Abstract

Background: Numerous genetic changes are associated with metastasis and invasion of cancer cells. To identify differentially expressed invasion-associated genes, we screened a panel of lung cancer cell lines (CL1–0, CL1–1, CL1–5, and CL1–5-F4 in order of increasing invasive activity) for such genes and selected one gene, collapsin response mediator protein-1 (CRMP-1), to characterize. Methods: We used a microarray containing 9600 gene sequences to assess gene expression in the cell panel and selected the differentially expressed CRMP-1 gene for further study. We confirmed the differential expression of CRMP-1 with northern and western blot analyses. After transfecting and overexpressing CRMP-1 in highly invasive CL1–5 cells, the cells were assessed morphologically and with an in vitro invasion assay. We used enhanced green fluorescent protein-tagged CRMP-1 and fluorescence microscopy to localize CRMP-1 intracellularly. CRMP-1 expression in 80 lung cancer specimens was determined by real-time quantitative reverse transcription–polymerase chain reaction (RT–PCR). All statistical tests were two-sided. Results: Expression of CRMP-1 was inversely associated with invasive activity in the cell panel, an observation confirmed by northern and western blot analyses. CRMP-1-transfected CL1–5 cells became rounded and had fewer filopodia and statistically significantly lower in vitro invasive activity than untransfected cells (all P<.001). During interphase, CRMP-1 protein was present uniformly throughout the cytoplasm and sometimes in the nucleus; during mitosis, CRMP-1 was associated with mitotic spindles, centrosomes, and the midbody (in late telophase). Real-time RT-PCR of lung cancer specimens showed that reduced expression of CRMP-1 was statistically significantly associated with advanced disease (stage III or IV; P = .010), lymph node metastasis (N1, N2, and N3; P = .043), early postoperative relapse (P = .030), and shorter survival (P = .016). Conclusions: CRMP-1 appears to be involved in cancer invasion and metastasis and may be an invasion-suppressor gene.

Metastasis, the spread of tumor cells from a primary tumor to secondary sites within the body, is a complicated process involving degradation of the basement membrane, invasion of the stroma, adhesion, angiogenesis, cell proliferation, and migration (1). Molecular aspects of metastasis are not clearly understood, and a variety of positive and negative factors may be involved (2), as well as numerous genetic changes. In addition, current clinical methods cannot accurately identify which patients will develop metastatic disease. Thus, improved methods to predict and to diagnose metastasis and improved treatments of metastasis are needed. One approach to solving this problem is to identify the genes controlling metastasis (3).

Cancers are a heterogeneous mass of neoplastic cells with various properties, including different metastatic activity (4). During carcinogenesis, some tumor cells probably acquire new phenotypes by the increased expression of metastasis-promoting genes or the decreased expression of metastasis-suppressor genes. Studies of differential gene expression in poorly metastatic cancer cells and highly metastatic cancer cells should identify genes associated with metastasis and, in fact, several metastasis-suppressor genes, such as NM23, KAI1, and KiSS-1, have been so identified (57). New high-throughput genomic technologies, such as DNA microarray technology, may facilitate gene-expression profiling of cancers. DNA microarray technology can analyze the expression of many genes simultaneously and has identified numerous differentially expressed genes (811). For example, Clark et al. (12) found that the small guanosine 5′-triphosphatase RhoC, when overexpressed, enhanced metastasis.

To identify genes associated with invasion, we used an established panel of human lung adenocarcinoma cell lines (13,14) with different invasive activities (CL1–0 cells and its sublines CL1–1, CL1–5, and CL1–5-F4 cells, in ascending order of activity), a complementary DNA (cDNA) microarray, and a colorimetric detection system (14,15). We selected a differentially expressed gene, collapsin response mediator protein-1 (CRMP-1), and determined whether its expression was associated with the invasive activity and metastasis of cancer cells.

Materials and Methods

Cell Lines and Culture Conditions

The panel of human lung adenocarcinoma cell lines, CL1–0, CL1–1, CL1–5, and CL1–5-F4, in ascending order of invasiveness, was established in our laboratory (13,14). Cells were grown in RPMI-1640 medium (Life Technologies, Inc. [GIBCO BRL], Rockville, MD) with 10% fetal bovine serum (FBS) (Life Technologies, Inc.) and 2 mMl-glutamine (Life Technologies, Inc.) at 37 °C in a humidified atmosphere of 5% CO2–95% air. CL1–0 is the parent cell line; CL1–1 and CL1–5 are sublines that were selected from CL1–0 cultures with a polycarbonate membrane coated with Matrigel (Collaborative Biomedical, Becton Dickinson Labware, Bedford, MA) in a Transwell invasion chamber (13). The cells that migrated through the membrane were harvested. The sublines from the first and fifth rounds of selection were designed CL1–1 and CL1–5, respectively (13). In the next selection step, CL1–5 cells were injected into the tail vein of severe combined immunodeficient mice, and cells from a lung tumor that arose were isolated and cloned. After four repeated in vivo selections, the cell line obtained was designated CL1–5-F4(14).

Adherent cells were detached from the culture dishes with trypsin–EDTA (Sigma, Deisenhofen, Germany). Before functional assays, only 0.02% EDTA was used to avoid damaging cell-surface antigens.

Microarray Analysis

Microarrays containing 9600 cDNA sequences from the Integrated Molecular Analysis of Genome and their Expression Consortium libraries (Research Genetics, Huntsville, AL) were prepared on nylon membranes by an arraying machine built inhouse as described previously (11,14,15). Experiments, in which a colorimetric detection method was used, were performed as described by Hong et al. (11) and Chen et al. (14,15). Briefly, each cell line was grown to a confluence of 80%, cells were cultured in fresh medium for 24 hours before messenger RNA (mRNA) was isolated, and the time was minimized between removal of cells from the incubator and lysis. Total cellular RNAs were extracted with RNAzol B (Biotecx Laboratories, Houston, TX), and mRNAs were isolated with oligotex-dT resin (Qiagen, Hilden, Germany). Purified mRNA (5 μg) derived from each cell line was reversed transcribed and labeled with biotin. The microarray carrying the double-stranded cDNAs was prehybridized in 7 mL of hybridization buffer (5× standard saline citrate [SSC], 0.1% N-lauroylsarcosine, 0.1% sodium dodecyl sulfate [SDS], 1% blocking reagent mixture manufactured by Roche Molecular Biochemicals, Mannheim, Germany, and salmon-sperm DNA [50 μg/mL]) at 68 °C for 1 hour. The biotin-labeled cDNA probes and hybridization solution (80 μL) containing human COT-1 DNA instead of salmon-sperm DNA were sealed with a microarray in a hybridization bag, and the bag was incubated at 95 °C for 2 minutes and then at 68 °C for 12 hours. After hybridization, the color reaction was initiated by incubating the membrane for 2 hours in 5 mL containing β-galactosidase-conjugated streptavidin (diluted 1 : 700; Life Technologies, Inc.), 4% polyethylene glycol (molecular weight, 8000; Sigma Chemical Co., St. Louis, MO), and 0.3% bovine serum albumin (BSA) in Tris-buffered saline (pH = 7.4). Color reactions were stopped by adding phosphate-buffered saline (PBS) containing 20 mM EDTA. The microarray image was digitized by use of a drum scanner (ScanView, Foster City, CA); image analysis and spot quantification were then done by the MuCDA program written inhouse and available via anonymous ftp at ftp://genestamp.ibms.sinica.edu.tw/marray/software. Gene expression data were processed by use of the protocol and program described by Iyer et al (10). Genes were grouped on the basis of expression profiles by the self-organizing maps algorithm, described by Tamayo et al. (16), and then genes whose expression profiles correlated either positively or negatively with the invasiveness of cell lines were identified. We selected CRMP-1 for further study.

Molecular Cloning and Plasmid Constructs

RNA from CL1–0 cells was reverse transcribed with SuperScript II reverse transcriptase (Life Technologies, Inc.) and random hexamer. cDNA encoding the entire human CRMP-1 coding region (GenBank accession number D78012) was amplified by polymerase chain reaction (PCR). Primer sequences were as follows: 5′ primer = 5′-CTCCGTCCGTGTCTCTATCC-3′ (nucleotides 24–43 of D78012); 3′ primer = 5′-CCTCCATCAGCACCAACTAAA-3′ (complementary to nucleotides 1955–1975). The reaction mixture was denatured at 94 °C for 30 seconds, annealed at 55 °C for 30 seconds, and extended at 72 °C for 3 minutes; this cycle was repeated 30 times. The 1952-base-pair CRMP-1 cDNA fragment was cloned into a TA vector, according to the manufacturer's instructions (pGEM-T-Easy cloning kit; Promega Corp., Madison, WI), and sequenced with an autosequencer (model ABI 377; PE Applied Biosystems, Foster City, CA). Sequence analysis showed 100% homology to the published sequence (17) for CRMP-1 cDNA.

pCIneo-CRMP-1 was created by inserting nucleotides 24–1975 of the CRMP-1 cDNA between the EcoRI and NotI sites of a pCI-neo mammalian expression vector (Promega Corp.) and used for transfection and expression of CRMP-1 in CL1–5 cells. Nucleotides 376–1975 of the CRMP-1 cDNA were inserted into the PstI and EcoRI sites of a pRSET C prokaryotic expression vector (Invitrogen Corp., Carlsbad, CA) to construct pRSET-CRMP-1, which was used to produce the protein to immunize mice. The coding region of CRMP-1 cDNA (nucleotides 151–1869) was amplified by PCR from pCIneo-CRMP-1 by use of the forward primer 5′-ATTGACTCGAGATGTCGTACCAGGGCAAGAA-3′ (nucleotides 151–170 from CRMP-1 sequence), which introduced an XhoI site (underlined), and the reverse primer 5′-ATATCGAATTCTCAACCGAGGCTGGTGAT-3′ (complementary to nucleotides 1852–1869), which introduced an EcoRI site (underlined). For protein localization study, the CRMP-1 fragment amplified in this manner was inserted inframe between the XhoI and EcoRI sites of a cytomegalovirus promoter-driven enhanced green fluorescent protein (EGFP) expression vector, pEGFP-C3 (Clontech, Palo Alto, CA), yielding pEGFP-CRMP-1, which produced the EGFP-tagged CRMP-1 protein. These constructs were isolated from plasmid clones and sequenced on both strands with an autosequencer (model ABI 377; PE Applied Biosystems).

Monoclonal Antibody (MAb) Production

pRSET-CRMP-1 was inserted into Escherichia coli BL21(DE3)pLysS cells. The CRMP-1 protein fused with polyhistidine, named His-CRMP-(amino acids 76–572), was obtained as inclusion bodies, which were solubilized. The fusion protein was then purified by SDS–polyacrylamide gel electrophoresis (PAGE) in 12.5% gels. Six-week-old BALB/c mice were given a subcutaneous injection of purified fusion protein in 0.1 mL of Freund's complete adjuvant (Life Technologies, Inc.) emulsified in 0.1 mL of sterile PBS; every 3 weeks, they then received subcutaneous booster injections of fusion protein in 0.1 mL of Freund's incomplete adjuvant (Life Technologies, Inc.) emulsified in 0.1 mL of sterile PBS. We checked the antibody titer by western blot analysis; preimmune serum was the negative control. To generate hybridomas, splenocytes from one immunized mouse were fused with mouse myeloma cells at a ratio of 10 : 3 in the presence of polyethylene glycol (molecular weight, 1500; Roch GmbH, Mannheim, Germany). Hybridomas were plated in 96-well plates in Dulbecco's modified Eagle medium–RPMI-1640 medium, 1 : 1 (vol/vol), supplemented with 15% NuSerum (Collaborative Research, Becton Dickinson Labware) and hypoxanthine–aminopterin–thymidine (Sigma Chemical Co., St. Louis, MO). Supernatants were screened by western blotting against the His-CRMP-(amino acids 76–572) fusion protein.

Northern and Western Blot Analysis

For northern blot analysis, each lane of an 0.8% agarose–formaldehyde gel was loaded with 20 μg of total RNA. After electrophoresis in 1× 3-(N-morpholino) propanesulphonic acid running buffer, containing 20 mM MOPS (pH 7.0), 5 mM sodium acetate, and 0.1 mM EDTA (pH 8.0), RNAs were blotted onto Hybond-N+ nylon membranes (Amersham, Buckinghamshire, U.K.) by the capillary method and were UV-cross-linked to the membranes. Blots were prehybridized in a solution of 5× SSC, 5× Denhardt's solution, 50 mM sodium phosphate (pH 6.2), salmon-sperm DNA (100 μg/mL), and 50% deionized formamide for 4 hours at 42 °C. Denhardt's solution (1×) contains 0.02% polyvinylpyrrolidone, 0.02% Ficoll, and 0.02% BSA. Membranes were then hybridized with 32P-labeled DNA probes synthesized by the Rediprime DNA labeling system (Amersham Life Science, Inc.). Hybridization was performed for 18 hours at 42 °C in a solution of 5× SSC, 5× Denhardt's solution, 10% dextran sulfate, 20 mM sodium phosphate (pH 6.2), salmon-sperm DNA (100 μg/mL), and 50% deionized formamide. Membranes were washed for two 15-minute periods in 2× SSC–0.5% SDS at room temperature, then for two 30-minute periods in 0.1× SSC–0.1% SDS at 52 °C, and then exposed to x-ray film overnight at –70 °C. The RNA in each lane was measured by comparing its signal intensity with that of the guanine nucleotide-binding protein β-subunit-like (Gβlike) probe (a housekeeping gene used as an internal control for RNA quantity) (18).

Proteins in the total cell lysate (5 μg of protein) were separated by SDS–PAGE in 12.5% gels and transferred to a polyvinylidene difluoride membrane (Immobilon-P membrane; Millipore Corp., Bedford, MA) by electrotransfer. After the blot was blocked in a solution of 5% skim milk, 0.1% Tween 20, and PBS, membrane-bound proteins were probed with MAbs, the membrane was washed, and the secondary antibody (horseradish peroxidase-conjugated anti-mouse immunoglobulin G antibody [Amersham]) was applied. Protein bands were detected with enhanced chemiluminescence (Amersham Life Science, Inc.) and x-ray film.

Transfection and Selection

pCIneo-CRMP-1 plasmids (5 μg) were transfected into 70% confluent CL1–5 cells with 20 U of LipofectAMINE reagent (Life Technologies, Inc.) as described previously (11). Other CL1–5 cells were transfected with the pCI-neo vector containing no insert (mock transfected) as a control. Gentamicin (G418; Life Technologies, Inc.) was added to 500 μg/mL for the selection of stable transfectants. Selection medium was changed every 3 days for a 3-week period. Clones of resistant cells were isolated and allowed to proliferate for further characterization. Integration of transfected plasmid DNA into chromosomal DNA was confirmed by northern blot analysis. For transient transfections, 70% confluent cultures of CL1–0 and CL1–5 cells were transfected with the pEGFP-CRMP-1 plasmid as above. Forty-eight hours later, living cells were examined directly and photographed with a Zeiss Axiphot epifluorescence microscope equipped with an MRC-1000 laser scanning confocal imaging system (Bio-Rad Laboratories, Rockville Center, NY) (11).

In Vitro Invasion Assay and Cell Growth Assay

The invasive activity of transfected clones was examined by use of a membrane invasion culture system (11), in which a polycarbonate membrane with 10-μm pores (Nucleopore Corp., Pleasanton, CA) coated with Matrigel (Collaborative Biomedical, Becton Dickinson Labware) at 5 mg/mL was placed between the upper and lower well plates of a membrane invasion culture system chamber. Cells were suspended in RPMI-1640 medium containing 10% NuSerum, and 2.5 × 104 cells were placed into each upper well of the chamber. After incubating for 48 hours at 37 °C, cells that had invaded through the coated membrane were removed from the lower wells with 1 mM EDTA in PBS and dot blotted onto a polycarbonate membrane with 3-μm pores. Blotted cells were stained with propidium iodine (Sigma Chemical Co., St. Louis, MO), and the number of cells in each blot was counted under a microscope at a magnification of ×50 by use of the Analytical Imaging Station software package (Imaging Research Inc., ON, St. Catharines, Canada). Each experiment was performed three times, and each sample was assayed in triplicate.

Cell growth was measured with the modified 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (19). Four thousand cells and 100 μL of culture medium per well of a 96-well plate were incubated for up to 4 days, and 10 μL of MTT (5 mg/mL) was added to each well and incubated for 4 hours at 37 °C. The reaction was stopped by adding 100 μL of 0.04 N HCl in isopropanol to each well, with vigorous mixing to solubilize colored crystals produced by the reaction. The absorbance at 570 nm relative to the absorbance at 630 nm, as the reference wavelength, was measured by a multiwell scanning spectrophotometer (Titertek Multiskan; Flow Laboratories, McLean, VA). Cell viability was examined by trypan blue dye exclusion. Each data point is the average of six determinations, and each experiment was repeated at least three times.

Filamentous Actin (F-actin) Staining

Cells grown on coverslips were washed three times in PBS, fixed for 10 minutes in 3.7% paraformaldehyde in PBS, and permeabilized for 10 minutes with 0.1% Triton X-100 in PBS. Nonspecific binding sites were blocked by a 15-minute incubation in 5% nonfat milk and PBS. After a 5-minute wash with PBS, cells were incubated with rhodamine-conjugated phalloidin at 5 U/mL (Molecular Probe, Eugene, OR) for 30 minutes. After additional PBS washes, cells were mounted on a slide in mounting medium (2% n-propyl gallate and 60% glycerol in PBS [pH 8.0]). Cells were examined and photographed with a Zeiss Axiphot epifluorescence microscope (Bio-Rad Laboratories) on Kodak T-max 400 film (Eastman Kodak Co., Rochester, NY).

Patients and Specimens

Eighty consecutive patients who underwent surgery for non-small-cell lung cancer (performed by Y.-C. Lee) at the National Taiwan University Hospital from September 1, 1994, through December 31, 1996, were included in the study. This investigation was performed after approval by the Institutional Review Board of the National Taiwan University Hospital. Written informed consent was obtained from all patients. None of the patients had received neoadjuvant chemotherapy or radiation therapy before surgery. Specimens of lung cancer tissue and adjacent normal lung tissue obtained at surgery were immediately snap-frozen in liquid nitrogen and stored at –80 °C until use. World Health Organization criteria (20) were used for histologic classification. Tumor size, local invasion, and lymph node metastasis were determined at pathologic examination. The final disease stage was determined by a combination of surgical and pathologic findings, according to the current tumor–node–metastasis system for lung cancer staging (21). Among the 80 patients, 51 were men and 29 were women (mean age [± standard deviation] = 62.9 ± 10.5 years), 38 of whom had squamous cell carcinoma and 42 of whom had adenocarcinoma. Nineteen of 51 men and 23 of 29 women had adenocarcinoma. The sex difference was statistically significant (P<.001, Fisher's exact test; two-sided) and consistent with the previous epidemiologic report in Taiwan (22). The surgical–pathology stage of disease was stage I in 31 patients, stage II in 19, stage III in 24, and stage IV in six. Tumor status was T1 in 15 patients, T2 in 42, T3 in nine, and T4 in 14. Forty-two patients had no lymph node metastasis (N0), and 38 had regional or mediastinal lymph node metastasis (N1 in 19 patients, N2 in 18 patients, and N3 in one patient). Follow-up data were obtained from the patients' medical charts and from our tumor registry service. Follow-up times, ranging from 6.5 to 70 months, lasted until June 30, 2000. Relapse time was calculated from the date of surgery to the date of detection of local recurrence or systemic metastasis. Survival time was calculated from the date of surgery to the date of death. Patients who died of postoperative complications within 30 days after surgery were excluded from the survival analysis. The relapse time ranged from 2 to 70 months (median, 17.2 months), and the survival time ranged from 6.5 to 70 months (median, 34.7 months).

Real-Time Quantitative Reverse Transcription (RT)–PCR

Total RNA was extracted from resected cancer tissue with an RNA extraction kit (RNeasy Mini Kit; Qiagen, Valencia, CA). The quality of RNA in samples was determined by electrophoresis through agarose gels and staining with ethidium bromide; 18S and 28S RNA bands were visualized with UV illumination. The samples used for the standard curve in the real-time quantitative RT–PCR were prepared by serial dilution of a specific RNA sample to contain 250, 50, 10, and 2 ng. The serially diluted samples were distributed into aliquots and stored at –80 °C until use. The primers used, based on the cDNA sequence of CRMP-1, were as follows: forward primer = 5′-CCACGATGATCATTGACCATGT-3′; reverse primer = 5′-AGGGAGTAATCACAGCAGGATTTG-3′. The sequence of the probe used to detect and quantify the RT–PCR product was 5′-AGCCTACTGACCTCTTTCGAGAAGTGGCA-3′ (23). The probe was labeled at the 5′ end with carboxyfluorescein and at the 3′ end with N,N,N′,N′-tetramethyl-6-carboxyrhodamine. The primers and probe used for quantitative RT–PCR of the TATA-box-binding protein (TBP) mRNA (internal control, GenBank accession number X54993) were as described by Bieche et al. (24). The identities of PCR products were confirmed by DNA sequencing. Each assay included samples to determine a standard curve, a no-template control, and total RNA samples in triplicate. The reaction conditions were as described previously (25). The fluorescence emitted by the reporter dye was detected online in real time with the ABI prism 7700 sequence detection system (PE Applied Biosystems). The threshold cycle (CT) is the fractional cycle number at which the fluorescence generated by cleavage of the probe exceeds a fixed level above baseline. For a chosen threshold, a smaller starting copy number results in a higher CT value. In this study, we used TBP mRNA as an internal control (24). The relative amount of tissue CRMP-1 mRNA, standardized against the amount of TBP mRNA, was expressed as –ΔCT = –[CT(CRMP–1)CT(TBP)]. The ratio of the number of CRMP-1 mRNA copies to the number of TBP mRNA copies was then calculated as 2–ΔCT × K, where K is a constant (25).

Statistical Analysis

Where appropriate, the data are presented as the mean ± standard deviation. All statistical analyses were performed with the Statistical Program for Social Sciences program package, version 8.0 (Chicago, IL). Comparisons of data between groups were made with Student's t test. The paired Kendall's W test was used to compare the –ΔCT of CRMP-1 mRNA expression between tumor samples and paired normal tissues. Fisher's exact tests and Student's t tests were used to compare the clinicopathologic characteristics of tumors (and patients) with high and low expression of CRMP-1 mRNA. Survival curves were obtained by the Kaplan–Meier method, and the difference in relapse time between groups with low and high expression of CRMP-1 was analyzed with the log-rank test (as was the difference in survival). All statistical tests were two-sided. P values of less than .05 were considered to be statistically significant.

Results

Identification of Differentially Expressed CRMP-1 mRNA by cDNA Microarray Technology

We used the cDNA microarray technique with colorimetric detection to identify differentially expressed genes associated with invasive activity in a panel of lung cancer cell lines, CL1–0, CL1–1, CL1–5, and CL1–5-F4, shown in increasing order of invasiveness. All experiments with the cDNA microarrays were performed in triplicate. Cell lines were grown independently in three cultures; for each culture, the entire process was carried out independently from RNA extraction to image analysis. The variation (mean ± standard deviation) in the triplicate experiments was 7.3% ± 3.3%. Cluster analysis of cDNA microarray data revealed that about 500 genes were positively or negatively associated with cancer cell invasiveness (14). Most of these genes were involved in angiogenesis, cell motility, adhesion, and proliferation. However, we found that the expression of CRMP-1 mRNA was negatively associated with cell invasiveness (Fig. 1, A) and have further characterized the expression of this gene.

Northern blot hybridization confirmed that CRMP-1 expression was much lower in CL1–5 and CL1–5-F4 cells than in CL1–0 and CL1–1 cells (Fig. 1, B). To confirm that the differential expression of CRMP-1 mRNA was also reflected by the expression of CRMP-1 protein, we used western blot analysis and MAb Y21, which recognized CRMP-1 but not other CRMPs (data not shown). We observed that the expression of CRMP-1 protein was also lower in CL1–5 and CL1–5-F4 than in CL1–0 and CL1–1 cells (Fig. 1, C).

Overexpression of CRMP-1 and Inhibition of Invasion In Vitro

To investigate whether the invasive phenotype and CRMP-1 expression were associated, CRMP-1 cDNA was inserted into the pCI-neo vector, the construct was transfected into CL1–5 cells, and five clones that stably expressed CRMP-1 were isolated. Transfected clones expressed higher levels of CRMP-1 transcription (Fig. 1, D) and translation (Fig. 1, E) products than the control clone transfected with pCI-neo alone. We used an in vitro-reconstituted basement membrane invasion assay to investigate whether CRMP-1 expression affected the invasive activity of cancer cells. After a 48-hour incubation, we noted a statistically significant reduction (40%–60%) in the invasive activity of CRMP-1-expressing clones compared with control clones (P<.001, Student's t test; Fig. 1, F). CRMP-1-transfected clone B5 expressed less CRMP-1 protein than did clone C6 (Fig. 1, E), but the invasion-suppression abilities of both clones were similar (Fig. 1, F), suggesting that there were other pathways that regulated the invasion ability of cancer cells or that the invasion-suppression ability of CRMP-1 protein was limited by an as yet unidentified protein that interacts with CRMP-1.

To determine how CRMP-1 might regulate the invasive ability of tumor cells, we first examined cell proliferation. We used a modified MTT assay to measure in vitro cell growth rates of CRMP-1-transfected cells and mock-transfected cells. In vitro proliferation was essentially unchanged by altering CRMP-1 expression (data not shown), ruling out the possibility that CRMP-1 is involved in regulating cell proliferation.

F-actin Polymerization in CRMP-1-Overexpressing Cells

F-actin is continuously polymerized and depolymerized in motile cells (26). Tumor-invasive activity is associated with the polymerized actin that is temporally and spatially found in lamellar protrusions in motile cells and plays a key role in cell motility (27). Phalloidin binds tightly to actin subunits in filaments but not to free actin monomers. We stained CL1–0, CL1–1, CL1–5, and CL1–5-F4 cells with rhodamine-conjugated phalloidin and examined the cells with fluorescence microscopy. There were fewer filopodia in CL1–0 cells than in CL1–5 or CL1–5-F4 cells (data not shown). We also stained the F-actin in transfected cell clones. CRMP-1-transfected cells had decreased invasive activity (Fig. 1, F) and a rounded morphology (Fig. 2, A). Mock-transfected cells had as many filopodia as parental CL1–5 cells, but CRMP-1-transfected CL1–5 cells had fewer filopodia (Fig. 2, A), closer to the number on CL1–0 cells.

Dynamic Distribution of CRMP-1 in the Cell Cycle and Association of CRMP-1 With Mitotic Spindle

To determine the intracellular localization of CRMP-1 in different stages of the cell cycle, we cloned the full coding region of CRMP-1 cDNA into a mammalian transfection vector so that an enhanced EGFP gene was fused to the CRMP-1 gene in CL1–0 and CL1–5 cells. Fluorescent images, obtained by laser scanning confocal microscopy, showed that EGFP-tagged CRMP-1 was distributed similarly in CL1–0 and CL1–5 cells. At interphase, CRMP-1 was diffusely distributed in the cytoplasm (Fig. 2, B), a pattern distinct from that of microtubules and F-actin. However, in some EGFP-CRMP-1-transfected cells, CRMP-1 was located in the cytoplasm and in the nucleus (Fig. 2, B). At prophase, some of CRMP-1 accumulated at the centrosomes (Fig. 2, B). During metaphase, most CRMP-1 was associated with the mitotic spindle and the centrosomes (Fig. 2, B). During anaphase, CRMP-1 was still associated with mitotic spindle (Fig. 2, B). During very late telophase, CRMP-1 was associated with the midbody (Fig. 2, B).

Association of CRMP-1 mRNA Expression in Lung Cancer Tissue, Postoperative Relapse, and Survival of Patients With Lung Cancer

Real-time quantitative RT–PCR was used to determine numbers of CRMP-1 transcripts in normal lung and lung cancer tissues from 80 patients with lung cancer. Expression of CRMP-1 mRNA in all 80 tumor samples was statistically significantly lower than in adjacent normal tissue (P<.001, paired Kendall's W test; two-sided). The –ΔCT value for the 80 tumor samples ranged from −5.67 to 3.73, with a mean of −1.94 (95% confidence interval [CI] = −2.41 to −1.47) and a median of −1.95. We arbitrarily used the median value to classify patients into high-expression or low-expression groups (Table 1). Low-expression patients were more likely than high-expression patients to have advanced disease (stage III or IV; P = .010) and lymph node metastasis (N1, N2, and N3; P = .043). The median duration to postoperative recurrence was also longer in high-expression patients (30.5 months; 95% CI = 5.6 to 55.5 months) than in low-expression patients (15.9 months; 95% CI = 8.3 to 23.5 months) (log-rank test, P = .030; Fig. 3, A). For high-expression patients, the probability of survival leveled off at .52, and they had a statistically significantly longer survival than low-expression patients (median survival, 27.9 months; 95% CI = 20.8 to 35.0 months; log-rank test, P = .016; Fig. 3, B).

Discussion

By using genome-wide cDNA microarray screening for invasion-associated genes, we identified CRMP-1, a potential invasion-suppressor gene. We found that expression of both CRMP-1 mRNA and protein was inversely associated with the invasive activity of a panel of lung cancer cell lines and that overexpression of CRMP-1 in a highly invasive cell line reduced its in vitro invasive activity and altered its F-actin cytoskeleton. In addition, we found that expression of CRMP-1 mRNA in lung cancer specimens was inversely associated with advanced disease, lymph node metastasis, early postoperative recurrence, and survival of patients with lung cancer.

Members of the CRMP family of phosphoproteins may mediate semaphorin/collapsin-induced growth cone collapse and are involved in axonal guidance and neuronal differentiation (2834). Five members of the CRMP gene family (CRMP-1, CRMP-2, CRMP-3, CRMP-4, and CRMP-5), encoding closely related 60- to 66-kd proteins, have been cloned (2834). The best-characterized function of these proteins remains the repulsive guidance of nerve axons (28), but the molecular mechanism is not yet characterized (28). CRMP family members have a 50%–70% amino-acid sequence homology (17,28,30) and are believed to form heterotetramers (30,35). However, each CRMP has a unique transcript and is expressed in a distinct pattern during development of the nervous system and nerve growth factor-induced neuronal differentiation (17,34,36). Transcription of CRMP genes is differentially regulated (3739). CRMP-2 is associated with neurofibrillary tangles in patients with Alzheimer's disease (40). CRMP-3 and CRMP-5 are recognized by autoantibodies from patients with small-cell lung cancers with paraneoplastic neurologic syndrome (41,42). In this study, we show that CRMP-1 may be involved in cancer cell invasion. Although we investigated the expression of CRMP-2, -3, -4, and -5 mRNAs in the lung cancer cell lines CL1–0, CL1–1, CL1–5, and CL1–5-F4 by using northern blot and RT–PCR analyses, we detected no association between these CRMPs and invasiveness.

Recent studies demonstrated that members of the semaphorin/collapsin families might control the movement of cells (43), and neuropilin–plexin complexes have been identified as semaphorin receptors (44). We used semiquantitative RT–PCR to determine the mRNA expression of semaphrin/collapsin-related proteins in CL1–0, CL1–1, CL1–5, and CL1–5-F4 cells (unpublished data) and found that expression of neuropilin-1 and neuropilin-2 was positively associated with invasiveness. These four cell lines expressed similar levels of plexin-A1, and expression of Sema3A was inversely associated with the invasiveness of these cells, consistent with reports that Sema3A inhibits migration of endothelial cells (45) and the branching morphogenesis of fetal mouse lung (46). Expression patterns of Sema3B, Sema3C, and Sema3F in CL1–0, CL1–1, CL1–5, and CL1–5-F4 cells were not associated with invasiveness. However, Sema3B and Sema3F genes have been proposed as tumor suppressor genes because they were located in the 3p21.3 region, where homozygous deletions have been identified in several small-cell lung cancers (4649). The localization and expression of Sema3F in lung cancer tissue and cell lines have suggested that it has a role in cell adhesion and cell migration (50), and Martin-Satué and Blanco (51) have found that Sema3C mRNA is overexpressed in metastatic human lung adenocarcinoma cell lines.

Invasion of cancer cells consists of a series of steps, including the attachment of the cells to an extracellular matrix, production of matrix-degrading enzymes, and increased cell motility (2). Actin plays a key role in various cell motility systems (27,52), including filopodia, highly dynamic actin-containing structures that protrude from the leading edges of motile cells. Filopodia are the first locomotory structures to appear in stimulated migratory cells and act as motors to pull the leading edges of the cell forward (5254). Thus, we investigated the effect of CRMP-1 on cell morphology and the actin cytoskeleton. Untransfected CL1–5 cells had an elongated morphology, but CRMP-1-transfected CL1–5 cells had a rounded morphology similar to that of CL1–0 cells. We detected changes in the distribution of F-actin and the loss of filopodia in CRMP-1-transfected CL1–5 cells, compared with untransfected CL1–5 cells, resulting in a pattern similar to that of CL1–0 cells. This is in contrast to the results of Gu and Ihara (55), who reported no apparent change in the organization of actin filaments in Neuro2a cells overexpressing CRMP-2. Thus, each CRMP family member may have a distinct biologic role.

The distribution of CRMP-1 changes dynamically during the cell cycle. In interphase, CRMP-1 was dispersed diffusely throughout the cytoplasm similar to the distribution of CRMP-2 (55) and was sometimes observed in the nucleus. To our knowledge, this is the first report of a CRMP family member localized to the nucleus. Although the CRMP-1 sequence does have a nuclear localization signal (bipartite = RKAFPEHLYQRVKIRNK beginning at amino acid 471) (56), the biologic activity of CRMP-1 in the nucleus is not clear. In mitotic cells, CRMP-1 is associated with microtubule-containing structures. Certain proteins, such as p53 and BRCA1, that associate with the centrosome, mitotic spindle, and midbody are known to regulate the cell cycle (57,58). Thus, CRMP-1 may interact with other proteins, may be a component of the mitotic machinery, and may participate in cell cycle regulation. The identification of partners for CRMP-1 in various cellular activities should provide further clues to its biologic roles and clarify the mechanisms underlying cancer invasion.

Acquisition of invasive properties is a key event in tumor progression. Large-scale gene expression analysis has allowed identification of the differentially expressed genes responsible for invasion, including CRMP-1. We found that expression of CRMP-1 affects the invasive activity of lung cancer cells in a functional assay, that CRMP-1 is differentially expressed in tumor tissues, and that its expression is related to tumor stage, lymph node metastasis, and survival of patients. These results indicate that CRMP-1 may be involved in the inhibition of tumor cell invasion and metastasis. Further investigation is needed to elucidate the mechanism by which CRMP-1 modulates carcinoma invasion and metastasis.

Table 1.

Clinicopathologic characteristics of tumors with high and low expression of collapsin response mediator protein-1 messenger RNA

 –ΔCT  
Characteristic < −1.95 ≥−1.95 P 
*Derived with Student's t test; other P values were derived with Fisher's exact test. All statistical tests were two-sided. SD = standard deviation. 
†Tumor stage, tumor status, and lymph node status were classified according to the International System for Staging Lung Cancer (21). 
Age, y, mean ± SD 62 ± 10 63 ± 11 .627* 
Sex, No. of patients    
    Male 24 27 .642 
    Female 16 13  
Stage,† No. of patients    
    I–II 19 31 .010 
    III–IV 21  
Tumor status,† No. of patients    
    T1–2 25 32 .137 
    T3–4 15  
Lymph node status,† No. of patients    
    N0 16 26 .043 
    N1–3 24 14  
Histology, No. of patients    
    Squamous cell carcinoma 16 22 .263 
    Adenocarcinoma 24 18  
 –ΔCT  
Characteristic < −1.95 ≥−1.95 P 
*Derived with Student's t test; other P values were derived with Fisher's exact test. All statistical tests were two-sided. SD = standard deviation. 
†Tumor stage, tumor status, and lymph node status were classified according to the International System for Staging Lung Cancer (21). 
Age, y, mean ± SD 62 ± 10 63 ± 11 .627* 
Sex, No. of patients    
    Male 24 27 .642 
    Female 16 13  
Stage,† No. of patients    
    I–II 19 31 .010 
    III–IV 21  
Tumor status,† No. of patients    
    T1–2 25 32 .137 
    T3–4 15  
Lymph node status,† No. of patients    
    N0 16 26 .043 
    N1–3 24 14  
Histology, No. of patients    
    Squamous cell carcinoma 16 22 .263 
    Adenocarcinoma 24 18  
Fig. 1.

Identification of differentially expressed collapsin response mediator protein-1 (CRMP-1) by complementary DNA (cDNA) microarray (panels A–C) and inhibition of invasion by overexpression of CRMP-1 (panels D–F). Panel A: Microarray images show higher expression of CRMP-1 in less invasive cell lines (cells lines in order of invasive activity are as follows: CL1–0 < CL1–1 < CL1–5 < CL1–5F4). Arrows = CRMP-1. Panel B: northern blot analysis of total RNA for CRMP-1. The 1.95-kilobase (kb) full-coding region of CRMP-1 cDNA was used as the probe. The 4.9-kb transcript of CRMP-1 is indicated. The guanine nucleotide-binding protein β-subunit-like (Gβ-like) probe was used as an internal control for RNA quantity. Panel C: western blot analysis of CRMP-1. We generated a specific monoclonal antibody, Y21, against CRMP-1 by use of the CRMP-1 fusion protein His-CRMP-(amino acids 76–572). Lanes 1–5 were probed with CRMP-1 monoclonal antibody Y21; lane 6 was probed with preimmune serum as the control. Lane 1 = CL1–0 cells; lane 2 = CL1–1 cells; lane 3 = CL1–5 cells; lane 4 = CL1–5-F4 cells; and lanes 5 and 6 = fusion protein His-CRMP (amino acids 76–572). Western blot analyses of three experiments produced the same results. M.W. = molecular weight; kD = kilodalton. Panel D: northern blot analysis of CRMP-1 expression in mock-transfected CL1–5 cells and CRMP-1-transfected CL1–5 cell clones (B5, C6, C8, C10, and C22). The 1.95-kb full-coding region of CRMP-1 cDNA was used as the probe. All five CRMP-1-transfected clones expressed CRMP-1 messenger RNA, but mock-transfected cells did not. The Gβ-like probe was used as an internal control for RNA quantity. Panel E: western blot analysis of CRMP-1 expression in mock-transfected CL1–5 cells and CRMP-1-transfected CL1–5 cell clones (B5, C6, C8, C10, and C22). CRMP-1 monoclonal antibody Y21 was used as the primary antibody. CRMP-1 protein was expressed in CRMP-1-transfected clones but not in mock-transfected cells. CL1–0 and CL1–5 cells were the positive and negative controls, respectively. M.W. = molecular weight; kD = kilodalton; arrow = 66-kD CRMP-1 proteins. Panel F: Expression of CRMP-1 suppressed the in vitro invasion activity of CL1–5 cells. To facilitate comparison of the relative invasiveness between mock-transfected clones (A2 and A5) and CRMP-1-transfected clones (B5, C6, C8, C10, and C22), values were normalized to the relative invasion activity of mock-transfected clone A5. Each clone was assayed in three experiments carried out in triplicate. CRMP-1-transfected clones (B5, C6, C8, C10, and C22), as a group, had statistically significantly reduced invasive activity (P<.001, by Student's t test; two-sided) compared with the mock-transfected clones (A2 and A5) as the other group. Error bars are 95% confidence intervals. All statistical tests were two-sided.

Fig. 1.

Identification of differentially expressed collapsin response mediator protein-1 (CRMP-1) by complementary DNA (cDNA) microarray (panels A–C) and inhibition of invasion by overexpression of CRMP-1 (panels D–F). Panel A: Microarray images show higher expression of CRMP-1 in less invasive cell lines (cells lines in order of invasive activity are as follows: CL1–0 < CL1–1 < CL1–5 < CL1–5F4). Arrows = CRMP-1. Panel B: northern blot analysis of total RNA for CRMP-1. The 1.95-kilobase (kb) full-coding region of CRMP-1 cDNA was used as the probe. The 4.9-kb transcript of CRMP-1 is indicated. The guanine nucleotide-binding protein β-subunit-like (Gβ-like) probe was used as an internal control for RNA quantity. Panel C: western blot analysis of CRMP-1. We generated a specific monoclonal antibody, Y21, against CRMP-1 by use of the CRMP-1 fusion protein His-CRMP-(amino acids 76–572). Lanes 1–5 were probed with CRMP-1 monoclonal antibody Y21; lane 6 was probed with preimmune serum as the control. Lane 1 = CL1–0 cells; lane 2 = CL1–1 cells; lane 3 = CL1–5 cells; lane 4 = CL1–5-F4 cells; and lanes 5 and 6 = fusion protein His-CRMP (amino acids 76–572). Western blot analyses of three experiments produced the same results. M.W. = molecular weight; kD = kilodalton. Panel D: northern blot analysis of CRMP-1 expression in mock-transfected CL1–5 cells and CRMP-1-transfected CL1–5 cell clones (B5, C6, C8, C10, and C22). The 1.95-kb full-coding region of CRMP-1 cDNA was used as the probe. All five CRMP-1-transfected clones expressed CRMP-1 messenger RNA, but mock-transfected cells did not. The Gβ-like probe was used as an internal control for RNA quantity. Panel E: western blot analysis of CRMP-1 expression in mock-transfected CL1–5 cells and CRMP-1-transfected CL1–5 cell clones (B5, C6, C8, C10, and C22). CRMP-1 monoclonal antibody Y21 was used as the primary antibody. CRMP-1 protein was expressed in CRMP-1-transfected clones but not in mock-transfected cells. CL1–0 and CL1–5 cells were the positive and negative controls, respectively. M.W. = molecular weight; kD = kilodalton; arrow = 66-kD CRMP-1 proteins. Panel F: Expression of CRMP-1 suppressed the in vitro invasion activity of CL1–5 cells. To facilitate comparison of the relative invasiveness between mock-transfected clones (A2 and A5) and CRMP-1-transfected clones (B5, C6, C8, C10, and C22), values were normalized to the relative invasion activity of mock-transfected clone A5. Each clone was assayed in three experiments carried out in triplicate. CRMP-1-transfected clones (B5, C6, C8, C10, and C22), as a group, had statistically significantly reduced invasive activity (P<.001, by Student's t test; two-sided) compared with the mock-transfected clones (A2 and A5) as the other group. Error bars are 95% confidence intervals. All statistical tests were two-sided.

Fig. 2.

Analysis of filamentous actin (F-actin) in collapsin response mediator protein-1 (CRMP-1)-transfected CL1–5 cells and mock-transfected CL1–5 cells (panel A) and subcellular localization of enhanced green fluorescent protein (EGFP)-tagged CRMP-1 in living CL1–0 cells in various phases of the cell cycle (panel B). Panel A, upper left: Phase-contrast micrography of clone C10, a representative CRMP-1-transfected clone, shows mostly rounded cells. Other CRMP-1-transfected clones (clones B5, C6, C8, and C22) also showed a similar morphology (data not shown). Upper right: Phase-contrast micrography of clone A2, a representative mock-transfected clone, shows elongated cells. Other mock-transfected clones also showed a similar morphology (data not shown). Lower left: Fluorescent staining of cells from clone C10. Actin filaments were detected with rhodamine-conjugated phalloidin, and cells seldom have filopodia (long, thin, needle-like projections protruding from the cell edge). Lower right: Fluorescent staining of F-actin in A2 cells shows numerous filopodia protruding from the cell surfaces. Scale bar = 10 μm. Panel B: CL1–0 cells were transiently transfected with the plasmid pEGFP-CRMP-1, carrying the fusion gene for EGFP-tagged CRMP-1, and the living cells were examined directly and photographed with a confocal microscope. Upper left: interphase cell with fluorescent CRMP-1 diffusely distributed in the cytoplasm. Upper middle: interphase cell with CRMP-1 localized in the nucleus and cytoplasm. Upper right: early prophase cell with CRMP-1 localized to the centrosomes. Lower left: metaphase cell with CRMP-1 associated with the mitotic spindle and centrosomes. Lower middle: anaphase cell with CRMP-1 associated with the mitotic spindle. Lower right: very late telophase cell with CRMP-1 localized at the midbody. Scale bar = 5 μm.

Fig. 2.

Analysis of filamentous actin (F-actin) in collapsin response mediator protein-1 (CRMP-1)-transfected CL1–5 cells and mock-transfected CL1–5 cells (panel A) and subcellular localization of enhanced green fluorescent protein (EGFP)-tagged CRMP-1 in living CL1–0 cells in various phases of the cell cycle (panel B). Panel A, upper left: Phase-contrast micrography of clone C10, a representative CRMP-1-transfected clone, shows mostly rounded cells. Other CRMP-1-transfected clones (clones B5, C6, C8, and C22) also showed a similar morphology (data not shown). Upper right: Phase-contrast micrography of clone A2, a representative mock-transfected clone, shows elongated cells. Other mock-transfected clones also showed a similar morphology (data not shown). Lower left: Fluorescent staining of cells from clone C10. Actin filaments were detected with rhodamine-conjugated phalloidin, and cells seldom have filopodia (long, thin, needle-like projections protruding from the cell edge). Lower right: Fluorescent staining of F-actin in A2 cells shows numerous filopodia protruding from the cell surfaces. Scale bar = 10 μm. Panel B: CL1–0 cells were transiently transfected with the plasmid pEGFP-CRMP-1, carrying the fusion gene for EGFP-tagged CRMP-1, and the living cells were examined directly and photographed with a confocal microscope. Upper left: interphase cell with fluorescent CRMP-1 diffusely distributed in the cytoplasm. Upper middle: interphase cell with CRMP-1 localized in the nucleus and cytoplasm. Upper right: early prophase cell with CRMP-1 localized to the centrosomes. Lower left: metaphase cell with CRMP-1 associated with the mitotic spindle and centrosomes. Lower middle: anaphase cell with CRMP-1 associated with the mitotic spindle. Lower right: very late telophase cell with CRMP-1 localized at the midbody. Scale bar = 5 μm.

Fig. 3.

Kaplan–Meier survival plots for patients with non-small-cell lung cancer, grouped according to collapsin response mediator protein-1 (CRMP-1) messenger RNA (mRNA) expression. The relative amount of tissue CRMP-1 mRNA, standardized against the amount of TATA-box-binding protein mRNA, was expressed as –ΔCT = –[CT(CRMP–1)CT(TBP)], where CT is the threshold cycle. Patients were included in the high-expression group when the – CT was –1.95 or greater (the median). Panel A: The difference in disease-free survival between the high- and low-expression patients is statistically significant (P = .030). All patients free of recurrence at their last follow-up are indicated by tick marks on the plot. At year 1, for high-expression patients, the 95% confidence interval (CI) is 0.60 to 0.90 for 24 patients at risk; for low-expression patients, the 95% CI is 0.40 to 0.73 for 19 patients at risk. At year 3, for high-expression patients, the 95% CI is 0.30 to 0.66 for 12 patients at risk; for low-expression patients, the 95% CI is 0.11 to 0.42 for eight patients at risk. Panel B: Difference in overall survival between the high- and low-expression patients is statistically significant (P = .016). All patients alive at their last follow-up are indicated by tick marks on the plot. At year 1, for high-expression patients, the 95% CI is 0.85 to 1.0 for 30 patients at risk; for low-expression patients, the 95% CI is 0.63 to 0.91 for 27 patients at risk. At year 3, for high-expression patients, the 95% CI is 0.46 to 0.79 for 19 patients at risk; for low-expression patients, the 95% CI is 0.28 to 0.61 for 15 patients at risk. All statistical tests were two-sided.

Fig. 3.

Kaplan–Meier survival plots for patients with non-small-cell lung cancer, grouped according to collapsin response mediator protein-1 (CRMP-1) messenger RNA (mRNA) expression. The relative amount of tissue CRMP-1 mRNA, standardized against the amount of TATA-box-binding protein mRNA, was expressed as –ΔCT = –[CT(CRMP–1)CT(TBP)], where CT is the threshold cycle. Patients were included in the high-expression group when the – CT was –1.95 or greater (the median). Panel A: The difference in disease-free survival between the high- and low-expression patients is statistically significant (P = .030). All patients free of recurrence at their last follow-up are indicated by tick marks on the plot. At year 1, for high-expression patients, the 95% confidence interval (CI) is 0.60 to 0.90 for 24 patients at risk; for low-expression patients, the 95% CI is 0.40 to 0.73 for 19 patients at risk. At year 3, for high-expression patients, the 95% CI is 0.30 to 0.66 for 12 patients at risk; for low-expression patients, the 95% CI is 0.11 to 0.42 for eight patients at risk. Panel B: Difference in overall survival between the high- and low-expression patients is statistically significant (P = .016). All patients alive at their last follow-up are indicated by tick marks on the plot. At year 1, for high-expression patients, the 95% CI is 0.85 to 1.0 for 30 patients at risk; for low-expression patients, the 95% CI is 0.63 to 0.91 for 27 patients at risk. At year 3, for high-expression patients, the 95% CI is 0.46 to 0.79 for 19 patients at risk; for low-expression patients, the 95% CI is 0.28 to 0.61 for 15 patients at risk. All statistical tests were two-sided.

Supported by grants NSC 89–2314-B001–006-M39 and NHRI89A1-PPLABADO1 from the National Science Council, Taiwan.

We thank Yi-Wen Chu, Marine Kao (National Health Research Institutes, Taiwan), and Kuan-Yu Chu (Institute of Biomedical Sciences, Taiwan) for technical assistance and Wen-Yi Shau (National Taiwan University) for statistical analysis.

References

1
Yoshida BA, Sokoloff MM, Welch DR, Rinker-Schaefer CW. Metastasis-suppressor genes: a review and perspective.
J Natl Cancer Inst
 
2000
;
92
:
1717
–30.
2
Woodhouse EC, Chuaqui RF, Liotta LA. General mechanisms of metastasis.
Cancer
 
1997
;
80
(8 Suppl):
1529
–37.
3
Kohn EC. Development and prevention of metastasis.
Anticancer Res
 
1993
;
13
:
2553
–9.
4
Fidler IJ, Kripke ML. Metastasis results from preexisting variant cells within a malignant tumor.
Science
 
1977
;
197
:
893
–5.
5
Steeg PS, Bevilacqua G, Kopper L, Thorgeirsson UP, Talmadge JE, Liotta LA, et al. Evidence for a novel gene associated with low tumor metastatic potential.
J Natl Cancer Inst
 
1988
;
80
:
200
–4.
6
Dong JT, Lamb PW, Rinker-Schaeffer CW, Vukanovic J, Ichikawa T, Isaacs JT, et al. KAI1, a metastasis suppressor gene for prostate cancer on human chromosome 11p11.2.
Science
 
1995
;
268
:
884
–6.
7
Lee JH, Miele ME, Hicks DJ, Phillips KK, Trent JM, Weissman BE, et al. KiSS-1, a novel human malignant melanoma metastasis-suppressor gene [published erratum appears in J Natl Cancer Inst 1997;89:1549]. J Natl Cancer Inst 1996;88:1731–7.
8
Ross DT, Scherf U, Eisen MB, Perou CM, Rees C, Spellman P, et al. Systematic variation in gene expression patterns in human cancer cell lines.
Nat Genet
 
2000
;
24
:
227
–35.
9
Khan J, Saal LH, Bittner ML, Chen Y, Trent JM, Meltzer PS. Expression profiling in cancer using cDNA microarrays.
Electrophoresis
 
1999
;
20
:
223
–9.
10
Iyer VR, Eisen MB, Ross DT, Schuler G, Moore T, Lee JC, et al. The transcriptional program in the response of human fibroblasts to serum.
Science
 
1999
;
283
:
83
–7.
11
Hong TM, Yang PC, Peck K, Chen JJ, Yang SC, Chen YC, et al. Profiling the downstream genes of tumor suppressor PTEN in lung cancer cells by complementary DNA microarray.
Am J Respir Cell Mol Biol
 
2000
;
23
:
355
–63.
12
Clark EA, Golub TR, Lander ES, Hynes RO. Genomic analysis of metastasis reveals an essential role for RhoC.
Nature
 
2000
;
406
:
532
–5.
13
Chu YW, Yang PC, Yang SC, Shyu YC, Hendrix MJ, Wu R, et al. Selection of invasive and metastatic subpopulations from a human lung adenocarcinoma cell line.
Am J Respir Cell Mol Biol
 
1997
;
17
:
353
–60.
14
Chen JJ, Peck K, Hong TM, Yang SC, Sher YP, Shih JY, et al. Global analysis of gene expression in invasion by a lung cancer model.
Cancer Res
 
2001
;
61
:
5223
–30.
15
Chen JJ, Wu R, Yang PC, Huang JY, Sher YP, Han MH, et al. Profiling expression patterns and isolating differentially expressed genes by cDNA microarray system with colorimetry detection.
Genomics
 
1998
;
51
:
313
–24.
16
Tamayo P, Slonim D, Mesirov J, Zhu Q, Kitareewan S, Dmitrovsky E, et al. Interpreting patterns of gene expression with self-organizing maps: methods and application to hematopoietic differentiation.
Proc Natl Acad Sci U S A
 
1999
;
96
:
2907
–12.
17
Hamajima N, Matsuda K, Sakata S, Tamaki N, Sakata M, Nonaka M. A novel gene family defined by human dihydropyrimidinase and three related proteins with differential tissue distribution.
Gene
 
1996
;
180
:
157
–63.
18
Shan B, Zhu X, Chen PL, Durfee T, Yang Y, Sharp D, et al. Molecular cloning of cellular genes encoding retinoblastoma-associated proteins: identification of a gene with properties of the transcription factor E2F.
Mol Cell Biol
 
1992
;
12
:
5620
–31.
19
Denizot F, Lang R. Rapid colorimetric assay for cell growth and survival: modifications to the tetrazolium dye procedure giving improved sensitivity and reliability.
J Immunol Methods
 
1986
;
89
:
271
–7.
20
The World Health Organization histological typing of lung tumors. 2nd ed. Am J Clin Pathol 1982;77:123–36.
21
Mountain CF. Revisions in the International System for Staging Lung Cancer.
Chest
 
1997
;
111
:
1710
–7.
22
Chen CJ, Wu HY, Chuang YC, Chang AS, Luh KT, Chao HH, et al. Epidemiologic characteristics and multiple risk factors of lung cancer in Taiwan.
Anticancer Res
 
1990
:
10
:
971
–6.
23
Torres R, Polymeropoulos MH. Genomic organization and localization of the human CRMP-1 gene.
DNA Res
 
1998
;
5
:
393
–5.
24
Bieche I, Onody P, Laurendeau I, Olivi M, Vidaud D, Lidereau R, et al. Real-time reverse transcription–PCR assay for future management of ERBB2-based clinical applications.
Clin Chem
 
1999
;
45
(8 Pt 1):
1148
–56.
25
Yuan A, Yang PC, Yu CJ, Chen WJ, Lin FY, Kuo SH, et al. Interleukin-8 messenger ribonucleic acid expression correlates with tumor progression, tumor angiogenesis, patient survival, and timing of relapse in non-small-cell lung cancer.
Am J Respir Crit Care Med
 
2000
;
162
:
1957
–63.
26
Symons MH, Mitchison TJ. Control of actin polymerization in live and permeabilized fibroblasts.
J Cell Biol
 
1991
;
114
:
503
–13.
27
Cooper JA. The role of actin polymerization in cell motility.
Annu Rev Physiol
 
1991
;
53
:
585
–605.
28
Quinn CC, Gray GE, Hockfield S. A family of proteins implicated in axon guidance and outgrowth.
J Neurobiol
 
1999
;
41
:
158
–64.
29
Goshima Y, Nakamura F, Strittmatter P, Strittmatter SM. Collapsin-induced growth cone collapse mediated by an intracellular protein related to UNC-33.
Nature
 
1995
;
376
:
509
–14.
30
Fukada M, Watakabe I, Yuasa-Kawada J, Kawachi H, Kuroiwa A, Matsuda Y, et al. Molecular characterization of CRMP5, a novel member of the collapsin response mediator protein family.
J Biol Chem
 
2000
;
275
:
37957
–65.
31
Minturn JE, Fryer HJ, Geschwind DH, Hockfield S. TOAD-64, a gene expressed early in neuronal differentiation in the rat, is related to unc-33, a C. elegans gene involved in axon outgrowth.
J Neurosci
 
1995
;
15
:
6757
–66.
32
Wang LH, Strittmatter SM. A family of rat CRMP genes is differentially expressed in the nervous system.
J Neruosci
 
1996
;
16
:
6197
–207.
33
Byk T, Dobransky T, Cifuentes-Diaz C, Sobel A. Identification and molecular characterization of Unc-33-like phosphoprotein (Ulip), a putative mammalian homolog of the axonal guidance-associated unc-33 gene product.
J Neurosci
 
1996
;
16
:
688
–701.
34
Gaetano C, Matsuo T, Thiele CJ. Identification and characterization of a retinoic acid-regulated human homologue of the unc-33-like phosphoprotein (hUlip) from neuroblastoma cells.
J Biol Chem
 
1997
;
272
:
12195
–201.
35
Wang LH, Strittmatter SM. Brain CRMP forms heterotetramers similar to liver dihydropyrimidinase.
J Neurochem
 
1997
;
69
:
2261
–9.
36
Byk T, Ozon S, Sobel A. The Ulip family phosphoproteins—common and specific properties.
Eur J Biochem
 
1998
;
254
:
14
–24.
37
Quach TT, Mosinger B Jr, Ricard D, Copeland NG, Gilbert DJ, Jenkins NA, et al. Collapsin response mediator protein-3/unc-33-like protein-4 gene: organization, chromosomal mapping and expression in the developing mouse brain.
Gene
 
2000
;
242
:
175
–82.
38
Matsuo T, Stauffer JK, Walker RL, Meltzer P, Thiele CJ. Structure and promoter analysis of the human unc-33-like phosphoprotein gene: E-box required for maximal expression in neuroblastoma and myoblasts.
J Biol Chem
 
2000
;
275
:
16560
–8.
39
Inagaki H, Kato Y, Hamajima N, Nonaka M, Sasaki M, Eimoto T. Differential expression of dihydropyrimidinase-related protein genes in developing and adult enteric nervous system.
Histochem Cell Biol
 
2000
;
113
:
37
–41.
40
Gu Y, Hamajima N, Ihara Y. Neurofibrillary tangle-associated collapsin response mediator protein-2 (CRMP-2) is highly phosphorylated on Thr-509, Ser-518, and Ser-522.
Biochemistry
 
2000
;
39
:
4267
–75.
41
Yu Z, Kryzer TJ, Griesmann GE, Kim K, Benarroch EE, Lennon VA. CRMP-5 neuronal autoantibody: marker of lung cancer and thymoma-related autoimmunity.
Ann Neurol
 
2001
;
49
:
146
–54.
42
Honnorat J, Byk T, Kusters I, Aguera M, Ricard D, Rogemond V, et al. Ulip/CRMP proteins are recognized by autoantibodies in paraneoplastic neurological syndromes.
Eur J Neurosci
 
1999
;
11
:
4226
–32.
43
Raper JA. Semaphorins and their receptors in vertebrates and invertebrates.
Curr Opin Neurobiol
 
2000
;
10
:
88
–94.
44
Takahashi T, Fournier A, Nakamura F, Wang LH, Murakami Y, Kalb RG, et al. Plexin-neuropilin-1 complexes form functional semaphorin-3A receptors.
Cell
 
1999
;
99
:
59
–69.
45
Miao HQ, Soker S, Feiner L, Alonso JL, Raper JA, Klagsbrun M. Neuropilin-1 mediates collapsin-1/semaphorin III inhibition of endothelial cell motility: functional competition of collapsin-1 and vascular endothelial growth factor-165.
J Cell Biol
 
1999
;
146
:
233
–42.
46
Ito T, Kagoshima M, Sasaki Y, Li C, Udaka N, Kitsukawa T, et al. Repulsive axon guidance molecule Sema3A inhibits branching morphogenesis of fetal mouse lung.
Mech Dev
 
2000
;
97
:
35
–45.
47
Xiang RH, Hensel CH, Garcia DK, Carlson HC, Kok K, Daly MC, et al. Isolation of the human semaphorin III/F gene (SEMA3F) at chromosome 3p21, a region deleted in lung cancer.
Genomics
 
1996
;
32
:
39
–48.
48
Roche J, Boldog F, Robinson M, Robinson L, Varella-Garcia M, Swanton M, et al. Distinct 3p21.3 deletions in lung cancer and identification of a new human semaphorin.
Oncogene
 
1996
;
12
:
1289
–97.
49
Sekido Y, Bader S, Latif F, Chen JY, Duh FM, Wei MH, et al. Human semaphorins A(V) and IV reside in the 3p21.3 small cell lung cancer deletion region and demonstrate distinct expression patterns.
Proc Natl Acad Sci U S A
 
1996
;
93
:
4120
–5.
50
Brambilla E, Constantin B, Drabkin H, Roche J. Semaphorin SEMA3F localization in malignant human lung and cell lines: a suggested role in cell adhesion and cell migration.
Am J Pathol
 
2000
;
156
:
939
–50.
51
Martin-Satue M, Blanco J. Identification of semaphorin E gene expression in metastatic human lung adenocarcinoma cells by mRNA differential display.
J Surg Oncol
 
1999
;
72
:
18
–23.
52
Mitchison TJ, Cramer LP. Actin-based cell motility and cell locomotion.
Cell
 
1996
:
84
:
371
–9.
53
Albrecht-Buehler G. Filopodia of spreading 3T3 cells. Do they have a substrate-exploring function?
J Cell Biol
 
1976
;
69
:
275
–86.
54
Chien CB, Rosenthal DE, Harris WA, Holt CE. Navigational errors made by growth cones without filopodia in the embryonic Xenopus brain.
Neuron
 
1993
;
11
:
237
–51.
55
Gu Y, Ihara Y. Evidence that collapsin response mediator protein-2 is involved in the dynamics of microtubules.
J Biol Chem
 
2000
;
275
:
17917
–20.
56
Robbins J, Dilworth SM, Laskey RA, Dingwall C. Two interdependent basic domains in nucleoplasmin nuclear targeting sequence: identification of a class of bipartite nuclear targeting sequence.
Cell
 
1991
;
64
:
615
–23.
57
Morris VB, Brammall J, Noble J, Reddel R. p53 localizes to the centrosomes and spindles of mitotic cells in the embryonic chick epiblast, human cell lines, and a human primary culture: an immunofluorescence study.
Exp Cell Res
 
2000
;
256
:
122
–30.
58
Hsu LC, White RL. BRCA1 is associated with the centrosome during mitosis.
Proc Natl Acad Sci U S A
 
1998
;
95
:
12983
–8.