcription imposed selection pressures (1). To be metastatic, tumor cells must of MMP-9 [via induction of cytosolic I␬B␣ (17)]. Metastin also complete a sequential, multistep cascade involving complex interac- induces excessive formation of focal adhesions and stress fibers in tions between tumor and host. Cells must navigate a series of obsta- hOT7T175-transfected B16/BL6 and induces the phosphorylation of cles, including immune surveillance, loss of original tissue context, FAK and paxillin (13), possibly through Rho (14). invasion through the basement membrane, lymphatic or hematoge- The primary objective of the current study was to identify the nous circulation, extravasation, and growth at the secondary site. Any upstream regulator(s) of KiSS1 on chr6. Additionally, microarray cell shed from the primary neoplasm that fails to complete all requisite hybridizations using metastatic (C8161) and nonmetastatic (neo6/ steps cannot form metastases (2, 3). C8161) variants of a human melanoma cell line could be useful in identifying some additional downstream regulators of KISS1. Received 8/6/02; accepted 11/14/02. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. MATERIALS AND METHODS 1 Supported by NIH Grants CA66128 (to D. R. W.), CA87728 (to D. R. W.), and CA88876 (to M. E. M.); the National Foundation for Cancer Research (to D. R. W.); a Cell Lines and Culture. C8161 is a human melanoma cell line that metasta- Grant-in-Aid for Scientific Research (C2-12670812) from the Japan Society for the sizes after orthotopic or i.v. injection into athymic mice (18). C8161.9 is a highly Promotion of Science (to N. H. and M. T.); a predoctoral fellowship from the Foreman metastatic subclone of C8161 (7). Nonmetastatic neo6/C8161.1, neo6/C8161.2, Foundation (to S. F. G. and D. R. W.); and the Jake Gittlen Memorial Golf Tournament (to D. R. W.). neo6/C8161.3, and neo6/C8161.6 were derived by MMCT using the 2 Present address: Laboratory of Biosystems and Cancer, National Cancer Institute, MCH262A1.D6 donor cell line (4). Metastatic neo6(del)(q16.3-q23)/C8161 hy- Building 37/Room 5046/MSC 4264, 37 Convent Drive, Bethesda, Maryland 20892. brids (neo6qdel/C8161.9.1, neo6qdel/C8161.9.2, neo6qdel/C8161.9.7, neo6qdel/ 3 Present address: Toyama Prefectural Central Hospital, 2-2-78 Nishinagae, Toyama, 930-8550 Japan. 4 Present address: Department of Bone Biology, Merck Research Laboratories, 6 The abbreviations used are: MMCT, microcell-mediated chromosome transfer; chr, WP26A-1000, West Point, Pennsylvania 19486. chromosome; DME-F12, 1:1 mixture of DMEM and Ham’s F-12 medium; cDME-F12, 5 To whom correspondence should be addressed at: Department of Pathology, Uni- mixture of DME-F12 and 5% fetal bovine serum; CMF-DPBS, Ca2ϩ/Mg2ϩ-free Dulbec- versity of Alabama at Birmingham, 1670 University Boulevard, Volker Hall G-038, co’s PBS; TRX, thioredoxin; VGP, vertical growth phase; MTD, mean tumor diameter; Birmingham, AL 35294-0019. E-mail: [email protected]; Phone: (205) 934-2956; RT-PCR, reverse-transcription PCR; RTQ, quantitative real-time RT-PCR; FAM, 6- Fax: (205) 934-1775. carboxyfluorescein; TAMRA, 6-carboxy-tetramethyl-rhodamine. 432

C8161.9.8) were prepared using a related chr6 microcell donor variant (11). using a hemacytometer. Cell numbers were determined daily for 10 days. Presence and retention of the added chromosome was verified by PCR of single- Morphology was recorded using an inverted microscope (Nikon Diaphot) sequence-length polymorphisms, karyotyping, and/or fluorescence in situ hybrid- equipped with a digital camera (DC Viewer Version 3.2.0.0; Leica Microsys- ization (see corresponding references in 7, 11 and 18). tems, Ltd., Heerbrugg, Germany). C8161-derived cells were cultured in DME-F12 (Invitrogen) supplemented In Vivo Experiments. Female athymic mice 3–4 weeks of age (Harlan with 5% fetal bovine serum (Atlanta Biologicals, Atlanta, GA), 1% nonessen- Sprague Dawley, Madison, WI) were maintained under the NIH and The tial amino acids (Invitrogen), and cDME-F12, but no antibiotics or antimy- Pennsylvania State University College of Medicine guidelines. The Institu- cotics. Neomycin-resistant cells were maintained in cDME-F12 containing 500 tional Animal Care and Use Committee approved all protocols. Food and water ␮g/ml of active G418 (Invitrogen). Cultures were maintained at 37°Cina were provided ad libitum. humidified atmosphere with 5% CO2. None of the cultures were infected with For in vivo experiments, C8161.9/pcDNA3 (P6 and 16), neo6/C8161.2 Mycoplasma spp. according to the PCR-based TaKaRa Mycoplasma detection (P20–21), neo6/C8161.6 (P10), C8161.9/TXNIP (P4–5), and C8161.9/CRSP3 kit (Panvera, Madison, WI). Cells were detached using a solution of 2 mM (P4–7) were detached at 70–90% confluence, rinsed with CMF-DPBS, and EDTA in CMF-DPBS (Invitrogen). detached witha2mM EDTA solution. To minimize aggregation, cells were Cell lines representing a continuum of melanoma progression toward in- diluted to the desired concentration in ice-cold HBSS and maintained on ice creasing malignancy were provided by Dr. M. Herlyn (Wistar Institute, Phil- until they were injected into the mice. Injections were performed using a adelphia, PA; Refs. 19–21). These cell lines are derived from normal mela- 27-gauge needle affixed to a 1-cc tuberculin syringe. nocytes, early vertical growth phase (VGP; WM793), VGP (WM115), and Orthotopic injections were made into the dorsolateral flank using 1 ϫ 106 metastases (WM239A, 1205-LU). cells in 0.1-ml ice-cold HBSS. Tumor size was measured weekly and geomet- RNA Isolation. Total RNA was isolated from cultured cells at 70–90% ric MTD was calculated (22). When MTD Ϸ 1.0 cm, tumors were surgically confluence or from tumor tissue after stabilization in RNAlater (Ambion, removed under Ketamine-Xylazine (80–85 mg/kg:14–16 mg/kg) anesthesia Austin, TX) and pulverization with a mortar and pestle. Poly(A) RNA was and the wounds closed with sterile stainless-steel 9-mm clips (Becton-Dick- isolated using Oligotex Suspension (Qiagen, Valencia, CA) according to the inson, Sparks, MD). Tumor tissue for RT-PCR analysis was obtained when manufacturer’s “batch” protocol. mRNA was precipitated using the Pellet tumors were 0.5–1.0 cm MTD to minimize impact of necrotic tissues. Metas- Paint coprecipitant (Novagen, Madison, WI), followed by resuspension in 10 tases were quantified 1 month after tumor removal. mM Tris (pH 8)-1 mM EDTA to a concentration of 50 ng/␮l. For experimental metastasis assays, cells (2 ϫ 105 in 0.2 ml of HBSS) were Microarray Hybridization. Poly(A) RNA was reverse transcribed with injected i.v. into the lateral tail veins of nude mice. Animals were killed 1 Cy3- and Cy5-end-labeled random 9-mers to generate fluorescent single- month after injection and their lungs removed for quantification of metastases. stranded cDNA probes. Probes were competitively hybridized to the Human To quantify lung metastases, lungs were removed and fixed in a diluted Unigene 1 microarray (Incyte Genomics, Palo Alto, CA), which contained Bouin’s solution (20% Bouin’s fixative in neutral buffered formalin). Grossly 9182 elements representing 8472 unique annotated genes or expressed se- visible metastases were scored using a dissecting stereomicroscope. Lungs that quence tagged clusters arrayed on glass slides. cDNAs on the array are PCR had metastases too numerous to count, or that had grown into a confluent layer, products of 500-5000 bp, with an average size of 1–1.2 kb. According to the were scored as having 200 metastases. manufacturer, the hybridization is able to detect Ϯ1.7-fold changes with 99% Northern Blot Analysis. Total RNA (10 ␮g/sample) was size separated on confidence, and when 200 ng of starting RNA is used to make a probe, 2 pg a 1% agarose gel containing 2.2 M formaldehyde at 80 V for 3.5 h. RNA was of RNA can usually be detected (ϳ1 in 100,000 copies). transferred onto a positively charged Hybond Nϩ nylon membrane (Amersham For Array 1, mRNA from C8161.9, passage 14 (P14), and neo6/C8161.2 Pharmacia Biotech, Uppsala, Sweden) using the Turboblotter system (P18) was used for the Cy3 and Cy5 probes, respectively. For Array 2, equal (Schleicher and Schuell, Keene, NH) and fixed by UV cross-linking (Strat- quantities of mRNA from four clones of neo6/C8161 (clones 1, 2, 3 and 6, at agene, La Jolla, CA). Full-length KISS1, TXNIP, or a PCR-generated fragment passages 16–28) were pooled for the Cy3 probe and four clones of neo6qdel/ of CRSP3 were radiolabeled ([␣32P]-dCTP) using the RediPrime II DNA C8161.9 (clones 1, 2, 7, and 8 at passages 8–10) were pooled for the Cy5 labeling mix (Amersham Pharmacia Biotech). Blots were prehybridized in probe. mRNA (600 ng) was used to create each probe. ExpressHyb solution (Clontech, Palo Alto, CA) containing 100 ␮g/ml dena- The identity of spotted DNA was verified by PCR. A failure in PCR testing tured salmon sperm DNA for 1 h. Labeled probe was added to the hybridiza- resulted in exclusion of the spot measurement. Genes were analyzed if Ն1 tion solution and incubated overnight at 68°C. Membranes were rinsed in two reading had a signal:background ratio Ն2.5, signal intensity Ͼ250 units for changes of low stringency wash buffer (2ϫ SSC ϩ 0.1% SDS) followed by one or both dyes, and a spot size of Ն40% of the spotted area. A total of 8083 2-min low-stringency washes for 15 min at 37–50°C and a high stringency array elements passed these quality control criteria for Array 1, and 7217 genes wash (0.1ϫ SSC ϩ 0.1% SDS) for 15 min at 50–60°C. The membranes were passed for Array 2. Cy5 and Cy3 data sets were normalized for median gene then exposed to Kodak BioMax MR X-ray film (Kodak, Rochester, NY). signal values to compensate for differences in labeling and detection effi- RTQ. RTQ was performed using the ABI Prism 7700 Sequence Detection ciency. System (Applied Biosystems, Foster City, CA) located at the Nucleic Acids Expression Vectors and Cloning. A full-length TXNIP clone in the Facility of the Penn State University Life Sciences Consortium. Relative pINCY vector was obtained from Incyte Genomics (No. 2888464). TXNIP mRNA levels were determined using the comparative CT (threshold cycle) was excised by enzymatic digestion with EcoRI and NotI (Promega, Madison, method. The expression of TXNIP, KISS1, and CRSP3 (CTS) was normalized WI) and subcloned into the pcDNA3 mammalian expression vector (Invitro- to an endogenous reference gene (G3PDH). Although G3PDH may exhibit gen). Full-length CRSP3 was cloned into the pcDNA3 vector. differential expression in some cancers (23–25), it is commonly used as a Transfection and in Vitro Growth Curves. pcDNA3-CRSP3 and standard. Our microarray experiments confirmed that expression of G3PDH pcDNA3-TXNIP were transfected into C8161.9 cells by electroporation using did not differ significantly among nonmetastatic and metastatic variants of the Gene Pulser II system (Bio-Rad Laboratories, Hercules, CA). Briefly, cells C8161. CTS was calculated by subtracting the CT value of the reference (CTR) ⌬ ⌬ ϭ Ϫ were detached, centrifuged, and resuspended in CMF-DPBS. Plasmid DNA from the CT value of the sample ( CT; CT CTS CTR). The relative (10 ␮g) was added to the cells and the mixture placed on ice for 5 min before expression (2Ϫ⌬⌬CT) to a calibrator was determined by subtracting the ␮ ϱ ⌬ ⌬ ⌬⌬ ϭ⌬ Ϫ⌬ electroporation. After electroporation (220 V, 960 Fd, ) cells were chilled CT(Calibrator) from the CT value [ CT CT CT(Calibrator)]. Samples on ice for 10 min before plating. One day later, 500 ␮g/ml G418 was added and internal controls were run in triplicate. and stable transfectants (21) were isolated and cloned by limiting dilution. Primer and fluorogenic probe sets for TXNIP, KISS1, CRSP3, and G3PDH For in vitro growth curves, C8161.9/TXNIP clones, C8161.9/pcDNA3, were designed using Primer Express V1.0 software (Applied Biosystems). The neo6/C8161.2, and neo6/C8161.6 were used at passages 4, 11, 20, and 10, TXNIP forward and reverse primers and fluorescent-labeled probe were respectively. For in vitro growth curves, C8161.9/CRSP3 clones, C8161.9/ 5Ј-GGCCTTAAAGGATGCGGACC-3Ј,5Ј-TCCAACAAACACCCCTGTA- pcDNA3, neo6/C8161.2, and neo6/C8161.6 were used at passages 8, 19, 23, TCA-3Ј, and 5Ј-ATCCTCAGCCAGCGCCCATG-3Ј, respectively. The se- and 13, respectively. Cells (1 ϫ 104) were seeded on 6-well tissue-culture quences of the CRSP3 forward and reverse primers and fluorescent-labeled plates containing 2 ml of cDME-F12 ϩ 500 ␮g/ml G418 per well. Six replicate probe were 5Ј-GCCCCAGCTTGACGTCTG-3Ј,5Ј-TGACAGGCAGTGA- wells per cell line per time point were detached with 2 mM EDTA and counted AATCAAAGAG-3Ј, and 5Ј-AACAGAGTGGGTTGGCTATCCATTCCG-3Ј, 433

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2003 American Association for Cancer Research. CRSP3/TXNIP/KISS1: METASTASIS SUPPRESSION PATHWAY respectively. The sequence of the KISS1 forward and reverse primers and ison, WI). All protocols were approved by both the Kanazawa University and fluorescent-labeled probe were 5Ј-TGCTGGTGCAGCGGG-3Ј,5Ј-CGA- Penn State University Institutional Review Boards. Patient identifiers were AGCGCAGGCCG-3Ј, and 5Ј-AGGACCTGCCGAACTACAACTGGAAC- removed for reporting of this experimental data. TCC-3Ј, respectively. The 5Ј fluorogenic reporter probe was FAM, and the 3Ј RT-PCR. Total RNA (2.5 ␮g) was reverse transcribed and amplified using fluorogenic quencher was BlackHole (Biosearch Tehnologies, Inc., Novato, the Advantage Titanium One-step RT-PCR kit (Clontech) according to the CA). The G3PDH forward and reverse primers and fluorescent-labeled probe manufacturer’s instructions. Reverse transcription proceeded at 50°C for 1 h, were 5Ј-GAAGGTGAAGGTCGGAGTC-3Ј,5Ј-GAAGATGGTGATGGGA- followed by 35 cycles of PCR (94°C, 45 s; 60°C, 45 s; 72°C, 60 s) and a final Ј Ј Ј Ј TTTC-3 , and 5 -CAAGCTTCCCGTTCTCAGCC-3 , respectively. The 5 extension step of 10 min at 45°C. TXNIP was amplified with the human- fluorogenic reporter probe was VIC (Biosearch Technologies, Inc.) and the specific primers 5Ј-CAAAGGGGTTTCCTCGATTTG-3Ј and 5Ј-TTGAAG- Ј 3 quencher fluorochrome was TAMRA. Primers were synthesized by GATGTTCCCAGAGG-3Ј yielding an expected product size of 422 bp. the Nucleic Acids Facility, and the probes obtained from Biosearch Technol- CRSP3 was amplified with the human-specific primers 5Ј-GGCTTTTCCTG- ogies, Inc. GCATGTTTA-3Ј and 5Ј-AACCAGTGTGGAAGTTTGCC-3Ј, giving an ex- RTQ was performed with the TaqMan Universal PCR Assay Mix in a pected product size of 540 bp. Human G3PDH was amplified a control 96-well reaction plate. G3PDH was amplified at the same time and used as a amplimer set (Clontech), giving an expected product size of 983 bp. The reference gene. Reverse transcription using the murine leukemia virus reverse products were resolved on a 1% agarose gel and visualized by ethidium transcriptase was performed at 42°C for 1 h, followed by a 5-min extension at bromide staining. 72°C. Each RT-PCR contained the following: 10 ␮M each TXNIP- and RT-PCR for clinical samples was performed as reported previously (12). G3PDH-specific primers, 1 ␮M each TXNIP/CRSP3/KISS1 and G3PDH flu- Analyses were made using a Prism 7700 sequence Detector (PE Biosystems). orogenic probe and 8 ␮l of the reverse transcriptase product. After incubation for 2 min at 50°C and 10 min at 95°C, PCR amplification was performed for The sequences of primers and fluorescence-labeled probes were as follows: Ј Ј Ј 40 cycles (95°C for 15 s and 60°C for 1 min) using the AmpliTaq Gold DNA KISS1: forward 5 -ACTCACTGGTTTCTTGGCAGC-3 , reverse 5 -AC- Ј Ј polymerase (Applied Biosystems). CTTTTCTAATGGCTCCCCA-3 , and probe 5 -FAM-ACTGCTTTCCTCT- Ј Ј Clinical Samples and Analysis of KISS1 and CRSP3 Expression. GTGCCACCCACT-TAMRA-3 ; and CRSP3: forward 5 -GCCCCAGC- Ј Ј Ј Twenty-four cutaneous melanoma specimens (11 primary and 13 metastasis) TTGACGTCTG-3 , reverse 5 -TGACAGGCAGTGAAATCAAAGAG-3 , Ј were obtained from 20 Japanese patients undergoing surgery at Kanazawa and probe 5 -FAM-AACAGAGTGGGTTGGCTATCCATTCCG-TAMRA- University Hospital. The specimens were frozen immediately in liquid nitrogen 3Ј; ␤-actin: forward 5Ј-TC ACCCACACTGTGCCCATCTACGA-3Ј, probe and used for RNA extraction. Total RNA was extracted from frozen tumors 5Ј-FAM-ATGCCCTCCCCCATGCCATCCTGCGT-TAMRA-3Ј, and re- using a silica-gel-based membrane system (Rneasy; Qiagen, Hilden, Germa- verse 5Ј-CAGCGGAACCGCTCATTGCCAATGG-3Ј. DNA standards were ny). To avoid false positives from contaminating genomic DNA, RNA samples generated by PCR amplification of gene products and the numbers of copies were treated with DNase I (Qiagen) for 15 min. The first strand cDNA were calculated. Normalization of samples was performed by dividing the synthesis was performed from 1 ␮g of total RNA using avian myeloblastosis number of copies of KISS-1 or DRIP-130 transcripts by the number of copies virus reverse transcriptase and oligodeoxythymidine primers (Promega, Mad- of ␤-actin transcripts.

Fig. 1. Differential expression of TXNIP in melanoma cells with differing metastatic potentials. A and B, distribution of expression by microarray hybridization. Scatter plot depicting the comparative expression of each transcript on the metastatic (x) and nonmetastatic (y) axes. Most transcripts are within the Ϯ2-fold difference indicated by the lines. Points representing genes more highly expressed in the metastatic cells are shown at the lower right, whereas those more highly expressed in the nonmetastatic cells are shown at the upper left. A, Array 1 (C8161.9 versus neo6/C8161.2). B, Array 2 (neo6qdel/C8161.pool versus neo6/C8161.pool). C, Northern blot of TXNIP in C8161-derived cell lines. Total RNA (10 ␮g/sample) was separated and probed with full-length TXNIP cDNA. Lanes: pcDNA3 ϭ vector-only transfectant of C8161.9; neo6 ϭ neo6/C8161 clones 1, 2, 3 and 6; 6qdel ϭ neo6qdel/C8161 clones 1, 2, 7 and 8; TXNIP ϭ C8161.9/TXNIP clones 3, 5, 11, and 13. D, TXNIP expression inversely correlates with melanoma progression. RTQ was used to determine relative mRNA levels, using the comparative CT (threshold cycle) method. Total RNA was extracted from cell lines derived from normal melanocytes, early VGP melanoma (WM793), VGP melanoma (WM115), and two metastatic melanomas (WM239A and 1205-LU). The expression of TXNIP was normalized to an endogenous reference gene (G3PDH). Data from two independent experiments (run in triplicate) were averaged and relative expression was plotted as a percent of melanocyte expression. E, TXNIP up-regulates KISS1 expression. RTQ was used to determine KISS1 expression levels in TXNIP transfected cells. KISS1 expression was normalized to G3PDH. The lack of a bar indicates that KISS 1 expression was not detectable. Relative expression (Y axis, logarithmic scale) is the fold-difference compared with neo6/C8161.1 cells. Parental cells (C8161) and vector-only controls (pcDNA3) are shown compared with neo6/C8161 (neo6) clones 1, 2, 3 and 6, neo6qdel/C8161.9 (6qdel) clones 1, 2, 7, and 8 and C8161.9/TXNIP (TXNIP) transfectant clones 3, 5, 11, and 13. 434

Statistical Analysis. The number of lung metastases was compared in Table 1 Differential gene expression observed in both cDNA hybridization microarrays a CRSP3 transfectants and control cells. A one-way ANOVA followed by the comparing metastatic and nonmetastatic variants of C8161 Student Newman-Keuls post-test was used to determine significance. Statisti- Fold-difference cal significance was defined as P Յ 0.05 using two-sided tests. The SE was C8161.9 6qdel calculated by dividing the SD by the square root of the sample size. For vs. vs. analysis of clinical samples, correlations between KISS1- and CRSP3-tran- Gene Accession no. Locus neo6.2 neo6b script expression by RT-PCR were compared by Spearman’s correlation test Down-regulated by chr6 (higher in metastatic cells) by ranks. Centromere protein E (312 kDa) NM_001813 4q24-q25 8.0 2.6 Dickkopf-1 like AF177394 10q11.2 6.7 3.2 RESULTS Procollagen-lysine, 2-oxoglutarate 5- NM_000935 3q23-q24 4.5 4.0 dioxygenase 2 Microarray Hybridization. To identify metastasis suppressor CDC2(cell division cycle 2) X05360 10q21.1 5.7 2.8 genes on chr6, a dual microarray strategy was used. In Array 1, highly Annexin A3 M20560 4q13-q22 4.7 2.5 MAD2 (mitotic arrest deficient, yeast, AW411207 4q27 4.3 2.6 metastatic subclone C8161.9 (4) was used in a competitive hybrid- homolog)-like 1 ization with metastasis-suppressed neo6/C8161.2. The vast majority Proteoglycan 1, secretory granule BE789807 10q22.1 3.5 3.3 Ͼ ␣ catenin-like 1 NM_003798 9q31.2 3.5 3.3 ( 96%) of genes showed similar expression – the mean fold- Protein kinase, cAMP-dependent, NM_002731 1p36.1 3.0 3.1 difference was 1.25 Ϯ 0.525 (mean Ϯ SD; Fig. 1A). To minimize the catalytic, ␤ impacts of clonal heterogeneity and gene expression attributable to Secreted frizzled-related protein 1 AF056087 8p12.p11.1 3.0 2.6 STAT induced STAT inhibitor-2 NM_003877 12q 3.1 2.3 gene dosage, Array 2 probes comprised equal mixtures of four clones each from neo6/C8161 and neo6qdel/C8161.9. Again, Ͼ96% of genes Up-regulated by chr6 (lower in metastatic cells) Ͻ Ϫ Ϫ showed 2-fold difference in expression. In metastatic cells, 218 TXNIP (Up-regulated by vitamin D3) AU121991 1 26.6 5.0 genes were expressed Ͼ2-fold higher. Fifty-two were expressed Ն2- a Twelve genes were expressed more than 3 standard deviations above or below the fold higher in the nonmetastatic cells. The average fold-difference was mean in both arrays. Mean fold-difference in both arrays is shown. Ϯ Ϯ b mRNA was isolated from a mixture of four cell clones from neo6/C8161 (neo6) and 1.25 0.316 (Fig. 1B). Applying a strict 3 SD threshold, the pool neo6qdel/C8161 (6qdel). See Materials and Methods for details. of candidate effectors was only 12 genes. Genes Expression and Validation. Eleven genes were more highly expressed in the metastatic cells (i.e., down-regulated by chr6; clones examined. Expression in neo6qdel/C8161 clones was detectable, Table 1). Three genes encoded proteins involved in cell replication: but significantly reduced compared with neo6/C8161 clones. The mag- CENP-E (centromere-associated protein E), CDC2, and MAD2 (mi- nitudes of differential expression were consistent between Northern blots totic arrest deficient 2 homologue-like 1). Six genes encoded proteins and microarrays. Verification of protein expression was not possible that are involved in signal transduction: DKK1 (dickkopf-1), SFRP1 because antibodies were not available. (secreted frizzled-related protein-1), ␣-catenin-like-1 annexin A3, TXNIP expression was evaluated relative to melanoma progression SOCS2 (STAT-induced STAT inhibitor-2), and the gene encoding the using a panel of cell lines representing different stages of tumor catalytic subunit C-␤ of PKA. Interestingly, the first three are in- progression using RTQ. Two independent experiments, each contain- volved in the WNT pathway. On the basis of their roles in extracel- ing samples in triplicate, were averaged. A strong inverse correlation lular signaling, embryonic development and neoplastic progression between melanoma progression and TXNIP expression was observed (26–28), it is conceivable that these genes may promote steps in the (Fig. 1D). metastatic cascade. Finally, two genes (PLOD2 and Proteoglycan Differential expression of multiple genes raised the question of 1/serglycin) are involved in adhesion or extracellular matrix inter- whether coordinate gene regulation and/or a common pathway might actions. involve TXNIP and KISS1. To determine whether TXNIP transfec- Surprisingly, none of the genes Ն3-fold more highly expressed in tants exhibited increased KISS1 expression, RTQ was performed nonmetastatic neo6/C8161 cells mapped to chr6. Thus, metastasis (Fig. 1E). KISS1 expression was undetectable in C8161, vector-only suppressors on chr6 were not directly identified by microarray, likely transfectants and in neo6qdel/C8161 cells. However, KISS1 was because of the constraints inherent in microarrays (array coverage detected in all neo6/C8161 clones and TXNIP transfectants. Thus, we ϳ14–25% of the genome). None of the known metastasis-suppressor hypothesized that TXNIP might be an upstream regulator of KISS1 genes KISS1, BRMS1, KAI1, and MKK4, were among the array expression. elements. Only one gene showed Ն3-fold higher expression in neo6/ Identification of CRSP3. Because TXNIP was mapped to chr1, C8161 cells in both arrays (Table 1). TXNIP (also known as VDUP1) the TXNIP and KISS1 regulatory gene(s) on chr6 remained unidenti- was expressed 27-fold higher in the Array 1, and 5-fold higher in fied. However, data from several other experiments assisted in iden- Array 2. TXNIP is a TRX-binding protein induced by 1␣,25- tification of candidate genes on chr6q. neo6qdel hybrids refined the dihydroxyvitamin D (calcitriol), intracellular Ca2ϩ, oxidative stress 3 metastasis suppressor locus to a region between markers D6S300 and and heat shock (29, 30). By binding thioredoxin (31) at the redox- D6S314 (ϳ45 Mbp; Ref. 11). Subsequently, introduction of chr6 with critical site (30, 32), TXNIP antagonizes the interactions between overlapping deletions and from revertants (i.e., rare metastases) iden- TRX and proliferation-associated peroxiredoxin 1 or apoptosis-stim- tified a correlation between loss of D6S457 and gain of metastatic ulating kinase (ASK-1/MAPKKK5). Association with metastasis is competence (ϳ10 cM).7 inferred because ASK1 phosphorylates and activates MKK4, a pros- The gene mapping nearest D6S457 was CRSP3 (DRIP130, Vitamin tate carcinoma metastasis suppressor (33). TRX controls cancer cell D Receptor Interacting Protein, 130 kDa). CRSP3 was a compelling growth through regulation of DNA synthesis and transcription factor candidate because it encodes a required element in a coactivator activity, and is over-expressed by, and stimulates the proliferation of, complex necessary for vitamin D receptor-regulated transcription many human solid tumor cells (reviewed in (34, 35). (36). CRSP3 is a component of other cofactor complexes, including TXNIP Expression. Differential expression of TXNIP was validated by RNA blotting using a panel of metastatic and nonmetastatic variants 7 M. E. Miele, M. Jewett, C. Paquette-Straub, S. Goldberg, J. Harms, G. Babu, C. of C8161 (Fig. 1C). TXNIP expression was virtually undetectable in the Morelli, F. Gualandi, P. Rimessi, G. Barbanti-Brodano, and D. R. Welch. Human vector-only control, but was readily detectable in all four neo6/C8161 melanoma metastasis-suppressor locus narrowed to 6q22.3-q23.3, submitted for publication. 435

Fig. 2. A, CRSP3 expression inversely correlates with melanoma progression. RTQ was used to determine relative mRNA levels using total RNA extracted from cell lines derived from different stages of melanoma progression (See Fig. 1D for details). CRSP3 expression was normalized to G3PDH. Relative expression is plotted as a percent of melanocyte expression. B, C, E, CRSP3 up-regulates TXNIP and KISS1. RNA blots (total RNA, 10 ␮g) measuring TXNIP (B) and KISS1 (C, E) mRNA expression. C8161.9 ϭ parental metastatic population; pcDNA3 ϭ vector-only; C8161.9/KISS1, a KISS1 transfectant; neo6 ϭ neo6/C8161 transfectant clones; 6qdel ϭ neo6qdel/C8161.9 transfectant clones; TXNIP ϭ TXNIP transfectant clones; CRSPmix ϭ mixed pools (uncloned) of CRSP3 transfectants; DRIP1-# ϭ isolated colonies (presumed clones) of CRSP3 transfectants. D, RTQ of CRSP3 Transfectants of C8161.9. RTQ was used to evaluate expression levels of CRSP3 relative to an endogenous control (G3PDH). Data are combined from two independent experiments of samples and controls run in triplicate. Expression has been normalized to vector-only (pcDNA3) transfectants, shown here as an expression level of 1. Samples represent the pcDNA3 vector-only control, two independently derived mixed pools of CRSP3 transfectants (CRSPmix), and clonal CRSP3 transfectants (clones 1-5, 1-9, 1-13, 1-20, and 2-8). E, RNA Blot (total RNA, 10 ␮g) of TXNIP expression in TXNIP transfectant clones. C8161.9 cells were transfected with TXNIP in the pcDNA3 vector. TXNIP was undetectable in the parental cells, highly expressed in the neo6/C8161 (neo6) clones, and exhibited variable expression in transfectant clones.

NAT and ARC, indicating a potential role in mediating transcription by several factors (30). Given the up-regulation of a vitamin D- responsive gene (TXNIP) in metastasis-suppressed neo6/C8161 cells, we hypothesized that loss of CRSP3 might affect metastasis by down-regulation of TXNIP. Furthermore, because CRSP3 is required for SP1-mediated transcription (37) and because the KISS1 promoter contains SP1 binding elements,8 additional evidence pointed toward CRSP3 being a strong candidate for the metastasis suppressor locus on human chr6q. CRSP3 Expression Is Inversely Correlated with Melanoma Progression. RTQ was performed to evaluate CRSP3 expression in the same panel of cell lines representing melanoma progression de- scribed above. CRSP3 expression also shows an inverse correlation with melanoma progression (Fig. 2A). Whether the higher level of expression observed for CRSP3 in early VGP is universal will require more detailed and extensive studies. Loss of expression occurs during VGP, the time where melanomas acquire metastatic potential. As with TXNIP, validation of protein expression was not possible because antibodies were unavailable. CRSP3 Up-regulates TXNIP and KISS1. If CRSP3 were indeed responsible for suppressing metastasis, we reasoned that it would do so by inducing expression of TXNIP and/or KISS1 (i.e., CRSP3 Fig. 3. TXNIP and CRSP3 do not suppress tumorigenicity of C8161. A and B, initial transfection should increase KISS1 and VDUP1 expression). C8161.9 growth rate for tumors is slowed by TXNIP (A) and CRSP3 (B). Cells (1 ϫ 106) were injected intradermally into the dorsolateral flanks of athymic mice. Measurements ceased cells were transfected with pcDNA3-CRSP3. By RNA blot TXNIP when primary tumors were surgically removed. C and D, TXNIP (C) and CRSP3 (D) were was undetectable in C8161.9 cells, but was highly expressed in still expressed in intradermal tumor tissue. RT-PCR was used to measure expression of CRSP3 and TXNIP in excised primary tumor. TXNIP expression was detected in each of the tumors. Likewise, CRSP3 expression was detected in each of the tumors. Note: TXNIP 8 R. S. Samant and D. R. Welch, unpublished observations. expression was detected in tumors formed by the CRSP3 transfectants (D). 436

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2003 American Association for Cancer Research. CRSP3/TXNIP/KISS1: METASTASIS SUPPRESSION PATHWAY neo6/C8161 clones, a mixed pool of CRSP3 transfectants, and TXNIP transfectants (Fig. 2B). Similarly, KISS1 was undetectable in C8161.9, vector-only transfectants, and neo6qdel/C8161 hybrids, but was expressed in a KISS1 transfectant clone of C8161.9, neo6/C8161 cells, and CRSP3 transfectants (Fig. 2, C and D). Biological Properties of TXNIP and CRSP3 Transfectants. To test whether CRSP3 and TXNIP would suppress metastasis, we transfected C8161.9 cells and isolated transfectants with various levels of expression (Fig. 2, E and F). Whereas CRSP3 mRNA was detectable in the vector-only control, it was expressed at significantly higher levels in the transfectants. CRSP3-transfectant clones 1-5, 1-9, 1-13, 1-20, and 2-8 were chosen for more extensive analysis. By RNA blot, TXNIP (1.2 Kb) was virtually undetectable in C8161.9, but was highly expressed in neo6/C8161 clones 2 and 6. TXNIP-transfectant clones 3, 5, 11, and 13 were chosen for further analysis.

Fig. 5. Expression of CRSP3 correlates with KISS1 expression in melanoma tumors. Total RNA was extracted from frozen melanoma specimens and RT-PCR was performed for CRSP3, KISS1, and ␤-actin. Expression of CRSP3 and KISS1 were each was normalized to ␤-actin. Relative expression was plotted and regression analysis performed.

Morphology of TXNIP or CRSP3 transfectants was not grossly affected in vitro (data not shown), consistent with previous observations involving neo6/C8161 and KISS1. No differences for growth in culture or saturation density were observed (data not shown). TXNIP and CRSP3 transfectants were injected at an orthotopic site in athymic mice and tumor growth was measured weekly. After tumors reached ϳ1 cm, they were removed, and the mice were allowed to recover 4 weeks before quantifying metastasis. The time to tumor removal was as follows: 32–36 days for C8161.9/pcDNA (vector control); 42–46 days for neo6/C8161.2, CRSP3-transfectant clones 1-13 and 1-20 and TXNIP transfectant clone 5; and 54–61 days for neo6/C8161.6, CRSP3 transfectant clones 1-5, 1-9, and 2-8 and TXNIP transfectant clones 3, 11, and 13 (Fig. 3, A and B). Although CRSP3 and TXNIP transfectants initially grew slower than vector-only controls, the growth rate was comparable with neo6/ C8161 cells. Once established, transfectants grew at rates comparable with parental or vector-only controls. TXNIP and CRSP3 expression was maintained during orthotopic tumor growth as assessed by RT- PCR (Fig. 3, C and D). Furthermore, CRSP3 was still functional because TXNIP expression was found in the tumors (Fig. 3D). CRSP3 and TXNIP transfectants formed significantly fewer metastases than controls. C8161.9/pcDNA3 mice developed an average of 111 metastases/lung. (Fig. 4, top). One neo6/C8161-injected mouse developed a single metastasis. Only two mice each in the C8161.9/ TXNIP and C8161.9/CRSP3 groups had detectable metastases (one metastasis per mouse). The low level of lung metastasis was impres- sive considering that most mice injected with CRSP3 or TXNIP cells recurred at the site of local tumor, consistent with observations that the tumors were still locally invasive. One mouse in the C8161.9/ CRSP3–1.9 group developed 23 lung metastases and also had axillary lymph node involvement. Tumor cells (2 ϫ 105) were also injected i.v. to assess metastasis. Two independent experiments were performed with comparable results. Fig. 4, B and C, shows data combined from both studies (n ϭ 16–17 mice/group) for TXNIP and CRSP3 transfectants, respec- Fig. 4. Metastasis is suppressed by TXNIP and CRSP3. A, cells (1 ϫ 106) were injected intradermally into the dorsolateral flanks of athymic mice. When mean tumor diameter tively. neo6/C8161 cells were used as positive controls. For the reached ϳ1 cm, tumors were removed. Mice were euthanized 4 weeks later and necrop- TXNIP experiment, vector-only transfectants formed lung metastases sied. Lung metastases were quantified as reported (22). Data are from a single represent- in all mice (mean ϭ 172). neo6/C8161.2 cells produced an unchar- ative experiment. Insets provide statistics for lung metastases. Bars represent number of lung metastases per animal Ϯ SEM. B and C, cells (2 ϫ 105) were injected i.v. into the acteristically and unexplained high number of lung metastases lateral tail veins of athymic mice. Mice were euthanized 4 (shown) to 8 weeks (data not (mean ϭ 5) with previously (or subsequently) unobserved incidence shown) later and necropsied. Surface lung metastases were quantified after fixation in Bouin’s fixative. Two independent experiments for TXNIP and CRSP3 transfectants are (11 of 16 mice). Nonetheless, suppression was significant. neo6/ combined. Bars represent the number of lung metastases per animal ϮSEM. C8161.6 developed no metastases. Only 8 of 64 mice injected with 437

Fig. 6. A proposed metastasis suppressor pathway. Solid lines represent the established relation- ships between proteins in the proposed pathway, broken lines represent related, potentially relevant, but nonvalidated interactions. Deletions of chr6 frequently occur in the region surrounding CRSP3. Loss of CRSP3 (or decreased expression, mutation, and so on) leads to an altered transcriptome attributable to disruption of CRSP3-containing transcription complexes. Among the downstream targets of CRSP3 (possibly through its role in the transcriptional modulation involving the vitamin D receptor) is TXNIP. Losses of CRSP3, therefore, result in diminished expression of TXNIP, thereby de- creasing the ratio of TXNIP to TRX. Thioredoxin is now able to bind ASK1 (apoptosis signaling kinase) and can modulate redox sensitive transcription of several other genes. Among the downstream targets of thioredoxin is the metastasis suppressor gene KISS1, which encodes a peptide ligand for a G-protein coupled receptor, hGPR54, which is proposed to affect motility and invasion. Ultimately, the effector molecules are not yet known.

C8161.9/TXNIP cells had any macroscopic metastases (maxi- attributable to unrelated cell lines) and the availability of multiple mum ϭ 2 per lung). CRSP3 likewise suppressed metastasis (Fig. 4C). clones (to minimize spurious results attributable to clonal hetero- Only 30% of mice had any metastases. Of those, only 10% (8 of 80) geneity). Array 1 compared two clones, C8161.9 and neo6/C8161.2, developed more than a single lesion. which were metastatic and nonmetastatic, respectively. Array 2 com- To compensate for in vivo latency of growth, mice injected with pared metastatic and nonmetastatic variants as well; however, equal CRSP3 clones were euthanized at 8 weeks, rather than 4 weeks, after proportions of mRNA from four clones were mixed to minimize the i.v. injection. Lung metastasis by two CRSP3 and two TXNIP trans- impact of clonal heterogeneity on interpretation. In both arrays, TX- fectants was significantly suppressed (data not shown). The lower NIP had the greatest differential expression (27- and 5-fold, respec- incidence of metastasis cannot, therefore, be explained by slower tively) between the populations. growth alone. TXNIP expression inversely correlated with melanoma progression Clinical Correlations. To address whether CRSP3 expression pre- and exerted an antimetastatic, but not tumor suppressing, effect upon dicts melanoma progression toward metastasis and whether CRSP3 transfection into melanoma cells. However, as with KISS1, TXNIP maps expression correlates with KISS1 expression in vivo, clinical samples to chromosome 1q, implying that regulatory elements on chr6 remained from 20 patients undergoing melanoma surgery at Kanazawa Univer- unidentified. Candidate genes on chr6 were evaluated taking into con- sity were examined by RTQ. KISS1 and CRSP3 were normalized to sideration updated genome maps, refinement of the metastasis suppressor ␤ -actin expression. Although the sample size was limited, there was a locus in revertants, information about transcription regulatory elements in 62.9% correlation between CRSP3 expression and KISS1 expression the KISS1 promoter and an association of TXNIP with vitamin D. (Fig. 5). This result was highly significant (P Ͻ 0.0023) by Spear- CRSP3 met all criteria and was evaluated for its metastasis-suppressor man’s correlation test by ranks. Although these results are consistent capabilities. CRSP3 transfectants were suppressed for metastasis but with the hypothesis that CRSP3 is a metastasis suppressor, a correla- remain tumorigenic. Furthermore, CRSP3 transfection resulted in up- tion coefficient of 63% must not be over-interpreted. Discordant data regulation of both TXNIP and KISS1. Taken together, these data implied do not account fully for presence of mutations within CRSP3, “con- that CRSP3 was the underlying defect resulting in acquisition of meta- tamination” by normal tissue, other interactors, or protein half-life. static potential. Loss or decreased expression of CRSP3 would result in diminished expression of TXNIP and KISS1, which would, in turn, allow DISCUSSION cells to metastasize. A model putting all three components is depicted in Metastasis suppressors are defined as genes that block metastasis Fig. 6. Because CRSP3 and TXNIP are known components of transcrip- without eradicating a cell’s ability to form a tumor (38).9 Several tional regulation, involvement of other genes remains a possibility. In- laboratories have identified metastasis suppressors using MMCT (re- deed, given the complex nature of metastasis regulation of other genes is viewed in 39–41). When we introduced chr6 into the C8161 and quite likely. MelJuSo, we observed significant suppression of metastasis without TXNIP: Possible Mechanisms of Suppression. A role for TXNIP tumor suppression. However, the metastasis suppressor gene identi- in controlling metastasis is consistent with previously published prop- fied by subtractive hybridization (i.e., KISS1) mapped to chr1, rather erties. TXNIP was first identified as a gene up-regulated in cells after than chr6 (7, 8), implying that a gene(s) on chr6 regulates an effector exposure to vitamin D3/calcitriol (29). A relationship between vitamin on chr1. This hypothesis was supported by two independent studies D3 and metastasis is possible, but is not addressed in the current study. mapping the putative regulatory gene to 6q16.3-q23 (11, 12). Whether exogenous vitamin D3 modulates C8161 metastasis by in- To complement the subtractive hybridization study, microarray ducing transcription of TXNIP remains to be determined. However, in analyses were performed. The experimental design took advantage of a single experiment using a single dose of calcitriol, we did not genetically related cell lines (to minimize the likelihood of differences observe an effect on in vitro morphology or proliferation.10

9 Note: there may be alterations in primary tumor growth kinetics, but masses still 10 S. F. Goldberg, M. E. Miele, N. Hatta, M. Takata, C. Paquette-Straub, L. P. develop and continue to express the metastasis suppressor genes. Freedman, and D. R. Welch, unpublished observations. 438

TXNIP could also exert effects independent of vitamin D3 (re- 4. Welch, D. R., Chen, P., Miele, M. E., McGary, C. T., Bower, J. M., Weissman, B. E., viewed in Ref. 42). Accumulating evidence suggests that redox reg- and Stanbridge, E. J. Microcell-mediated transfer of chromosome 6 into metastatic human C8161 melanoma cells suppresses metastasis but does not inhibit tumorige- ulation is important in gene regulation. TRX can control cell growth nicity. Oncogene, 9: 255–262, 1994. via regulation of DNA synthesis and transcription factor activity. 5. Miele, M. E., Robertson, G., Lee, J. H., Coleman, A., McGary, C. T., Fisher, P. B., Therefore, TXNIP binding to TRX at its active site would antagonize Lugo, T. G., and Welch, D. R. Metastasis suppressed, but tumorigenicity and local invasiveness unaffected, in the human melanoma cell line MelJuSo after introduction TRX interactions with other proteins which, in turn, could affect of human chromosomes 1 or 6. Mol. Carcinog., 15: 284–299, 1996. multiple downstream effectors. In the present report, we show that 6. Goldberg, S. F., Harms, J. F., Quon, K., and Welch, D. R. Metastasis-suppressed enforced expression of TXNIP increases KISS1 expression. It is C8161 melanoma cells arrest in lung but fail to proliferate. Clin. Exp. Metastasis, 17: 601–607, 1999. reasonable to speculate that TXNIP, via TRX binding, is an upstream 7. Lee, J-H., Miele, M. E., Hicks, D. J., Phillips, K. K., Trent, J. M., Weissman, B. E., regulator of KISS1. However, because TXNIP was expressed at low and Welch, D. R. KiSS-1, a novel human malignant melanoma metastasis-suppressor levels in metastatic cells, whereas KISS1 expression was undetect- gene. J. Natl. Cancer Inst., 88: 1731–1737, 1996. 8. Lee, J-H., Miele, M. E., Hicks, D. J., Phillips, K. K., Trent, J. M., Weissman, B. E., able, such an explanation is not straightforward. Perhaps, a threshold and Welch, D. R. KiSS-1, a novel human malignant melanoma metastasis-suppressor of TXNIP must be exceeded to inactivate all of the TRX in the cell. gene [Erratum]. J. Natl. Cancer Inst., 89: 1549, 1997. CRSP3: Possible Mechanisms of Metastasis Suppression. In 9. Lee, J-H., and Welch, D. R. Identification of highly expressed genes in metastasis- suppressed chromosome 6/human malignant melanoma hybrid cells using subtractive C8161, loss or mutation of either KISS1 or TXNIP is unlikely. Rather, hybridization and differential display. Intl. J. Cancer, 71: 1035–1044, 1997. dysregulation of their expression by genes on chr6 is implied by the 10. West, A., Vojta, P. J., Welch, D. R., and Weissman, B. E. Chromosome localization data. Whether a single gene on chr6 is responsible for coordinate and genomic structure of the KiSS-1 metastasis suppressor gene (KISS1). Genomics, 54: 145–148, 1998. regulation of TXNIP and KISS1, or whether KISS1 regulation is 11. Miele, M. E., Jewett, M. D., Goldberg, S. F., Hyatt, D. L., Morelli, C., Gualandi, F., (in)direct, remains to be determined. Our working hypothesis is that Rimessi, P., Hicks, D. J., Weissman, B. E., Barbanti-Brodano, G., and Welch, D. R. CRSP3 is the beginning of a common cascade regulating both TXNIP A human melanoma metastasis-suppressor locus maps to 6q16.3-q23. Intl. J. Cancer, 86: 524–528, 2000. and KISS1. CRSP3 is part of vitamin D receptor-related coactivator 12. Shirasaki, F., Takata, M., Hatta, N., and Takehara, K. Loss of expression of the complexes, indicating a potential role in mediating transcription by metastasis suppressor gene KiSS1 during melanoma progression and its association several factors (reviewed in Ref. 43). Conceivably, loss or mutation of with LOH of chromosome 6q16.3-q23. Cancer Res., 61: 7422–7425, 2001. 13. Ohtaki, T., Shintani, Y., Honda, S., Matsumoto, H., Hori, A., Kanehashi, K., Torao, CRSP3 could explain the breadth of differential expression observed Y., Kumano, S., Takatsu, Y., Matsuda, Y., Ishibashi, Y., Watanabe, T., Asada, M., between metastatic and nonmetastatic cells. Yamada, T., Suenaga, M., Kitada, C., Usuki, S., Kurokawa, T., Onda, H., Nishimura, O., and Fujino, M. Metastasis suppressor gene KiSS1 encodes peptide ligand of a In summary, we demonstrated inverse correlations between meta- G-protein-coupled receptor. Nature (Lond.), 411: 613–617, 2001. static potential of human melanomas and expression of CRSP3, TX- 14. Kotani, M., Detheux, M., Vandenbogaerde, A., Communi, D., Vanderwinden, J. M., NIP, and KISS1. Although further validation of CRSP3 and TXNIP in Le Poul, E., Brezillon, S., Tyldesley, R., Suarez-Huerta, N., Vandeput, F., Blanpain, C., Schiffmann, S. N., Vassart, G., and Parmentier, M. The metastasis suppressor melanoma samples will be required, the data imply that these mole- gene KiSS-1 encodes kisspeptins, the natural ligands of the orphan G protein-coupled cules are involved in the regulation of metastasis. The data also receptor GPR54. J. Biol. Chem., 276: 34631–34636, 2001. implicate a complex and intricate network whereby gene expression 15. Muir, A. I., Chamberlain, L., Elshourbagy, N. A., Michalovich, D., Moore, D. J., Calamari, A., Szekeres, P. G., Sarau, H. M., Chambers, J. K., Murdock, P., can be regulated by these genes. Further genomic and proteomic Steplewski, K., Shabon, U., Miller, J. E., Middleton, S. E., Darker, J. G., Larminie, analysis of these cells may permit the elucidation of specific tran- C. G. C., Wilson, S., Bergsma, D. J., Emson, P., Faull, R., Philpott, K. L., and scriptional and regulatory pathways that are altered in metastasis. Harrison, D. C. AXOR12: a novel human G protein-coupled receptor, activated by the peptide KiSS-1. J. Biol. Chem., 276: 28969–28975, 2001. Importantly, we have provided preliminary evidence that the three 16. Hori, A., Honda, S., Asada, M., Ohtaki, T., Oda, K., Watanabe, T., Shintani, Y., functional metastasis suppressors (CRSP3, TXNIP, and KISS1) are Yamada, T., Suenaga, M., Kitada, C., Onda, H., Kurokawa, T., Nishimura, O., and serially regulated. This data suggests a compelling, albeit incomplete, Fujino, M. Metastin suppresses the motility and growth of CHO cells transfected with its receptor. Biochem. Biophys. Res. Comm., 286: 958–963, 2001. model for a melanoma metastasis suppression pathway (Fig. 6), 17. Yan, C. H., Wang, H., and Boyd, D. D. KiSS-1 represses 92-kDa type IV collagenase among the first such pathways elucidated. According to this model, expression by down- regulating NFkB binding to the promoter as a consequence of deletions on chr6, which are frequent in melanoma, result in loss of IkBa-induced block of p65/p50 nuclear translocation. J. Biol. Chem., 276: 1164– 1172, 2001. CRSP3 expression. As a required component in several transcriptional 18. Welch, D. R., Bisi, J. E., Miller, B. E., Conaway, D., Seftor, E. A., Yohem, K. H., coactivator complexes, loss of CRSP3 leads to an altered transcrip- Gilmore, L. B., Seftor, R. E. B., Nakajima, M., and Hendrix, M. J. C. Characterization tome, including both negative and positive changes in gene expres- of a highly invasive and spontaneously metastatic human malignant melanoma cell line. Intl. J. Cancer, 47: 227–237, 1991. sion. Among the downstream targets of CRSP3 is TXNIP, resulting in 19. Herlyn, M., Thurin, J., Balaban, G., Bennielli, J. L., Herlyn, D., Elder, D. E., Bondi, diminished expression of TXNIP. Thus, the intracellular ratio of E., Guerry, D. P., Nowell, P., Clark, W. H., and Koprowski, H. Characteristics of TXNIP to TRX is decreased, leading to increased TRX activity and cultured human melanocytes isolated from different stages of tumor progression. Cancer Res., 45: 5670–5676, 1985. failure to activate KISS1 expression. Still unanswered are questions 20. Graeven, U., and Herlyn, M. In vitro growth patterns of normal human melanocytes related to the mechanisms by which KISS1 and TXNIP actually and melanocytes from different stages of melanoma progression. J. Immunother., 12: mediate the metastasis suppression. 199–202, 1992. 21. Valyi-Nagy, I., Shih, I. M., Gyorfi, T., Greenstein, D., Juhasz, I., Elder, D. E., and Herlyn, M. Spontaneous and induced differentiation of human melanoma cells. Int. J. Cancer, 54: 159–165, 1993. ACKNOWLEDGMENTS 22. Welch, D. R. Technical considerations for studying cancer metastasis in vivo. Clin. Exp. Metastasis, 15: 272–306, 1997. We wish to acknowledge Dr. Meenhard Herlyn (Wistar Institute, Philadel- 23. Lau, W. Y., Lai, P. B. S., Leung, M. F., Leung, B. C., Wong, N., Chen, G., Leung, phia) for providing the panel of melanoma cell lines. We are grateful to all T. W., and Liew, C. T. Differential gene expression of hepatocellular carcinoma using cDNA microarray analysis. Oncol. Res., 12: 59–69, 2000. members of the Welch lab for helpful advice and discussions. 24. Thellin, O., Zorzi, W., Lakaye, B., De Borman, B., Coumans, B., Hennen, G., Grisar, T., Igout, A., and Heinen, E. Housekeeping genes as internal standards: use and limits. J. Biotechnol., 75: 291–295, 1999. REFERENCES 25. Savonet, V., Maenhaut, C., Miot, F., and Pirson, I. Pitfalls in the use of several “housekeeping” genes as standards for quantitation of mRNA: the example of thyroid 1. Welch, D. R., and Tomasovic, S. P. Implications of tumor progression on clinical cells. Anal. Biochem., 247: 165–167, 1997. oncology. Clin. Exptl. Metastasis, 3: 151–188, 1985. 26. Polakis, P. Wnt signaling and cancer. Genes Dev., 14: 1837–1851, 2000. 2. Fidler, I. J. The biology of cancer metastasis or, “you cannot fix it if you do not know 27. Smalley, M. J., and Dale, T. C. Wnt signaling and mammary tumorigenesis. how it works.” Bioessays, 13: 551–554, 1991. J. Mammary Gland Biol. Neoplasia, 6: 37–52, 2001. 3. Fidler, I. J., and Radinsky, R. Search for genes that suppress cancer metastasis. 28. Taipale, J., and Beachy, P. A. The hedgehog and Wnt signaling pathways in cancer. J. Natl. Cancer Inst., 88: 1700–1703, 1996. Nature (Lond.), 411: 349–354, 2001. 439

29. Chen, K. S., and DeLuca, H. F. Isolation and characterization of a novel cDNA from 36. Rachez, C., Lemon, B. D., Suldan, Z., Bromleigh, V., Gamble, M., Na¨a¨r, A. M., HL-60 cells treated with 1, 25-dihydroxyvitamin D-3. Biochim. Biophys. Acta, 1219: Erdjument-Bromage, H., Tempst, P., and Freedman, L. P. Ligand-dependent tran- 26–32, 1994. scription activation by nuclear receptors requires the DRIP complex. Nature (Lond.), 30. Junn, E., Han, S. H., Im, J. Y., Yang, Y., Cho, E. W., Um, H. D., Kim, D. K., Lee, 398: 824–828, 1999. K. W., Han, P. L., Rhee, S. G., and Choi, I. Vitamin D3 up-regulated protein 1 37. Ryu, S., Zhou, S., Ladurner, A. G., and Tjian, R. The transcriptional cofactor complex mediates oxidative stress via suppressing the thioredoxin function. J. Immunol., 164: CRSP is required for activity of the enhancer-binding protein Sp1. Nature (Lond.), 6287–6295, 2000. 397: 446–450, 1999. 31. Yamanaka, H., Maehira, F., Oshiro, M., Asato, T., Yanagawa, Y., Takei, H., and 38. Welch, D. R., and Rinker-Schaeffer, C. W. What defines a useful marker of metas- Nakashima, Y. A possible interaction of thioredoxin with VDUP1 in HeLa cells detected tasis in human cancer?. J. Natl. Cancer Inst., 91: 1351–1353, 1999. in a yeast two-hybrid system. Biochem. Biophys. Res. Comm., 271: 796–800, 2000. 39. Yoshida, B. A., Sokoloff, M., Welch, D. R., and Rinker-Schaeffer, C. W. Metastasis- 32. Nishiyama, A., Matsui, M., Iwata, S., Hirota, K., Masutani, H., Nakamura, H., suppressor genes: a review and perspective on an emerging field. J. Natl. Cancer Inst., Takagi, Y., Sono, H., Gon, Y., and Yodoi, J. Identification of thioredoxin-binding 92: 1717–1730, 2000. protein-2/vitamin D(3) up-regulated protein 1 as a negative regulator of thioredoxin function and expression. J. Biol. Chem., 274: 21645–21650, 1999. 40. Welch, D. R., and Goldberg, S. F. Molecular mechanisms controlling human mela- 33. Yoshida, B. A., Dubauskas, Z., Chekmareva, M. A., Christiano, T. R., Stadler, W. M., noma progression and metastasis. Pathobiology, 65: 311–330, 1997. and Rinker-Schaeffer, C. W. Mitogen-activated protein kinase kinase 4/stress-acti- 41. Welch, D. R., and Wei, L. L. Genetic and epigenetic regulation of human vated protein/Erk kinase 1 (MKK4/SEK1), a prostate cancer metastasis suppressor breast cancer progression and metastasis. Endocrine-related Cancer, 5: 155–197, gene encoded by human chromosome 17. Cancer Res., 59: 5483–5487, 1999. 1998. 34. Powis, G., and Montfort, W. R. Properties and biological activities of thioredoxins. 42. Nakamura, H., Nakamura, K., and Yodoi, J. Redox regulation of cellular activation. Annu. Rev. Pharmacol. Toxicol., 41: 261–295, 2001. Annu. Rev. Immunol., 15: 351–369, 1997. 35. Powis, G., and Montfort, W. R. Properties and biological activities of thioredoxins. 43. Rachez, C., and Freedman, L. P. Mediator complexes and transcription. Curr. Opin. Ann. Rev. Biophys. Biomol. Struct., 30: 421–455, 2001. Cell Biol., 13: 274–280, 2001.

440

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2003 American Association for Cancer Research. Melanoma Metastasis Suppression by Chromosome 6: Evidence for a Pathway Regulated by CRSP3 and TXNIP

Steven F. Goldberg, Mary E. Miele, Naohito Hatta, et al.

Cancer Res 2003;63:432-440.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/63/2/432

Cited articles This article cites 41 articles, 10 of which you can access for free at: http://cancerres.aacrjournals.org/content/63/2/432.full#ref-list-1

Citing articles This article has been cited by 26 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/63/2/432.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/63/2/432. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.