Genome-wide gene expression profiling of ... - Wiley Online Library

Download PDF
2 downloads 36496 Views 485KB Size Report
Comment
Dec 13, 2005 - 22q13.1. NM_004262. HAT. J3.17. 4q13.2. U79275. HSU79275. J3.18 .... microarray analysis, while black ..... GH, Seo JS, Kim CK, Kim YW.
Int. J. Cancer: 118, 2461–2469 (2006) ' 2005 Wiley-Liss, Inc.

Genome-wide gene expression profiling of cervical cancer in Hong Kong women by oligonucleotide microarray Yick-Fu Wong1*, Tak-Hong Cheung1, George S.W. Tsao2, Keith W.K. Lo1, So-Fan Yim1, Vivian W. Wang3, Macy M.S. Heung1, Samuel C.S. Chan1, Loucia K.Y. Chan1, Tina W.F. Ho1, Katherine W.Y. Wong1, Chen Li4, Yu Guo4, Tony K.H. Chung1 and David I. Smith5 1 Department of Obstetrics & Gynaecology, The Chinese University of Hong Kong, Hong Kong 2 Department of Anatomy, Hong Kong University, Hong Kong 3 Department of Obstetrics, Gynecology and Reproductive Medicine, Brigham and Women’s Hospital, Harvard Medical School, USA 4 Department of Bioinformatics, Harvard School of Medicine, USA 5 Department of Experimental Pathology, Mayo Foundation School of Medicine, USA An analysis of gene expression profiles obtained from cervical cancers was performed to find those genes most aberrantly expressed. Total RNA was prepared from 29 samples of cervical squamous cell carcinoma and 18 control samples, and hybridized to Affymetrix oligonucleotide microarrays with probe sets complementary to over 20,000 transcripts. Unsupervised hierarchical clustering of the expression data readily distinguished normal cervix from cancer. Supervised analysis of gene expression data identified 98 and 139 genes that exhibited >2-fold upregulation and >2-fold downregulation, respectively, in cervical cancer compared to normal cervix. Several of the genes that were differentially regulated included SPP1 (Osteopontin), CDKN2A (p16), RPL39L, Clorf1, MAL, p11, ARS and NICE-1. These were validated by quantitative RT-PCR on an independent set of cancer and control specimens. Gene Ontology analysis showed that the list of differentially expressed genes included ones that were involved in multiple biological processes, including cell proliferation, cell cycle and protein catabolism. Immunohistochemical staining of cancer specimens further confirmed differential expression of SPP1 in cervical cancer cells vs. nontumor cells. In addition, 2 genes, CTGF and RGS1 were found to be upregulated in late stage cancer compared to early stage cancer, suggesting that they might be involved in cancer progression. The pathway analysis of expression data showed that the SPP1, VEGF, CDC2 and CKS2 genes were coordinately differentially regulated between cancer and normal. The present study is promising and provides potential new insights into the extent of expression differences underlying the development and progression of cervical squamous cell cancer. This study has also revealed several genes that may be highly attractive candidate molecular markers/targets for cervical cancer diagnosis, prognosis and therapy. ' 2005 Wiley-Liss, Inc. Key words: gene expression; cervical cancer; microarray

Although cervical cancer is highly curable when detected early, it remains the leading cause of gynecological cancer death worldwide. It is the most common malignancy of the genital tract in Hong Kong women.1 Ninety five percent of cervical cancers are squamous cell in origin and the vast majority of them are associated with infection with oncogenic subtypes of human papillomavirus (HPV).2 However, HPV infection alone is insufficient for malignant transformation.3 Other genetic events independent or in conjunction with HPV infection are required. Previous molecular studies on tumor-related genes have been hampered because only a few genes can be studied at any one time. Interactions between different genes or genetic pathways can be overlooked.4 Gene expression profiling is a powerful tool to identify and target markers associated with disease or therapy. The simultaneous analysis of numerous messenger RNA expression patterns as well as their relationship with biologic functions is a powerful tool to begin to understand the biological process of cancer development and to identify important genes involved in this process.5 Such RNA-based global analyses of gene expression have led to the identification of large numbers of dysregulated Publication of the International Union Against Cancer

genes in many different types of human cancers. However, there have been relatively few reports in cervical cancers.6–10 In this study, we studied a set of 29 cervical squamous cell carcinoma, using genome-wide gene expression profiling on the Affymetrix U133A chips containing probes for over 20,000 human gene transcripts. Selected differentially expressed genes were further validated using real-time RT-PCR and immunohistochemistry. In addition, signaling pathways potentially involved in cervical tumorigenesis were also analyzed. The results obtained from the present study, therefore, lay the groundwork for future analysis of these potential markers/targets for clinical utility in the diagnosis, prognosis and treatment of cervical cancer. Material and methods Patients and samples A total of 47 specimens including 29 cervical cancers and 18 normal cervix were studied using oligonucleotide microarrays. For validation of genes of interest revealed by gene expression profiling, an additional set of tissue samples including 60 cervical cancer and 30 normal cervical epithelium specimens were used in a real-time RT-PCR assay. Before surgical or radiotherapy treatment, a representative part of the tumor tissue was initially dissected, and the specimens were harvested within 30 min of resection from the cancer patient. About half of the tissue specimens were snap-frozen in liquid nitrogen and stored at 280°C until use for RNA extraction. The remaining specimen was embedded in OCT compound for frozen section and staining with hemotoxylin and eosin, and then was used to corroborate the histological diagnosis and to determine the proportion of tumor to stroma tissue in the tumor specimen collected. All cervical cancers used in this study were cervical squamous cell carcinoma. Each case of cervical cancer was staged according to the International Federation of Gynecology and Obstetrics (FIGO) criteria (Table I).11 Normal cervix surface epithelium layer specimens were collected from patients who underwent hysterectomy for benign conditions, and treated the same as cancer tissue. All of the patients involved in this study were Hong Kong Chinese and recruited at the Department of Obstetrcis and Gynaecology, Prince of Wales Hospital, The Chinese University of Hong Kong, Hong Kong. Informed consent was obtained from each patient. The study was approved by the Clinical Research Ethics Committee of The Chinese University of Hong Kong. Grant sponsor: Research Grant Council of the Hong Kong Special Administration Region; Grant number: CUHK4084/01M. *Correspondence to: Department of Obstetrics and Gynaecology, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, Hong Kong. Fax: 1852-2636-0008. E-mail: [email protected] Received 21 June 2005; Accepted after revision 23 August 2005 DOI 10.1002/ijc.21660 Published online 13 December 2005 in Wiley InterScience (www. interscience.wiley.com).

2462

WONG ET AL.

TABLE I – CLINICO-PATHOLOGIC CHARACTERISTICS OF 29 PATIENTS WITH CERVICAL SQUAMOUS CELL CARCINOMA Case code

C280 C303 C312 C326 C337 C359 C385 C407 C409 C421 C430 C447 C455 C505 C521 C531 C555 C558 C569 C647 C651 C668 C686 C689 C692 C695 C697 C701 C703

Clinical HPV test HPV16 HPV18 HPV31 HPV33 HPV52 HPV58 stage

I I I I I III III III IV IV II III II I I II I II I II II III I IV III I IV I I

1 1 1 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

1 1 1 1 1 2 1 2 1 1 1 1 1 1 1 2 2 2 1 2 1 1 1 2 1 1 1 1 1

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 1 2 2 2 2 2 2 2 2 2

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 1 2 2 2 2 2 2 2 2 2 2 2

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 1 2 2 2 2 2 2 2 2 2 2 2 2

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 1 2 2 2 2 2

RNA extraction The proportion of malignant cells in all 29 tumors used for microarray hybridization was more than 80%. Total RNA was extracted from the cervical cancer and normal cervix tissue samples using RNeasy Mini Kit, according to the manufacturer’s protocol (Qiagen, Valencia, CA). For the real-time RT-PCR assay, microdissection was performed in all 60 cancer and 30 normal control specimens to enrich target cells, i.e., tumor cells and normal epithelial cells, to a purity of more than 90%. For the manual microdissection, 10-lm thick frozen tissue sections were prepared. Target cells were collected under dissection microscope using a continuous section stained by H&E as guidance. The quality of all RNA samples was monitored by gel electrophoresis on 1% Denaturing Agarose Gels to check 28S and 18S ribosomal RNA to ensure that degraded samples were not used. DNA extraction For HPV infection, detection in the 29 tumor specimens that were assayed with microarray, DNA from corresponding frozen tumor samples was prepared by proteinase K digestion and phenol/chloroform extraction. The DNA integrity was examined by a b-globin PCR assay yielding a human b-globin product of 268 bp, using primers GH20 (50 -GAAGAGCCAAGGACAGGTAC-30 ) and PC04 (50 -CAACTTCATCCACGTTCACC-30 ). Microarray Quality RNAs obtained from 29 cervical cancers and 18 normal cervix were analyzed using the Human U133A (HG-U133A) GeneChip1 arrays that were purchased from Affymetrix, Santa Clara, CA. This chip probes for over 20,000 transcripts. Double-stranded complementary DNA (cDNA) and biotinylated complementary RNA (cRNA) were synthesized from total RNA and hybridized onto microarrays. The array hybridization, washing and staining procedures were performed according to the manufacturer’s protocols. All data used were derived from Affymetrix 5.0 software. GeneChip 5.0 output files were given as a signal that represents the difference between the intensities of the sequence-specific perfect match probe set and the mismatch probe set or as detection of

present, marginal or absent signals, as determined by the program’s algorithm. Gene arrays were scaled to an average signal of 1,500 and then analyzed independently. Gene expression data analysis After raw image DAT data files were initially processed to create CEL files, DNA-Chip analyzer (dCHIP) (Version 1.3) (www.dCHIP.org) was used for data quality checking and highlevel analysis. A fold-change analysis was performed in which the ratio of the geometric means of the expression intensities of the relevant gene fragments was computed. This ratio was reported as the fold change (up or down). Confidence intervals and p-values on the fold change were also calculated with the use of a 2-sided Welch modified 2-sample t test. p-values of 0.01 or less were considered significant. In this study, only those with differences between cancer and normal control >2.0-fold (90% lower bound) were considered to be differentially expressed in cancer. Real-time RT-PCR For validation of array results by quantitative PCR, cDNA was prepared from each tumor and normal control in which microdissection was carried out to enrich target cells. Quantitative realtime RT-PCR analysis was performed using a fluorescence temperature cycler ABI Prism 7900 Sequence Analyzer (Applied Biosystems, Forster City, CA). Each reaction was run in duplicate. The comparative Ct method was used to calculate the amount of amplification as specified by the manufacturer. In brief, 1 lg of total RNA from each sample was reverse-transcribed using SuperScript II Reverse Transcriptase (Invitrogen, Carlsbad, CA). Reverse-transcribed RNA samples (1 ll from 20 ll of total volume corresponding to 50 ng of total RNA) were amplified using Assay-On-Demand Taqman probes, primers and TaqMan Universal PCR Master Mix (Applied Biosystems, Forster City, CA) to produce PCR products specific for 8 genes including SPP1 (Osteopontin), CDKN2A (p16), RPL39L, Clorf1, MAL, p11, ARS and NICE-1, respectively. The primers and probes were purchased from Applied Biosystems (Forster City, CA). Relative expression values of each gene were determined from a standard curve, using a universal RNA sample (Clontech, Palo Alto, CA). Samples were amplified with GAPDH primers for determination of the relative starting amount of cDNA in each sample, and all genes were normalized to that amount. Negative controls without template were produced for each run. Immunostaining To evaluate whether the differential SPP1 mRNA expression detected by array was reflected in corresponding decreases in the SPP1 protein, protein expression was evaluated by standard immunohistochemical staining on formalin-fixed tumor tissue from 47 archival cervical squamous cell carcinoma and 5 normal cervix specimens. After pretreatment with 10 mM citrate buffer (pH 6.0) using a microwave, the tissue sections (5 lm thick) were incubated with an anti-SPP1 polyclonal antibody prepared by ProSci (Poway, CA) at 1:1,000 dilution. Slides were subsequently labeled with streptavidin-biotin (Dako, Glostrup, Denmark), stained with diaminobenzidine and counterstained with hematoxylin. The intensity of staining was graded as 0 (not greater than negative control), 11 (light), or 21 (heavy). HPV detection HPV infection and genotype were determined by HPV consensus PCR and HPV type specific PCR, as described previously.12 Only HPV consensus PCR positive specimens were subjected to HPV-type specific PCRs, including HPV-16, -18, -31, -33, -52 and -58. For PCRs of HPV-16, -18, -31, and -33, we adopted conventional methods while for PCRs of HPV-52 and -58, we used the primer pair: 50 -GCATTCATAGCACTGCCAC-30 ; and 50 -GCCTCTACTTCAAACCAGCC-30 , corresponding to positions of the sense strand 761–779 and antisense strand 909–928, and yielding a 168-

GENOME-WIDE GENE EXPRESSION PROFILING OF CERVICAL CANCER IN HONG KONG WOMEN

2463

FIGURE 1 – Dendrogram is plotted from hierarchical clustering of 497 probe sets in 47 cervical samples, and divides them into 2 major groups: cancer group and normal control group. The sample is labeled by sample code. C, squamous cell carcinoma; N, normal cervix.

FIGURE 2 – Upregulated genes in cervical cancer (C) compared to normal cervix (N).

bp amplimer, and the primer pair: 50 –CATGTACCATTGTGTGCCC-30 and 50 -ACCGCTTCTACCTCAAACC-30 , corresponding to positions of the sense strand 833–851 and antisense strand 932–950, yielding a 118-bp amplimer, respectively. The products of the reaction were electrophoresed on 2% agarose gels. Results Gene expression data obtained from all 47 microarrays were normalized, and then the corresponding model-based expression index was calculated using the perfect match-only model. Analysis of the data from each chip in dChip showed that 5 of the 47 microarrays had more variability in gene expression ratios than the rest. However, analysis of all 47 microarrays together did not show any aberrant behavior related to these 5 microarrays. Accordingly, such 5 microarrays were not excluded in further high-level evaluation. To better examine the possibility that the 47 cervical tissue samples could be separated into 2 distinct sets on the basis of geneexpression profiles, unsupervised hierarchical clustering was performed. This analysis revealed 497 of 22,283 probe sets satisfying a filtering criteria, i.e., the rate of probe set call in the arrays used was 20%, and showed a distinct separation between cancers and normal controls (Fig. 1). For supervised analysis, a t test with p-values 2.0 (90% lower bound) and mean difference >100 between cancer and control samples. From that, 271 probe sets had a 0% median false discovery rate (FDR) assessed by 50 permutations, indicating that these genes might be potentially biologically interesting. Among these 271 transcripts identified corresponding to known genes, 38 genes were represented by more than 1 transcript (range, 2–4). When these were accounted for, 237 individual genes were identified. Of the 237 differentially regulated genes, 98 were upregulated and 139 were downregulated in the 29 cervical squamous cell carcino-

mas, i.e., more were underexpressed (58%) in cervical cancer compared to normal cervix than the overexpressed (42%). Figures 2 and 3 show the up and downregulated genes in cervical cancer compared to normal cervix. When a more stringent criteria was adopted, we found 102 genes had >2-fold change of expression in at least two-thirds of cervical cancer compared to normal. Of these 102 significantly differentially expressed genes from this study, 18 have previously been reported to be differentially expressed in cervical cancer (Table II). They showed a similar pattern of expression in the present study, as that reported in previously published findings. The remaining 84 genes were identified as being associated with cervical cancers for the first time. Table III lists the novel differentially expressed genes by >2-fold in at least two-thirds of cervical cancer cases. In this set of cervical cancers, there were 13 cases that were stage I, 6 stage II, 6 stage III and 4 stage IV, respectively (Table I). When early stage (I–II) cases were compared to late stage (III– IV) cases, 2 genes, connective tissue growth factor (CTGF) (Accession No. M92934) and regulator of G-protein signaling 1 (RGS1) (Accession No. NM_002922), were found to be specifically upregulated in late stage cancer with 3.01-fold and 2.39-fold increased expression, respectively. To validate the microarray results, a total of 8 genes including 3 upregulated, SPP1, CDKN2A and RPL39L, and 5 downregulated, Clorf10, MAL, P11, ARS and NICE-1 genes, were selected for real-time RT-PCR analysis on an independent set of 60 cancer and 30 normal samples. The expression differences for both the overexpressed genes and underexpressed genes in cancers as compared to normal samples almost mirrored the microarray data (Fig. 4). As Kim and coworkers found increased GAPDH gene expression in cervical cancers, we did a comparison of GAPDH gene with 18S gene to be used for the determination of relative starting amount of cDNA in 10 RNA samples, including 6 cancers and 4 normal controls. The result did not show any differences in differential expression of SPP1 detected in these samples using either 1

2464

WONG ET AL.

FIGURE 3 – Downregulated genes in cervical cancer (C) compared to normal cervix (N).

house-keeping gene, GAPDH or 18S for the normalization (data in detail is not shown). Gene ontology classification of the differentially expressed in cervical cancer identified 2 major molecular function subgroups, including genes associated with structural molecular activity and peptidase activity; and 6 major biological process subgroups associated with (i) cell proliferation (n 5 38), (ii) cell cycle (n 5 30), (iii) protein catabolism (n 5 20), (iv) proteolysis and peptidolysis (n 5 20), (v) immune response (n 5 26) and (vi) response to stress (n 5 28). Genes related to cell proliferation represented the largest

cluster, accounting for up to 16% (38/237) of the modified genes. The complete list of genes classified by GO is available on request. SPP1 protein levels in paraffin-embedded block sections of 47 cervical squamous cell carcinomas were assessed by immunohistochemical staining. Most positive staining observed in cancer tissue was localized to cellular membrane and cytoplasm. Photomicrograph of SPP1 is shown as example of the immunostaining experiments (Fig. 5). There was strong signal from SPP1 in cancer cells in the tumor tissue. In contrast, little or no

2465

GENOME-WIDE GENE EXPRESSION PROFILING OF CERVICAL CANCER IN HONG KONG WOMEN TABLE II – GENES PREVIOUSLY REPORTED AS MODIFIED IN CERVICAL CANCER AND IDENTIFIED IN THE CURRENT STUDY Chromosomal location

GenBank accession

Gene symbol

Fold change

U38945 BC002700 NM_004526 NM_001071 NM_001793 BC001188 NM_002276 AA807529 AF022375 NM_004428

CDKN2A KRT7 MCM2 TYMS CDH3 TFRC KRT19 MCM5 VEGF EFNA1

9p21 12q12-q13 3q21 18p11.32 16q22.1 3q29 17q21.2 22q13.1 6p12 1q21-q22

BG170541 NM_000930 NM_000421 NM_002274 NM_005547 NM_001562 NM_000697 NM_002371

MET PLAT KRT10 KRT13 IVL IL18 ALOX12 MAL

7q31 8p12 17q21 17q12-q21.2 1q21 11q22.2-q22.3 17p13.1 2cen-q13

6.65 3.71 3.06 2.96 2.58 2.54 2.52 2.24 2.12 2.1 2.01 2.01 22.48 22.87 23.03 23.69 210.66 214.37

References

Wang et al. [13] Smedts et al. [14] Ishimii et al. [15] Suzuki et al. [16] Han et al. [17] Lloyd et al. [18] Mittal et al. [19] Williams et al. [20] Gaffney et al. [21] Chen et al. [22]; Wu et al. [23] Baykal et al. [24] Shimada et al. [25] Carrilho et al. [26] Carrilho et al. [26] Xu et al. [27] Cho et al. [28] Nigam et al. [29] Hatta et al. [30]

TABLE III – NOVEL DIFFERENTIALLY EXPRESSED GENES BY >2-FOLD IN ATLEAST TWO-THIRDS OF CERVICAL CANCER CASES IDENTIFIED BY U133A OLIGONUCLEOTIDE ARRAYS GenBank accession

Gene symbol

Fold change

Chromosomal location

GenBank accession

Gene symbol

Fold change

Chromosomal location

M83248 NM_005101 NM_021105 NM_018098 NM_000935 NM_005531 NM_016448 NM_001034 NM_002800 NM_005844 BF973387 BE563442 NM_005046 NM_000114 NM_007196 BC002710 AA181179 NM_006853 AA292789 NM_004605 NM_024046 NM_006393 AV724216 AF062006 NM_016433 AF133207 AL049698 U92457 AF007162 AF124790 NM_003113 NM_002705 NM_005975 NM_003986 NM_001557 NM_023067 NM_007268 U24056 NM_004262 U79275 NM_016321 NM_001423

SPP1 G1P2 PLSCR1 ECT2 PLOD2 IFI16 RAMP RRM2 PSMB9 HCG9 ESTs IL1RN KLK7 EDN3 KLK8 KLK10 CAMKK2 KLK11 FLJ11029 SULT2B1 MGC8407 NEBL NDRG4 LGR5 GLTP HSPB8 MLLT4 GRM4 CRYAB ESR2 SP100 PPL PTK6 BBOX1 IL8RB FOXL2 VSIG4 KCNJ4 HAT HSU79275 RHCG EMP1

10.08 4.64 3.86 3.71 3.71 3.17 3.1 2.89 2.85 22.03 22.08 22.11 22.12 22.19 22.38 22.41 22.43 22.5 22.5 22.52 22.55 22.56 22.57 22.58 22.59 22.61 22.62 22.67 22.7 22.71 22.77 22.87 22.9 23.04 23.05 23.06 23.1 23.1 23.17 23.18 23.19 23.19

4q21-q25 1p36.33 3q23 3q26.1-q26.2 3q23-q24 1q22 1 2p25-p24 6p21.3 6p21.3 – 2q14.2 19q13.41 20q13.2-q13.3 19q13.3-q13.4 19q13.3-q13.4 12q24.2 19q13.3-q13.4 17q23.2 19q13.3 3p21.31 10p12 16q21-q22.1 12q22-q23 12q24.11 12q24.23 6q27 6p21.3 11q22.3-q23.1 14q 2q37.1 16p13.3 20q13.3 11p14.2 2q35 3q23 Xq12-q13.3 22q13.1 4q13.2 12q13.1 15q25 12p12.3

NM_016610 X05421 S76346 NM_014456 NM_006079 BC000059 NM_024508 AI923984 NM_005416 NM_014058 AA548647 AI770004 NM_030819 NM_006945 X07695 AL049699 NM_006518 NM_002084 NM_024888 NM_001432 NM_002771 AF009664 NM_001785 U63296 NM_019598 NM_003843 NM_025087 NM_003245 NM_006846 S70004 NM_001942 NM_015596 NM_006025 N26005 NM_006121 NM_004669 BF671400 NM_019060 AB059408 NM_020427 NM_016190 NM_006061

TLR8 KRT3 RUNX1 PDCD4 CITED2 CELSR1 ZBED2 SPRR1A SPRR3 DESC1 UPK1A EPB41L3 MGC11335 SPRR2B KRT4 ME1 SPRR2C GPX3 FLJ11535 EREG PRSS3 PRSS2 CDA HPGD KLK12 SCEL FLJ21511 TGM3 SPINK5 GYS2 DSG1 KLK13 P11 PPP1R3C KRT1 CLIC3 ESTs C1orf42 HOP SLURP1 C1orf10 CRISP3

23.21 23.22 23.29 23.31 23.33 23.35 23.43 23.44 23.48 23.58 23.68 23.73 23.8 23.86 23.94 23.95 24 24.16 24.32 24.49 24.53 24.53 24.56 25.03 25.1 25.13 25.28 25.36 25.59 25.8 26.24 26.38 26.45 26.93 26.96 27.69 27.77 29.08 29.27 211.36 214.92 238.61

Xp22 12q12-q13 21q22.3 10q24 6q23.3 22q13.3 3q13.13 1q21-q22 1q21-q22 4q13.2 19q13.13 18p11.32 16q22.1 1q21-q22 12q12-q13 6q12 1q21-q22 5q23 19p13.3 4q13.3 9p11.2 7q34 1p36.2-p35 4q34-q35 19q13.3-q13.4 13q22 4p12-p11 20q11.2 5q32 12p12.2 18q12.1 19q13.3-q13.4 12q13.1 10q23-q24 12q12-q13 9q34.3 – 1q21 4q11-q12 8q24.3 1q21 6p12.3

signal from SPP1 was detected in stroma cells in the tumor as well as normal cervix tissues. Of 47 tumor samples stained with anti-SPP1 antibody, 28 were scored as 21, 14 were 11 and 5 were 0.

We examined HPV infection and HPV genotype in 29 cervical cancer samples. HPV infection was detected in 28 of 29 cancers. Of the 28 HPV-positive samples, 22 were typed as HPV-16 and 2 as HPV-18. We identified 1 specimen that was positive for HPV-

2466

WONG ET AL.

FIGURE 4 – Real-time PCR analysis of the 8 altered genes in cervical cancer. Gray bars represent the average fold change obtained with microarray analysis, while black bars represent that obtained with the RT-PCR. All values are presented as relative fold changes with upregulation > 0 and downregulation < 0.

FIGURE 5 – Paraffin-embedded sections of cervical squamous cell carcinoma after staining with anti-SPP1 (a) and H&E (b).

31, 1 for HPV-52, 1 for HPV-58 and 1 specimen had an unknown HPV type. None of the specimens were typed as HPV-33-positive in this set of cervical cancers (Table I). Comparison of gene expression profiles between the cervical cancers infected with HPV-16 and those infected with non-HPV-16 HPVs revealed that 36 genes with alterations of expression more than 3-fold were shared in both types of cancer. However, 23 genes altered in expression were specifically associated with cancers infected with HPV-16. These genes are mainly related to DNA replication and protein modification revealed by gene ontology analysis. Five genes were associated with cancers infected with non-HPV-16 HPV infection (Table IV). These genes are mainly involved in cell growth and apoptosis. To identify signaling pathways that are associated with cervical tumorigenesis, we analyzed our microarray expression data using PathwayAssist (Version 2.5 Trial) (Iobion Informatics LLC, La Jolla, CA). After importing 237 differentially expressed genes into PathwayAssist, we obtained 1 signaling pathway associated with cervical squamous cell carcinoma. The expression data showed the SPP1, VEGF, CDC2 and CKS2 genes as coordinately differentially regulated between cancer and normal. These genes encode for proteins that are part of a signaling pathway associated with

tumor cell ‘‘DNA fragmentation, cell survival and apoptosis’’ as well as regulation of signal transduction. SPP1, CDC2 and CKS2 have not been previously identified, as being dysregulated in cervical cancer. The interacting genes revealed in this study constitute a potentially important signaling pathway involved in cervical cancer (Fig. 6). Discussion There have been some reports on the use of microarrays to identify genes that are aberrantly expressed in cervical tumors as compared to normal cervix.7–10 Although these studies have provided useful information, but they have all been limited by a number of factors, including heterogeneous specimens, insufficient numbers of tumor specimens analyzed, and microarrays with relatively few features present on them. We have performed a more global assessment of differential gene expression in cervical tumors, and obtained expression profiles from the most common histologic subtype of cervical cancer, squamous cell carcinoma as compared to its normal counterpart. This study revealed 237 genes with greater than 2-fold increases or decreases in gene expression in cervical cancers as compared to

GENOME-WIDE GENE EXPRESSION PROFILING OF CERVICAL CANCER IN HONG KONG WOMEN

normal cervix, while 102 genes were with >2-fold changes of expression in at least two-thirds of the cancer. Eighteen of these 102 genes were previously reported as being differentially expressed in cervical tumors, whereas the remainder had not been described in any reports on expression profiling of cervical cancer. Twelve of the 18 previously described genes were upregulated in cervical cancer, including p16 (CDKN2A) that had a 6.65-fold increase. Wang et al. reported that p16 overexpression was significantly higher in cervical CIN (75%) and in squamous cell carcinoma (75%) than normal or inflamed cervices.13 Trunk et al. reported p16-positive dysplastic cells could be detected in 95% of high grade CIN lesions; however, in slides obtained from patients

with nonsuspicious smears, relatively few p16-positive cells were found.31 Overexpression of p16, therefore, may be as a biomarker helpful for the identification of dysplastic cervical epithelial cells on histologic slides as well as in cervical smears. Another gene previously described as overexpressed and verified in our study is keratin 7 (KRT7). Smedts et al. observed increasing expression of KRT7 with increased grade of CIN.14 The remaining 10 overexpressed genes previously observed are MCM2, TYMS, CHD3, TRFC, KRT19, MCM5, VEGF, EFNA1, MET and PLAT. They have all been reported to be related to either cervical tumorigenesis or its progression.15–25 Further clarification of their functional role in cervical malignancy would be warranted. Six previously

TABLE IV – DIFFERENTIALLY EXPRESSED GENES IN HPV-16 INFECTED AND NON-HPV-16 HPV INFECTED CERVICAL CANCER IDENTIFIED BY U133A OLIGONUCLEOTIDE ARRAYS GeneBank accession

Gene symbol

In HPV-16 infected case M83248.1 NM_005101.1 BC002700.1 NM_021105.1 NM_000935.1 NM_001071.1 NM_004526.1 NM_001034.1 NM_016448.1 NM_014456.1 BC000059.1 AA548647 NM_001139.1 NM_004262.1 NM_002705.1 NM_002274.1 U79275.1 NM_014058.1

SPP1 G1P2 KRT7 PLSCR1 PLOD2 TYMS MCM2 RRM2 RAMP PDCD4 CELSR1 UPK1A ALOX12B HAT PPL KRT13 HSU79275 DESC1

NM_016321.1 AL356504 AI923984 S70004.1 NM_005416.1

RHCG LOC400786 SPRR1A GYS2 SPRR3

In non-HPV-16 HPV infected case NM_004052.2 BNIP3 NM_005046.1 KLK7 T50399 NM_002178.1

HBA2 IGFBP6

NM_002963.2

S100A7

FIGURE 6 – Potential signaling pathways involved in cervical cancer.

2467

Biological process

Fold change

Antiapoptosis Protein modification Biogenesis Phospholipid scrambling Protein modification Nucleic acid metabolism DNA replication DNA replication – Apoptosis Cell adhesion – Lipid metabolism Epidermis development – Epidermis development – Proteolysis and peptidolysis Ammonium transport – Epidermis development Glycogen biosynthesis –

12.56 6.72 5.27 4.36 4.24 3.98 3.64 3.63 3.53 23.85 24 24.06 24.15 24.29 24.39 24.41 24.42 24.84

Apoptosis Proteolysis and peptidolysis – Regulation of cell growth Epidermis development

24.58 24.98

25.28 25.32 25.47 26.06 26.38

25.86 25.94 210.26

2468

WONG ET AL.

described underexpressed genes are keratin 10 (KRT10), KRT13, IVL, interleukin-18 (IL-18), AOLX12 and MAL. The expression of 2 types of Keratin, KRT10 and KRT13 was reduced with cervical tumorigenesis.26 IVL expression was reduced in CIN, while expression of IL-18 was reversely correlated to HPV oncogenic E6 gene expression in cervical cells.27,28 Activity of AOLX12 product as well as MAL mRNA level were significantly decreased in cervical cancers.29,30 The repeated identification of the above 18 genes as aberrantly regulated in cervical cancers using multiple platforms suggests that these genes may form the basis for the development of diagnostic markers or screening methods to detect this disease at an earlier and potentially curable stage of cervical cancer. The majority of the significantly aberrantly regulated genes (84 of 102) identified in this study have not been previously reported in cervical cancers. One of the upregulated genes, SPP1, encodes an extracellular glycosylated bone phosphoprotein, osteopontin. There is considerable interest in the role of SPP1 in human tumorigenesis, especially since it appears to be a marker of transformed epithelial cells.32 It has also been recognized that SPP1 seems to act as a survival factor.33 We found that SPP1 had a 10.08-fold increased expression in cervical cancer when compared to normal control. Real-time PCR confirmed the overexpression of the SPP1 mRNA and immunohistochemistry revealed similar overexpression of the SPP1 protein. However, the relationship between SPP1 overexpression and clinical outcomes remains unclear. When the time of follow-up in all cases of cervical cancer investigated reaches 2 years, we might further analyze the correlation of aberrant expression of above genes of interest revealed in this study with the clinico-pathologic characteristics of cervical squamous cell carcinoma. One of the most challenging aspects of expression profiling is to determine the biological interaction and relevance of large numbers of differentially expressed genes. We used PathwayAssist software to identify clusters of interacting genes that were coordinately up or downregulated and to identify signaling pathways that contribute to cervical tumorigenesis. This work identified 4 genes that interact with each other to presumably induce cervical tumorigenesis. SPP1 was discussed earlier. VEGF mRNA and its protein were dominantly expressed in the cytoplasm of malignant cells in cervical cancer with positivity of 79% and 70%, respectively.21 Overexpressed cytosolic VEGF was shown to be an independent prognostic factor of overall survival in cervical cancer.21 The third gene, cell division cycle 2 (CDC2), is derived from chromosomal band 10q21.1. Its expression in this set of cervical cancers was increased 2.7-fold compared with that of the normal tissue. Upregulation of CDC2 associated with increased malignancy has previously been reported in colon cancer.34 The fourth interacting gene is CKS2 (the CDC28 protein kinase regulatory subunit 2), which was found to be increased 2.23-fold in cervical cancers as compared to normal tissue. The CKS2 gene was found to be expressed in different patterns through the cell cycle in the cervical cancer cell line HeLa, which reflects the specialized role for this encoded protein. Quantitative PCR experiments confirmed that CKS2 was expressed at significantly higher levels in colon tumors with metastasis.35 These data should help in understanding the progression of cervical tumors and facilitate prediction of their metastatic potential. In this study, we have identified 23 genes that were frequently up or downregulated in cervical cancer infected with HPV16, but not in cancers infected with other oncogenic HPV sub-types. We also identified 5 genes as aberrantly regulated only in nonHPV16-positive cervical cancers. These genes may contribute to a better understanding of the carcinogenesis and progression of cervical tumors that are infected with different types of HPV. We may have found SPP1 upregulation in this set of HPV 16 positive cervical cancer. Particularly, we are interested to know how about the specificity of this type of HPV infection related to abnormal expression of SPP1 in cervical cancer. We may have this answer

when we obtain the information from an undergoing study in a large number of tumor cases. We could not, however, make a comparison in this study between HPV-positive and HPV-negative cervical cancers, as there was only 1 HPV-negative case in this set of cervical tumors. To uncover genes associated with the progression of cervical cancer, we analyzed and compared the gene expression profiles between 15 stage I–II and 14 stage III–IV cervical tumors. The late stage cervical tumors had distinct expression profiles from the early-stage tumors, although there were a number of genes that were similarly aberrantly regulated in early and late-stage cancers. We did identify 2 genes, CTGF (connecting tissue growth factor) and RGS1 (regulator of G protein signaling-1), which were upregulated in late stage cancers as compared to early stage cancers. These 2 genes have not previously been associated with cervical cancer or its progression. However, it had previously been reported that CTGF mRNA levels were significantly higher (p < 0.01) in hepatocellular cancer tissues than in pre-cancerous and normal liver tissues.36 Cox regression analysis showed CTGF expression had a significant correlation with patient survival in patients with primary glioma.37 The RGS1 gene is located within chromosomal band 1q31 and the protein encoded by this gene is localized to the plasma membrane. RGS1 can desensitize a variety of chemotactic receptors, including the receptors for N-formylmethionyl-leucyl-phenylalanine, leukotriene B4 and C5a.38 Demonstration of increased expression of CTGF and RGS1 in late stage cervical cancer should contribute to a better understanding of progression of cervical tumors, and facilitate prediction of their metastatic potential. This does provide important clues as to which genes are most promising for additional evaluation. Furthermore, an overexpressed gene that is not expressed in normal tissues and that encodes for membrane bound proteins holds the most promise for use as specific targets for antibody or cell-mediated vaccines for cervical cancer. RGS1 may thus be a potential target for the development of a cell-mediated vaccine or antibody-based immunotherapy against cervical cancers expressing these antigens. We have, therefore, identified a number of aberrantly regulated genes in cervical cancer. However, the detailed mechanism(s) of differential regulation of the genes identified in this study remains unclear. Certainly, many of the genes are differentially expressed due to altered transcriptional activity, but some may be the result of amplification or deletion since they reside at chromosomal locations that have been previously been reported to be altered in cervical cancer. Several upregulated genes, including MCM2, PLSCR1, ECT2 and PLOD2, are located within the region between 3q21 and 3q26, a chromosomal region previously reported to be amplified in cervical cancer. The downregulated genes IL8RB, SP100, MGC8407, FLJ21511 and CRYAB located at 2q35, 2q37.1, 3p21.31, 4p12-p11 and 11q22.3-q23.1, respectively, all occur at chromosomal locations previously reported to be frequently lost in cervical cancers.39–43 Determining the mechanisms responsible for dysregulation of these genes will require LOH or array CGH analysis on the same specimens subjected to expression profiling. In conclusion, we have identified differences in gene expression between cervical squamous cell carcinoma and normal cervical epithelium in an analysis of a large number of manually macrodissected tumor and normal specimens. We have identified a number of aberrantly regulated genes that were previously described in cervical cancers, and a much larger list of genes that were not previously associated with cervical cancer development. Alterations in gene expression in cervical tumors may yield clues to their pathogenesis. Analysis of a larger number of cervical cancer specimens and correlation with biological phenotypes with gene expression patterns may identify clinically meaningful characteristics of this common gynecologic malignancy. This could provide important clues to develop novel diagnostic and prognostic markers as well as strategies for efficient prevention and therapy for cervical cancer.

GENOME-WIDE GENE EXPRESSION PROFILING OF CERVICAL CANCER IN HONG KONG WOMEN

Acknowledgements We are grateful to the staff of the Mayo Clinic Microarray Core Lab for performing the gene-expression experiments. We also

2469

thank the Gynaecologic Cancer Research Laboratory and Lee Hysan Clinical Research Laboratories, The Chinese University of Hong Kong, for providing reagents and equipment.

References 1. 2. 3. 4. 5. 6. 7. 8.

9.

10. 11. 12.

13.

14.

15. 16. 17. 18. 19. 20.

21.

22. 23. 24.

Hong Kong Cancer Registry (http://www.ha.org.hk/cancerreg). Cancer Incidence and Mortality in Hong Kong (1998–1999). Hospital Authority, Hong Kong. zur Hausen H. Human papillomavirus infections—a major cause of human cancers. Biochem Biophys Acta 1996;1288:55–78. Mayor S. Human papillomavirus classified as carcinogenic. BMJ 1996;313:70. Kieback DG, Fischer D-C. Gene expression profile in endometrioid endometrial carcinoma. Gynecol Oncol 2001;83:175–6. Wadlow R, Ramaswamy S. DNA microarrays in clinical cancer research. Curr Mol Med 2005;5:111–20. Shim C, Zhang W, Rhee CH, Lee JH. Profiling of differentially expressed genes in human primary cancer by complementary DNA expression array. Clin Cancer Res 1998;4:3040–50. Cheng Q, Lau WM, Tay SK, Chew SH, Ho TH, Hui KM. Identification and characterization of genes involved in the carcinogenesis of human squamous cell cervical carcinoma. Int J Cancer 2002;98:419–26. Kitahara O, Katagiri T, Tsunoda T, Harima Y, Nakamura Y. Classification of sensitivity or resistance of cervical cancers to ionizing radiation according to expression profiles of 62 genes selected by cDNA microarray analysis. Neoplasia 2002;4:295–303. Chen Y, Miller C, Mosher R, Zhao X, Deeds J, Morrissey M, Bryant B, Yang D, Meyer R, Cronin F, Gostout BS, Smith-McCune K, et al. Identification of cervical cancer markers by cDNA and tissue microarrays. Cancer Res 2003;63:1927–35. Ahn WS, Bae SM, Lee JM, Namkoong SE, Han SJ, Cho YL, Nam GH, Seo JS, Kim CK, Kim YW. Searching for pathogenic functions to cervical cancer. Gynecol Oncol 2004;93:41–8. International Federation of Obstetrics and Gynecology. News. J Gynecol Obstet 1987;25:87. Lo KWK, Wong YF, Chang MKM, Li JCB, Poon JS, Wnag VW, Zhu SN, Zhang TM, He ZG, Wu QL, Li GD, Tam JSL, et al. Prevalence of human papillomavirus in cervical cancer: a multicenter study in China. Int J Cancer 2002;327–31. Wang JL, Zheng BY, Li XD, Angstrom T, Lindstrom MS, Wallin KL. Predictive significance of the alterations of p16INK4A, p14ARF, p53, and proliferating cell nuclear antigen expression in the progression of cervical cancer. Clin Cancer Res 2004;10:2407–14. Smedts F, Ramaekers F, Link M, Lauerova L, Troyanovsky S, Schijf C, Vooijs GP. Detection of keratin subtypes in routinely processed cervical tissue: implications for tumour classification and the study of cervix cancer aetiology. Virchows Arch 1994;425:145–55. Ishimi Y, Okayasu I, Kato C, Kwon HJ, Kimura H, Yamada K, Song SY. Enhanced expression of Mcm proteins in cancer cells derived from uterine cervix. Eur J Biochem 2003;270:1089–101. Suzuki M, Ohwada M, Tamada T. Thymidylate synthase activity in normal and malignant gynecologic tissues. Asia-Oceania J Obstet Gynaecol 1992;18:379–85. Han AC, Edelson MI, Peralta SA, Knudsen KA, Lifschitz-Mercer B, Czernobilsky B, Rosenblum NG, Salazar H. Cadherin expression in glandular tumors of the cervix. Cancer 2000;89:2053–8. Lloyd JM, O’Dowd T, Driver M, Tee DE. Demonstration of an epitope of the transferrin receptor in human cervical epithelium—a potentially useful cell marker. J Clin Pathol 1984;37:131–5. Mittal KR, Demopoulos RI, Goswami S. Patterns of keratin 19 expression in normal, metaplastic, condylomatous, atrophic, dysplastic, and malignant cervical squamous epithelium. Am J Clin Pathol 1992;98:419–23. Williams GH, Romanowski P, Morris L, Madine M, Mills AD, Stoeber K, Marr J, Laskey RA, Coleman N. Improved cervical smear assessment using antibodies against proteins that regulate DNA replication. Proc Natl Acad Sci USA 1998;95:14932–7. Gaffney DK, Haslam D, Tsodikov A, Hammond E, Seaman J, Holden J, Lee RJ, Zempolich K, Dodson M. Epidermal growth factor receptor (EGFR) and vascular endothelial growth factor (VEGF) negatively affect overall survival in carcinoma of the cervix treated with radiotherapy. Int J Radiat Oncol Biol Phys 2003;56:922–8. Cheng WF, Chen CA, Lee CN, Wei LH, Hsieh FJ, Hsieh CY. Vascular endothelial growth factor and prognosis of cervical carcinoma. Obstet Gynecol 2000;96:721–6. Wu D, Suo Z, Kristensen GB, Li S, Troen G, Holm R, Nesland JM. Prognostic value of EphA2 and EphrinA-1 in squamous cell cervical carcinoma. Gynecol Oncol 2004;94:312–19. Baykal C, Ayhan A, Yuce K, Ayhan A. Overexpression of the c-Met/ HGF receptor and its prognostic significance in uterine cervix carcinomas. Gynecol Oncol 2003;88:123–9.

25. Shimada H, Ikeuchi M, Suwa M, Michimoto T, Ono Y, Tanada S, Himeno K, Hoshino T, Takashima E. Study on the plasma plasminogen activators in patients with malignant gynecologic tumors. Nippon Sanka Fujinka Gakkai Zasshi—Acta Obstet Gynaecol Jap 1989;41: 1353–9. 26. Carrilho C. Alberto M. Buane L. David L. Keratins 8, 10, 13, and 17 are useful markers in the diagnosis of human cervix carcinomas. Hum Pathol 2004;35:546–51. 27. Xu XC, Mitchell MF, Silva E, Jetten A, Lotan R. Decreased expression of retinoic acid receptors, transforming growth factor beta, involucrin, and cornifin in cervical intraepithelial neoplasia. Clin Cancer Res 1999;5:1503–8. 28. Cho YS, Kang JW, Cho M, Cho CW, Lee S, Choe YK, Kim Y, Choi I, Park SN, Kim S, Dinarello CA, Yoon DY. Down modulation of IL18 expression by human papillomavirus type 16 E6 oncogene via binding to IL-18. FEBS Lett 2001;501:139–45. 29. Nigam S, Kumar GS, Sutherland M, Schewe T, Ikawa H, Yamasaki Y, Ueda N, Yamamoto S. Metabolic suppression of platelet-type 12lipoxygenase in human uterine cervix with invasive carcinoma. Int J Cancer 1999;82:827–31. 30. Hatta M, Nagai H, Okino K, Onda M, Yoneyama K, Ohta Y, Nakayama H, Araki T, Emi M. Down-regulation of members of glycolipid-enriched membrane raft gene family, MAL and BENE, in cervical squamous cell cancers. J Obstet Gynaecol Res 2004;30:53–8. 31. Trunk MJ, Dallenbach-Hellweg G, Ridder R, Petry KU, Ikenberg H, Schneider V, von Knebel Doeberitz M. Morphologic characteristics of p16INK4a-positive cells in cervical cytology samples. Acta Cytol 2004;48:771–82. 32. Senger DR, Wirth DF, Hynes RO. Transformed mammalian cells secrete specific proteins and phosphoproteins. Cell 1979;16:885–93. 33. Denhardt DT, Giachelli C, Rittling SR. Role of osteopontin in cellular signaling and toxicant injury. Ann Rev Pharmacol Toxicol 2001;41: 723–49. 34. Goke R, Barth P, Schmidt A, Samans B, Lankat-Buttgereit B. Programmed cell death protein 4 suppresses CDK1/cdc2 via induction of p21(Waf1/Cip1) Am J Physiol Cell Physiol 2004;287:C1541–6. 35. Li M, Lin YM, Hasegawa S, Shimokawa T, Murata K, Kameyama M, Ishikawa O, Katagiri T, Tsunoda T, Nakamura Y, Furukawa Y. Genes associated with liver metastasis of colon cancer, identified by genome-wide cDNA microarray. Int J Oncol 2004;24:305–12. 36. Ding X, Wang W. Expressions of cysteine-rich61, connective tissue growth factor and Nov genes in hepatocellular carcinoma and their clinical significance. World J Gastroenterol 2004;10:3414–18. 37. Xie D, Yin D, Wang HJ, Liu GT, Elashoff R, Black K, Koeffler HP. Levels of expression of CYR61 and CTGF are prognostic for tumor progression and survival of individuals with gliomas. Clin Cancer Res 2004;15:10:2072–81. 38. Denecke B, Meyerdierks A, Bottger EC. RGS1 is expressed in monocytes and acts as a GTPase-activating protein for G-protein-coupled chemoattractant receptors. J Biol Chem 1999;274:26860–8. 39. Aubele M, Zitzelsberger H, Schenck U, Walch A, Hofler H, Werner M. Distinct cytogenetic alterations in squamous intraepithelial lesions of the cervix revealed by laser-assisted microdissection and comparative genomic hybridization. Cancer 1998;84:375–9. 40. Heselmeyer K, Macville M, Schrock E, Blegen H, Hellstrom AC, Shah K, Auer G, Ried T. Advanced-stage cervical carcinomas are defined by a recurrent pattern of chromosomal aberrations revealing high genetic instability and a consistent gain of chromosome arm 3q. Genes Chromosomes Cancer 1997;19:233–40. 41. Kirchhoff M, Rose H, Petersen BL, Maahr J, Gerdes T, Lundsteen C, Bryndorf T, Kryger-Baggesen N, Christensen L, Engelholm SA, Philip J. Comparative genomic hybridization reveals a recurrent pattern of chromosomal aberrations in severe dysplasia/carcinoma in situ of the cervix and in advanced-stage cervical carcinoma. Genes Chromosomes Cancer 1999;24:144–50. 42. Hidalgo A, Schewe C, Petersen S, Salcedo M, Gariglio P, Schluns K, Dietel M, Petersen I. Human papilloma virus status and chromosomal imbalances in primary cervical carcinomas and tumour cell lines. Eur J Cancer 2000;36:542–8. 43. Umayahara K, Numa F, Suehiro Y, Sakata A, Nawata S, Ogata H, Suminami Y, Sakamoto M, Sasaki K, Kato H. Comparative genomic hybridization detects genetic alterations during early stages of cervical cancer progression. Genes Chromosomes Cancer 2002;33:98–102.