Skip to main content
NCBI home page
The American Journal of Pathology logoLink to The American Journal of Pathology
. 2005 Jul;167(1):223–231. doi: 10.1016/s0002-9440(10)62967-7

Follicular Thyroid Tumors with the PAX8-PPARγ1 Rearrangement Display Characteristic Genetic Alterations

Ludovic Lacroix *†, Vladimir Lazar *, Stefan Michiels *‡, Hugues Ripoche §, Philippe Dessen §, Monique Talbot †, Bernard Caillou , Jean-Pierre Levillain ||, Martin Schlumberger †**, Jean-Michel Bidart †||
PMCID: PMC1603430  PMID: 15972966

Abstract

Follicular thyroid carcinomas (FTC) arise through oncogenic pathways distinct from those involved in the papillary histotype. Recently, a t(2;3)(q13;p25) rearrangement, which juxtaposes the thyroid transcription factor PAX8 to the peroxisome proliferator-activated receptor (PPAR) γ1, was described in FTCs. In this report, we describe gene expression in 11 normal tissues, 4 adenomas, and 8 FTCs, with or without the PAX8-PPARγ1 translocation, using custom 60-mer oligonucleotide microarrays. Results were confirmed by quantitative real-time polymerase chain reaction of 65 thyroid tissues and by immunohistochemistry. Statistical analysis revealed a pattern of 93 genes discriminating FTCs, with or without the translocation, that were morphologically undistinguishable. Although the expression of thyroid-specific genes was detectable, none appeared to be differentially regulated between tumors with or without the translocation. Differentially expressed genes included genes related to lipid/glucose/amino acid metabolism, tumorigenesis, and angiogenesis. Surprisingly, several PPARγ target genes were up-regulated in PAX8-PPARγ-positive FTCs such as angiopoietin-like 4 and aquaporin 7. Moreover many genes involved in PAX8-PPARγ expression profile presented a putative PPARγ-promoter site, compatible with a direct activity of the fusion product. These data identify several differentially expressed genes, such as FGD3, that may serve as potential targets of PPARγ and as members of novel molecular pathways involved in the development of thyroid carcinomas.

Follicular cell-derived thyroid tumors include benign adenomas, papillary and follicular carcinomas, two entities considered as differentiated carcinomas, and anaplastic carcinomas. Although clinically benign thyroid nodules are common in the general population, thyroid carcinomas are infrequent tumors, and both the putative relationships between adenomas and carcinomas and the mechanisms of thyroid oncogenesis are not clearly understood.1

In thyroid tumors, genetic abnormalities have been largely substantiated. Fusion of the tyrosine-kinase domain of the RET gene and the 5′ domain of various genes (RET/PTC) and activating mutation of the B type Raf kinase (BRAF) gene constitute the two major independent and nonoverlapping genetic alterations detected in papillary carcinomas.2,3 A high prevalence of activating mutations of all three RAS genes has been reported in follicular neoplasms.4 Chromosomal imbalances are also frequent in FTCs;5 and the t(2;3)(q13;p25) rearrangement, yielding a PAX8-PPARγ1 fusion gene, is found in 10 to 63% FTCs,6 but also in some follicular adenomas.7–9 These observations suggest that FTC proceeds from at least two distinct oncogenic events, namely RAS mutations and PAX8-PPARγ1 rearrangement.10

Peroxisome proliferator-activated receptor γ (PPARγ) belongs to the nuclear hormone receptor superfamily and plays a critical role in the differentiation of adipocytes and in the regulation of fat metabolism.11 The involvement of PPARγ in the development of tumors, including thyroid carcinomas, is still debated.12,13 The anti-proliferative effects of PPARγ agonists have been demonstrated in thyroid carcinoma cell lines14,15 and either induce cell-cycle arrest or promote cell-death. Furthermore, thyroid tumors not harboring PAX8-PPARγ translocation displayed decreased PPARγ gene expression,16–18 and this suggests that decreased activity of PPARγ might contribute to carcinogenesis. In FTCs, the PAX8-PPARγ1 fusion oncogene appears to act through a dominant-negative effect on the transcriptional activity of wild-type PPARγ1, inhibiting agonist-induced transactivation.6 Recently, in vitro experiments demonstrated that the fusion oncoprotein contributes to the malignant transformation by acting on several pathways, some of which are normally regulated by PPARγ.19

To explore the physiopathological mechanisms associated to the presence of the PAX8-PPARγ1 rearrangement, we analyzed gene profiles of normal thyroid tissues, follicular adenomas and follicular thyroid carcinomas using custom-designed 60-mer oligonucleotides microarrays of 22,000 features, representing about 17,000 distinct genes. Transcriptional changes associated with the presence or the absence of the PAX8-PPARγ1 rearrangement revealed a 93-gene discriminating pattern. Changes in the expression of several genes of this molecular signature were confirmed by real-time quantitative real-time polymerase chain reaction (Q RT-PCR) in a large series of thyroid tissues and at the protein level, by immunohistochemistry on a tissue array.

Materials and Methods

Tissue Samples

Sixty-five thyroid tissue samples were selected after histological analysis and classified according to World Health Organization recommendations (Table 1).20 Normal contralateral thyroid tissues were obtained from patients with a unifocal tumor. All specimens were frozen at −80°C in isopentane and stored in liquid nitrogen directly after surgical resection and until RNA extraction. Informed consent was obtained from all patients. Twenty-three of these 65 samples were used for microarray experiments. All thyroid samples were obtained in euthyroid subjects, as assessed by serum thyroid stimulating hormone (TSH) concentrations in the normal range at the time of surgery. Complete clinical and histological information concerning samples used in this study is available in supplementary Table S1 (http://ajp.amjpathol.org).

Table 1.

Characteristics of Thyroid Tumors

Q RT-PCR Microarray Histological features
Normal thyroid 17 11
tissue
Follicular 16 4 Typical 7
adenoma Atypical 9
Follicular 27 8 MIF 7
carcinoma
Anaplastic 5 0 WIF 16
carcinoma Hürthle 4
Open in a new tab

Q RT-PCR and microarray columns represent the number of samples included in the experiments. All tumors used in microarray experiments were included in the Q RT-PCR experiments. Staging is based on the 2002 TNM Classification (American Joint Committee on Cancer, 2002). MIF, minimally invasive; WIF, widely invasive. 

Total RNA Preparation and Reverse Transcription

Total RNA was isolated from frozen tissue samples using Trireagent (Sigma-Aldrich, Saint Louis, MO) and purified on Rneasy columns (Qiagen, Hilden, Germany) according to manufacturer’s protocols. Quality of RNA preparation, based on the 28S/18S ribosomal RNAs ratio, was assessed using the RNA 6000 Nano Lab-On-chip as developed on the Agilent 2100 Bioanalyzer device (Agilent Technologies, Palo Alto, CA). All specimens included in this study displayed a ratio of 28S to 18S higher than 1.5 (average 1.8). RNA samples were frozen in nuclease-free water (Promega Corporation, Madison, WI). One microgram of total RNA was reverse-transcribed by Moloney murine leukemia virus reverse transcriptase in the presence of random primers (Applied Biosystems, Foster City, CA). For each sample, the PAX8-PPARγ1 translocation detection assay was performed as previously described.18 Briefly, the PCR reaction was performed with several primers previously described.6,7 PCR products were visualized using 2% agarose gel. For each sample with detectable PCR product, all visible bands were purified and directly sequenced to confirm the presence of PAX8-PPARγ1 translocation. The efficiency of reverse transcription was controlled with amplification of several genes with the same cDNA (with quantitative real-time PCR).

Microarray Analysis

A pool composed by equal amount of total RNA from each tissue sample was used as the RNA reference. Five-microgram aliquots of total RNA from each sample and from the reference pool were used to generate labeled antisense cRNA with T7 RNA polymerase. Labeling of cRNAs was performed with cyanine 3 (Cy3)-CTP for all samples and cyanine 5 (Cy5)-CTP for the RNA reference (Perkin Elmer NEN, Boston, MA). Reverse transcription, linear amplification, cRNA labeling, and purification were performed using the Agilent Linear amplification kit. The cRNA concentration and Cy3-CTP or Cy5-CTP incorporation were assessed using an UV-visible spectrophotometer. We used custom-designed 60-mer oligonucleotide microarrays of 22,000 features, representing 16,840 known unique genes developed by Agilent Technologies. Hybridization was performed during 17 hours at 60°C, with 1 μg of Cy3-labeled cRNA of each sample mixed to the same amount of Cy5-labeled cRNA reference.

The Feature Extraction software (Agilent Technologies) was used to quantify intensity of fluorescent images and to normalize results using the local background subtraction option as recommended for oligo-microarray procedures. Files used for statistical analysis contained, for each tumor sample, the list of 22,000 features associated to a set of values including log ratio compared with reference, P value of log ratio, and intensities. To evaluate the reliability of data, linearity and intrassay reproducibility were checked as follows. First, RNA reference labeled with either Cy3 or Cy5 was hybridized to compare the efficiencies of incorporation for each cyanine and to evaluate the intensity noise. Second, one sample processed in duplicate assessed the reproducibility. All data obtained from microarray analysis have been submitted to Array Express at the European Bioinformatics Institute (http://www.ebi.ac.uk/arrayexpress/). ArrayExpress (at the European Bioinformatics Institute) is a public repository for microarray data, which is aimed at storing well-annotated data in accordance with Microarray Gene Expression Data recommendations (http://www.mged.org). Gene functions were determined using gene ontology database FatiGO (http://fatigo.bioinfo.cnio.es), Panther Ontology (http://myscience.appliedbiosystems.com), and On Line Mendelian Inheritance In Man description (http://www.ncbi.nlm.nih.gov/).

Statistical Analysis

Microarray data analysis was performed using the Resolver software (Rosetta Inpharmatics, Kirkland, WA). All data were filtered to eliminate low-intensity value under 200 arbitrary units for both colors, a threshold determined on the basis of the linearity test. Each selected gene had at least a twofold change, with a P value less than 1% in a minimum of four independent samples. Using this procedure, 1859 genes passed the filter. For the unsupervised clustering, a hierarchical agglomerative algorithm that pairs samples according to their similarity was used. Analysis of variance (ANOVA) was performed on the microarray data and on the Q RT-PCR data. For each individual gene, one-way ANOVA tests were performed to compare results obtained from normal tissues and FTCs and from FTCs presenting or not presenting the translocation. Two-way ANOVA tests were applied for the Q RT-PCR data to compare average gene expression values between FTCs with and without the translocation while adjusting for the histological type.

Analysis of Promoter Sequences

To search potential peroxisome proliferator response element (PPRE) sequences within the upstream 2-kb sequence located before the coding determining sequence of the selected genes, a homology matrix was designed on the basis of nine well-characterized PPRE sequences of the human PPARγ promoter, including those regulating acyl-coenzyme A oxidase, apolipoprotein C-III, carnitine palmitoyltransferase 1B, ATP-binding cassette D2 (2 PPRE), aquaporin (AQ7), nuclear receptor LXR, and acyl-coA binding protein genes. The homology matrix associates a value to each nucleotide at each position of the consensus PPRE (AGGTCA N AGGTCA) sequence. This value is related to the frequency of each nucleotide in the nine PPRE analyzed, and when the frequency is null, a negative value is associated. A score between −7 and 98 was attributed to each sequence analyzed. The threshold value for this score was determined as the minimum value of the 95% confidence interval of average of score obtained by the nine well-characterized PPRE. The number of putative promoters found in our gene set was compared with the number of putative PPRE expected by chance and determined in 500 randomly selected gene sets. More information is available in supplementary Table S3 (http://ajp.amjpathol.org).

Real-Time Quantitative PCR

Oligonucleotide primers and Taqman probes specific for PPARγ, PAX8A, FGD3, decorin (DCN), RPLP0, PPIA, or 28S were designed to be intron spanning using the PrimerExpress computer software (Applied Biosystems). Sequences were from the GenBank database, and the oligonucleotides were purchased from MWG Biotech (Courtaboeuf, France). Primers and probes for thyroid-specific genes SLC5A5, TPO, and TG were already published;21 and 18S, ANGPTL4, acetyl-Coenzyme A acyltransferase 1 (ACAA1), PDE8, and HBP17 were obtained from Assays-On-Demand (Applied Biosystems). Q-PCRs were performed on equivalent of 10 ng per total RNA per tube in a final volume of 18 μl and developed as previously described.18 The reference pool, corresponding to the microarray RNA reference, was used as a calibrator (1× sample). Normalization was assessed by a combination of four housekeeping genes (18S, 28S, RPL0, and PPIA).22

Immunohistochemistry and Tissue Array Construction

A tissue array, including all of the samples analyzed in the microarray experiment, was constructed using a tissue-arrayer device (Alphelys, Plaisir, France) with a 1-mm needle. Quadruplicate samples were prepared both from the tumor part and from the normal thyroid tissue at distance from the tumor. A total of 128 spots by antibody were analyzed.

Immunohistochemistry was performed on formalin-fixed paraffin-embedded 5-μm sections as described previously.18 Two independent observers carried out the immunohistochemical analysis, considering both the percentage of positive cells (noted as percentage of stained cells) and the intensity of staining (noted from 0 to 4 +). Antibody against PPARγ was a mouse monoclonal antibody γ, which recognizes the COOH terminus of the protein (Santa Cruz, CA). A 26-amino acid peptide and a 22-amino acid peptide, spanning the COOH-terminal portions of ACAA1 and FGD3, respectively, were synthesized by a conventional solid-phase method using a model 432A peptide synthesizer (Applied Biosystems). The identity and purity of each peptide were verified by 1) amino acid analysis on an α LKB analyzer (LKB, Rockville, MD); 2) HPLC; and 3) sequence analysis using ES-TOF mass spectrometry (Micromass Quattro LCT, Villeurbanne, France). The peptides were then conjugated to keyhole limpet hemocyanin using benzidine as the coupling agent, and two rabbits were immunized by intradermal injections of each synthetic peptide-carrier conjugate. After three subsequent boosts at 3-week intervals, animals were bled, and their sera were tested in an enzyme-linked immunosorbent assay.

Results

Detection of the PAX8-PPARγ1 Translocation

The PAX8-PPARγ1 rearrangement was detected in 4 of 23 (17%) FTCs and 1 of 16 (6%) benign hypofunctioning follicular adenomas (FTAs), as previously described.18 In the five positive samples, RT-PCR and sequencing revealed four isoforms generated by the fusion of different exons of PAX8 (exon 7; exons 7 and 8; exons 7, 8, and 9; and exons 7 and 9) with exon 1 of PPARγ1. None of the normal thyroid tissues (n = 17), Hürthle-cell carcinomas (n = 4), and anaplastic carcinomas (n = 5) displayed any evidence of the translocation.

Analysis and Differential Gene Expression

Microarray experiments, based on the dual-color technology were performed on 23 thyroid samples, including 8 FTCs, 4 FTAs, and 11 normal contralateral tissues. Three FTCs and one FTA harbored PAX8-PPARγ1 rearrangement. Using the Resolver software, an intensity- and fold changed-based filtering approach selected 1859 features from the 22,000 present on the 60-mer oligonucleotides microarray for further analysis. The nonsupervised hierarchical clustering performed on these features is presented as a hyperbolic view in Figure 1. One cluster included 8 of 11 normal thyroid tissue samples. Two normal contralateral tissues (NT6 and NT9), presenting lymphocytic infiltration at histological examination, segregated separately and were associated with the presence of serum anti-thyroid antibodies. Most FTAs and FTCs clustered together. One carcinoma sample, namely FTC17, displayed a quite different expression profile compared with those of the FTCs and corresponded to a particularly aggressive tumor. Interestingly, a peculiar cluster contained all of the four PAX8-PPARγ-positive tumors. This distinct cluster underlines the existence of a specific gene expression profile associated with the translocation.

Figure 1.

Figure 1

Open in a new tab

Hierarchical clustering of thyroid tumors. Classification is determined using the 1859 filtered features, with an agglomerative algorithm based on Pearson correlation coefficient. The figure represents a hyperbolic lens view of the classification tree. The red plot is the root of the clustering tree, the blue plots are classification nodes, and the brown plots represent the samples (NT, non tumoral tissues; t(2;3), presence of PAX8-PPARγ translocation in the sample).

Gene Expression Profile Associated with Follicular Thyroid Carcinomas

To explore gene expression profiles related to tumors, an ANOVA test was performed for each of the 1859 filtered genes. Normal thyroid tissues displayed 61 genes differentially expressed at P < 0.01 (N/T− set) when compared with PAX8-PPARγ-negative FTCs and 116 genes when compared with PAX8-PPARγ-positive FTCs at P < 0.01 (N/T+ set). These sets of genes are described in supplementary Table S2 (http://ajp.amjpathol.org). Only 27 genes were differently regulated in both sets of FTCs versus normal tissues (Figure 2). Some of these genes were already found to be up-regulated in human FTCs, such as transforming growth factor α (TGFA), or down-regulated, such as gelsolin, tumor necrosis factor receptor superfamily, member 11b, or trefoil factor 3.23 Interestingly, increased creatine kinase, mitochondrial 1 and decreased DCN gene expressions were also reported in a murine model of follicular carcinoma.24

Figure 2.

Figure 2

Open in a new tab

Profile comparison and ontology analysis. Distribution of the ontology functions of the genes differentially expressed between normal tissues and FTCs with or without the PAX8-PPARγ translocation. Each gene was associated to an ontology cluster determined by using three gene description databases (FatiGO, Panther Ontology, and On Line Mendelian Inheritance In Man). The percentage value on the horizontal axis corresponds to the proportion of genes belonging to an ontology cluster compared with all genes of N/T+ set or N/T− set. Percentage on the left of axis (green) corresponds to down-regulated genes, whereas percentage on the right of axis (red) corresponds to up-regulated genes.

The low number of genes in common between those two sets might indicate separate oncogenesis pathways. When genes were ordered by ontology, N/T− and N/T+ gene sets displayed important differences (Figure 2). Microarray analysis revealed that the N/T− set mainly included down-regulated genes (84%) as previously described.25 This included genes involved in cell signaling (CRABP1, DDR2, and DPP6), immunity (CCL21, CXCL12, and DF), cellular metabolism, and enzymes (FGL2 and PGM5). Interestingly, the N/T+ set mainly included up-regulated genes (76%) such as genes involved in cellular energetic metabolism (ACAA1, AK3L1, fructose(1,6) bisphosphatase (FBP1), DGAT2, and AQ7), cell signaling (GDF5, RDC1, DPP4, and FGD3), cellular growth (GADD45G, MYCL1, and TNFRSF21), and transcription (NFE2L3 and PPARγ). Surprisingly, the statistical analysis did not select the thyroid-specific genes present on the microarray, including NIS, TPO, TG, TSH-R, and DUOX2, as differently expressed between FTCs and normal tissue samples.

Gene Expression Profile Associated with the PAX8-PPARγ Translocation

To further investigate the expression profile associated with the PAX8-PPARγ-rearrangement, we compared these two groups of follicular tumors by the ANOVA test (P < 0.01). A set of 93 genes (T−/T+ set) was identified (Figure 3; supplementary Table S2 (http://ajp.amjpathol.org)). Interestingly, a 26-fold up-regulated PPARγ gene expression was detected in the positive follicular tumors. Because the entire PPARγ1 open reading frame is present in the rearranged product, wild-type PPARγ and PAX8-PPARγ transcripts were undistinguishable by the corresponding 60-mer oligonucleotide. In accordance with previous results, PPARγ expression level appears to be the most specific difference between those two tumor groups.8,16,18

Figure 3.

Figure 3

Open in a new tab

Clustering of tumors presenting or not the PAX8-PPARγ1 translocation using the T−/T+ gene set. The clustering is based on the T−/T+ set of genes determined by ANOVA. Red or green color scale represented respectively up- and down-regulation of genes in comparison with the reference. Each line corresponds to a gene and each column to a tumor (t(2;3), presence of the PAX8-PPARγ translocation in the samples). The HUGO gene symbol is associated to the average fold change value ratio between FTCs with or without the translocation.

Several genes, known as PPARγ positively regulated targets, were up-regulated in PAX8-PPARγ-positive tumors, including angiopoietin-like 4 and aquaporin 7, which displayed a 23-fold and 4-fold gene expression increase, respectively. Expression of genes involved in cellular metabolism, particularly those associated to lipid, glucose, and amino acid metabolisms, was modified: ACAA1, a gene involved in the β-oxidation process; DGAT2, an acyl CoA diacylglycerol acyltransferase; and FBP1, involved in neo-glucogenesis, were increased by a fourfold, eightfold, and sevenfold, respectively. Several genes associated to tumorigenesis, such as MYCL1 (homolog to v-myc), TGFA, and neuregulin 1, and to cell signaling, such as FGD3, a putative guanine nucleotide exchange factor, were also up-regulated. Endothelial cell-specific molecule 1 (ESM1), placental growth factor (PGF), and HBP17 (heparin-binding growth factor binding protein also name FGF-BP1), all genes known to activate angiogenesis, displayed increased gene expression. Metallothionein MT1G and MT1H, overexpressed in several types of carcinomas, were increased by 14-fold and 7-fold, respectively. Finally, several genes with unknown function (FLJ11154, FLJ20489, and FLJ41259) were up-regulated more than ninefold.

Analysis of Gene Promoter Sequences

We first examined whether any variation of gene expression could be related to its chromosomal localization. The selected genes spread all over the genome, and no significant association between their position and their expression was found (data not shown). The in silico analysis of the promoter of the selected genes indicated that putative PPRE were detectable in 68 promoters of the 93 genes in T−/T+ set (73%). Among genes with putative PPRE, we also observed genes already proposed to be regulated by PPARγ such as ANGPTL4, AQ7, ACAA1, APOA1, or FGD3. The frequency of putative PPRE in T+/T− set was significantly higher than expected by chance (P value <0.001). All PPRE sequences are reported in the supplementary Table S3 (http://ajp.amjpathol.org).

Quantitative Gene Expression Analysis

Q RT-PCR was carried out to confirm changes in the expression of 15 selected genes on a larger series of 65 thyroid tumor samples, including the 23 samples analyzed in the microarray experiments. Five tumors harbored the PAX8-PPARγ1 rearrangement (four FTCs and one FTA). The values were normalized using the mean of the four housekeeping genes (18S, 28S, RPLP0, and PPIA)22 and compared with the pool of all samples. First, we compared the expression values in the 23 samples tested both by Q RT-PCR and by microarray experiments. A significant correlation of all genes tested was observed (P < 0.01). Second, analysis of thyroid-specific genes TG, TPO, and NIS/SLC5A5 indicated that they were expressed in all tissues excepted in anaplastic carcinomas. Although they were not retained by the microarray analysis, their expression was significantly down-regulated in follicular carcinomas compared with normal tissue (P < 0.05, P < 0.01, and P < 0.001, respectively), as previously described.21 NIS/SLC5A5 and TPO also appeared to be slightly decreased in FTCs compared with FTAs (P < 0.05 and P < 0.01). However, their expression was not significantly different between tumors harboring or not the translocation.

Third, 12 genes selected from microarray analysis were analyzed by Q RT-PCR in a large series of thyroid tumors (Figure 4). The ANOVA was adjusted both for the tumor type (FTA or FTC) and for the presence or absence of the translocation. In tumors positive for the translocation when compared with negative tumors, Q RT-PCR analysis clearly confirmed the overexpression of PPARγ, ANGPTL4, ACAA1, HBP17, and FGD3 genes (P < 0.001) and TGFA (P = 0.003). No significant changes between gene expression of positive and negative tumors for translocation were found for PAX8A (P = 0.084), PDE8B (P = 0.088), and ESM1 (P = 0.092). The expression was significantly different between FTAs and FTCs only for CRAPB1 gene (P < 0.002). Finally, TGFA (P = 0.001), PDE8B (P = 0.03), CRABP1 (P < 0.001), ESM1 (P < 0.001), and DCN (P < 0.001) gene expression was significantly decreased in both follicular benign and malignant tumors when compared with normal thyroid tissues.

Figure 4.

Figure 4

Open in a new tab

Validation of gene expression in follicular thyroid tumors. Box plots for PPARγ, PAX8A, ACAA1, ANGPTL4, HBP17, FGD3, TGFA, PDE8, CRABP1, ESM1, and DCN gene expression as measured by real-time quantitative PCR normalized with four housekeeping genes. Gene expression levels are reported according to the diagnosis (N, normal; FTA, hypofunctioning FTAs; ATC, anaplastic thyroid carcinomas). For tumors without the translocation, the box shows the limits of the middle half of the data, and the line inside the box represents the median. Whiskers are drawn to represent the standard span of the 5th/95th percentile for all. Circles correspond to tumors with PAX8-PPARγ1 translocation. Horizontal barswith * or ** represent significant P values between N and FTC, and vertical barswith * or ** represent significant P values between tumors with or without the translocation.

Immunohistochemistry

Tissue array allowed the immunohistochemistry of all normal, benign, and malignant thyroid tissue specimens tested in microarray experiments. Using PPARγ antibody, a strong immunostaining was found in nuclei of all tumors displaying the PAX8-PPARγ1 translocation (Figure 5, A and B). Using the anti-ACAA1 antibody, a strong immunostaining was observed in the cytoplasm of tumor cells presenting the PAX8-PPARγ1 translocation. The staining showed a characteristic localization, consisting in small or large intracytoplasmic positive vacuoles. Large positive vacuoles were preferentially located in the apical part of tumor cells (Figure 5, C and D). In normal thyroid tissue, a weak staining was observed in a minority of cells with a lower number of vacuoles. Immunohistochemistry with the anti-FGD3 antibody indicated that the FGD3 protein was heterogeneously distributed in the tumor. In tumors with the rearrangement, a more intense staining was observed; FGD3 was localized at the lateral membrane without any staining at the basal or apical membranes (Figure 5, E and F).

Figure 5.

Figure 5

Open in a new tab

PPARγ, ACAA1, and FGD3 immunostaining in thyroid carcinomas presenting the PAX8-PPARγ translocation. PPARγ immunostaining is presented at magnification ×100 (A) and ×200 (B). A strong positivity is observed in the nuclei of tumor cells (long arrows). Nuclei of endothelial cells are not stained (small arrows). For ACAA1 immunostaining at magnification ×100 (C), numerous positive intracytoplasmic vacuoles are observed in the majority of tumor cells. For ACAA1 immunostaining at magnification ×400 (D), positive large vacuoles are present in the apical part of tumor cells (long arrows); smaller vacuoles are indicated by small arrows (L, lumen of tumor follicle). FGD3 immunostaining is presented at magnification ×100 (E) and ×400 (F). Heterogeneity of the staining is observed in the tumor follicle. FGD3 localizes at the lateral membrane (long arrows), and no staining is observed at basal or apical membrane in translocated tumors.

Discussion

Identifying molecular genetic mechanisms is particularly challenging in thyroid oncogenesis because differentiated thyroid tumors present with distinct benign and malignant histotypes with different behaviors. Only a few large-scale genomics studies have been devoted to follicular tumors.23,26 In the present study, we have undertaken expression profiling of FTCs that appear as histologically homogeneous and do or do not harbor the PAX8-PPARγ1 rearrangement.

Nonsupervised hierarchical clustering performed on results obtained for FTAs and FTCs was not efficient to distinguish benign and malignant tumoral tissues, probably because of the low number of samples included in the analysis. However, statistical analysis revealed distinct patterns between normal tissues and follicular tumors. Those sets of genes included several interesting genes such as SELE, CRABP1, gelsolin, DCN, or PPARγ that have been previously identified during thyroid tumor progression in a mouse model and in human tumors.17,24,25 The small leucine-rich proteoglycan decorin is associated with negative regulation of cell growth and has a prominent role in TGF-β and EGFR activation pathways that contribute to its role in cellular proliferation, angiogenesis, and immunomodulation.24,27 The expression of thyroid-specific genes, such as TG, TPO, TSH-R, or NIS, was detectable in the microarray experiments and by quantitative PCR. None of them appeared to be differentially regulated between tumors harboring or not harboring the translocation.

Overall, in the FTC signature, a peculiar group including all of the PAX8-PPARγ1-positive tumors clustered separately. Several lines of evidence indicate that angiogenic factors are involved in neoplastic growth and aggressiveness of thyroid tumors. Vascular endothelial growth factor (VEGF), VEGF-C, and angiopoietin-2 and their tyrosine kinase receptors are overexpressed in thyroid carcinomas.28 Follicular carcinomas with PPARγ rearrangement exhibited vascular invasion.29 In our study, increased gene expression of ESM1, PGF, and HBP17 is consistent with the overexpression of ANGPTL4 or with the down-regulation of tissue inhibitor of metalloproteinase 3 (TIMP3). ESM1, a proteoglycan secreted by endothelial cells under the control of inflammatory cytokines, plays a role in the adhesion process and angiogenesis.30,31 Both PGF and TIMP3 regulate inter- and intramolecular cross-talks between the VEGF receptor tyrosine kinases, and TIMP3 inhibits downstream signaling and angiogenesis independently of its MMP-inhibitory activity. Although the function of ANGPTL4 is not completely understood, it is known to promote angiogenesis in response to hypoxia.32

In solid tumors, functional properties of only a few fusion genes have been characterized.33–35 Recently, it was demonstrated that the transforming properties of the PAX8-PPARγ1 fusion oncoprotein depend in part on the inhibition of wild-type PPARγ.19 This is in line with the down-regulation of PPARγ expression in several tumors not harboring the PAX8-PPARγ translocation, including thyroid cancers.17 In our study, the finding of an up-regulation of the expression of several known or suspected PPARγ target genes leads to question the assumed mechanism of action of the chimeric product as having a dominant-negative effect. ANGPTL4, which is positively regulated by PPARγ agonist, was found to be one of the most up-regulated in PAX8-PPARγ FTCs. Aquaporin 7, involved in water channel activity, was also detected as up-regulated in a recent report.36 Expression of multiple genes involved in lipid (ACAA1), glucose (FBP1), and amino acid metabolisms was increased in tumors presenting the translocation in line with recent results. (During the reviewing of this paper, Lui et al37 described a molecular clustering associated with the PAX8-PPARγ translocation. ACAA1(AF035295), PPARG(L40904), FBP1(U21931), ABCC3(AF085692), FBN1(X63556), and ATP10B(AB018258) are shared in the two studies.) Furthermore, putative PPRE were detectable in promoter sequences of these genes, suggesting that their expression could be directly regulated by PPARγ. Finally, among genes with putative assigned function, the increased expression of FGD3 gene, a recently discovered member of the FGD1 family,38 deserves further investigation into its potential role in the oncogenic process associated with the PAX8-PPARγ1 rearrangement. FGD1 encodes a guanine nucleotide exchange factor (GEF) that specifically activates the Rho GTPase Cdc42, and interacts with cortactin in Arp2/3 complex-mediated actin assembly.39 Like RAS, Rho GTPases serve similar functions in signal transduction and for the establishment of cadherin-dependent cell-cell contact, and aberrant GTPase function has been implicated in cancer development.40,41

In summary, we report, for the first time, a global gene expression analysis of follicular thyroid carcinomas bearing the PAX8-PPARγ1 translocation using Agilent oligonucleotide microarrays, quantitative RT-PCR, and immunohistochemistry on tissue microarrays. The specific pattern associated with the PPARγ rearrangement is in line with previous hypotheses suggesting an independent pathway.10 Moreover, the differentially expressed genes would be of value both for discovering new potential targets of PPARγ and for identifying molecular pathways involved in the development of thyroid carcinomas.

Supplementary Material

Supplemental Material
amjpathol_167_1_223__index.html (1.2KB, html)

Acknowledgments

We thank Alain Deroussan for expert assistance in mass spectrometry. We also thank the Prof. Gilles Vassal and the Genomic Group at Institut Gustave-Roussy for their continuous support.

Footnotes

Address reprint requests to Jean-Michel Bidart, Institut Gustave-Roussy, 39, rue Camille Desmoulins, 94805 Villejuif, France. E-mail: bidart@igr.fr.

Supported by grants from Electricité de France, Commissariat à l’Energie Atomique LRC-29V, Association pour la Recherche sur le Cancer, Ligue contre le Cancer (Comité Val de Marne), and LIPHA Santé. L.L. is a recipient from the French Endocrine Society-Merck LIPHA Santé grant.

Supplemental material for this article appears on http://ajp.amjpathol.org.

References

  1. Fagin JA. Molecular genetics of tumors of thyroid follicular cells. Braverman LE, Utiger RD, editors. Philadelphia: Lippincott Williams & Wilkins; 2000:pp 886–898. [Google Scholar]
  2. Bongarzone I, Butti MG, Coronelli S, Borrello MG, Santoro M, Mondellini P, Pilotti S, Fusco A, Della Porta G, Pierotti MA. Frequent activation of ret protooncogene by fusion with a new activating gene in papillary thyroid carcinomas. Cancer Res. 1994;54:2979–2985. [PubMed] [Google Scholar]
  3. Xing M, Vasko V, Tallini G, Larin A, Wu G, Udelsman R, Ringel MD, Ladenson PW, Sidransky D. BRAF T1796A transversion mutation in various thyroid neoplasms. J Clin Endocrinol Metab. 2004;89:1365–1368. doi: 10.1210/jc.2003-031488. [DOI] [PubMed] [Google Scholar]
  4. Fagin JA. Minireview: branded from the start-distinct oncogenic initiating events may determine tumor fate in the thyroid. Mol Endocrinol. 2002;16:903–911. doi: 10.1210/mend.16.5.0838. [DOI] [PubMed] [Google Scholar]
  5. Roque L, Rodrigues R, Pinto A, Moura-Nunes V, Soares J. Chromosome imbalances in thyroid follicular neoplasms: a comparison between follicular adenomas and carcinomas. Genes Chromosomes Cancer. 2003;36:292–302. doi: 10.1002/gcc.10146. [DOI] [PubMed] [Google Scholar]
  6. Kroll TG, Sarraf P, Pecciarini L, Chen CJ, Mueller E, Spiegelman BM, Fletcher JA. PAX8-PPARgamma1 fusion oncogene in human thyroid carcinoma [corrected]. Science. 2000;289:1357–1360. doi: 10.1126/science.289.5483.1357. [DOI] [PubMed] [Google Scholar]
  7. Nikiforova MN, Biddinger PW, Caudill CM, Kroll TG, Nikiforov YE. PAX8-PPARgamma rearrangement in thyroid tumors: RT-PCR and immunohistochemical analyses. Am J Surg Pathol. 2002;26:1016–1023. doi: 10.1097/00000478-200208000-00006. [DOI] [PubMed] [Google Scholar]
  8. Marques AR, Espadinha C, Catarino AL, Moniz S, Pereira T, Sobrinho LG, Leite V. Expression of PAX8-PPAR gamma 1 rearrangements in both follicular thyroid carcinomas and adenomas. J Clin Endocrinol Metab. 2002;87:3947–3952. doi: 10.1210/jcem.87.8.8756. [DOI] [PubMed] [Google Scholar]
  9. Cheung L, Messina M, Gill A, Clarkson A, Learoyd D, Delbridge L, Wentworth J, Philips J, Clifton-Bligh R, Robinson BG. Detection of the PAX8-PPAR gamma fusion oncogene in both follicular thyroid carcinomas and adenomas. J Clin Endocrinol Metab. 2003;88:354–357. doi: 10.1210/jc.2002-021020. [DOI] [PubMed] [Google Scholar]
  10. Nikiforova MN, Lynch RA, Biddinger PW, Alexander EK, Dorn GW, II, Tallini G, Kroll TG, Nikiforov YE. RAS point mutations and PAX8-PPAR gamma rearrangement in thyroid tumors: evidence for distinct molecular pathways in thyroid follicular carcinoma. J Clin Endocrinol Metab. 2003;88:2318–2326. doi: 10.1210/jc.2002-021907. [DOI] [PubMed] [Google Scholar]
  11. Desvergne B, Wahli W. Peroxisome proliferator-activated receptors: nuclear control of metabolism. Endocr Rev. 1999;20:649–688. doi: 10.1210/edrv.20.5.0380. [DOI] [PubMed] [Google Scholar]
  12. Koeffler HP. Peroxisome proliferator-activated receptor gamma and cancers. Clin Cancer Res. 2003;9:1–9. [PubMed] [Google Scholar]
  13. Michalik L, Desvergne B, Wahli W. Peroxisome-proliferator-activated receptors and cancers: complex stories. Nat Rev Cancer. 2004;4:61–70. doi: 10.1038/nrc1254. [DOI] [PubMed] [Google Scholar]
  14. Ohta K, Endo T, Haraguchi K, Hershman JM, Onaya T. Ligands for peroxisome proliferator-activated receptor gamma inhibit growth and induce apoptosis of human papillary thyroid carcinoma cells. J Clin Endocrinol Metab. 2001;86:2170–2177. doi: 10.1210/jcem.86.5.7493. [DOI] [PubMed] [Google Scholar]
  15. Martelli ML, Iuliano R, Le Pera I, Sama I, Monaco C, Cammarota S, Kroll T, Chiariotti L, Santoro M, Fusco A. Inhibitory effects of peroxisome poliferator-activated receptor gamma on thyroid carcinoma cell growth. J Clin Endocrinol Metab. 2002;87:4728–4735. doi: 10.1210/jc.2001-012054. [DOI] [PubMed] [Google Scholar]
  16. Aldred MA, Morrison C, Gimm O, Hoang-Vu C, Krause U, Dralle H, Jhiang S, Eng C. Peroxisome proliferator-activated receptor gamma is frequently downregulated in a diversity of sporadic nonmedullary thyroid carcinomas. Oncogene. 2003;22:3412–3416. doi: 10.1038/sj.onc.1206400. [DOI] [PubMed] [Google Scholar]
  17. Marques AR, Espadinha C, Frias MJ, Roque L, Catarino AL, Sobrinho LG, Leite V. Underexpression of peroxisome proliferator-activated receptor (PPAR)gamma in PAX8/PPARgamma-negative thyroid tumours. Br J Cancer. 2004;91:732–738. doi: 10.1038/sj.bjc.6601989. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lacroix L, Mian C, Barrier T, Talbot M, Caillou B, Schlumberger M, Bidart JM. PAX8 and peroxisome proliferator-activated receptor gamma 1 gene expression status in benign and malignant thyroid tissues. Eur J Endocrinol. 2004;151:367–374. doi: 10.1530/eje.0.1510367. [DOI] [PubMed] [Google Scholar]
  19. Gregory Powell J, Wang X, Allard BL, Sahin M, Wang XL, Hay ID, Hiddinga HJ, Deshpande SS, Kroll TG, Grebe SK, Eberhardt NL, McIver B. The PAX8/PPARgamma fusion oncoprotein transforms immortalized human thyrocytes through a mechanism probably involving wild-type PPARgamma inhibition. Oncogene. 2004;23:3634–3641. doi: 10.1038/sj.onc.1207399. [DOI] [PubMed] [Google Scholar]
  20. Hedinger C, Williams E, Sobin L. Histological typing of thyroid tumours. Berlin: Springer-Verlag; 1988 [Google Scholar]
  21. Lazar V, Bidart JM, Caillou B, Mahe C, Lacroix L, Filetti S, Schlumberger M. Expression of the Na+/I- symporter gene in human thyroid tumors: a comparison study with other thyroid-specific genes. J Clin Endocrinol Metab. 1999;84:3228–3234. doi: 10.1210/jcem.84.9.5996. [DOI] [PubMed] [Google Scholar]
  22. Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A, Speleman F. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol. 2002;3:1–11. doi: 10.1186/gb-2002-3-7-research0034. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Barden CB, Shister KW, Zhu B, Guiter G, Greenblatt DY, Zeiger MA, Fahey TJ., III Classification of follicular thyroid tumors by molecular signature: results of gene profiling. Clin Cancer Res. 2003;9:1792–1800. [PubMed] [Google Scholar]
  24. Ying H, Suzuki H, Zhao L, Willingham MC, Meltzer P, Cheng SY. Mutant thyroid hormone receptor beta represses the expression and transcriptional activity of peroxisome proliferator-activated receptor gamma during thyroid carcinogenesis. Cancer Res. 2003;63:5274–5280. [PubMed] [Google Scholar]
  25. Aldred MA, Ginn-Pease ME, Morrison CD, Popkie AP, Gimm O, Hoang-Vu C, Krause U, Dralle H, Jhiang SM, Plass C, Eng C. Caveolin-1 and caveolin-2,together with three bone morphogenetic protein-related genes, may encode novel tumor suppressors down-regulated in sporadic follicular thyroid carcinogenesis. Cancer Res. 2003;63:2864–2871. [PubMed] [Google Scholar]
  26. Cerutti JM, Delcelo R, Amadei MJ, Nakabashi C, Maciel RM, Peterson B, Shoemaker J, Riggins GJ. A preoperative diagnostic test that distinguishes benign from malignant thyroid carcinoma based on gene expression. J Clin Invest. 2004;113:1234–1242. doi: 10.1172/JCI19617. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Stander M, Naumann U, Wick W, Weller M. Transforming growth factor-beta and p-21: multiple molecular targets of decorin-mediated suppression of neoplastic growth. Cell Tissue Res. 1999;296:221–227. doi: 10.1007/s004410051283. [DOI] [PubMed] [Google Scholar]
  28. Bunone G, Vigneri P, Mariani L, Buto S, Collini P, Pilotti S, Pierotti MA, Bongarzone I. Expression of angiogenesis stimulators and inhibitors in human thyroid tumors and correlation with clinical pathological features. Am J Pathol. 1999;155:1967–1976. doi: 10.1016/S0002-9440(10)65515-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. French CA, Alexander EK, Cibas ES, Nose V, Laguette J, Faquin W, Garber J, Moore F, Jr, Fletcher JA, Larsen PR, Kroll TG. Genetic and biological subgroups of low-stage follicular thyroid cancer. Am J Pathol. 2003;162:1053–1060. doi: 10.1016/S0002-9440(10)63902-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Lassalle P, Molet S, Janin A, Heyden JV, Tavernier J, Fiers W, Devos R, Tonnel AB. ESM-1 is a novel human endothelial cell-specific molecule expressed in lung and regulated by cytokines. J Biol Chem. 1996;271:20458–20464. doi: 10.1074/jbc.271.34.20458. [DOI] [PubMed] [Google Scholar]
  31. Aitkenhead M, Wang SJ, Nakatsu MN, Mestas J, Heard C, Hughes CC. Identification of endothelial cell genes expressed in an in vitro model of angiogenesis: induction of ESM-1, (beta)ig-h3, and NrCAM. Microvasc Res. 2002;63:159–171. doi: 10.1006/mvre.2001.2380. [DOI] [PubMed] [Google Scholar]
  32. Yoon JC, Chickering TW, Rosen ED, Dussault B, Qin Y, Soukas A, Friedman JM, Holmes WE, Spiegelman BM. Peroxisome proliferator-activated receptor gamma target gene encoding a novel angiopoietin-related protein associated with adipose differentiation. Mol Cell Biol. 2000;20:5343–5349. doi: 10.1128/mcb.20.14.5343-5349.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Allander SV, Illei PB, Chen Y, Antonescu CR, Bittner M, Ladanyi M, Meltzer PS. Expression profiling of synovial sarcoma by cDNA microarrays: association of ERBB2, IGFBP2, and ELF3 with epithelial differentiation. Am J Pathol. 2002;161:1587–1595. doi: 10.1016/S0002-9440(10)64437-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Tonon G, Modi S, Wu L, Kubo A, Coxon AB, Komiya T, O’Neil K, Stover K, El-Naggar A, Griffin JD, Kirsch IR, Kaye FJ. t(11;19)(q21;p13) translocation in mucoepidermoid carcinoma creates a novel fusion product that disrupts a Notch signaling pathway. Nat Genet. 2003;33:208–213. doi: 10.1038/ng1083. [DOI] [PubMed] [Google Scholar]
  35. Schaefer KL, Brachwitz K, Wai DH, Braun Y, Diallo R, Korsching E, Eisenacher M, Voss R, Van Valen F, Baer C, Selle B, Spahn L, Liao SK, Lee KA, Hogendoorn PC, Reifenberger G, Gabbert HE, Poremba C. Expression profiling of t(12;22) positive clear cell sarcoma of soft tissue cell lines reveals characteristic up-regulation of potential new marker genes including ERBB3. Cancer Res. 2004;64:3395–3405. doi: 10.1158/0008-5472.CAN-03-0809. [DOI] [PubMed] [Google Scholar]
  36. Thomas D, Lizyness M, Sanders D, Koenig R, Misek D, MacDonald J, Kuick R, Nikiforov Y, Hanash S, Giordano TJ. Gene expression signature of PAX8-PPARg1 positive follicular thyroid carcinoma. Thyroid. 2004;14:728. [Google Scholar]
  37. Lui WO, Foukakis T, Liden J, Thoppe SR, Dwight T, Hoog A, Zedenius J, Wallin G, Reimers M, Larsson C. Expression profiling reveals a distinct transcription signature in follicular thyroid carcinomas with a PAX8-PPARgamma fusion oncogene. Oncogene. 2005;24:1467–1476. doi: 10.1038/sj.onc.1208135. [DOI] [PubMed] [Google Scholar]
  38. Pasteris NG, Nagata K, Hall A, Gorski JL. Isolation, characterization, and mapping of the mouse Fgd3 gene, a new faciogenital dysplasia (FGD1; Aarskog Syndrome) gene homologue. Gene. 2000;242:237–247. doi: 10.1016/s0378-1119(99)00518-1. [DOI] [PubMed] [Google Scholar]
  39. Kim K, Hou P, Gorski JL, Cooper JA. Effect of Fgd1 on cortactin in Arp2/3 complex-mediated actin assembly. Biochemistry. 2004;43:2422–2427. doi: 10.1021/bi036173t. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Braga VM, Machesky LM, Hall A, Hotchin NA. The small GTPases Rho and Rac are required for the establishment of cadherin-dependent cell-cell contacts. J Cell Biol. 1997;137:1421–1431. doi: 10.1083/jcb.137.6.1421. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Sahai E, Marshall CJ. RHO-GTPases and cancer. Nat Rev Cancer. 2002;2:133–142. doi: 10.1038/nrc725. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Material
amjpathol_167_1_223__index.html (1.2KB, html)
amjpathol_167_1_223__1.pdf (179.5KB, pdf)

Articles from The American Journal of Pathology are provided here courtesy of American Society for Investigative Pathology

RESOURCES