Down-Regulation of DUSP6 Expression in Lung Cancer —Its Mechanism and Potential Role in Carcinogenesis

Abstract

Our preliminary studies revealed that oncogenic KRAS (KRAS/V12) dramatically suppressed the growth of immortalized airway epithelial cells (NHBE-T, with viral antigen-inactivated p53 and RB proteins). This process appeared to be a novel event, different from the so-called premature senescence that is induced by either p53 or RB, suggesting the existence of a novel tumor suppressor that functions downstream of oncogenic KRAS. After a comprehensive search for genes whose expression levels were modulated by KRAS/V12, we focused on DUSP6, a pivotal negative feedback regulator of the RAS-ERK pathway. A dominant-negative DUSP6 mutant, however, failed to rescue KRAS/V12-induced growth suppression, but conferred a stronger anchorage-independent growth activity to the surviving subpopulation of cells generated from KRAS/V12-transduced NHBE-T. DUSP6 expression levels were found to be weaker in most lung cancer cell lines than in NHBE-T, and DUSP6 restoration suppressed cellular growth. In primary lung cancers, DUSP6 expression levels decreased as both growth activity and histological grade of the tumor increased. Loss of heterozygosity of the DUSP6 locus was found in 17.7% of cases and was associated with reduced expression levels. These results suggest that DUSP6 is a growth suppressor whose inactivation could promote the progression of lung cancer. We have here identified an important factor involved in carcinogenesis through a comprehensive search for downstream targets of oncogenic KRAS.

Oncogenic mutations of KRAS occur very early in carcinogenesis of the lung and affect even premalignant lesions.1,2 It is therefore likely that additional genetic and/or epigenetic alterations are necessary for lung neoplasms to develop into advanced form of cancers. The inactivation of putative tumor suppressors, p53 and p16INK4A/RB protein, and activation of telomerase, are generally accepted as crucial to the promotion carcinogenesis.1,3 A recent study, however, reported that these alterations are not enough to provide primary bronchial epithelial cells with a fully malignant phenotype in vitro, 4 and suggested further alterations would be necessary for the progression of lung cancer.

DUSP6, dual-specificity phosphatase 6, is a putative negative feedback regulator for the RAS-ERK pathway, playing physiologically important role(s) in the maintenance of cellular homeostasis in response to growth factors.5,6,7,8,9,10,11 Disruption of this feedback loop could therefore result in neoplastic, and even malignant, transformation by providing cells with enhanced growth activity. Actually, a series of previous studies reported the involvement of DUSP6 down-regulation caused by the hypermethylation of its promoter in the progression of pancreatic cancers.12,13,14,15,16 To our knowledge, however, there have been only a few reports investigating the significance of DUSP6 in other types of cancers including lung cancers.4,17

Our preliminary study revealed that oncogenic KRAS (KRAS/V12) dramatically suppressed growth of immortalized airway epithelial cells whose p53 and RB proteins were inactivated by viral antigens. This seemed a novel event different from so-called premature senescence that is induced by p53 or RB pathway,3 and suggested there might exist novel tumor suppressors downstream of oncogenic KRAS. Thus, we have been interested in identifying such suppressors. Also, our recent study demonstrated that insulin-like growth factor-binding proteins 2 and 4, downstream targets of oncogenic KRAS, suppress the growth of cancer cells in a negative-feedback manner and that disruption of this feedback loop, through their promoter’s hypermethylation, is possibly involved in the progression of lung cancers.18 It is worth noting that growth suppressors lie hidden downstream of the signal transduction pathway of an oncogene. Thus, we were prompted to search for downstream targets of oncogenic KRAS comprehensively, to further identify potential growth suppressors. Among genes whose expression was modulated by oncogenic KRAS, this study has focused on DUSP6, and elucidated its role in the KRAS-induced growth suppression and in carcinogenesis of the lung.

Materials and Methods

Cell Culture

An immortalized human airway epithelial cell line (16HBE14o, Simian virus 40-transformed human bronchial epithelial cells), described by Cozens et al,19 was kindly provided by Grunert DC (California Pacific Medical Center Research Institute) via Kaneko T (Yokohama City University School of Medicine). A subclone of 16HBE14o cells, described as NHBE-T in this paper, was used for experiments. Immortalized airway epithelial cell lines (HPL1D and HPL1A, Simian virus 40-transformed human small airway epithelial cells) were established by Masuda et al.20 Human lung cancer cell lines (A549, H358, and H1299) and the human embryonic kidney cell line (HEK293T) were purchased from American Type Culture Collection (ATCC, Manassas, VA). Human lung cancer cell lines, LC2/ad, Lu134A, Lu140, and PC9, were obtained form Riken Cell Bank (Tsukuba, Japan), and Immuno-Biological Laboratories Co. (Gunma, Japan), respectively. Human lung cancer cell lines, TKB1, TKB2, TKB4, TKB5, TKB6, TKB7, TKB8, TKB12, TKB15, TKB16, TKB17, and TKB20 were kindly provided by Kamma H (Kyorin University School of Medicine, Tokyo Japan). The cells were cultured and grown in DMEM (Sigma Aldrich, St. Louis, MO) (NHBE-T, HPL1D, HPL1A, A549, H358, PC9, LC2/ad, TKB1, TKB2, TKB4, TKB5, TKB6, TKB7, TKB8, TKB2, and H1299) or RPMI1640 medium (Sigma) (Lu130, Lu134A, Lu140, HEK293T, TKB12, TKB15, TKB16, and TKB17) supplemented with 10% heat-inactivated fetal bovine serum (Sigma), 100 units/ml of penicillin (Sigma), and 100 μg/ml of streptomycin (Sigma).

Cell Growth Assays

Cells (2.5 × 10⁵) were seeded onto a 10-cm culture dish (Iwaki, Tokyo, Japan), and grown to a semiconfluent state for 5 to 7 days. The cells were counted, and 2.5 × 10⁵ cells were seeded again onto a 10-cm dish. Several passages were repeated in the same manner. The sum of population doublings (PDLs) at each point, was calculated by the formula ΣPDL_n = log₂ (count_n/2.5 × 10⁵) + ΣPDL_n−1.

Colony Formation Assays

Cells (1.0 × 10⁴ or 5.0 × 10⁴) were seeded onto a 10-cm culture dish (Iwaki), and grown for optimal period (8 to 14 days). The cells were fixed with methanol and Giemsa-stained, and colonies visible in scanned photographs were counted.

Soft Agar Colony Formation Assays

Cells (1.25 × 10⁴) were grown in 1 ml of DMEM-based 0.3% agar (agar noble; Becton Dickinson, Sparks, MD) containing 10% fetal bovine serum in a 3.5-cm culture dish (Iwaki) for 5 weeks. The agar was fixed with a buffered 4% paraformaldehyde solution, and colonies visible in scanned photographs were counted.

Plasmid Construction

cDNA encoding wild-type KRAS (KRAS/G12) and mutated KRAS (KRAS/V12) were PCR-amplified (PrimeSATR HS DNA polymerase, Takara Bio Inc., Kyoto, Japan) with primers having BamH1 restriction sites (forward (F), 5′-CTCCGCGGATCCAAGCTTGCTGAAA-3′ and reverse (R), 5′-AGGGGCGGATCCTCATTACATAA-3′; the BamH1 restriction sites are in italics). Plasmid vectors, pcDNA3.1-KRAS/G12 and -KRAS/V12 described previously,2 were used as templates. The amplified fragments were digested with BamH1 (Takara), and inserted into pQCXIH retroviral provirus vectors (BD Clontech, Palo Alto, CA). cDNA encoding DUSP6 variant 2 (Gene Bank Accession # BC003143) was PCR-amplified (PrimeSATR HS DNA polymerase, Takara) with primers (F, 5′-GCTCGACCCCCATGATAGATACG-3′ and R, 5′-CATTCCAGCAAGGAGGGATGTGG-3′), using cDNA reverse-transcribed (SuperScript III, Invitrogen, Carlsbad, CA) from NHBE-T-derived polyadenylated RNA as a template. After the adenylation of 3′ terminals with TaqDNA polymerase (Takara), the amplified fragment was subcloned into the pT7Blue vector (Novagen, Darmstadt, Germany). The DUSP6 cDNA was PCR-amplified (PrimeSATR HS DNA polymerase, Takara) again using pT7Blue/DUSP6 as a template with mismatch primers having BamH1 restriction sites (F, 5′-GATTGCGGATCCGGCATGATAGATACGCTCAG-3′ and R, 5′-TCCAGCAAGGAGGGATCCGGGGTCTTTCAC-3′; BamH1 restriction sites are in italics). After digestion with BamH1 (Takara), the fragment was inserted into pQCXIP (BD Clontech). A dominant negative mutant of DUSP6 that had lost its phosphatase activity (DUSP6/C293S: Cysteine at position-293 was converted to Serine)8 was generated using a site-directed mutagenesis kit (Stratagene, La Jolla, CA) with mismatch primers (5′-GTCTTGGTACATAGCTTGGCTGGCATTAG-3′ and 5′-CTAATGCCAGCCAAGCTATGTACCAAGAC-3′: mismatch bases are in italics). A retroviral provirus vector (psiDKI, Takara) expressing siRNA targeting two positions of DUSP6 mRNA (5′-CCAACCAGAATGTATACCA-3′ [+1103 to +1121] and 5′-GAACTGTGGTGTCTTGGTA-3′ [+855 to +873]) and the same sized siRNA with scramble sequences were constructed according to the manufacturer’s instruction (Takara). Candidate promoter regions of −1539 to +24, −1030 to + 24, and −539 to + 24 at the DUSP6 gene locus (NC_000012) (position of the first base (A) of the start codon [ATG] was determined as + 1), were PCR-amplified (PrimeSATR HS DNA polymerase, Takara) with primer sets, F-1539, 5′-GGTCTTGCGGGAGGACTT-3′ and r + 24, 5′-CACGGGTCTGAGCTATCT-3′; F-1030, 5′-CAAACAGACCTGGGCCTTTA-3′ and r + 24, 5′-CACGGGTCTGAGCTATCT-3′; F-511, 5′-GAGACGCTCGCTGTTTGTATC-3′ and r + 24, 5′-CACGGGTCTGAGCTATCT-3′, using genomic DNA extracted from non-tumor parts of surgically resected lung as a template. After adenylation of the 3′ terminals with TaqDNA polymerase (Takara), the products were subcloned into pT7Blue (Novagen). They were then cut out with HindIII (Takara) and KpnI (Takara), and inserted into pGL4.1 (Promega, Madison, WI). The accuracy of all of the constructs was verified by DNA sequencing (Dye-deoxy DNA sequencing kit, Amersham Life Science, Piscataway, NJ).

Retroviral-Mediated Gene Transfer

The pQCXIP(H)-based expression vectors and pDKsi based-knockdown vectors designed as described above and the pCL10A1 retrovirus packaging vector (Imgenex, San Diego, CA) were cotransfected into HEK293T cells with Lipofectamine 2000 reagent (Invitrogen). At 24 hours after the transfection, conditioned medium was recovered as a viral solution. Desired genes were introduced by incubating cells with the viral solution containing 10 μg/ml of polybren (Sigma) for 24 hours. Cells stably expressing desired genes were selected with 1.0 μg/ml of puromycin (or 500 μg/ml hygromycin B or 1000 μg/ml of neomycin [G418]) (Invitrogen) for 3 days. The pooled clones were used for biological analyses and expression profiling.

Expression Profiling

Gene expression in the empty vector-, KRAS/G12-, and KRAS/V12-transduced NHBE-T cells was comprehensively evaluated with an U133A human gene chip microarray (Affymetrix, San Diego, CA) according to the manufacturer’s recommendations. In brief, total RNA was extracted from the cells immediately after completion of the selection (5 days post-transduction), and was reverse transcribed with Super Script II reverse transcriptase (Invitrogen). Complementary RNA was transcribed in vitro using T7 RNA polymerase (Invitrogen) and concomitantly labeled with a fluorescent dye. The labeled complementary RNA was fragmented by acetic acid treatment, and hybridized to the cDNA array. The hybridized signals were scanned with a Gene Chip Scanner 3000 (Affymetrix), and the signal values of each probe were globally normalized according to manufacture’s recommendation. It was checked that differences in signal values of house keeping genes including β-actin (ACTB) and glyceraldehyde-3-phosphate dehydrogenase among the transfectants were no more than 1.5-fold. Probes showing an absence call were excluded. Then, a hierarchical clustering analysis (Ward’s Method) was performed with computer software (GeneSpring, Agilent technology, Palo Alto, CA). Probes (= genes) whose signal values were elevated or reduced more than fivefold by the KRAS/V12 transduction in comparison with both the empty vector transduction and the KRAS/G12 transduction were extracted by using GeneSpring (Agilent Technology).

Quantitative Reverse Transcription-PCR

First-strand cDNA was synthesized from total RNA using the SuperScript First-Strand Synthesis System according to the protocols of the manufacturer (Invitrogen). The cDNA generated was used as a template in real-time PCR with SYBR Premix EXTaq (Takara) and run on a Thermal Cycler DICE real-time PCR system (Takara). The primer set used for the detection of DUSP6 was F, 5′-AACAGGGTTCCAGCACAGCAG-3′; R, 5′-GGCCAGACACATTCCAGCAA-3′. That for ACTB was F, 5′-TGGCACCCAGCACAATGAA-3′; R, 5′-CTAAGTCATAGTCCGCCTAGAAGCA-3′. The means and standard deviations of the copy number of DUSP6 normalized to the value for ACTB mRNA were statistically obtained from triplicate reactions.

Western Blotting

The cells grown to subconfluence were solved with extraction buffer, as described elsewhere.2 After centrifugation, supernatants were recovered as protein extracts. The extracts were mixed with equal volumes of 2×SDS buffer, and then boiled. The samples were subjected to SDS-polyacrylamide gel electrophoresis, and transferred onto polyvinylidene difluoride membranes (Amersham). The membranes were incubated with nonfat dry milk in 0.01 M/L Tris-buffered saline containing 0.1% Tween-20 to block non-immunospecific protein binding, and then with 0.1 μg/ml of primary antibody against either KRAS (Santa Cruz, Santa Cruz, CA), DUSP6 (Abnova), ERK (Cell Signaling Technology, Beverly, MA), phosphorylated ERK (Cell Signaling Technology), or β-actin (Sigma). After washing with 0.01 M/L Tris-buffered saline containing 0.1% Tween-20, the membranes were incubated with animal-matched horseradish peroxidase-conjugated secondary antibodies (Amersham). Immunoreactivity was visualized with the enhanced chemiluminescence system (ECL, Amersham).

Treatment with 5-Azacytidine and Trichostatin A

Cells were treated either with 10 μmol/L of 5-azacytidine (AZA, Sigma) for 72 hours by exchanging the medium everyday or with 300 ng/ml of trichostatin A (TSA, Wako, Osaka, Japan) for 24 hours. In addition, cells were treated also with AZA for 48 hours and then with a combination of AZA and TSA for an additional 24 hours.

Bisulfate Conversion-Based DNA Sequencing

Genomic DNA was extracted from cultured cells as described previously,2 and subjected to bisulfate conversion treatment using a MethylEasy DNA bisulfate modification kit (Human Genetic Signatures, Macquarie Park, Australia) according to the manufacturer’s instructions. The region from −960 to −266 of the DUSP6 gene, including the promoter locus, was PCR-amplified (TaqHS DNA polymerase, Takara) with three sets of primers (F1, 5′-GAGTTGGGTTTTTAAAGTGGTAAATA-3′ and R1, 5′-CAAAACACATAAACCAAAACACTTC-3′; F2, 5′-GAAATATGAGATAATTGAAGTGTTTTGG-3′ and R2, 5′-CAACTCCTCAATAAATACAAACAAC-3′; F3, 5′-TGTTTGTATTTATTGAGGAGTTGTTT-3′ and R3, 5′-CTACCCAAATAATTTTTATTCCTCC-3′), and subcloned into pT7Blue (Novagen). Eight clones were randomly chosen, and nucleotide conversions at CpG sites were searched for, by DNA sequencing (Dye-deoxy DNA sequencing kit, Amersham).

Bisulfate Conversion-Based Methylation-Specific PCR

The region from +544 to +627 in intron 1 of the DUPS6 gene, reported to be highly methylated in pancreatic cancers,16 was PCR-amplified (TaqHS DNA polymerase, Takara) with primers specific for either methyl (F, 5′-GTAGGGGTCGCGAATCGCGC-3′ and R, 3′-ACCGCCGTAACCCGCAACCG-3′) or un-methyl DNA (F, 5′-GTAGGGGTTGTGAATTGTGT-3′ and R, 5′-AACCACCAATACCCACAACCA-3′)16 using bisulfate-converted DNA as a template.

Luciferase Reporter Assays

The pGL4.1-based vectors bearing the 5′-untranslated region of the DUSP gene (described in the section Plasmid Construction) and pGL4.7-TK Renilla (Promega, Madison) were co-transfected into NHBE-T cells with Lipofectamine 2000 (Invitrogen). After 24 hours, cells were solved, and luminosity was measured using a Dual Luciferase Reporter Assay system (Promega) on a luminometer (TD-20/20, Turner BioSystems, Sunnyvale, CA). The luminosity derived from each pGL4.1-based vector was corrected with that from pGL4.7-TK Renilla, to normalize the transfection efficiency. Luminosity relative to that of an empty pGL4.1 vector was determined as promoter activity.

Bromodeoxyuridine Pulse Labeling

Cells were incubated in the ordinal medium (DMEM with 10% fetal bovine serum) containing 10 μmol/L of bromodeoxyuridine (BrdU, Sigma) for 45 minutes, thereafter fixed with 4% paraformaldehyde, and then treated with 1N HCl to denature DNA. Incorporated BrdU was visualized by immunocytochemistry with a specific antibody against BrdU (Becton Dickinson, San Jose, CA) using an EnVSION detection system (DakoCytomation, Dako, Ely, UK). A thousand or more cell nuclei were counted, and the percentage of positively labeled cells was determined as a pulse-labeling index.

Flow Cytometry

Cells were harvested by trypsinization, fixed with cold 70% ethanol, and treated with 10 mg/ml of RNase A (Sigma). The cells were suspended in 50 μg/ml of propidium iodide. DNA content was measured with a flow cytometer (Epics Elite; Beckman Coulter, Fullerton, CA).

Primary Lung Cancers

All cases examined were of lung cancer patients who underwent surgical resection at Kanagawa Cardiovascular and Respiratory Disease Center Hospital between 2001 and 2008. Informed consent for research use was obtained from all of the subjects providing materials. One hundred and seventy-two cases (adenocarcinomas [ADC, 107], squamous cell carcinomas [SQC, 51], and large cell carcinomas [LCC, 14]), were examined for KRAS codon 12 mutations by restriction endonuclease-mediated selective PCR21 as described below. Twenty-two cases (ADC,18 LCC,3 and SQC1) were found to have a mutated KRAS (12.8%). The cases with KRAS mutations and 74 cases of non-small cell lung cancer without a KRAS mutation, a total of 96 cases, were used for further analyses. ADCs were classified into histological subtypes according to the criteria of the World Health Organization.22 If the case involved a mixed type, the major component was described as the subtype in the present study.

REMS-PCR Analysis for Detection of KRAS Codon 12 Mutations

Genomic DNA was extracted from fresh frozen lung cancer tissues or paraffin sections by methods described previously.2,21 For detection of KRAS codon 12 mutations, the RESMS PCR described by Ward et al was partially modified. Briefly, the reaction mixture, a final volume of 5 μl, containing 0.5 μl of 10 ng/μl DNA solution (1 ng/μl), 0.5 μl of 10×PCR buffer (1×), 0.4 μl of each 2.5 mmol/L dNTP (0.2 mmol/L), 0.04 μl of 50 μmol/L primes (three primer sets described by Ward et al, each 0.4 μmol/L), 0.025 μl of TaqHS DNA polymerase (Takara), 0.4 μl of BstN1 (New England BioLabs, Northbrook, IL) (0.8 unit/μl), and 2.935 μl of distilled water (final volume, 5 μl), was PCR-amplified on a Thermal Cycler Dice (Takara) with the following program; 94°C for 2 minutes (one cycle), 92°C for 20 seconds, 60°C for 3 minutes (35 cycles), held at 4°C. The product was resolved by electrophoresis in 5% agar. A representative result is shown in supplementary Figure S1 (see supplemental Figure S1 at http://ajp.amjpathol.org).

Immunohistochemistry

Maximal tumor sections were subjected to immunohistochemistry. Formalin-fixed tissue sections (4 μm thick) were deparaffinized and rehydrated, and incubated with 3% hydrogen peroxide, followed by 5% goat serum to block endogenous peroxidase activities and non-immunospecific protein binding. Antigen retrieval treatment, boiling in citrated buffer (0.01 M/L, pH 6.0), was used for the detection of Ki-67 antigen to restore the masked epitope. Then the sections were incubated with primary antibody against either DUSP6 (Abnova, Taipei, Taiwan) or Ki-67 (MIB1, Dako). Immunoreactivity was visualized with an Envision detection system (Dako), and the nuclei were counterstained with hematoxylin. The specificity of the anti-DUSP6 antibody was confirmed by an experiment with sections cut from paraffin-embedded HEK293T cells that were transfected with the pQCXIP/DUSP6 expression vectors (see supplemental Figure S2 at http://ajp.amjpathol.org). Expression levels of DUSP6 in lung cancer cells were subdivided into four categories, negative (0), faint (1), moderate (2), and strong (3). Faint expression was defined as a level similar to that in bronchial epithelial cells. Strong expression was defined as unequivocally greater signal intensity than that of bronchial epithelial cells, and moderate expression was defined as a level in between faint and strong expression. The expression score was determined as the average level. Lung cancers with a score of less than 1, and equal to or more than 1, were classified as reduced expressers and retained expressers, respectively. A labeling index of MIB1 was calculated as a proportion of positive cells by counting 500 to 1000 cancer cells. The lung cancers with a labeling index of less than 0.15 and with 0.15 or more were classified as low and high expressers, respectively.

Loss of Heterozygosity Analysis

H&E-stained tissue sections were checked microscopically. If a tumor was medullary, the tumorous and non-tumorous parts were macroscopically dissected with a razor blade. If a tumor contained abundant interstitial tissue, the tumorous part was captured using the PALM-MCB laser microdissection system (Carl Zeiss, Jena, Germany). The method of DNA extraction from formalin-fixed, paraffin-embedded tissue sections was previously described elsewhere.2 Loss of heterozygosity (LOH) was evaluated by restriction fragment length polymorphism analysis. Briefly, a 168-bp fragment covering a polymorphic region of exon 1 (340 (t/g); 114 (leucine/valine) was PCR-amplified using 5′-GCGACTGGAACGAGAATACGGG-3′ and 3′- TTTCAATCCACGCGCCCCG-5′, with HotStar Taq polymerase (Stratagene, La Jolla, CA) according to the manufacturer’s instruction. Subsequently, the PCR product was digested with BsiHAKI (New England BioLabs), and solved by electrophoresis in 5% agar. In this assay system, only the Leucine allele, but not the Valine allele, is specifically cut into two fragments of 135 and 33 bp.

Search for Mutations

Cells were harvested by trypsinization, and subsequently subjected to DNA extraction as described elsewhere.2 H&E-stained tissue sections (acetone- or ethanol-fixed, paraffin-embedded tissue) were checked microscopically. If a tumor was medullary, the tumorous and the non-tumorous parts were macroscopically dissected with a razor blade. If a tumor contained abundant interstitial tissue, the tumorous part was captured using the PALM-MCB laser microdissection system (Carl Zeiss, Jena, Germany). The method of DNA extraction from paraffin-embedded tissue section was previously described elsewhere.2 A fragment of the coding exons (exons 1 to 3) was PCR-amplified using HotStar-TaqDNA polymerase (Stratagene) according to the manufacturer’s instruction. PCR products were purified with EXO-SAP-IT (Amersham), and subjected to a dye-terminator reaction using a Big Dye version 3.1 kit (Applied BioSystems, Foster City, CA). Primers used for PCR and DNA sequencing were as described previously.12

Statistical Analysis

Differences in cell cycle fractions, DUSP6 expression scores or allelic status among clinicopathologic subgroups were analyzed with Student’s t-test or the one-way analysis of variance (Game-Howell’s test or Tukey’s HSD test). The correlation of DUSP6 expression levels with clinicopathologic parameters, KRAS gene mutations, MIB index levels, and allelic status was analyzed with Fisher’s exact test. P values less than 0.05 were considered significant.

Results

Comprehensive Search for Downstream Targets of Oncogenic KRAS

Forced expression of oncogenic KRAS (KRAS/V12) dramatically suppressed the growth of NHBE-T cells (Figure 1, A–D) with flattened and vacuolated morphological changes (Figure 1E). The results led us to expect potential growth suppressor(s) downstream of oncogenic KRAS. A comprehensive evaluation of gene expression revealed the KRAS/V12-transduced cells to have a different expression profile from empty vector (mock)- and KRAS/G12-transduced cells, as the KRAS/V12-transduced cells were classified into the most distant branch on a dendrogram described from a hierarchical clustering analysis (see supplemental Figure S3 at http://ajp.amjpathol.org). To identify the factors determining such a difference, we extracted sets of genes up- and down-regulated genes more than fivefold in the KRAS/V12-transduced cells compared with the mock-transduced cells (Table 1 and 2). Among up-regulated genes, DUSP6, a putative negative feedback regulator for the KRAS-ERK pathway, was detected with two independent probes and showed more than a ninefold difference in both. We therefore focused on DUSP6 as a potential suppressor involved in the oncogenic KRAS-induced growth suppression or the carcinogenesis of the lung.

Figure 1.

Biological effects of oncogenic KRAS on NHBE-T cells. Empty vector (pQCXIH; MOCK), KRAS/G12 (G12), or KRAS/V12 (V12) was retrovirally transduced into NHBE-T cells. After a 4-day selection process, the surviving populations were harvested and counted, and 1.0 × 10⁴ cells were reseeded onto a 10-cm dish. After 14 days, the cells were fixed with methanol and stained with Giemsa (A). The means and standard deviations (error bars) of colony counts from triplicate experiments are presented (B). The cells harvested immediately after the selection process were examined for the expression of KRAS and ACTB by Western blotting (C). The selected cells were grown and passed several times. The means and standard deviations (error bars) of cumulated PDLs from triplicate experiments are presented (D). Cells (2.5 × 10⁴) after the selection process were re-seeded onto chamber slides (LAB-TEK, Electron Microscopy Science; Hatfield, PA) cultured for 48 hours, then fixed with 90% ethanol, and stained with the Papanicolaou method (E; ×400).

Table 1.

Up-Regulated Genes

Common	Genbank	Map	Mock		KRAS/G12		KRAS/V12		Ratio V12/MOCK	Ratio V12/G12
			Signal value	Flags	Signal value	Flags	Signal value	Flags
PPBP	R64130	4q12-q13	0.06	A	0.10570161	A	24.379782	P	406.3297	230.6472153
NAV3	NM_014903	12q14.3	0.17999999	A	0.30749556	A	32.909496	P	182.8305435	107.0242965
IL1RL1	NM_003856	2q12	0.12306067	A	1	A	17.37906	P	141.2235119	17.37906
ESM1	NM_007036	5q11.2	0.17	A	0.4035879	A	22.06809	P	129.8122941	54.67976121
SERPINB2	NM_002575	18q21.3	0.35317335	P	0.99999994	P	36.968746	P	104.6759219	36.96874822
CCL3	NM_002983	17q11-q21	0.41999996	A	0.93209594	A	43.92214	P	104.5765338	47.12190893
MMP1	NM_002421	11q22.3	0.8234009	P	1	P	82.57412	P	100.2842236	82.57412
NICE-1	NM_019060	1q21	0.95156735	A	1	A	82.501305	P	86.70043692	82.501305
IL24	NM_006850	1q32	0.20504226	P	0.99999994	P	14.915695	P	72.74449179	14.91569589
IL1B	NM_000576	2q14	0.58934176	P	1	P	32.343746	P	54.88113722	32.343746
PI3	NM_002638	20q12-q13	0.27	A	0.32671407	A	14.2768345	P	52.87716481	43.69825426
IL1B	M15330	2q14	0.64027035	P	1	P	23.309114	P	36.40511231	23.309114
S100A7	NM_002963	1q21	0.29999998	A	0.62460035	A	9.610644	P	32.03548214	15.38686938
HSD17B2	NM_002153	16q24.1-q24.2	0.16	A	0.23062168	A	5.040771	P	31.50481875	21.85731628
FOS	BC004490	14q24.3	0.48000002	A	0.9513144	A	11.526349	P	24.01322608	12.11623518
STEAP	NM_012449	7q21	0.43252343	P	1	P	7.8730335	P	18.20255957	7.8730335
DUSP6	BC003143	12q22-q23	0.8425659	P	1	P	12.893464	P	15.30261787	12.893464
EGR3	NM_004430	8p23-p21	0.49825794	A	1	A	6.581242	P	13.20850401	6.581242
MMP3	NM_002422	11q22.3	0.8211574	P	1	P	10.295607	P	12.53792148	10.295607
SCG2	NM_003469	2q35-q36	0.607316	P	1	P	7.09176	P	11.67721582	7.09176
KRTHA4	NM_021013	17q12-q21	0.6581211	P	1	P	7.3771305	P	11.20938153	7.3771305
DUSP6	BC003143	12q22-q23	0.9104498	P	1	P	9.899106	P	10.87276421	9.899106
CSF2	M11734	5q31.1	0.82838905	P	1	P	8.029184	P	9.692527925	8.029184
CXCL2	M57731	4q21	0.7698449	P	1	P	6.89791	P	8.960129501	6.89791
EGR1	NM_001964	5q31.1	1.7957487	P	1.3774424	P	12.828883	P	7.144030231	9.313553147
MME	NM_007287	3q25.1-q25.2	0.83048177	P	1	P	5.7785997	P	6.958129496	5.7785997
GEM	NM_005261	8q13-q21	0.5474987	A	0.99999994	P	3.7885349	P	6.919714878	3.788535127
DOCK4	NM_014705	7q31.1	0.9751374	P	1	P	6.589132	P	6.757131867	6.589132
BMP2	AA583044	20p12	0.77180034	A	1	A	5.088037	P	6.592426482	5.088037

Mock, empty vector-transduced; KRAS/G12-transduced; KRAS/V12-transduced NHBE-T cells; Common, common gene name; Map, Chromosomal locus; Genbank, gene bank accession number. Flags indicate whether gene expression is present (P) or absent (A). 

Table 2.

Down-Regulated Genes

Common	Genbank	Map	Mock		KRAS/G12		KRAS/V12		Ratio V12/MOCK	Ratio V12/G12
			Signal value	Flags	Signal value	Flags	Signal value	Flags
CMKOR1	AI817041	2q37.3	2.897788	P	1	A	0.13582303	A	0.046871279	0.13582303
STS	NM_000351	xp22.32	3.4030955	P	1	A	0.19202565	A	0.056426759	0.19202565
CCL2	S69738	17q11.2-q21.1	2.5081918	P	1	P	0.14537899	P	0.057961672	0.14537899
SCAND2	NM_022050	15q25-q26	0.38724586	P	0.18206042	A	0.023738872	A	0.06130181	0.130390076
SCAND2	NM_022050	15q25-q26	2.1270185	P	1	A	0.13039011	A	0.061301822	0.13039011
PDGFRL	NM_006207	8p22-p21.3	1.8670229	P	1	P	0.12245405	A	0.065587867	0.12245405
COL5A1	AI130969	9q34.2-q34.3	1.3539611	P	0.99999994	P	0.09460237	P	0.069870818	0.094602376
ITPR2	AI963873	12p11	4.62	P	0.7495205	A	0.35317504	A	0.076444814	0.471201308
TSPAN-2	BF129969	1p13.1	1.825053	P	1	P	0.14110127	P	0.077313519	0.14110127
LASS4	NM_024552	19p13.3	0.97412205	P	0.7371226	A	0.07814046	A	0.080216293	0.106007413
EPHA4	NM_004438	2q36.1	3.301421	P	1	A	0.2928391	A	0.088700926	0.2928391
NEDD9	AL136139	6p25-p24	1.4222593	P	1	P	0.1354199	P	0.095214635	0.1354199
NEDD9	U64317	6p25-p24	2.1407213	P	1	P	0.20491455	A	0.095722199	0.20491455
METAP2	NM_006838	12q23.1	1.5916058	P	0.99999994	P	0.15723512	A	0.098790241	0.157235129
HOXC8	NM_022658	12q13.3	0.4796673	P	0.46980467	A	0.04846686	A	0.10104266	0.103163853
SGEF	AU118383	3q25.2	2.7925768	P	1	A	0.28845292	A	0.103292744	0.28845292
MEST	NM_002402	7q32	1.0919998	P	1	P	0.12080742	P	0.110629526	0.12080742
	M78162		2.2374306	P	1	A	0.25059348	A	0.11200056	0.25059348
RHOBTB3	N21138	5q15	1.592752	P	1	P	0.17884138	P	0.112284511	0.17884138
MARCKS	AW163148	6q22.2	1.7614398	P	1	P	0.19830465	A	0.112580998	0.19830465
PAPSS2	AW299958	10q23-q24	1.7204325	P	1	P	0.20847437	P	0.121175559	0.20847437
OLR1	AF035776	12p13.2-p12.3	1.2022126	P	0.99999994	P	0.15065517	P	0.125314915	0.150655179
GALNT4	NM_003774	12q21.3-q22	4.326977	P	0.99999994	A	0.5778095	A	0.133536531	0.577809535
FBLN1	Z95331	22q13.31	1.1362673	P	1	P	0.16118747	P	0.141857	0.16118747
FGL2	NM_006682	7q11.23	2.5700002	P	0.55733573	A	0.3745796	A	0.145750806	0.672089694
NR2F1	AI951185	5q14	1.3209105	P	1	P	0.19696297	P	0.149111518	0.19696297
NR2F1	AI951185	5q14	5.6848426	P	4.3037305	P	0.84767556	P	0.149111527	0.196962974
SYCP1	NM_003176	1p13-p12	2.3125896	P	0.99999994	A	0.3712496	A	0.160534148	0.371249622
TNFSF10	AW474434	3q26	2.61	P	0.6342096	A	0.43879327	A	0.168120027	0.691874216
SCNN1A	NM_001038	12p13	2.845504	P	1	P	0.48475575	A	0.170358485	0.48475575

Mock, empty vector-transduced; KRAS/G12-transduced; KRAS/V12-transduced NHBE-T cells; Common, common gene name; Map, Chromosomal locus; Genbank, gene bank accession number. Flags indicate whether gene expression is present (P) or absent (A). 

DUSP6 Expression in Immortalized Cells and Cancer Cells

The KRAS/V12-transduced NHBE-T cells showed higher levels of DUSP6 mRNA and protein than either the mock- or KRAS/G12-transduced cells (Figure 2, A–B), confirming the results of the microarray analyses (Table 1 and 2). On the other hand, the lung cancer cells, TKB14, TKB1, TKB2, and TKB8, showed considerably low mRNA levels, less than 0.3-fold that in NHBE-T (Figure 2A; see also supplemental Figure S4 at http://ajp.amjpathol.org). Despite having KRAS mutations, A549 and H358 also showed lower mRNA levels (0.15- and 0.7-fold, respectively) than the KRAS/V12-transduced cells (Figure 2A). The protein levels were lower in all lung cancer cell lines examined, except for TKB2, than in NHBE-T (Figure 2C). Interestingly, the levels of mRNA and protein were not parallel, as TKB20 and H1299 cells expressing very high levels of DUSP6 mRNA (about 39- and 13-fold in comparison with NHBE-T, respectively) showed very low protein levels (0.05-fold in TKB20 in comparison with NHBE-T, and undetectable in H1299) (Figure 2, A and C). These findings suggested translational and/or post-translational regulation23,24 to be one of the mechanisms causing the reduction of DUSP6 protein expression in lung cancer cells.

Figure 2.

Expression of DUSP6 in immortalized and cancer cell lines. Copy numbers of DUSP6 and ACTB mRNA were measured by quantitative reverse transcription-PCR. The means and standard deviations (error bars) of the mRNA expression of DUSP6 normalized to that of ACTB obtained from triplicate experiments are presented (A). DUSP6 and ACTB protein expression in immortalized (B, upper panel) and lung cancer cells (C, upper panel) was analyzed by Western blotting. Representative results are shown. Levels of DUSP6 and ACTB protein were semiquantified with a densitometer (NIH Image; National Institute of Mental Health at Bethesda, MD). The protein expression of DUSP6 was normalized to that of ACTB. The expression level relative to that in MOCK (B, lower panel) to NHBE-T (C, lower panel) cells is presented. IMC, immortalized airway cells; MOCK, empty vector-transduced NHBE-T; G12, KRAS/G12-transduced NHBE-T; V12, KRAS/V12-transduced NHBE-T; adenocarcinoma, ADC; squamous cell carcinoma, SQC; large cell carcinoma, LCC; small cell carcinoma, SCC.

Effect of DUSP6 on Growth Activity of Immortalized and Cancer Cells

Forced expression of wild-type DUSP6, but not a dominant negative mutant (C293S),8 markedly suppressed the growth of NHBE-T cells (Figure 3, A–D). A serial transduction of KRAS/V12 markedly suppressed growth also of the C293S-transudced NHBE-T (Figure 3, E and F), although the C293S-transduced cells showed about a 1.5-fold increase in colony formation compared with the mock-transfected cells (Figure 3, E and F). The level of restoration of colony formation by a dominant negative suppression on DUSP6 activity was too small to consider that DUSP6 is an essential factor to induce the growth suppression as downstream target of oncogenic KRAS. Interestingly, KRAS/V12 activated the ERK pathway, as well as suppressed growth, and C293S further activated ERK in a dominant negative manner (Figure 3G), but did not enhance the KRAS-induced growth suppression. Therefore, it was speculated that oncogenic KRAS could modulate targets other than the ERK1/2 pathway for the growth suppression. On the other hand, the surviving subpopulations generated from the KRAS/V12-transduced cells (MOCK/V12 in Figure 3H) were no more affected with the growth suppression, suggesting that the surviving subpopulation potentially having some (epi-) genetic alterations, which made the cells resistant to the KRAS/V12-induced suppression, was selected. Moreover, the surviving subpopulations generated from the C293/V12-transduced cells exhibited gain of anchorage-independent activity in soft agar (Figure 3, I and J; see also supplemental Figure S5 at http://ajp.amjpathol.org), although no enhancement in growth activity was observed under anchorage-supplied culture conditions (on the ordinary poly-l-lysine-coated plastic dish) (Figure 3H). The result indicated that DUSP6’s inactivation further conferred a growth advantage to the subpopulation surviving from the oncogenic KRAS-induced growth suppression under the specific condition, implying that DUSP6’s inactivation might be involved especially in the progression process of KRAS-mediated carcinogenesis. Actually, forced expression of DUSP6 markedly suppressed the growth of A549, a lung cancer cell line bearing a KRAS mutation (G12S) (Figure 4, A–D), and significantly attenuated the incorporation of BrdU (Figure 4E, upper panel; Table 3). The result of the flow cytometric analysis that the fractions at S phase decreased and those at G1 phase increased confirmed the former observation (Figure 4E, lower panel; Table 3). No sub-G1 fraction, which indicates the induction of apoptosis, was observed (Figure 4E, lower panel; Table 3). Similar results were obtained from experiments with H1299 and TKB1, a cell line without KRAS mutations (see supplemental Figure S6 at http://ajp.amjpathol.org). Taken together, these results suggested that DUSP6 could function as a growth suppressor by preventing cell cycle progression and implied that it could prevent carcinogenesis of the lung with or without KRAS mutations. However, exogenously transduced genes often induce artificial phenotypes due to the excessive amount not being physiologically relevant. So, we checked levels of expression of exogenously transduced DUSP6 in the cancer cell lines and compared these with a level of endogenous expression in NHBE-T. The levels of exogenous expressions in the cancer cell lines were higher than that of endogenous expression in NHBE-T, but seemed not far excessive (see supplemental Figure S7 at http://ajp.amjpathol.org), although it may be impossible to exactly judge whether such levels of the exogenous expression were physiologically relevant or not. It is a limitation of the transduction experiments. We, therefore, performed a knockdown experiment on NHBE-T and A549 cells to further verify a potential significance of the DUSP6’s down-regulation in the KRAS/V12-induced growth suppression and in the biological properties of cancer cells using a stable siRNA expression system. However, we did not obtain the expected result in its effect of on KRAS/V12-induced growth suppression, and growth activity of NHBE-T in ordinary culture and in soft agar (see supplemental Figure S8 at http://ajp.amjpathol.org). Also, we failed to find any significant change in growth activity of A549 cells in ordinary culture and in soft agar (data not shown). This inconsistency was speculated to be due to a level of reduction too small to exhibit significant change in biological activity, as an efficiency of the knockdown seemed not enough to overcome the KRAS-mediated up-regulation of DUSP6 expression (see supplemental Figure S8 at http://ajp.amjpathol.org). To verify more satisfactorily significances of DUSP6 in the oncogenic KRAS-induced growth suppression and in the biological properties of cancer cells, further improvement of experimental system might be required.

Figure 3.

Biological effects of DUSP6 on NHBE-T cells. Empty vector (pQCXIP; MOCK), DUSP6, or DUSP6/C293S (C293S)8 was retrovirally transduced into NHBE-T cells. Three days after selection with puromycin, the surviving cells were harvested and counted, and 1.0 × 10⁴ cells were re-seeded onto a 10-cm dish. After 14 days, the cells were fixed with methanol and stained with Giemsa (A). The means and standard deviations (error bars) of colony counts from triplicate experiments are presented (B). The cells harvested immediately after the selection process were examined for expression of DUSP6 and ACTB by Western blotting (C). The cells after the selection process were grown and passed several times. The means and standard deviations (error bars) of cumulated PDLs from triplicate experiments are presented (D). The puromycin-selected MOCK- and DUSP6/C293 (C293)-transduced cells were serially transduced with KRAS/V12 (V12). Three days after selection with hygromycin B, the cells were harvested and counted, and 5.0 × 10⁴ cells were re-seeded onto a 10-cm dish. After 14 days, the cells were fixed with methanol and stained with Giemsa (E). The means and standard deviations (error bars) of counts of colonies visible in scanned photographs from triplicate experiments are presented (F). The subpopulations generated from the MOCK/V12- or C293S/V12- serially transduced NHBE-T cells and the empty pQCXIP/pQCXIH-transduced NHBE-T cells (MOCK/MOCK) were examined for the expression of phosphorylated ERK1/2 (pERK1/2), ERK1/2, KRAS, and ACTB by Western blotting. Representative results are shown (G). The subpopulations were grown and passed several times under ordinal culture conditions. The means and standard deviations (error bars) of cumulated PDLs from triplicate experiments are presented (H). Cells (1.25 × 10⁴) of the subpopulations were cultured and grown in 0.3% soft agar for 5 weeks, then fixed with a buffered 4% paraformaldehyde solution (I). The means of colony counts and standard deviations (error bars) from triplicate experiments are presented (J).

Figure 4.

Biological effects of DUSP6 on A549 lung cancer cells. Empty vector (pQCXIP; MOCK) or DUSP6 was retrovirally transduced into A549 cells. Three days after the selection process, cells were harvested and counted, and 1.0 × 10⁴ cells were re-seeded onto a 10-cm dish. After 12 days, the cells were fixed with methanol and stained with Giemsa (A). The means and standard deviations (error bars) of colony counts from triplicate experiments are presented (B). The cells harvested immediately after the selection process were examined for the expression of DUSP6 and ACTB by Western blotting. Representative results are shown (C). The selected cells were cultured, grown, and passaged. The means and standard deviations (error bars) of cumulated PDLs from triplicate experiments are presented (D). The cells (2.5 × 10⁴) were seeded onto chamber slides (LAB-TEK 2-well chamber, Electron Microscopy Science) and cultured for 48 hours, then pulse-labeled with BrdU, fixed, and immunostained for BrdU for determining a labeling index (E, upper panels; LI, BrdU labeling index). The cells (2.5 × 10⁵) after the selection process were seeded onto a 10-cm dish and grown for 5 days, then harvested and fixed with 70% ethanol, and subjected to RNase A treatment. The cells were stained with propidium iodide, and cell cycle fractions were analyzed with a flow cytometer (E, lower panels).

Table 3.

Effect of DUSP Expression on Cell Cycle Progression

	Mock	DUSP6	P value
sub-G1	0.0 ± 0.0	0.0 ± 0.0	1.000
G1	66.4 ± 1.9	73.9 ± 1.0	0.048
S	15.5 ± 1.0	8.9 ± 1.1	0.031
G2/M	17.1 ± 1.2	18.1 ± 1.0	0.330
BrdU Index	16.5 ± 0.7	5.6 ± 1.3	0.011

Mock, empty vector-transduced; DUSP6, DUSP6-transduced cells. Mean± Standard deviation from three independent experiments is shown; P value calculated with Student’s t-test. 

Deficiency of Oncogenic KRAS-Mediated Transactivation and Translation of DUSP6 in Lung Cancer Cells

Exogenous expression of KRAS/V12 only slightly modulates DUSP6 expression at the mRNA or protein levels in either A549 (KRAS-mutated) or H1299 cells (KRAS-intact) (see supplemental Figure S9 at http://ajp.amjpathol.org), suggesting transcriptional and/or translational deficiency to be one of the mechanisms involved in the reduction of DUSP6 expression observed in lung cancer cells.

Analysis of CpG Methylation in the DUSP6 Promoter

Treatment with a DNA methyl-transferase inhibitor (AZA) alone, or in combination with a histone deacetylase (TSA), restored DUSP6 mRNA expression in all of the cancer cell lines examined, although the extent of restoration differed among the cells (Figure 5A). In NHBE-T, however, TSA-treatment, but not AZA-treatment, restored the expression, and the combination had no more of an effect than TSA alone (Figure 5A). These results implied the involvement of the promoter’s hypermethylation (and/or histone deacetylation) in the pathological suppression of DUSP6 transcription in the lung cancer cells, but not necessarily in its regulation in NHBE-T. To test this notion, we attempted to determine the promoter region of DUSP6. The results of luciferase reporter assays indicated that the promoter region lay between −1030 and +24 with strong elements between −1030 and −510, as the construct bearing −1030 to +24 showed markedly greater reporter activity than that bearing −511 to + 24 (Figure 5B). On the other hand, an in silico analysis using computer software (Genomatix, Munichi, Germany), indicated that the promoter was within the locus from −960 to −266, well consistent with the murine DUSP6 promoter demonstrated by Ekerot et al.6 We thus focused on this locus and searched for CpG methylation by bisulfate conversion-based DNA sequencing. Unexpectedly, almost all of the CpG sites were unmethylated in the cell lines examined (Figure 5C). We further examined the cell lines for methylation at the locus from +544 to +627, where a significant association between methylation status and mRNA expression was reported in pancreatic cancers.16 We used the same methylation-specific PCR analysis that was described previously.16 Although this locus was methylated in some lung cancer cell lines (Figure 5D), no correlation between the methylation status and DUSP6 mRNA expressions was found (Figure 2A and Figure 5D). We therefore speculated that the restoration of mRNA expression mediated by the inhibitors (Figure 5A) could be a secondary event resulting from the restoration of silenced potential transactivators, changes in the modification of histones, or other factors involved in the transcription process, although the involvement of hypermethylation of another site on the DUSP6 gene locus was not completely excluded.

Figure 5.

Methylation status of the DUSP6 promoter and intron 1. Cells treated with vehicle (Vcl), AZA, TSA, or a combination of AZA and TSA were examined for restoration of DUSP6 mRNA expression by quantitative reverse transcription-PCR. The copy number of DUSP6 mRNA normalized to that of β-actin was determined as the DUSP6 expression level in each cell sample. Then, relative expression normalized to the level of the untreated control (Ctl) in each cell line that was obtained from triplicate experiments, are plotted (y axis) (A). Luciferase reporter constructs, empty vector, −511 to +24, −1030 to +24, and −1539 to +24 (position of the first base of the start codon was determined as +1), were transfected into NHBE-T. Luciferase activities were measured and normalized as described in the Materials and Methods. The means and standard deviations (error bars) of the ratio of activity to that for empty vector (VECTOR) from duplicated experiments are presented (B). The region from −960 to −266 in the DUSP6 gene locus was PCR-amplified using sodium bisulfate-modified genomic DNA as a template. Eight subclones of PCR products were analyzed for methylation status by DNA sequencing. Open (○) and filled (•) circles indicate un-methylated and methylated cytosine at indicated CpG sites, respectively (C). The region from +544 to +627 in intron 1 of the DUPS6 gene was PCR-amplified with primers specific for either methyl or un-methyl DNA16 using sodium bisulfate-modified genomic DNA as a template. The products were resolved by electrophoresis in 3% agar and stained with ethidium bromide (D).

DUSP6 Expression in Primary Lung Cancers

For 96 cases of primary lung cancer, DUSP6 expression was examined by immunohistochemistry, and the correlation with growth activity, KRAS mutations, and other pathological parameters, was analyzed. DUSP6 expression was semiquantified according to the scoring system described in the Materials and Methods. The growth activity was evaluated as a MIB1 labeling index. Representative results of immunohistochemistry for DUSP6 and Ki-67 (MIB1) are shown in Figure 6. Statistical analysis revealed a significant inverse correlation between DUSP6 expression and growth activity (Table 4) (confirmed with Spearmen’s correlation test; ρ = −0.324, Z = −3.141, P = 0.0016). Significant inverse correlations were also observed between DUSP6 expression and the histological grade of ADC, T factor, and stage (Table 4). No difference in DUSP6 expression was found between cases with and without KRAS mutations (Table 4).

Figure 6.

Expression of DUSP6 and Ki-67 in primary lung cancers. The expression of DUSP6 (left panels) and Ki-67 (MIB1, right panels) in normal bronchial epithelia and primary lung cancers was examined by immunohistochemistry. Levels of DUSP6 expression were subdivided into four categories; negative (level 0, left bottom panel), faint (level 1, left fourth panel), moderate (level 2, left third panel), and strong (level 3, left second panels). Faint expression was defined as a level similar to that of bronchial epithelial cells (left top panel). Strong expression was defined as unequivocally greater signal intensity than that of bronchiolar epithelial cells (left second panel), and moderate expression was defined as a level in between faint and strong (left fourth panel). Proportions of Ki-67-expressing cells were determined as a labeling index of MIB1 (MIB1 index) as described in the Materials and Methods. ADC, adenocarcinoma; WEL, well differentiated; MOD, moderately differentiated; BAC, bronchioloalveolar carcinoma; ACI, acinar carcinoma; SOL, solid carcinoma with mucin. Magnification is × 400 for each.

Table 4.

Correlation between DUSP6 Expression and Clinicopathologic Status in Primary Lung Cancers

	Reduced (Score <1)	Retained (Score >1)	Score (M ± SD)
*MIB1 Labeling Index (96)
Low Level (<15%) (34)	6 (17.6%)	28 (82.4%)	1.416 ± 0.564†
High Level (≥15%) (62)	37 (59.7%)	25 (40.3%)	0.898 ± 0.536†
KRAS Codon12 (90)
Mutant type (22)	8 (36.4%)	14 (63.6%)	1.119 ± 0.693
Wild-type (68)	32 (47.1%)	36 (52.9%)	1.057 ± 0.575
*Histology (96)
ADC (65)	18 (27.7%)	47 (72.3%)	1.272 ± 0.558‡
*Grade
WEL (37)	5 (13.5%)	32 (86.5%)	1.446 ± 0.506§
MOD (20)	8 (40.0%)	12 (60.0%)	1.075 ± 0.521
POR (8)	5 (62.5%)	3 (37.5%)	0.869 ± 0.548§
Subtype
BAC (35)	6 (25.7%)	29 (74.3%)	1.458 ± 0.433
ACI (11)	4 (36.4%)	7 (63.6%)	0.982 ± 0.541
PAP (12)	4 (16.7%)	8 (83.3%)	1.183 ± 0.242
SOL (7)	3 (42.9%)	4 (57.1%)	0.950 ± 0.492
SQC (26)	21 (80.8%)	5 (19.2%)	0.754 ± 0.451‡
Grade
WEL (6)	5 (16.7%)	1 (83.3%)	0.750 ± 0.464
MOD (16)	13 (81.3%)	3 (18.7%)	0.660 ± 0.457
POR (4)	3 (75.0%)	1 (25.0%)	1.100 ± 0.606
LCC (5)	4 (80.0%)	1 (20.0%)	0.310 ± 0.486†
*T Factor (96)
T1 (36)	10 (27.8%)	26 (72.2%)	1.274 ± 0.625§
T2 (41)	21 (48.8%)	20 (51.2%)	1.012 ± 0.640
T3 (6)	3 (50.0%)	3 (50.0%)	1.133 ± 0.383
T4 (13)	10 (76.9%)	3 (23.1%)	0.746 ± 0.288§
N Factor (96)
N0 (61)	22 (36.1%)	39 (63.9%)	1.148 ± 0.637
N1 (11)	7 (63.6%)	4 (36.4%)	0.955 ± 0.468
N2 (18)	10 (55.6%)	8 (44.4%)	0.939 ± 0.609
N3 (6)	4 (66.7%)	2 (33.3%)	1.067 ± 0.476
*Stage (96)
1A (31)	6 (19.4%)	25 (80.6%)	1.350 ± 0.587¶
1B (25)	14 (56.0%)	11 (44.0%)	0.891 ± 0.659
2A (2)	2 (100.0%)	0 (0.0%)	0.500 ± 0.283
2B (10)	4 (40.0%)	6 (60.0%)	1.180 ± 0.454
3A (10)	4 (40.0%)	6 (60.0%)	1.140 ± 0.509
3B (16)	12 (75.0%)	4 (25.0%)	0.831 ± 0.406¶
4 (2)	1 (50.0%)	1 (50.0%)	1.100 ± 0.283

^*P < 0.005, Fisher’s exact test; 

^†P < 0.001, Student’s t-test; 

^‡P = 0.001, Tukey’s HSD test; 

^§P = 0.001, Games-Howell’s test; 

^¶P = 0.0012 Games-Howell’s test. ADC, adenocarcinoma; SQC, squamous cell carcinoma; LCC, large cell carcinoma; WEL, well differentiated; MOD, moderately differentiated; POR, poorly differentiated carcinomas; BAC, bronchioalveolar carcinoma; ACI, aciner adenocarcinoma; PAP, papillary adenocarcinoma; SOL, solid adenocarcinoma with mucin production. 

Allelic Imbalance of the DUSP6 Locus in Primary Lung Cancers

Among 121 cases examined, 64 cases (52.8%) were heterogeneous and informative, 22 cases (18.2%) were homozygous having the leucine allele, and 35 cases (28.9%) were homozygous having the valine allele. The allele frequency was 44.6% and 55.4% in the leucine and valine type, respectively. Among the informative cases (46 ADC, 16 SQC, and 2 LCC), LOH, evaluated by restriction fragment length polymorphism analysis, was found in 11 cases (17.2%; 10 ADC, 1 SQC, 0 LCC) (Table 5). A representative result of the restriction fragment length polymorphism analysis is shown in Figure 7. Analysis of the correlation between allelic status and immunohistochemical expression of DUSP6 revealed tumors with LOH to have significantly weaker expression than those without LOH, especially in adenocarcinomas (Table 5). No statistically significant association was found with the other analytical subjects, including histological type (ADC [21.7%, 10/46], SQC [6.3%, 1/16], LCC [0.0%, 0/2]; Fisher’s exact test, P = 0.2972), stage (1A [13.9%, 5/36], 1B [15.3%, 2/13], 2B [25.0%, 1/4], 3A [14.2%, 1/7], 3B [50.9%, 2/4]; Fisher’s exact test, P = 0.4712), MIB1 index (LOH 0.184 ± 0.216, retained 0.234 ± 0.219; Student’s t-test, P = 0.4966), and so on (data not shown).

Table 5.

Correlation between DUSP6 Allelic Status and Expression

	RTN	LOH	P value
Whole (64)	1.158 ± 0.664 (53)	0.755 ± 0.664 (11)	0.182
ADC (46)	1.542 ± 0.664 (36)	0.820 ± 0.664 (10)	0.022
SQC (16)	0.327 ± 0.094 (15)	0.110 ± 0.000 (1)	NA
LCC (2)	0.500 ± 0.707 (2)	NA	NA

RTN, retained; LOH, loss of heterozygosity; NA, not available; ADC, adenocarcinoma; SQC, squamous cell carcinoma; LCC, large cell; LCC, large cell carcinoma; (), indicates number of cases. P value calculated with Student’s t-test. 

Figure 7.

LOH analysis. A 168-bp fragment covering a polymorphic region of exon 1 (340 (t/g); 114 (leucine/valine) was PCR-amplified, subsequently digested with BsiHAKI, and then solved by electrophoresis in 5% agar. The leucine allele (Lu) is cut into 135 (and 33) −bp fragments (faster migrating band). The valine allele is resistant to BsiHAKI digestion (slower migrating band). A representative result is shown. Upper panel, without BsiHAKI digestion; lower panel, with BsiHAKI digestion. NIN, not informative (homozygous); RTN, allele is retained; LOH, loss of hemi allele; N, non-tumoral part; T, tumoral part.

Gene Mutation in the Protein-Coding Exons of DUSP6 Gene

Among 21 lung cancer cell lines (described in the Materials and Methods) and 48 primary lung cancers (36 ADC, 10 SQC, and 2 LCC) examined, no novel base substitution or deletion was found (data not shown), although known genetic polymorphisms were seen in some of the cancer cell lines and primary lung cancers (data not shown).

Discussion

It has been generally accepted that RAS is an oncogene, as innumerable studies reproducibly demonstrated that mutated RAS (either KRAS or HRAS) provided immortalized murine fibroblastic cells (NIH3T3) with malignant phenotypes such as contact inhibition-ignoring and anchorage-independent growth activities, or tumorigenicity in immunodeficient animals.4,25,26 However, mutated RAS (HRAS/V12) has been known to induce severe growth suppression, regarded as premature senescence, in some types of cells, such as early-passage murine and human fibroblasts,27,28,29 and small airway epithelial cells.2 The present study demonstrated a similar event in the KRAS/V12-transduced NHBE-T cells (Figure 1E). Previous studies reported p16INK4A and/or p14ARF, transactivated by ETS2 and/or DMP1 through the RAS-ERK (or-p38MAPK) pathway, to be essential effectors inducing premature senescence, as confirmed by the findings that a blockade of the p16INK4A/RB and/or p14ARF/p53 pathways by E1A (adenovirus) or large T (Simian virus 40) antigens eliminated this event.27,28,29,30,31,32 The NHBE-T used in this study, however, is Simian virus 40-immortalized, expressing large T antigens. The KRAS/V12-induced growth suppression seemed a novel event different from so-called premature senescence. We therefore examined other cell lines for this event. Transduction of KRAS/V12 induced severe growth suppression in the cells without KRAS mutations (H1299 [p53-inactivated, p16-inactivated, KRAS-intact, see supplemental Figure S10 at http://ajp.amjpathol.org], TKB1 [p53-inactivated, p16-inactivated, KRAS-intact, data not shown], and primary small airway epithelial cells, data not shown), but only modest suppression in the cells with KRAS mutations (A549 [p53-intact, p16-inactivated, KRAS-mutated, see supplemental Figure S11 at http://ajp.amjpathol.org] and H358 [p53-inactivated, p16inactivated, KRAS-mutated, data not shown]). These results indicated that the KRAS/V12-induced growth suppression was not unique to NHBE-T, suggested the p16INK4A/RB and/or p14ARF/p53 pathways not to be essential for such suppression, and implied that only clones overcoming suppression would be able to develop into lung cancers having KRAS mutations. Therefore, it would be of great interest to identify the effector(s) inducing such growth suppression, because they could be potential tumor suppressors playing crucial role(s) in oncogenic KRAS-mediated carcinogenesis, especially at the initiating step. A comprehensive search for downstream targets of oncogenic KRAS (KRAS/V12) uncovered several candidates (listed in Tables 1 and 2). The present study focused on DUSP6, but failed to prove it indispensable to induce oncogenic KRAS-induced growth suppression, since a dominant negative DUSP6 (C293S) failed to rescue the suppression significantly (Figure 3, E–G), and oncogenic KRAS did not increase DUSP6 expression despite the induction of growth suppression in H1299 cells (see supplemental Figure S9 at http://ajp.amjpathol.org). The potential involvement of the other factors listed in the tables is currently being investigated in detail. On the other hand, the subpopulation generated from NHBE-T serially transduced with a dominant-negative DUSP6 (C293S) and KRAS/V12 acquired higher anchorage-independent growth activity than that generated from cells transduced with KRAS/V12 alone (Figure 3, I and J, see also supplemental Figure S5 at http://ajp.amjpathol.org), suggesting that DUSP6 promotes the progression of lung carcinogenesis by maintaining stable activation of the ERK-mediated oncogenic pathway. The observations that forced expression of DUSP6 markedly suppressed the growth of lung cancer cell lines (Figure 4; see supplemental Figure S6 at http://ajp.amjpathol.org) and that the expression decreased with an increase in growth activity, as well as the progression of grade and stage of lung cancers (Figure 6 and Table 4) support our notion. Interestingly, early stage or low-grade lesions tended to express DUSP6 at high levels compared with the normal bronchial epithelia (Figure 6 and Table 4). This finding was comparable with previous observations in pancreatic cancers.13,14 Both in the lung and in the pancreas, KRAS mutations are very early events, which can be found even in premalignant lesions.1,2,15 It is supposed that premalignant or early lesions still maintain a DUSP6-mediated negative feedback response that interferes with oncogenic stimuli from mutated KRAS. However, in lung cancers, KRAS mutations are not so frequent, and found only in 5% to 30% of cases.1,2,33 As a substitute for KRAS mutations, either EGFR or BRAF oncogenes are mutated mutually exclusively in 10% to 50%34,35 and 0% to 3%34 of cases, respectively. Mutations in any of these three oncogenes are estimated to occur in up to 80% of lung cancers. EGFR, KRAS, and BRAF activate a common signaling pathway in this order, which subsequently activates the MEK-ERK cascade.36 Thus, oncogenic stimuli from EGFR, KRAS, or BRAF potentially excite a DUSP6-mediated negative-feedback response in the same way. Actually, a previous report demonstrated that DUSP6 was also transactivated by transduction of an oncogenic mutant of EGFR.4 This observation suggests that DUSP6 acts as a growth suppressor not only in oncogenic KRAS-mediated carcinogenesis, but also in EGFR- (and BRAF)-mediated carcinogenesis. The result that DUSP6 levels did not differ between the lung cancers with KRAS mutations and those without KRAS mutations (Table 4) seems supportive of this notion.

Aside from the role of DUSP6 in carcinogenesis of the lung, we tried to elucidate the mechanisms mediating the reduction of its expression. The methylation status of the promoter region and intron 1 (+544 to +627),16 mutations in the coding exons, and allelic imbalance (LOH) were investigated, because these are pivotal mechanisms inactivating many known tumor suppressors.37 We, however, failed to find hypermethylation of the promoter, a correlation between the methylation status of intron 1 and the expression, or mutations, in any of the lung cancer cell lines (Figure 5). Therefore, we have considered that hypermethylation, at least of the regions examined here, and gene mutation are not the major cause of the down-regulation of DUSP6 expression in lung cancers. On the other hand, LOH was found in 17.2% of primary lung cancers and was inversely associated with DUSP6 expression in ADC (Figure 7 and Table 5), suggesting LOH to be one of the mechanisms mediating the reduction in expression. However, the 50% decrease in gene dosage alone seems insufficient to explain the severe reduction (Figures 4 and 6) and/or transcriptional deficiency observed (see supplemental Figure S9 at http://ajp.amjpathol.org) in some of the cell lines and primary lung cancers. A recent study demonstrated a result suggesting the involvement of several transcription factors, such as FKHD, ETS2, INκB, and SOX5, in the regulation of DUSP6 expression.6 ETS2 is known to be activated via the MEK-ERK cascade,38,39 and is a pivotal effector inducible of premature senescence as described above.27,28,29,30,31,32 On the other hand, FKHD is a downstream target of PI3K, which could be activated indirectly by oncogenic EGFR or KRAS.2,36 It is speculated that potential alterations of such factors might be an additional mechanism causing the severe reduction and/or transcriptional deficiency of DUSP6 in lung cancers. Further detailed investigation will be required to test this notion.

In summary, the present study has suggested that DUSP6 could be a growth suppressor whose down-regulation is involved in the progression process of lung cancers. We hope our efforts will contribute to a better understanding of carcinogenesis of the lung.