Comprehensive gene expression profiling of ...

Endocrine-Related Cancer (2004) 11 843–854

Comprehensive gene expression profiling of anaplastic thyroid cancers with cDNA microarray of 25 344 genes M Onda, M Emi, A Yoshida 1, S Miyamoto, J Akaishi, S Asaka, K Mizutani, K Shimizu 2, M Nagahama 3, K Ito 3, T Tanaka 4,5 and T Tsunoda 4 Department of Molecular Biology, Institute of Gerontology, Nippon Medical School, 1-396, Kosugi-cho, Nakahara-ku, Kawasaki 211-8533, Japan 1 Kanagawa Prefectural Cancer Center, 1-1-2, Nakao, Asahi-ku, Yokohama 241-0815, Japan 2 Department of Surgery, Nippon Medical School and 3Ito Hospital, 4-3-6, Jinguumae, Shibuya-ku, Tokyo 150-8308, Japan 4 Laboratory for Medical Informatics, RIKEN, 22-7-1, Suehiro-cho, Tsurumi-ku, Kanagawa 230-0045, Japan 5 Laboratory of Molecular Medicine, Human Genome Center, Institute of Medical Science, University of Tokyo, Tokyo 108-8639, Japan (Requests for offprints should be addressed to M Emi; Email: [email protected])

Abstract Little is known about the genetic mechanisms of anaplastic thyroid cancer (ATC). This is the most virulent of all human malignancies, and it is believed to result from transformation of differentiated thyroid cancers. To identify a set of genes involved in the development of ATC, we investigated expression profiles of 11 cell lines derived from ATC using a cDNA microarray representing 25 344 genes. Semi-quantitative RT-PCR experiments carried out for some genes that had shown altered expression on the microarray verified frequent over-expression of destrin, HSPA8, stathmin, LDH-A, ATP5A1, PSMB6, B23, HDP-1 and LDH-B, and frequent under-expression of thyroglobulin, PBP and c-FES/FPS genes among the cell lines and also among ten primary ATCs. In addition to mRNA expression studies, up-regulation of GDI2, destrin and stathmin were confirmed with immunohistochemical analysis. The extensive list of genes identified provides valuable information towards understanding the development of ATC, and provides a source of possible biomarkers for diagnosis and/or molecular targets for the development of novel drugs to treat ATC. Endocrine-Related Cancer (2004) 11 843–854

Introduction Thyroid cancers are classified as medullary, papillary, follicular or anaplastic. The medullary type derives from parafollicular C cells; papillary, follicular and anaplastic cancer originate from follicular cells of the thyroid gland. Anaplastic thyroid cancer (ATC) is thought to arise mainly from a background of differentiated (papillary or follicular) cancer, on the basis of clinicopathological observations that ATCs are often accompanied by such cells, and because anaplastic tumors tend to arise in patients who had previously been treated for differentiated cancer of the thyroid (Nakamura et al. 1992, Ozaki et al. 1999, Kitamura et al. 2000). The clinical behavior of ATC is markedly distinct from other types of thyroid cancer. It is one of the most virulent cancers of all human malignancies (Sherman

2003), with a mean survival time among patients of less than 1 year after diagnosis, regardless of treatment (Passler et al. 1999, Voutilainen et al. 1999). Differences in biological characteristics among thyroid tumors might be explained by variations in the pattern of sequential somatic mutations among genes that participate in the mechanisms of growth and differentiation. Although mutation of TP53 (Kitamura et al. 2000) and b-catenin (Garcia-Rostan et al. 1999) are observed in some ATCs, the former probably inactivating a tumor suppressor and the latter activating an oncogenic function, the underlying molecular mechanism involved in this type of thyroid cancer is poorly understood. Using 11 anaplastic cancer cell lines (ACLs) and ten primary ATCs, we investigated gene-expression profiles on a cDNA microarray consisting of 25 344 genes. The

Endocrine-Related Cancer (2004) 11 843–854 1351-0088/04/011–843 # 2004 Society for Endocrinology Printed in Great Britain

DOI:10.1677/erc.1.00818 Online version via http://www.endocrinology-journals.org

Onda et al.: cDNA microarray of anaplastic thyroid cancer tumors displayed remarkably characteristic profiles that should be useful for molecular diagnosis, for prediction of prognosis and for identifying potential target molecules for novel drugs to treat this type of cancer.

photometer, and its quality was checked by formaldehyde-agarose gel electrophoresis.

Preparation of cDNA microarray

Materials and methods ATC cell lines Eleven cell lines that were derived from ATCs, 8305c, 8505c, ARO, FRO, TTA1, TTA2, TTA3, KTA1, KTA2, KTA3 and KTA4, were used for this study. These cell lines were maintained with Dulbecco’s modified Eagle’s medium (Invitrogen, Carlsbad, CA, USA) for 8305c and 8505c, minimum essential medium for ARO and FRO and RPMI 1640 for the other seven lines. All media contained 10% fetal bovine serum but no antibiotics. The cells were cultured in a 378C incubator under 5% CO2 atmosphere.

Patients and specimens Primary ATC and non-cancerous thyroid tissues were excised from ten patients who underwent surgery at the Ito Hospital, Tokyo, Japan; the samples were frozen immediately and stored at 808C. All patients had given informed consent according to guidelines approved by the Institutional Research Board. All tumor specimens that we analyzed contained more than 70% tumor cells. To equalize the tumor-associated deviation in gene expression from normal thyroid, an equal amount of a mixture of RNAs from the ten non-cancerous tissues was used as a normal control for competitive hybridizations on the microarray.

RNA extraction and RNA amplification Each tissue was homogenized with TRIZOL reagent (Invitrogen) according to the manufacturer’s instructions for RNA extraction. One microgram of extracted RNA was electrophoresed on a 3.0% formaldehyde denaturing gel to eliminate degenerated RNA. Samples with 28S/18S ratios greater than 1.5 were selected for subsequent purification using RNeasy kits (QIAGEN, Valencia, CA, USA) to eliminate contamination with DNA. Total RNA was prepared for microarray analysis using T7 RNA polymerase-based amplification with MessageAmp aRNA kits (Ambion, Austin, TX, USA). In the first round, 5 mg aliquots of total RNA were used as templates for amplification, then 2mg aliquots of first-amplified RNA (aRNA) became the templates for second-round amplification. After the second-round amplification, aRNAs were purified with RNeasy purification kits, the amount of each aRNA was measured by a spectro-

844

The microarray contains 27 648 genes and expression sequence tags (ESTs) were generated with Spot Ready DNA for microarray (Amersham Biosciences Corp., Piscataway, NJ, USA). This cDNA panel was spotted using microarray spotter generation III (Amersham Biosciences Corp.) on microarray slide type 7 (Amersham Biosciences Corp.) and it was then cross-linked by u.v.

Labeling of aRNA and competitive hybridization To generate hybridization probes, a 3 mg aliquot of each second-round aRNA was rendered fluorescent with the amino allyl cDNA labeling kit (Ambion), following the manufacturer’s protocol. Probes derived from cell lines and from the normal thyroid gland pool were labeled respectively with Cy5 or Cy3 mono-reactive dye (Amersham International plc, Amersham, Bucks, UK). To eliminate incorporated dye, the labeled probes were cleaned up with QIAquick PCR purification kits (Qiagen). Fluorescent labeled probe (15 pmol) from each cell line was mixed with the Cy3-labeled normal control in 4microarray hybridization buffer (Amersham Biosciences Corp.) and de-ionized formamide. The probe mixtures were hybridized for 12 h at 408C, then washed once with 0.1SSC, 0.2% SDS for 5 min and twice for 10 min in the same washing solution. All procedures were performed with an automated slide processor system (Amersham Biosciences Corp.) After hybridization, fluorescent signals were scanned with GenePix 4000 (Amersham International plc and data were collected by GenePix Pro 3.0 software (Amersham International plc). Scanned signals were normalized by a global method (Manos & Jones 2001, Yang et al. 2002).

Analysis of data We performed random permutation tests to distinguish genes that were expressed differently between ACL and normal thyroid gland (Kitahara et al. 2002). The criteria for selection of discriminating genes were (1) signal/noise ratio of the gene greater than 3.0 in at least ten cell lines, (2) P value in a random permutation test lower than 0.0001 and (3) expression in cancer at least twofold stronger than normal (over-expressed) or half that of normal thyroid tissue (under-expressed). Genes that fitted these criteria were considered significant for discrimination.

www.endocrinology-journals.org

Endocrine-Related Cancer (2004) 11 843–854 Semi-quantitative RT-PCR for ACL and ATC To confirm the microarray results we performed semiquantitative RT-PCR (SQ-PCR) analysis of genes selected according to the following criteria: (1) the P value in permutation tests after microarray analysis was below 0.000001 and (2) expression in ACL was either threefold stronger or one-third weaker than in the normal thyroid gland. In addition to these selected genes, expression of p53 gene was evaluated in ATC samples because p53 is considered to be one of the key genes in ATC carcinogenesis. Five normal thyroid samples served as a control. cDNA was reverse transcribed from 10 mg of each total RNA in the usual manner. To adjust the amount of transcribed cDNA, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was selected as an internal control and SQ-PCR experiments were done as previously described, after adjustment of concentrations (Ono et al. 2000). The primer sequences for GAPDH were 50 -GGAAGGT GAAGGTCGGAGT-30 (forward) and 50 -TGGGTGG AATCATATTGGAA-30 (reverse). Sequence information was collected from the NCBI GenBank (http://www.ncbi.nlm.nih.gov/), and all primers were designed with primer 3 software (http://wwwgenome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi). Information about the PCR primers is available upon request to the corresponding author. SQ-PCR experiments were performed with 1 ml cDNA for the template, 5U Takara EX Taq (Takara, Otsu, Japan), 1PCR buffer (10 mM Tris–HCl, 50 mM KCl and 1.5 mM MgCl2) and reverse primers in 30 ml of total reaction mixture. PCR conditions for each gene were optimized in their respective linear phases of amplification. For evaluation of differences in gene expression between ACL/ATC and normal thyroid gland, 10 ml of each SQ-PCR product was electrophoresed on a 2.0% agarose gel and stained with ethidium bromide. After staining, the density of each sample spot was measured by AlphaImager 3300 (AlphaIonotech, San Leandro, CA, USA) with background revision. A 16 bit imaging score was acquired from each sample. All SQ-PCR experiments were duplicated. We applied Student’s t-tests to the results of the SQ-PCR assay; P values smaller than 0.05 were considered statistically significant. All statistical procedures were archived by Statview version 5.0 software (SAS Institute Inc., Cary, NC, USA).

Immunohistochemical analysis To determine a correlation between RNA expression and protein expression, we performed immunohistochemical analyses using ten paraffin-embedded samples of primary ATC in three selected genes. GDI2 was significantly overexpressed with microarray analysis, stathmin and destrin


were significantly over-expressed with microarray analysis and SQ-PCR in both ACL and ATC cases. All samples were collected at the Ito Hospital, Tokyo, Japan. Antibodies for this assay were available for GDI2 (1:200 dilution; ProteinTech Group, Inc., Chicago, IL, USA), Op-18 (stathmin) (1:100 dilution; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) and destrin/ADF (1:100 dilution; Sigma, St Louis, MO, USA). Antigens were microwaved prior to immunostaining with VECTASTAIN Elite ABC kits (Vector Laboratories Inc., Burlingame, CA, USA) and Dako ENVISION kits (Dako Corporation, Carpinteria, CA, USA) following the manufacturers’ instructions. The sections were counterstained with hematoxylin, and then scanned at low power to identify areas that were evenly stained. Estimates of the numbers of positive cells were scored as follows: negative, 0%; 1, 1–10%; 2, 11–25%; 3, 26–50%; 4, >50% positive (Saiz et al. 2002). Two independent investigators performed the estimation.

Results Over-expressed genes in ACL Permutation tests selected 31 genes and ESTs were upregulated in ACL compared with normal thyroid gland. Those genes, along with their accession numbers, speculated function and chromosomal position are shown in Table 1. Twenty-four of these genes have known or suggested functions. Among the over-expressed group were genes encoding small nuclear ribonucleoprotein, stathmin (Op-18) and DNA topoisomerase III, all apparently related to mechanisms of cell growth, were up-regulated in ACL. On the other hand, metabolismrelated genes such as ATP5A1, ATP synthase and ODC1 were also over-expressed in ACL; however, these result might simply reflect the activated cell dynamics in the immortalized cells. In other categories, several genes encoding ribosomal protein were expressed dominantly in ACL, although a literature search revealed no implied connection between expression levels of those genes and thyroid cancer, especially in ATC, the anaplastic form in particular.

Under-expressed genes in ACL The 56 genes that were under-expressed in ACL in comparison with the normal thyroid gland are listed in Table 2. Fifty-one of these were known genes, with a wide distribution of speculated functions. For example, ubiquitin-activating enzyme E1 is thought to be a tumorsuppressor gene, while the proto-oncogenes encoding cfes/fps and human receptor protein tyrosine phosphatase hPTP-J precursor were unexpectedly under-expressed in

845

Onda et al.: cDNA microarray of anaplastic thyroid cancer Table 1 List of over-expressed genes in ACL Gene name FLJ39361 fis, clone PEBLM 2004733 GD12 GDP dissociation inhibitor 2 Ribosomal protein L9 DSTN-destrin (actin depolymerizing factor) Small nuclear ribonucleoprotein poylpeptide A (SNPRA) IMAGE:1141693 50 similar to gb:X 16869 elongation factor 1-a 1 (eef1a1) Tyrosine 3-mono-oxygenase/tryptophan 5-monooxygenase activation protein, poly peptide Scaffold attachment factor A (SAF-A) IMAGE:47833 30 similar to gb:K00558 tubulin a-1 chain Hums3 mRNA for 40S ribosomal protein s3 HSP A8 heat shock 70 kDa protein 8 Stathmin ELAM-1 ligand fucosyltransferase (ELFT) NPD017 Lactate dehydrogenase-A (LDH-A, EC 1.1.1.27) MRL3 mRNA for ribosomal protein L3 homolog Heterogeneous nuclear ribonucleoprotein complex K Vascular proton-ATPase subunit M9.2 NEDD5: neural precursor cell expressed, developmentally down regulated 5 HL23 ribosomal protein homolog NAD+-specific isocitrate dehydrogenase b subunit precursor (IDH3B) Hums3 mRNA for 40S ribosomal protein s3 ATP5A1 PSMB6, proteasome subunit Y Annexin A2 hB23 gene for B23 nucleophosmin KIAA0683 protein Similar to helix-destabilizing protein (HDP-1) DNA topoisomerase III Lactate dehydrogenase B (LDH-B) Ornithine decarboxylase (ODC1)

ACL. These results suggested either that the proposed functions of the listed gene are variable, or the genes themselves were altered in ATC cell lines. Moreover, genes encoding laminin B2 chain and PLOD3, said to relate to cell structure, were down-regulated in ACL. A gene expression portrait of all 87 genes with altered expression in the cell lines is shown in Fig. 1.

Validation of microarray data and comparison of ACLs with ATCs To validate the microarray results, we performed SQ-PCR experiments using samples from 11 cancer cell lines and ten primary ATC tumors, as well as tissues from five normal thyroid glands, and evaluated gene expression after normalization of signals according to the expression of GAPDH. The genes to be tested by SQ-PCR were

846

Accession number

Function

Position

AK096680 NM_001494 U09953 NM_006870 NM_004596.1

Unknown Miscellaneous Ribosomal protein Cell structure Cell proliferation

5p14 10p15 4p13 20p11.23 19q13.1

AA706503

Unknown

9q34

NM_003406 AF068846 H11622 X55715 BC016660 X53305 M58596 AF271783 NM_005566 P09001 S74678 Y15286

Signal transduction Transcription Unknown Ribosomal protein Miscellaneous Cell proliferation Enzyme Unknown Metabolism Miscellaneous Transcription Miscellaneous

8q23.1 1q44 12q13 11q13.3-q13.5 12q23 1p36.1-p35 11q21 Xq22.1 11p15 3q21-q23 9q21.32-q21.33 5q35.2

NM_004404 X55954

Miscellaneous Ribosomal protein

2q37 17q

U49283 X55715 BT007209 D29012 D00017 M26697 NM_016111 XM_210678 U43431 Y00711 M16650

Miscellaneous Ribosomal protein Metabolism Prosome Miscellaneous Miscellaneous Unknown Unknown Replication Metabolism Metabolism

20p13 11q13.3-q13.5 18q12-q21 17p13 15q21-q22 5q35 16p13.3 6p22.1 17p12-p11.2 12p12.2-p12.1 2p25

chosen according to the criteria of (1) a P value below 0.000001 in the permutation test of microarray results and (2) expression levels in tumor at least threefold stronger or one-third weaker than that of normal thyroid gland tissue, as shown in Materials and methods. Figures 2 and 3 show SQ-PCR results for 12 of the over-expressed genes (Fig. 2A for ACL, Fig. 2B for ATC) and 15 under-expressed genes (Fig. 3A for ACL, Fig. 3B for ATC). All of the results corroborated our microarray data, with statistical significance evaluated by Student’s t-test. Of the 12 genes that were significantly over-expressed in ACL, destrin, HSPA8, stathmin, LDH-A, ATP5A1, PSMB6, B23, HDP-1 and LDH-B were commonly overexpressed both in primary ATC tumors and ACL. NPD017, NAD+ specific isocitrate dehydrogenase-b subunit precursor (IDH3B) and ANXA2 were overexpressed only in cell lines. Of the 15 significantly


Carcinoembryonic antigen (CGM2) Cell cycle-regulated factor p78 Proliferating-cell nucleolar protein P120 Splicing factor, arginine/serine-rich 7 (SFRS7) M-phase phosphoprotein, mpp5 KH type splicing regulatory protein (KSRP) Aquaporin Laminin B2 chain lysylhydoxylase isoform 3 (PLOD3) CD34 Histone deacetylase 3 (HDAC3) Estrogen sulfotransferase (SULTIA3) Methylmalonate semialdehyde dehydrogenase (M M SDH) NAD (H)-specific isocitrate dehydrogenase g subunit (IDH3G) Heat shock protein (hsp 70) gene Thyroglobulin (TG) IgG Fc receptor hFcRn (FCGRT) Arg/Abl-interacting protein ArgBP2a (ArgBP2a) Protein tyrosine kinase related mRNA sequence Phosphatidylethanolamine-binding protein (PBP) Glucose-6-phosphate dehydrogenase Multiple exostosis-like protein (EXTL) Carbonic anhydrase precursor (CA 12) Acyl-CoA thioester hydrolase Clones 23667 and 23775 zinc finger protein WD repeat protein HAN11 (AN11) TTF-I interacting peptide 20 ADP-ribosylation factor nkat9 ATP-driven ion pump (ATP1AL1) Secretory protein (P1.B) Neuronal nitric oxide synthase (NOS1) Sex hormone-binding globulin (SHBG) Metaxin 1 (MTX1) Butyrophilin (BTF2) Parathymosin HBB hemoglobin, b Receptor protein tyrosine phosphatase hPTP-J precursor c-fes/fps proto-oncogene (FES) Serine-threonine phosphatase (PP5) PSMC1 proteasome 26S subunit, ATPase, 1 26S proteasome subunit p97

Gene name

Table 2 List of under-expressed genes in ACL

L31792 AF068007 M32110 NM_006276 X98261 U94832 D63412 J03202 AF046889 S53911 U66914 L25275 M93405 U69268 NP_005336 X05615 U12255 AF049884 L05148 X75252 M24470 U67191 AF037335 U91316 U90919 U94747 AF000560 NM_001661 L76672 U02076 L15203 U17327 X05403 NM_002455 U90550 M24398 AF117710 U73727 X06292 U25174 NM_002802 D78151

Accession number Cell adhesion Cell cycle Cell proliferation Cell proliferation Cell proliferation Cell proliferation Cell structure Cell structure Cell structure Cell surface antigen Chromosome structure Enzyme Enzyme Enzyme Heat shock protein Hormone Immune response Kinase Kinase Kinase inhibitor Metabolism Metabolism Metabolism Metabolism Miscellaneous Miscellaneous Miscellaneous Miscellaneous Miscellaneous Miscellaneous Miscellaneous Miscellaneous Miscellaneous Miscellaneous Miscellaneous Miscellaneous Miscellaneous Oncogene Oncogene Phosphatase Prosome Proteasome

Function

19q13.2 12q13.12 12p13 2p22 10q11 19p13.3 18q11.2-q12.1 1q31 7q22 1q32 5q31 16p11.2 14q24.3 Xq28 6p21.3 8q24.2 19q13.3 4q35.1 2q12 12q24.23 Xq28 1p36.1 15q22 1p36.31-p36.11 14q24.3 17q24.1 19q13.4 17q12 19q13.4 13q12.1-q12.3 21q22.3 12q24.2-q24.31 17p13-p12 1q21 6p22.1 17q12-q22 11p15 1p35.3-p35.1 15q26.1 19q13.3 14q32.11 Chr.3

Position



847

848 X64707 U89326 M84739 U64105 U78798 X63679 X04450 M30496 L13852 W69766 AL137302 NM_021941 AL035364 AI160936

Accession number

Gene in bold locates on chromosomal region where loss of heterozygosity was previously reported in ATC.

BBC1 Bone morphogenetic protein receptor type I (ALK-6) Autoantigen calreticulin (CALR) Guanine nucleotide exchange factor p115-RhoGEF (ARHGEF1) TNF receptor associated factor 6 (TRAF6) TRAMP protein CD1a antigen (CD1a gene) UCHL3 ubiquitin carboxyl-terminal esterase L3 (ubiquitin thiol esterase) Ubiquitin-activating enzyme E1-related protein IMAGE:343903 TEX27 testis expressed sequence 27 C21orf97 chromosome 21 open reading frame 97 HS747E2A hypothetical protein HS747E2A IMAGE:1715637 30 similar to gb:J00117 choriogonadotropin b chain precursor

Gene name

Table 2 continued

Ribosomal protein Signal transduction Signal transduction Signal transduction Signal transduction Signal transduction Tumor antigen Ubiquitination Ubiquitination, tumor suppressor Unknown Unknown Unknown Unknown Unknown

Function

16q24.3 4q23-q24 19p13.2 19q13.13 11p12 Chr.8 1q21-q23 13q21.33 3p21 16q22 6pter-p22.3 21q22 22q12.1 19q13.3

Position

Onda et al.: cDNA microarray of anaplastic thyroid cancer



Genes AK096680 NM_001494 NM_006870 U09953 R25535 AA706503 M86400 AF068846 H11622 X55715 BC016660 X53305 M58596 AF271783 NM_005566 H05820 S74678 Y15286 NM_004404 X55954 U49283 X55715 BT007209 D29012 BT007432 M26697 NM_016111 XM_210678 U43431 Y00711 M16650 X05615 U89326 U90919 M32110 R39171 M84739 U12255 X04450 D63412 L25275 U94747 L13852 X75252 AF049884 NM_002802 M30496 U64105 U73727 X98261 W69766 X64707 L05148 L31792 M24470 M93405 U25174 H06378 D78151 U78798 X06292 AF000560 U66914 S53911 U69268 U67191 AF068007 AF037335 J03202 H28951 L76672 AF046889 AL137302 X63679 U91316 U02076 U17327 NM_021941 L15203 U94832 X05403 AL035364 NM_002455 AI160936 U90550 M24398 AF117710

Cell Line

8305c 8505c ARO FRO TTA1 TTA2 TTA3 KTA1 KTA2 KTA3 KTA4 Under-expressed genes LOW

Over-expressed genes HIGH

Expression level Figure 1 Expression portrait of 87 genes with significantly altered expression in ACL, according to microarray analysis; 31 of them were up-regulated and 56 were down-regulated in the cell lines. The signal indicator reflects the expression intensity of each gene in each ACL sample. The darkest red at the right-hand end of the bar indicates expression in ACL that is three times as strong as that in normal thyroid; the darkest green at the far left of the bar indicates a level of expression a third of normal thyroid.

under-expressed genes in ACL, thyroglobulin, PBP and cfes/fps proto-oncogene, were also under-expressed in primary tumor. Down-regulation of calreticulin (CALR), hFcRn (FCGRT), estrogen sulfotransferase (SULT1A), HAN11, p115-RhoGEF (ARHGEF1), methylmalonate semialdehyde dehydrogenase (MMSDH), TRAF6, TTF-1, CD34, NAD (H)-specific isocitrate dehydrogenase-g subunit (IDH3G), C21orf97 and sex hormone-binding globulin (SHBG), were limited to cancer cell lines. With regard to the p53 gene, the expression was slightly decreased (0.92-fold) in ACLs but it did not reach statistical significance with microarray analysis. We also tested p53 gene expression in ten ATC samples with SQPCR. The decreased expression was seen in ATC compared with normal thyroid tissue ðP ¼ 0:03Þ (Fig. 4).


Immunohistochemical analysis for GDI2, destrin and stathmin To evaluate the correlation between microarray profiling and protein expression, we performed immunohistochemical analysis of three antigenic proteins for which antibodies were commercially available (GDI2, stathmin and destrin). GDI2 was highly expressed (scored over 3) in seven of the ten primary ATC tumors. Stathmin (Op18) was also highly expressed in primary ATCs, with six tumors scoring over 3. Destrin was stained in five cases, mainly in cytoplasm. The representative results of these experiments are shown in Fig. 5. On the other hand, all genes showed lower expression (staining) in part of the normal thyroid tissue. Table 3 summarizes the results of immunostaining analysis of GDI2, stathmin and destrin in

849

Onda et al.: cDNA microarray of anaplastic thyroid cancer Normal M N1 N2 N3 N4 N5 1 2

ACL 3 4 5

Normal

6 7

8 9 10 11

ACL

M N1 N2 N3 N4 N5 1 2 3 4 5

6 7 8 9 10 11

DSTN

p=1.97E-9

TG

p=2.7E-16

HSPA8

p=6.67E-8

CALR

p=1.62E-6

Stathmin

p=2.5E-6

FCGRT

p=1.06E-6

NPD017

p=5.47E-5

LDH-A

p=2.34E-8

IDH3B

p=0.0014

ATP5A1

p=9.81E-10

PSMB6

p=3.6E-6

ANXA2

p=1.87E-11

B23

p=8.11E-11

HDP-1

p=1.85E-9

LDH-B

p=5.7E-6

SULT1A3

p=8.76E-5

HAN11

p=0.0043

PBP

p=9.08E-14

ARHGEF1

p=0.05

MMSDH

p=5.14E-8

TRAF6

p=0.0032

FES

p=2.23E-8

TTF-1

p=2.3E-6

CD34

p=9.73E-10

IDH3G

p=2.35E-5

GAPDH

Normal M N1 N2 N3 N4 N5 1

C21orf97

p=0.00014

SHBG

p=1.05E-5

ATC 2

3

4

5

6

7

8

9

GAPDH

10

DSTN

p=0.04

HSPA8

p=0.0004

Stathmin

p=0.02

LDH-A

p=0.001

ATP5A1

p=0.006

PSM B6

p=0.0003

B23

p=0.001

HDP-1

p=0.001

LDH-B

p=0.005

Normal M N1 N2 N3 N4 N5 1 2

ATC 3 4 5 6

7

8

9 10

TG

p=0.0004

PBP

p=0.0004

FES

p=0.008

GAPDH

GAPDH

Figure 2 (A) Confirmation by SQ-PCR of up-regulation of 12 genes in ACL. Expression of GAPDH constituted an internal control. The intensity of each sample was measured and evaluated with a 16 bit image and adjustment of background. P values are the results of t-tests. M, size marker; N1-N5, normal thyroid; lane 1, cell line 8305c; lane 2, 8505c; lane 3, ARO; lane 4, FRO; lane 5, TTA1; lane 6, TTA2; lane 7, TTA3; lane 8, KTA1; lane 9, KTA2; lane 10, KTA3; lane 11, KTA4. (B) SQ-PCR results in primary ATCs. Nine of the 12 genes overexpressed in ACL were also significantly over-expressed in all of the primary tumors; P values are the results of t-tests. Ex, 10x.

all ten primary ATC cases. Consistent correlation between cDNA microarray data, SQ-PCR data and immunohistochemistry provides solid verification for the present

850

Figure 3 (A) SQ-PCR assay of 15 genes that were significantly under-expressed in ACL, using GAPDH as an internal control. The intensity of each sample was measured and evaluated with a 16 bit image and adjustment of background; P values shown are the result of t-tests. M, size marker; N1-N5, normal thyroid; lane 1, cell line 8305c; lane 2, 8505c; lane 3, ARO; lane 4, FRO; lane 5, TTA1; lane 6, TTA2; lane 7, TTA3; lane 8, KTA1; lane 9, KTA2; lane 10, KTA3; lane 11, KTA4. (B) SQ-PCR images showing down-regulation of three genes in primary ATCs. Ex, 10x.

study. In addition, this is the first report of over-expression in RNA and protein levels of these three genes in human thyroid malignancies as far as we can find in the literature.

Discussion Among human cancers, ATC is one of the most aggressive and has the highest potential for malignancy; at present,


Endocrine-Related Cancer (2004) 11 843–854 Normal M N1 N2 N3 N4 N5 1

p53

ATC 2

3

4

5

6

7

8

9 10 P=0.03

GAPDH

Figure 4 Results of SQ-PCR experiment for p53 gene expression analysis with ATC samples. Expression of p53 in ATC was weakly decreased compared with normal thyroid tissue; P ¼ 0:03 (Student’s t-test).

the 1-year survival rate among patients with this disease is only a few percent (Passler et al. 1999, Voutilainen et al. 1999). Although several clinical approaches have been tried, no effective therapeutic strategy has yet been established for ATC. Few studies on the molecular aspect of this disease have been done because ideal biological material is hard to obtain, since most ATC tumors are not subject to surgical intervention. For this study we did have access to 11 established ACLs for analysis on a cDNA microarray, which allowed us to construct a molecular portrait of ATC-derived cells. This is the first report to document a comprehensive gene expression profile of ATC. At present, p53 is the most well-known gene that might be responsible for ATC carcinogenesis (Kitamura et al. 2000); however, the expression level of p53 did not alter significantly between ACL and normal thyroid tissue with microarray analysis. In ATC samples, expression of p53 was weakly decreased. In addition, point mutations of p53 were found in five ACLs (TTA1: codon 72, CGC (wild type)-CTC (mutant); KTA2: codon 158, AAC-GAC; 8305c: codon 248, CGGGGG; ARO: codon 273, CGT-CAT; KTA3: codon 276, GCC-CCC) All mutations were missense mutations but there was no specific mutation spectrum in ACLs. Hence it is difficult to explain the carcinogenic mechanism of only p53 single gene alterations. It is therefore necessary to examine gene expression alterations widely through the human genome to discover the novel genes responsible for ACL/ATC carcinogenesis. Although some discrepancies are likely to exist between cell lines and primary ATCs, in general cell lines are thought to reflect, to a large degree, the characteristics of their tumors of origin; for example, a relationship between gene expression profiles of cell lines and primary tumor has been confirmed in bladder cancer (Sanchez-Carbayo et al. 2002). Our microarray analysis identified the up-regulation of 31 genes in the panel of 11 ACLs. Some of the over-expressed genes encoded ribosomal proteins, and several others encoded metabolism-related proteins such as lactate dehydrogenase A, B or ATP5A1 (ATP synthase); however, these may have been over-expressed simply as a result of active dynamism


of the immortalized cells. Among the 31 genes in this list, GDI2 (located on 10p15) binds and solubilizes several membrane-associated Rab proteins in a GDP/GTPdependent manner (Chinni et al. 1998). Amplification of chromosome 10p has been observed in a small number of head and neck cancers (Speicher et al. 1995). Our study showed up-regulation and over-expression of GDI2 in both ACL and ATC, suggesting that GDI2 contributes to carcinogenesis of ATC. The gene encoding destrin behaved similarly in our experiments. Destrin is important for actin remodeling, endocytosis, polarized cell growth and cellular activation (Moriyama et al. 1990, Yahara et al. 1996). Chromosome 20p, where the destrin gene locates, is often amplified in ovarian cancer cell lines (Watanabe et al. 2001). Our results suggest that destrin might be a previously unsuspected participant in carcinogenesis. For its part, stathmin (Op18) is a member of a novel class of microtubule-destabilizing proteins that regulate the dynamics of microtubule polymerization and depolymerization (Mistry & Atweh 2002). Stathmin protein appears to have oncogenic potential, because it is widely expressed in various kinds of human cancers, including leukemia (Ghosh et al. 1993), prostate cancer (Friedrich et al. 1995) and breast cancer (Curmi et al. 2000), and because inhibition of stathmin can decrease the rate of proliferation of K562 erythroleukemic cells (Luo et al. 1994). In the clinical setting, some anti-cancer drug regimens are designed to inhibit microtubule assembly and arrest cells in mitosis, or to promote assembly of microtubules and stabilize tubulin polymers by preventing their depolymerization (Mistry & Atweh 2002). Over-expression of annexin II (15q21-22) reflects poor prognosis of colorectal and gastric cancers (Emoto et al. 2001a,b). Our results appear to corroborate an oncogenic role of annexin II in ATC. On the other hand, 56 genes, including hypothetical protein and ESTs, were significantly down-regulated in ACLs on our microarray. Although one of them, thyroglobulin, serves as a tumor marker in differentiated thyroid cancers (Hoang-Vu et al. 1992), its expression was significantly decreased in ACL. Expression of TTF-1 (interacting peptide 20) also decreased in this study; the TTF-1 gene encodes a transcription factor that contributes to expression of thyroid-specific proteins like thyroglobulin (Fabbro et al. 1994). These results implied that a drastic abolition of normal thyroidal function had occurred in the tissue — giving rise to the parent ATCs, and suggested that thyroglobulin cannot be used as a marker for anaplastic thyroid tumors. PBP was also under-expressed. PBP, alternatively known as Rafmediated activation inhibitor protein (RKIP), is expressed in prostate cancers but not in metastatic foci derived from

851

Onda et al.: cDNA microarray of anaplastic thyroid cancer

GDI2

Stathmin

X 200

X 200

Destrin

X 200 Figure 5 Representative images from immunohistochemical analyses of GDI2, stathmin and destrin, confirming over-expression of these proteins in primary ATC materials. GDI2 protein was diffuse in cytoplasm; stathmin was expressed in cytoplasm and the periphery of nuclei; destrin was expressed only in cytoplasm.

those tumors. In other reports, over-expression of RKIP in C4-2B cells decreased cell invasion in vitro and inhibited lung metastasis (Fu et al. 2003). Thus PBP has shown a tumor-suppressive function in prostate cancer; our data suggest that PBP might be a tumor suppressor for human ATC as well. Of the 56 under-expressed genes listed in Table 2, 22 genes are located in chromosomal regions where we previously detected LOH in > 20% of informative ATCs. In particular, CD34 (1q32), SHBG

(17p13-p12) and autoantigen CALR (19p13.2) locate on the chromosomal position which showed frequent LOH in ATC (40%, 44% and 36% respectively) (Kitamura et al. 2000). These genes should be investigated as to potential function of tumor suppression in thyroid tissue. Cell lines generally reflect the character of their tumors of origin but it is always possible for the nature of a cell line to change during immortalization. It is therefore necessary to evaluate expression profiles of primary tumors as well,

Table 3 Summary of immunohistochemical analysis. Numbers show the staining score Case ATH-1 ATH-2 ATH-3 ATH-4 ATH-5 ATH-6 ATH-7 ATH-10 ATH-11 ATH-13

GD12

Stathmin

Destrin

2 4 Negative 4 1 3 4 4 3 2

3 3 1 4 1 1 4 4 2 4

4 4 2 3 1 1 1 4 3 2

Percentage of positive staining cells; 0%, negative; 1–10%, 1; 11–25%, 2; 26–50%, 3; > 50% positive, 4.

852


Endocrine-Related Cancer (2004) 11 843–854 but because ATC is so aggressive that there is little chance for surgical intervention which would provide fresh tissue for analysis. For the study reported here we did have access to ten precious ATC samples, and we were able to perform SQ-PCR with selected genes. Of 12 genes that had been significantly over-expressed in ACL, nine (destrin, HSPA8, stathmin, LDH-A, ATP5A1, PSMB6, B23, HDP-1 and LDH-B) were also over-expressed in primary ATCs. Of those, LDH-A and -B are isoenzymes, and LDH-A is reported to increase in tumors of all origins (Liu et al. 2003). Generally speaking, up-regulation of metabolic enzymes such as LDH-A or -B and ATP5A1 is likely to be the result of high metabolic activity of ACL, not a cause of carcinogenesis. B23 (alternatively NPM1) is a nucleolar phosphoprotein that is more abundant in tumors than in normal cells. Functionally, it relates to chromosomal translocation, and its over-expression has been confirmed in acute myeloid leukemia by microarray analysis (Jhanwar et al. 1984). HSPA8, PSMB6 and HDP-1, located at 12q23, 17p13 and 6p22.1 respectively, do not have any known functions, especially in cancer. On the other hand, upregulation of NPD017, IDH3B and ANXA2 was found in ACLs only, suggesting that the activating changes might have been acquired in the process of immortalization. Of the 15 genes that were selected from the list of under-expressed elements in ACL only three (thyroglobulin, PBP and FES) were under-expressed in primary tumors as well. Down-regulation of thyroglobulin might reflect the loss of normal thyroidal function in ATC. As the gene encoding PBP locates at 12q24-23, where LOH is frequent in ATCs (Kitamura et al. 2000), the combined evidence suggests that PBP might have tumor-suppressive function in thyroid tissue. The FES gene, at 15q26.1 (Lionberger & Smithgall 2000), exhibits strong expression in hematopoietic cells of the myeloid lineage and may regulate chronic myelogenous leukemia (Lionberger & Smithgall 2000). The meaning of the down-regulation of this apparent proto-oncogene in ACL and ATC is unclear; however, it might have a novel function in ATC carcinogenesis apart from the one it has in other situations. The work reported here has yielded useful information for understanding the molecular mechanism(s) involved in ATC, and has revealed several novel genetic alterations that could be responsible for ATC carcinogenesis. The cause of up-regulation or down-regulation of these genes, and genetic or epigenetic change should be determined. Further study is required.

Acknowledgements We thank N Tsuruta, K Shimizu, M Tanaka and J Sato for excellent secretarial assistance. We also thank Mrs R


Foltz for critical reading of this manuscript. Cell lines 8305c and 8505c were obtained from the Japanese Collection of Research Bioresources Cell Bank. ARO and FRO, established by Dr G J F Juillard (University of California–Los Angeles, Los Angeles, CA, USA), were kindly given by Dr H Nanba (Nagasaki University, Nagasaki, Japan). The others cell lines were established by A Y. This work was supported by special grants for Strategic Advanced Research on Cancer from the Ministry of Education, Science, Sports and Culture of Japan and by a Research for the Future Program Grant of the Japan Society for the Promotion of Science.

References Chinni SR, Brenz M & Shisheva A 1998 Modulation of GDPdissociation inhibitor protein membrane retention by the cellular redox state in adipocytes. Experimental Cell Research 242 373–380. Curmi PA, Nogues C, Lachkar S, Carelle N, Gonthier MP, Sobel A, Lidereau R & Bieche I 2000 Overexpression of stathmin in breast carcinomas points out highly proliferative tumours. British Journal of Cancer 82 142–150. Emoto K, Sawada H, Yamada Y, Fujimoto H, Takahama Y, Ueno M, Takayama T, Uchida H, Kamada K, Naito A et al. 2001a Annexin II overexpression is correlated with poor prognosis in human gastric carcinoma. Anticancer Research 21 1339–1345. Emoto K, Yamada Y, Sawada H, Fujimoto H, Ueno M, Takayama T, Kamada K, Naito A, Hirao S & Nakajima Y 2001b Annexin II overexpression correlates with stromal tenascin-C overexpression: a prognostic marker in colorectal carcinoma. Cancer 92 1419–1426. Fabbro D, Di Loreto C, Beltrami CA, Belfiore A, Di Lauro R & Damante G 1994 Expression of thyroid-specific transcription factors TTF-1 and PAX-8 in human thyroid neoplasms. Cancer Research 54 4744–4749. Friedrich B, Gronberg H, Landstrom M, Gullberg M & Bergh A 1995 Differentiation-stage specific expression of oncoprotein 18 in human and rat prostatic adenocarcinoma. Prostate 27 102–109. Fu Z, Smith PC, Zhang L, Rubin MA, Dunn RL, Yao Z & Keller ET 2003 Effects of raf kinase inhibitor protein expression on suppression of prostate cancer metastasis. Journal of the National Cancer Institute 95 878–889. Garcia-Rostan G, Tallini G, Herrero A, D’Aquila TG, Carcangiu ML & Rimm DL 1999 Frequent mutation and nuclear localization of beta-catenin in anaplastic thyroid carcinoma. Cancer Research 59 1811–1815. Ghosh PK, Anderson J, Cohen N, Takeshita K, Atweh GF & Lebowitz P 1993 Expression of the leukemia-associated gene, p18, in normal and malignant tissues; inactivation of expression in a patient with cleaved B cell lymphoma/ leukemia. Oncogene 8 2869–2872. Hoang-Vu C, Dralle H, Scheumann G, Maenhaut C, Horn R, von zur Muhlen A & Brabant G 1992 Gene expression of differentiation- and dedifferentiation markers in normal and

853

Onda et al.: cDNA microarray of anaplastic thyroid cancer malignant human thyroid tissues. Experimental and Clinical Endocrinology 100 51–56. Jhanwar SC, Neel BG, Hayward WS & Chaganti RS 1984 Localization of the cellular oncogenes ABL, SIS, and FES on human germ-line chromosomes. Cytogenetics and Cell Genetics 38 73–75. Kitahara O, Katagiri T, Tsunoda T, Harima Y & Nakamura Y 2002 Classification of sensitivity or resistance of cervical cancers to ionizing radiation according to expression profiles of 62 genes selected by cDNA microarray analysis. Neoplasia 4 295–303. Kitamura Y, Shimizu K, Tanaka S, Ito K & Emi M 2000 Allelotyping of anaplastic thyroid carcinoma: frequent allelic losses on 1q, 9p, 11, 17, 19p, and 22q. Genes, Chromosomes and Cancer 27 244–251. Lionberger JM & Smithgall TE 2000 The c-Fes protein-tyrosine kinase suppresses cytokine-independent outgrowth of myeloid leukemia cells induced by Bcr-Abl. Cancer Research 60 1097–1103. Liu ZJ, Peng WC, Yang X, Huang JF, Zhang XB, Zhang Y & Maekawa M 2003 Relative mRNA expression of the lactate dehydrogenase A and B subunits as determined by simultaneous amplification and single strand conformation polymorphism. Relation with subunit enzyme activity. Journal of Chromatography B Analytical Technologies in the Biomedical and Life Sciences 793 405–412. Luo XN, Mookerjee B, Ferrari A, Mistry S & Atweh GF 1994 Regulation of phosphoprotein p18 in leukemic cells. Cell cycle regulated phosphorylation by p34cdc2 kinase. Journal of Biological Chemistry 269 10312–10318. Manos EJ & Jones DA 2001 Assessment of tumor necrosis factor receptor and Fas signaling pathways by transcriptional profiling. Cancer Research 61 433–438. Mistry SJ & Atweh GF 2002 Role of stathmin in the regulation of the mitotic spindle: potential applications in cancer therapy. Mt Sinai Journal of Medicine 69 299–304. Moriyama K, Nishida E, Yonezawa N, Sakai H, Matsumoto S, Iida K & Yahara I 1990 Destrin, a mammalian actindepolymerizing protein, is closely related to cofilin. Cloning and expression of porcine brain destrin cDNA. Journal of Biological Chemistry 265 5768–5773. Nakamura T, Yana I, Kobayashi T, Shin E, Karakawa K, Fujita S, Miya A, Mori T, Nishisho I & Takai S 1992 p53 gene mutations associated with anaplastic transformation of

854

human thyroid carcinomas. Japanese Journal of Cancer Research 83 1293–1298. Ono K, Tanaka T, Tsunoda T, Kitahara O, Kihara C, Okamoto A, Ochiai K, Takagi T & Nakamura Y 2000 Identification by cDNA microarray of genes involved in ovarian carcinogenesis. Cancer Research 60 5007–5011. Ozaki O, Ito K, Mimura T & Sugino K 1999 Anaplastic transformation of papillary thyroid carcinoma in recurrent disease in regional lymph nodes: a histologic and immunohistochemical study. Journal of Surgical Oncology 70 45–48. Passler C, Scheuba C, Prager G, Kaserer K, Flores JA, Vierhapper H & Niederle B 1999 Anaplastic (undifferentiated) thyroid carcinoma (ATC). A retrospective analysis. Langenbeck’s Archives of Surgery 384 284–293. Saiz AD, Olvera M, Rezk S, Florentine BA, McCourty A & Brynes RK 2002 Immunohistochemical expression of cyclin D1, E2F-1, and Ki-67 in benign and malignant thyroid lesions. Journal of Pathology 198 157–162. Sanchez-Carbayo M, Socci ND, Charytonowicz E, Lu M, Prystowsky M, Childs G & Cordon-Cardo C 2002 Molecular profiling of bladder cancer using cDNA microarrays: defining histogenesis and biological phenotypes. Cancer Research 62 6973–6980. Sherman SI 2003. Thyroid carcinoma. Lancet 361 501–511. Speicher MR, Howe C, Crotty P, du Manoir S, Costa J & Ward DC 1995 Comparative genomic hybridization detects novel deletions and amplifications in head and neck squamous cell carcinomas. Cancer Research 55 1010–1013. Voutilainen PE, Multanen M, Haapiainen RK, Leppaniemi AK & Sivula AH 1999 Anaplastic thyroid carcinoma survival. World Journal of Surgery 23 975–978; discussion 978–979. Watanabe T, Imoto I, Kosugi Y, Ishiwata I, Inoue S, Takayama M, Sato A & Inazawa J 2001 A novel amplification at 17q2123 in ovarian cancer cell lines detected by comparative genomic hybridization. Gynecologic Oncology 81 172–177. Yahara I, Aizawa H, Moriyama K, Iida K, Yonezawa N, Nishida E, Hatanaka H & Inagaki F 1996 A role of cofilin/destrin in reorganization of actin cytoskeleton in response to stresses and cell stimuli. Cell Structure and Function 21 421–424. Yang YH, Dudoit S, Luu P, Lin DM, Peng V, Ngai J & Speed TP 2002 Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Research 30 e15.
