Frequent inactivation of SPARC by promoter hypermethylation in colon cancers

Authors

  • Eungi Yang,

    1. Department of Pathology, Yonsei University College of Medicine, Seoul 120-752, Korea
    2. Brain Korea 21 Projects for Medical Science, Yonsei University College of Medicine, Seoul 120-752, Korea
    Search for more papers by this author
  • Hyun Ju Kang,

    1. Department of Pathology, Yonsei University College of Medicine, Seoul 120-752, Korea
    2. Brain Korea 21 Projects for Medical Science, Yonsei University College of Medicine, Seoul 120-752, Korea
    Search for more papers by this author
  • Kwi Hye Koh,

    1. Department of Pathology, Yonsei University College of Medicine, Seoul 120-752, Korea
    2. Brain Korea 21 Projects for Medical Science, Yonsei University College of Medicine, Seoul 120-752, Korea
    Search for more papers by this author
  • Hwanseok Rhee,

    1. Brain Korea 21 Projects for Medical Science, Yonsei University College of Medicine, Seoul 120-752, Korea
    Search for more papers by this author
  • Nam Kyu Kim,

    1. Department of Surgery, Yonsei University College of Medicine, Seoul 120-752, Korea
    Search for more papers by this author
  • Hoguen Kim

    Corresponding author
    1. Department of Pathology, Yonsei University College of Medicine, Seoul 120-752, Korea
    2. Brain Korea 21 Projects for Medical Science, Yonsei University College of Medicine, Seoul 120-752, Korea
    • Department of Pathology, Yonsei University College of Medicine, CPO Box 8044, Seoul, Korea
    Search for more papers by this author
    • Fax: +82-2-363-5263


Abstract

Epigenetic modification of gene expression plays an important role in the development of human cancers. The inactivation of SPARC through CpG island methylation was studied in colon cancers using oligonucleotide microarray analysis and methylation specific PCR (MSP). Gene expression of 7 colon cancer cell lines was evaluated before and after treatment with the demethylating agent 5-aza-2′-deoxycytidine (5Aza-dC) by oligonucleotide microarray analysis. Expression of SPARC was further examined in colon cancer cell lines and primary colorectal cancers, and the methylation status of the SPARC promoter was determined by MSP. SPARC expression was undetectable in 5 of 7 (71%) colorectal cancer cell lines. Induction of SPARC was demonstrated after treatment with the demethylating agent 5Aza-dC in 5 of the 7 cell lines. We examined the methylation status of the CpG island of SPARC in 7 colon cancer cell lines and in 20 test set of colon cancer tissues. MSP demonstrated hypermethylation of the CpG island of SPARC in 6 of 7 cell lines and in all 20 primary colon cancers, when compared with only 3 of 20 normal colon mucosa. Immunohistochemical analysis showed that SPARC expression was downregulated or absent in 17 of 20 colon cancers. A survival analysis of 292 validation set of colorectal carcinoma patients revealed a poorer prognosis for patients lacking SPARC expression than for patients with normal SPARC expression (56.79% vs. 75.83% 5-year survival rate, p = 0.0014). The results indicate that epigenetic gene silencing of SPARC is frequent in colon cancers, and that inactivation of SPARC is related to rapid progression of colon cancers. © 2007 Wiley-Liss, Inc.

Methylation of CpG islands located within the promoter region of tumor-related genes is an epigenetic modification frequently associated with the development of human cancers.1 Aberrant methylation of CpG-rich promoter regions results in inactivation of genes in cancer cells through transcriptional silencing.2 In colorectal carcinomas, epigenetic silencing has been reported for a number of genes; aberrant CpG methylation was found for hMLH1 in 10–20% of carcinomas, for p16 in 20–30%, and for APC in 10–20%. VHL, MGMT and COX2 are also frequently silenced in colorectal cancer.3 Identification of selectively hypermethylated genes in colon cancers may improve our understanding of the role of epigenetic inactivation in colorectal tumorigenesis and may be valuable as a specific diagnostic marker. The combination of high throughput microarray analysis and the induction of methylated genes by demethylating agents can be used to identify genes that are silenced by aberrant hypermethylation and establish a panel of epigenetic markers for specific cancers.4, 5 In this study, we used oligonucleotide microarrays to identify genes that are differentially expressed in colon cancer cells after treatment with a demethylating agent. We identified 70 genes with functions potentially related to tumorigenesis that contain CpG islands in their promoter region and are expressed after treatment with a demethylating agent. We report the identification of secreted protein, acidic and rich in cysteine (SPARC), as a putative colon cancer related gene that is silenced by aberrant hypermethylation in colon cancer cells.

Material and methods

Cell lines and tissue samples

Cell lines were obtained from either the American Type Culture Collection (ATCC; http://www.atcc.org) or the Korean Cell Line Bank (KCLB; http://cellbank.snu.ac.kr). DLD-1, HCT8, HCT116, LOVO and HT29 were grown in RPMI medium supplemented with 10% fetal bovine serum (Life Technologies, Grand Island, NY), 100 U/ml penicillin and 100 mg/ml streptomycin at 37°C in the presence of 5% CO2. CCD18-Co, RKO and WiDr were cultured in minimum essential medium supplemented with 10% fetal bovine serum (Life Technologies) and antibiotics (100 U/ml penicillin and 100 mg/ml streptomycin). Fresh tissue samples were collected from 20 patients with colorectal cancer as a test set. In each case, grossly normal mucosa remote from the tumor was included as a control. A total of another 292 colon cancer cases were selected as a validation set, and used for tissue microarray construction. For tissue microarray construction, colorectal cancer specimens were obtained from the archives of the Department of Pathology at Yonsei University, Seoul, Korea. Samples were selected from consecutively identified cases at the Gastrointestinal Tumor Working Group Tissue Bank at Yonsei University Medical Center (Seoul, Korea) between December 1996 and November 1999, and were subjected to immunohistochemical analysis for SPARC using tissue microarray. All cases were assessed independent of any knowledge of molecular status. Histological factors were recorded as follows: (i) histological grade (well/moderately/poorly differentiated), (ii) growth pattern (pushing/mixed/irregular) and (iii) peritumoral lymphoid reaction (absent/present). Tumors with less than 10% SPARC expression were regarded as negative, 10–30% as weak positive and greater than 30% as strong positive. The Kaplan–Meier survival curve of 292 stage I, II, III and IV colorectal cancer patients according to SPARC expression was calculated using SPSS version 11 (Chicago, IL).

Authorization for the use of these tissues for research purposes was obtained from the Institutional Review Board of Yonsei University of College of Medicine. Some of the fresh specimens were supported by the Liver Cancer Specimen Bank from the National Research Resource Bank Program of the Korea Science and Engineering Foundation in the Ministry of Science and Technology. Fresh tissues were obtained immediately after surgical excision and stored at −70°C prior to DNA and RNA extraction. Tumor specimens were microdissected on a cryostat and fractionated to enrich the tumor cell populations more than 70%. Genomic DNA was prepared using the sodium dodecyl sulfate-proteinase K and phenol–chloroform extraction method. For the RT-PCR analysis, 6 out of 20 cases of colon cancer test samples were selectively microdissected with the laser capture microdissection technique. Tumor cells were selectively procured from H&E-stained slides without any contamination from normal cells, using a 30G1/2 hypodermic needle (Becton Dickinson, Franklin Lakes, NJ) affixed to a micromanipulator, as described previously.6

5-Aza-2′-deoxycytidine (5Aza-dC) treatment

Seven colon cancer cell lines (DLD-1, HCT8, HCT116, LOVO, RKO, HT29 and WiDr) were treated with a demethylating agent, 5Aza-dC (Sigma Chemical, St Louis, MO). Exponentially growing cells were seeded in T-75 culture flasks. After overnight incubation cells were exposed continuously to 5 μm 5Aza-dC for 6 days, with a change of drug and culture medium every 48 hr.

Microarray formulation

High-density spotted-oligonucleotide microarrays were manufactured at the array core facility in the Department of Pathology, College of Medicine of the Catholic University of Korea. The human Oligolibrary™ was purchased from Compugen/Sigma-Genosys and consisted of 18,861 oligonucleotides, representing 18,664 LEADS™ clusters plus 197 controls of glyceraldehyde-3-phosphate dehydrogenase. Thus, a total of 18,861 synthesized 60-mer oligos were robotically printed and processed.

RNA preparation and hybridization

Total mRNA was extracted from cells grown to 80–90% confluence using TRIZOL reagent (Invitrogen, San Diego, CA). Twenty microgram RNA was used for synthesis of cDNA target molecules.7 Target cDNA and Universal Human Reference RNA (Stratagene, La Jolla, CA) were hybridized to an oligonucleotide microarray containing 18,664 probe sets representing 18,664 unique (LEADS™) genes, and the array was scanned using GenePix scanners. Expression values for each gene were calculated using GenePix Pro 4.0 analysis software. Some hybridization were carried out in duplicate with fluororeversal to compensate for the different chemical properties of the fluorescent dye molecules and for potential biases associated with normalization.

Classification of colorectal cancer cell lines by molecular pattern analysis

Unsupervised hierarchical clustering analysis was used to analyze gene expression in 7 colorectal cancer cell lines (with or without 5Aza-dC treatment) and 2 normal colonic mucosae. We selected a data set of genes that satisfied the filtering criteria: genes having more than 60% of log-transformed ratio values presenting across all arrays were chosen, and genes with less than 0.3 standard deviations of log-transformed ratio were discarded. The selected gene data set was then applied to average linkage hierarchical clustering analysis using the uncentered correlation similarity metric method in Cluster version 2.20, and the resulting expression map was visualized with Treeview version 1.60 (http://rana.lbl.gov/EisenSoftware.htm).

Semiquantitative RT-PCR

First-strand cDNA was synthesized from 1 mg total RNA using random hexamer primers (Qiagen, Valencia, CA) and M-MLV reverse transcriptase (Invitrogen) according to the manufacturers' instructions. Twenty nanogram cDNA was used in each RT-PCR reaction. β-Actin was used as a quantitative internal control. Primer sets were designed to contain an exon–exon junction, and were as follows: SPARC: 5′-TGA TGA GAC AGA GGT GGT GG-3′ (sense), 5′-AAG TGG CAG GAA GAG TCG AAG-3′ (antisense) and β-actin, 5′-ACA GAG TAC TTG CGC TCA GGA G-3′ (sense) and 5′-TGT ATG CCT CTG GTC GTA CCA C-3′ (antisense). Five microliters aliquots of the RT-PCR product were subjected to 2% agarose gel electrophoresis and stained with ethidium bromide.

Isolation and bisulfite treatment of genomic DNA

Genomic DNA was isolated from fresh frozen tissue by overnight incubation in 100 μg/ml proteinase K (Sigma) and 0.1% SDS (Sigma) at 55°C with subsequent phenol/chloroform extraction and isopropanol precipitation. Herring sperm DNA (1 μg; Promega, Madison, WI) was added as a carrier. Two micrograms of DNA was denatured in 0.2 mol/l NaOH for 10 min at 37°C in 50 μl total volume. Thirty microlitres of 10 mmol/l hydroquinone (Sigma) and 520 μl 3.5 mol/l sodium bisulfite (pH 5.0; Sigma) was added. After 16 hr at 50°C, the DNA was purified and incubated in 0.3 mol/l NaOH for 5 min at room temperature. After ethanol precipitation, DNA was dissolved in 40 μl Tris-EDTA.8

Sequencing analysis of bisulfite treated genomic DNA

A 221-bp fragment of the 5′ region of the SPARC gene was amplified using the primers ATT TAG TTT AGA GTT TTG AGT GG (sense) and ACA AAA CTT CCC TCC CTT AC (antisense). PCR conditions were as follows: 95°C for 5 min; 40 cycles of 95°C for 20 sec, 60°C for 20 sec and 72°C for 30 sec; and a final extension of 4 min at 72°C. After incubation with exonuclease I and shrimp alkaline phosphatase (USB, Cleveland, OH), PCR products were sequenced using the SequiTherm Excel II cycle sequencing kit as recommended by the manufacturer (Epicentre Technologies, Madison, WI). Products of the sequencing reactions were resolved by electrophoresis and visualized by autoradiography. The methylation status of each sample was evaluated by determining the percentage of the intensity of the cytosine band versus the thymine band for each CpG site.9

Methylation-specific polymerase chain reaction

Methylation status of the SPARC gene was also determined by MSP as described previously.10 Bisulfite treated DNA (1 μg) was amplified using primers specific for either methylated or unmethylated DNA under the following conditions: 95°C for 5 min; 40 cycles of 95°C for 20 sec, 62°C for 20 sec and 72°C for 30 sec; and a final extension of 4 min at 72°C. Primer sequences were TTT TTT AGA TTG TTT GGA GAG TG (sense) and AAC TAA CAA CAT AAA CAA AAA TAT C (antisense) for unmethylated reactions (132 bp product), and GAG AGC GCG TTT TGT TTG TC (sense) and AAC GAC GTA AAC GAA AAT ATC G (antisense) for methylated reactions (112 bp product). Five microliters PCR product was loaded onto 3% agarose gels and visualized by ethidium bromide staining.9

Western blot analysis

Whole lysates were prepared from cell lines using lysis buffer [50 mM Tris (pH 7.4), 1% Triton X-100, 5 mM EDTA, 1 mM KCl, 140 mM NaCl, 2 mM MgCl2, 1 mM phenylmethylsulfonyl fluoride, 1 mM sodium fluoride, 1% aprotinin, 1 μM leupeptin and 1 mM sodium ortho-vanadate]. Twenty microgram of total protein lysate was loaded into each lane, size-fractionated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane that was blocked with tris-buffered saline-Tween 20 containing 5% skim milk. Membranes were incubated with primary antibodies, SPARC (AON-5031; Haematologic Technologies, Essex Junction, VT) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Trevigen, Gaitherburg, MD) for 2 hr at room temperature. After washing, membranes were incubated with HRP-conjugated secondary antibody (Santacruz Biotech, Santa Cruz, CA), washed and developed with ECL-Plus (Amersham Pharmacia Biotech, UK).

Immunohistochemistry

Colorectal carcinoma tissue arrays containing 292 carcinomas and matched normal mucosae were constructed from formalin-fixed and paraffin-embedded tissues (IsuAbxis, Seoul, Korea). Sections (5 μm) were cut onto coated slides and deparaffinized using routine techniques. Antigen retrieval was performed in 10 mM sodium citrate buffer (pH 6.0) heated at 95°C in a steamer for 20 min. After blocking endogenous peroxidase activity with a 3% aqueous H2O2 solution for 5 min, sections were incubated with anti-SPARC monoclonal antibody (Haematologic Technologies) at a final concentration of 4 μg/ml for 60 min. Labeling was detected with the Envision Plus Detection Kit (DAKO, Carpinteria, CA), as recommended by the manufacturer, and all sections were counterstained with hematoxylin. SPARC expression was detected in the mesenchymal and stromal cells, normal colonic epithelical cells and infrequently in colon cancer cells.10 The extent of immunostaining was categorized based on the proportion of cancer cells with SPARC expression: <10%, negative; 10–30%, weak positive; and >30%, strong positive.11 Comparisons according to SPARC expression treated these as 2 groups: without SPARC expression (negative) versus with SPARC expression (weak positive and strong positive).

Results

Identification of genes induced by 5Aza-dC in colon cancer celllines

The gene expression profiles in 7 colon cancer cell lines before and after induction by 5Aza-dC were determined using oligonucleotide microarrays. The expression profiles of 2 normal colonic mucosae were used as controls. The relative expression of each gene was measured by comparing its expression ratio to that of Universal Human Reference RNA. We initially performed molecular pattern analysis to determine whether our spotted-oligoarray system was able to identify colorectal carcinoma cell lines, colon cancer cell lines treated with 5Aza-dC and normal mucosa by molecular profiling. Unsupervised hierarchical clustering analysis of 7 colorectal cancer cell lines treated with and without 5Aza-dC, and 2 normal colonic mucosae was performed according to gene expression profile. Genes that passed the filtering criteria were selected (genes expressed in more than 60% of all arrays were selected, and genes with standard deviation of less than 0.3 were discarded). A total of 6,490 genes were selected and applied to average linkage hierarchical clustering analysis using the uncentered correlation similarity metric method. When we compared the gene expression in 7 colon cancer cell lines before and after 5Aza-dC treatment, there was a substantial increase (>2-fold) in signal intensities of 83 transcripts in DLD-1, 99 transcripts in HCT116, 57 transcripts in HCT8, 30 transcripts in HT29, 62 transcripts in LOVO 17 transcripts in RKO and 35 transcripts in WiDr after 5Aza-dC treatment. Overall, 226 transcripts were upregulated >2-fold in at least 1 of the 7 cancer cell lines after drug treatment. Notably, this large panel of genes, including several testis cancer antigens such as G antigens, is consistent with previous reports of genes inducible by 5Aza-dC.12 Although several genes have been reported to be induced by 5Aza-dC treatment independent of the methylation status of their 5′ CpG islands, many of the genes identified here represent potential targets for aberrant methylation in colorectal cancer. Among the genes that were upregulated after 5Aza-dC treatment, genes known to be related to tumorigenesis (such as those involved in cell cycle regulation, induction of apoptosis, differentiation or transcriptional regulation) were examined for the presence of CpG islands in their promoters. In total, 70 genes met these criteria and are listed in Table I according to the expression fold change after 5Aza-dC treatment (Table I).

Table I. Novel Target Genes that Changed to 5-aza Induced Genes
SymbolChromosomal locationFunction
SPARC5q31.3-q32Cell adhesion growth
FSTL13q13.33Function as a negative regulator of cell growth
CCL202q33-q37Chemotaxis
TFF321q22.3Digestion
MAFK7p22Transcription factor
ALB4q11-q13Trensport
CKB14q32Amino acid and derivative metabolism
HMGB113q12Bind DNA with low specificity
IGFBP417q12-q21.1Signal transduction
ACTN419q13Invasive growth
IGSF411q23.2Virulence
GAGE1Xp11.4-p11.2Cellular defense response
LGALS122q13.1Apoptosis
HOXA107p15-p14Transcription regulation
SGK6q23Protein phosphorylation
CEACAM119q13.2Immune response
DUSP15q34Cell cycle
DEFB18p23.2-p23.1Nitrogen metabolism
ID120q11Transcription regulation from Pol II promoter
APP21q21.3Cell adhesion
MUC11q21Cholesterol catabolism
MGST24q28.3Leukotriene metabolism
MSNXq11.2-q12Cell motility
APOH17q23-qterComplement activation, classical pathway
EMP112p12.3Ionic insulation of neurons by glial cells
SPRY213q22.1Histogenesis and organogenesis
DUSP15q34Cell cycle
TGFB15q31Cell proliferation
MXI110q24-q25Transcription regulation
ELF31q32.2Transcription from Pol II promoter
SPOCK5q31Cell adhesion
BTG112q22Negative control of cell proliferation
MUT6p21Transport
GMNN6p22.1Cell cycle arrest
CDC45L22q11.21DNA replication initiation
CRIP17q11.23Zinc binding
ATF31q32.3Transcription regulation
TYR11q14-q21Malanin biosynthesis from tyrosine
IFI3019p13.1Immune response
ICB-11p35.2Cell adhesion
MLPH2q37.3Transport
ST511p15RNA dependent DNA replication
NCF17q11.23Superoxide metabolism
BTC4q13-q21Cell cycle
RAC317q25.3Cell growth and/or maintenance
OLFM19q34.3Signal transduction
GPR307p22Chemotaxis
PMP2217p12-p11.2Ionic insulation of neurons by glial cells
ABCB101q42Transport
EPLIN12q13Transcription regulation from Pol II promoter
LGALS419q13.13Cell adhesion
INSIG17q36Metabolism; cell proliferation;
MX121q22.3Signal transduction
GPC3Xq26.1Cell growth and/or maintenance
EGR15q31.1Transcription regulation
LEF14q23-q25Transcription regulation
SIAT13q27-q28Humoral immune response; protein modification
HCLS13q13Protein modification
MDK11p11.2Cell cycle control
PTP4A16q12DNA replication
FGFR34p16.3ATP binding
MCAM11q23.3Cell adhesion
KLF34p14Repression of transcription from Pol II promoter
CDKN1A6p21.2Cell cycle
JUNB19p13.2Transcription regulation
APBB111p15Signal transduction
CCND36p21Cell cycle control
WFDC220q12-q13.2Signal transduction
DNER2q37.1Cell adhesion
APOB2p24-p23Signal transduction

Identification of SPARC as a gene that is inactivated in colon cancer cell lines

Among the 70 genes identified by oligonucleotide microarray analysis, expression of SPARC (secreted protein acidic and rich in cystein) was suppressed 10.3 to 30.8 fold in all 7 colon cancer cell lines compared with normal mucosa. A 3.1 to 8.7 fold induction of SPARC after 5Aza-dC treatment was observed in DLD-1, HCT8, HCT116 and HT29 cell lines, and LOVO and WiDr cells also showed a slight increase in SPARC expression (Table II). RKO cells did not show an increased level of SPARC expression after treatment with 5Aza-dC. These results indicate that expression of SPARC is suppressed through epigenetic inactivation in colon cancers.

Table II. Expression Level (Log Ratio) of SPARC with or Without 5-aza-2′-Deoxycytidine Treatment in Colorectal Cancer Cell Lines
 Log ratio valueFold change (Cell line + 5-aza: Cell line)Log ratio valueFold change (Normal tissue/Cell line)
Cell lineCell line + 5-azaNormal tissue
DLD-1−4.65−1.538.7−1.1011.7
HCT8−3.69−2.053.1−1.1012.9
HCT116−4.94−2.874.2−1.1030.8
LOVO−4.03−4.121.0−1.1016.3
RKO−3.37−3.550.9−1.1010.3
HT29−4.15−1.974.5−1.1017.8
WiDr−4.59−4.311.2−1.1024

We next performed RT-PCR and Western blot analysis to examine the expression of SPARC mRNA and protein in 7 colon cancer cell lines and a normal colonic fibroblast cell line, CCD-18Co. The SPARC transcript was detectable in CCD-18Co, but undetectable in 5 (71%) of the 7 colon cancer cell lines, and protein expression was evident only in CCD-18Co (Figs. 1a and 1b). These results confirm the difference in SPARC expression between most colon cancer cell lines and fibroblasts. PCR-LOH analysis at the SPARC (5q31.3-q32) gene locus was performed on 20 colorectal carcinoma test samples (Fig. 1c), demonstrating that inactivation was not due to allelic deletion, since only 2 (Cases 2 and 9) out of 20 cases showed allelic loss at SPARC.

Figure 1.

RT-PCR analysis, Western blot analysis and PCR-LOH of the SPARC gene in colorectal cell lines. (a) RT-PCR analysis of SPARC mRNA expression in 7 colorectal cancer cell lines (DLD-1, HCT8, HCT116, LOVO, RKO, HT29 and WiDr) and normal control cell line (CCD-18Co). SPARC mRNA expression was reduced in 5 colorectal cancer cell lines (DLD-1, HCT116, LOVO, RKO and WiDr). β-Actin was amplified as an internal control. (b) Western blot analysis of SPARC expression in 1 normal control and 7 colon cancer cell lines. SPARC expression is severely suppressed in colon cancer cell lines. (c) PCR-LOH analysis at the SPARC gene locus (5q31.3-q32) in 20 colorectal carcinoma samples. Two cases out of 20 (Cases 2 and 9) showed allelic loss at the SPARC locus. (d) RT-PCR analysis of the SPARC gene in colorectal cancer cell lines after treatment with 5-aza-2′-deoxycytidine. SPARC gene expression was induced in 5 colorectal cancer cell lines (DLD-1, HCT8, HCT116, LOVO and HT29). β-actin was amplified as in internal control. + and − denote with and without 5Aza-dC treatment, respectively.

SPARC mRNA expression was measured before and after 5Aza-dC treatment using RT-PCR analysis. Before treatment the cell lines DLD-1, HCT116, LOVO, RKO and WiDr did not express SPARC mRNA, while HCT8 and HT29 expressed low levels of SPARC mRNA (Fig. 1d). Following treatment with 5Aza-dC, SPARC mRNA was induced in DLD-1, HCT8, HCT116, LOVO and HT29, but not in RKO or WiDr cells.

Methylation analysis of the SPARC gene in colon cancer cell lines and primary colorectal cancer samples

The 5′ region of the SPARC gene contains a CpG-rich sequence fulfilling the criteria of CpG islands (%GC = 61.9, ObsCpG/ExpCpG = 0.602, length = 291)13 (Figs. 2a and 2b). The methylation status of the CpG sites in the SPARC promoter region was determined in 7 colon cancer cell lines and 20 colon cancer test samples using MSP.

Figure 2.

DNA sequencing analysis of CpG islands in the SPARC gene after bisulfite modification of genomic DNA isolated from colorectal cancer cell lines. (a) Distribution of CpG dinucleotides (vertical lines) in the 5′ region of the SPARC gene showing a CpG-rich sequence (CpG island) spanning from exon 1 to intron 1. CpG island start = 1955, end = 2245, %GC = 61.9. ObsCpG/ExpCpG = 0.602, length = 291. (b) Promoter sequence of human SPARC. The 16 CpG sites are underlined. (c) MSP analysis of the SPARC gene in colorectal cancer cell lines. The promoter region of the SPARC gene was methylated in 7 cell lines (DLD-1, HCT8, HCT116, LOVO, RKO, HT29 and WiDr). Lanes labeled M and U denote products amplified by primers recognizing methylated and unmethylated sequence respectively. (d) The methylation status of 16 CpG sites was examined in several cell lines. ○, unmethylated site; equation image, partially methylated site; •, methylated site.

We initially performed MSP using primers specific for unmethylated and methylated DNA. Among the 16 CpG sites, the 3rd and 4th sites were included in the sense primer and the 9–11th sites were included in the antisense primer (Fig. 2b). MSP for methylated DNA gave no product in the normal human colon fibroblast cell line, CCD18-Co. Six colon cancer cell lines, DLD-1, HCT116, LOVO, RKO, HT29 and WiDr, gave PCR products only with primers for methylated DNA, indicating fully methylated CpG islands. The HCT8 cell line gave products from both primers specific for methylated and unmethylated DNA, indicating a hemimethylated status of CpG sites (Fig. 2c). Direct sequencing analysis of bisulfite modified PCR products was used to determine whether any CpG sites were preferentially methylated in the 7 cell lines that were found to contain methylated DNA (Fig. 2b). In the 5 cell lines (DLD-1, HCT116, LOVO, RKO and WiDr) with no SPARC expression, complete methylation was found in all 16 CpG sites in HCT116, RKO and WiDr, while full methylation at CpG 1-15 and hemimethylation at the 16th CpG sites were found in DLD-1 and LOVO (Fig. 2d). In the 2 cell lines with low level of SPARC expression (HCT8 and HT29), full methylation was present at the 1–6th and 9–16th CpG sites and hemimethylation was found at the 7th and 8th CpG sites in HCT8, while full methylation at 1–6th and 8–16th CpG sites and unmethylation at the 7th CpG site was found in HT29 (Fig. 2d).

Genomic DNA was isolated from primary colorectal samples and subjected to methylation-specific PCR analysis (Figs. 3a and 3b). CpG methylation in SPARC was detected in all of the 20 colorectal cancer samples. In contrast, methylation was detected in only 3 of the corresponding 20 normal colonic mucosa samples, suggesting that CpG methylation of the SPARC gene is a frequent event in colorectal cancer. We measured expression of SPARC in 6 of the 20 samples by semiquantitive RT-PCR of the extraction of RNA with the laser capture microdissction microscope and found that SPARC is suppressed in 5 out of 6 colon cancers (Fig. 3c). We could not perform the RT-PCR in remaining 14 cases because adequate amount of samples were not available in these 14 cases. We could not perform Western blotting analysis, because adequate amount of protein could not be obtained by our laser capture dissection method. Subsequent immunohistochemical analysis demonstrated that SPARC is expressed in normal colonic epithelial cells, stromal cells and some of the tumors (Figs. 4a, 4b and 4c). SPARC was diffusely expressed in the cytoplasm in both colonic epithelial and tumor cells (Fig. 4d). Immunohistochemistry without primary antibody application did not give any positive signal (data not shown). When we evaluated SPARC expression of colon cancer cells in our 20 colon tumors, SPARC expression was absent in 5 cases, suppressed in 12 cases and intense in 3 cases.

Figure 3.

MSP analysis of the SPARC gene in 20 primary human colorectal carcinoma samples (CRC) and the corresponding 20 normal colonic mucosa samples (NL). (a) Representative examples of the MSP assay in primary tumors and their adjacent normal colonic mucosa. M, methylated; U, unmethylated; T, colorectal carcinoma samples; N, normal colonic mucosa samples; DLD-1, positive control; CCD-18Co, negative control. (b) Open boxes, gray boxes and black boxes indicate unmethylation, partial methylation and homozygous methylation, respectively. (c) RT-PCR analysis of SPARC mRNA expression in 6 out of 20 colon cancer samples and matched normal mucosae, showing that SPARC expression is severely suppressed in 5 of 6 colon carcinomas (1, 3, 4, 17 and 19).

Figure 4.

Immunohistochemical analysis of SPARC expression in colonic mucosae and colorectal carcinomas. (a) SPARC expression in colonic epithelial cells: strong expression of SPARC in the cytoplasm of normal colonic epithelial cells was found in most of the colonic mucosae. (b) SPARC expression in tumor stromal cells: strong expression of SPARC in the stromal cells around the carcinoma cells was evident in most of the cancer samples. (c) Strong expression of SPARC in tumor cells: stromal expression of SPARC in carcinoma cells was found in some tumors. (d) Strong expression of SPARC in tumor cells: SPARC is strongly expressed in the cytoplasm of some carcinoma cells. (e) Weak expression of SPARC in tumor cells, low magnification: weak SPARC expression is evident in carcinoma cells. (f) Weak SPARC expression in tumor cells, high magnification: strong expression of SPARC in the stromal cells is evident when compared with the weak expression in cancer cells. (g) No SPARC expression in carcinoma cells, low magnification: no SPARC expression was frequently found in the tumor cells. (h) No SPARC expression in cancer cells, high magnification: in cases where cancer cells were negative for SPARC expression, intense expression of SPARC was frequently found in stromal cells.

Inactivation of SPARC expression is related to poor patient outcome

The expression of SPARC protein was examined in the validation set of 292 primary colorectal cancers by immunostaining (Figs. 4c4h and Table III). Similar to the result of test samples, SPARC expression was not detected in 81 cases (27.7%), weak SPARC expression was found in 165 cases (56.5%) and strong SPARC expression was found in 46 cases (15.8%). In most cases the stromal cells near to or surrounding the cancer cells showed strong expression. Suppression of SPARC is not related to any clinicopathological factors except mucin production, which is more frequent in cases without SPARC expression (20% vs. 9% p = 0.011). Comparing the 5-year survival rate of patients according to SPARC expression levels reveals significant differences; 5-year survival was 56.79% in patients with negative expression, 70.91% in those with weak positive expression and 93.48% in patients with strongly positive expression (p = 0.0001, Fig. 5a). Overall, patients with SPARC expression (weak positive and strong positive) showed a better survival rate than patients lacking SPARC expression (75.83% vs. 56.79% 5-year survival rate, p = 0.0014, Fig. 5b).

Figure 5.

The Kaplan-Meier survival curve of 292 colorectal cancer patients according to SPARC expression. The survival curve shows that lack of SPARC expression correlated with poor prognosis of colorectal carcinomas. (a) Five year survival rate of colorectal carcinoma patients without SPARC expression (Group 0), low SPARC expression (Group 1) and strong SPARC expression (Group 2) was 56.79% or 70.91%, 93.48%, respectively (p = 0.0001). (b) Overall, patients with SPARC expression (weak positive and strong positive) showed better survival rate than patients without SPARC expression (5-year survival rate 75.83 versus 56.79%, p = 0.0014).

Table III. Comparison of the Clinicopathological Features of 292 Colorectal Carcinoma Patients According to the Presence of SPARC Expression
FeaturesNumber of carcinomas without SPARC expression (%)Number of carcinomas with SPARC expression (%)p value
  • 1

    Microsatellite instability (MSI) was determined by the mobility shift of PCR products using 5 microsatellite loci (BAT25, BAT26, D2S123, D5S346 and D17S250) in tumors. Tumors showing instability in two or more markers were classified as high MSI, those showing it in one marker as low MSI, and those showing no instability as microsatellite-stable.

Age (years)
 <60 years40 (49.4)112 (53.1)0.851
 ≥60 years41 (50.6)99 (46.9)
Gender
 Male47 (58.0)115 (54.5)0.480
 Female34 (42.0)96 (45.5)
Stage
 I6 (7.4)26 (12.3)0.398
 II37 (45.7)80 (37.9)
 III26 (32.1)80 (37.9)
 IV12 (14.8)25 (11.9)
Size
 ≥6.0 cm38 (46.9)96 (45.5)0.336
 <6.0 cm43 (53.1)115 (54.5)
Growth pattern
 Pushing10 (12.3)35 (16.6)0.955
 Mixed36 (44.4)98 (46.4)
 Irregular35 (43.2)78 (37.0)
Differentiation
 Well13 (16.0)41 (19.4)0.983
 Moderate37 (45.7)101 (47.9)
 Poor31 (38.3)69 (32.7)
Location
 Right colon18 (22.2)51 (24.2)0.725
 Left colon63 (77.8)160 (75.8)
Mucin production
 Absent65 (80.2)192 (91.0)0.011
 Present16 (19.8)19 (9)
Lymphoid reaction
 Absent33 (40.7)87 (41.2)0.247
 Present48 (59.3)124 (58.8)
MSI status1
 High10 (12.3)26 (12.3)0.578
 Low or stable71 (87.7)185 (87.7)

Discussion

Hypermethylation of DNA promoter CpG islands results in transcriptional silencing of genes on the inactivated X chromosome, imprinted genes and exogenous integrated genes. Aberrant DNA methylation is now recognized as a common and important event in cancer development. In colorectal carcinomas, silencing of the expression of certain genes by CpG island methylation is associated with carcinogenesis.14 To identify genes that are targets for inactivation by CpG methylation in colon carcinomas, colon cancer cell lines were treated with a methylation inhibitor. Using DNA microarray analysis, we identified 70 genes with functions putatively related to tumorigenesis that contain CpG islands and were induced following 5Aza-dC treatment. From these 70 genes, we selected SPARC for further analysis because: (i) SPARC was suppressed in 5 of the 7 colon cancer cell lines, (ii) induction of SPARC was observed in 5 of the 7 cell lines after 5Aza-dC treatment and (iii) SPARC is known to be involved in tumor cell growth, differentiation and metastasis.

SPARC is a calcium binding glycoprotein of 35 kDa, and is the most common noncollagenous protein in bone. SPARC is known to be predominantly located in the nonorganic component of bone and adheres collagen and calcium; however, the role of SPARC in these tissues has not been clearly elucidated.15, 16, 17, 18 Originally, SPARC was thought to be a glycoprotein involved in the formation of the intracellular calcified matrix. However, many other diverse functions have since been reported, including: interaction with the extracellular matrix, epithelial cells, and stromal cells18; cell motility; cell adhesion; tumor invasion; and proliferation of normal and tumor cells.19, 20, 21, 22SPARC is also known to inhibit DNA synthesis and stimulate production of TGF-β.23 Based on this findings, it is possible to speculate that downregulation of SPARC may affect the behavior of colon cancer cells, although the specific mechanism remains to be determined.

The expression of SPARC in cancer tissues and functional analysis of the SPARC gene in tumor cell lines has been widely studied. The expression of SPARC in both normal and tumor cells is highly dependent on tumor type. In many cancers, upregulation of SPARC has been reported in the peritumoral stromal cells, and this fibroblast-derived SPARC is believed to have diverse effects on the biology of tumor cells.9SPARC has been reported to enhance the invasive capacity of prostate and breast cancer cells, glioma cells, esophageal, bladder and metastatic hepatocellular cancers, invasive meningiomas and malignant melanomas.24, 25, 26, 27, 28, 29, 30, 31, 32

In this study, we demonstrated that expression of SPARC is downregulated in colorectal carcinomas. Similar suppression of SPARC has been reported in lung cancer and pancreatic cancer.9, 24 This suppression may be related to the rapid growth of tumors, since SPARC has an antiproliferative function through the modulation of cell cycle regulatory proteins or growth factors.33

When we evaluated the relationship between SPARC suppression and clinicopathologic factors in 292 colorectal carcinomas, suppression of SPARC was not associated with age, sex, tumor site, MSI status or tumor stage. However, SPARC suppression was associated with decreased mucin production and with poor prognosis. This result raises the possibility that SPARC expression may be used as a biomarker for selecting chemotherapy in colon cancer patients with modest progression.

We demonstrated that the downregulation of SPARC is related to aberrant methylation of CpG islands in the promoter region. SPARC was shown to be methylated in all of the 7 colon cancer cell lines and in 20 colon cancer tissues but in only 3 of the 20 matched normal colonic muscosal samples. These findings suggest that methylation of SPARC is a common event in colorectal carcinogenesis. Analysis of the methylation sites showed that most of the CpG sites were methylated in the cell lines with SPARC inactivation. However, we found that the 7th CpG site is commonly unmethylated in the 2 cell lines with SPARC expression, suggesting that the 7th CpG site is particularly important in the regulation of SPARC expression.

In conclusion, we showed that SPARC expression is frequently suppressed in colon cancer cell lines in conjunction with aberrant DNA methylation, and that suppression of SPARC expression is related to poor patient outcome. These results suggest that suppression of SPARC may be related to colorectal cancer development and progression.

Ancillary