Clinical Cancer Research Bridging the Lab and the Clinic in Cancer Medicine Infection and Cancer: Biology, Therapeutics, and Prevention
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Giordano, T. J.
Right arrow Articles by Koenig, R. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Giordano, T. J.
Right arrow Articles by Koenig, R. J.
Clinical Cancer Research Vol. 12, 1983-1993, April 2006
© 2006 American Association for Cancer Research

Human Cancer Biology

Delineation, Functional Validation, and Bioinformatic Evaluation of Gene Expression in Thyroid Follicular Carcinomas with the PAX8-PPARG Translocation

Thomas J. Giordano1, Amy Y.M. Au4, Rork Kuick2, Dafydd G. Thomas1,3, Daniel R. Rhodes1, Kenneth G. Wilhelm, Jr.3, Michelle Vinco1, David E. Misek2, Donita Sanders1, Zhaowen Zhu5, Raffaele Ciampi5, Samir Hanash2,6, Arul Chinnaiyan1, Roderick J. Clifton-Bligh4, Bruce G. Robinson4, Yuri E. Nikiforov5 and Ronald J. Koenig3

Authors' Affiliations: Departments of 1 Pathology, 2 Pediatrics, and 3 Internal Medicine, University of Michigan Medical School, Ann Arbor, Michigan; 4 Cancer Genetics Unit, Kolling Institute of Medical Research, University of Sydney, New South Wales, Australia; 5 Department of Pathology, University of Cincinnati College of Medicine, Cincinnati, Ohio; and 6 Fred Hutchinson Cancer Research Center, Seattle, Washington

Requests for reprints: Thomas J. Giordano, Department of Pathology, UH 2G332/0054, Ann Arbor, MI 48109-0054. Phone: 734-936-6776; Fax: 734-763-4095; E-mail: Giordano{at}umich.edu.


   Abstract
Top
 Abstract
Materials and Methods
 Results
 Discussion
 References
 
A subset of follicular thyroid carcinomas contains a balanced translocation, t(2;3)(q13;p25), that results in fusion of the paired box gene 8 (PAX8) and peroxisome proliferator-activated receptor {gamma} (PPARG) genes with concomitant expression of a PAX8-PPAR{gamma} fusion protein, PPFP. PPFP is thought to contribute to neoplasia through a mechanism in which it acts as a dominant-negative inhibitor of wild-type PPAR{gamma}. To better understand this type of follicular carcinoma, we generated global gene expression profiles using DNA microarrays of a cohort of follicular carcinomas along with other common thyroid tumors and used the data to derive a gene expression profile characteristic of PPFP-positive tumors. Transient transfection assays using promoters of four genes whose expression was highly associated with the translocation showed that each can be activated by PPFP. PPFP had unique transcriptional activities when compared with PAX8 or PPAR{gamma}, although it had the potential to function in ways qualitatively similar to PAX8 or PPAR{gamma} depending on the promoter and cellular environment. Bioinformatics analyses revealed that genes with increased expression in PPFP-positive follicular carcinomas include known PPAR target genes; genes involved in fatty acid, amino acid, and carbohydrate metabolism; micro-RNA target genes; and genes on chromosome 3p. These results have implications for the neoplastic mechanism of these follicular carcinomas.

Well-differentiated thyroid carcinoma, excluding medullary carcinoma, is divided into papillary, follicular, and oncocytic cell types. Papillary carcinoma is the most common form of thyroid cancer. It has several recognized histologic subtypes that share activating mutations of genes within the RET/RAS/BRAF/MAPK signaling pathway (1, 2), although each papillary carcinoma subtype has a distinct gene expression signature (3). Follicular carcinoma is either associated with mutations of the RAS gene family (4) or a distinctive translocation of chromosomes 2 and 3 that results in fusion of paired box gene 8 (PAX8) with peroxisome proliferator-activated receptor {gamma} (PPARG; ref. 5). Unlike the other thyroid carcinomas, the spectrum of mutations present in oncocytic carcinoma remains largely elusive, although mutations of GRIM-19 (NDUFA13), a gene involved in mitochondrial metabolism and regulation of cell death, recently have been identified in a minority of oncocytic carcinomas (6).

PAX8 encodes a transcription factor that is expressed at high levels in thyrocytes. It is necessary for normal thyroid development and it directs the expression of many thyroid-specific genes. PPARG encodes a nuclear hormone receptor transcription factor whose activity is related to adipocyte differentiation (7–9), lipid and carbohydrate metabolism (10), and cellular proliferation and differentiation. PPAR{gamma} is expressed at very low levels in normal thyroid and has no known function in that organ. In follicular carcinomas with the t(2;3)(q13;p25) translocation, the promoter and 5' coding region of PAX8 are fused in-frame with the coding region of PPAR{gamma}1, resulting in a fusion protein designated PPFP. PPFP is expressed at high levels in PAX8-PPARG translocation-positive follicular carcinomas and is thought to play an oncogenic role through several mechanisms (11, 12).

Despite their low incidence compared with papillary carcinomas, follicular carcinomas with the PAX8-PPARG translocation are of great interest for several reasons. First, balanced translocations in epithelial tumors are uncommon compared with mesenchymal and hematologic tumors, for reasons that are not entirely clear. Defining the distinctive characteristics of thyroid tumors that favor chromosomal rearrangements would potentially be very informative. Second, as a nuclear receptor, PPAR{gamma} displays myriad cellular functions, including potential roles in neoplastic transformation. A better understanding of its role in carcinogenesis is needed, and follicular carcinomas with this translocation should serve as a relevant model. Finally, molecularly targeted treatments for follicular carcinoma are needed; thus, an improved understanding of their biology and relevant therapeutic targets might be gained by elucidation of the consequences of the PAX8-PPARG translocation. With these rationales in mind, we derived the gene expression signature of a cohort of follicular carcinomas with this translocation and did supporting studies to corroborate the signature. Transfection experiments showed that the promoters of several genes in the PAX8-PPARG signature are activated by PPFP in a manner that partially overlaps the functions of PAX8 or PPAR{gamma}. Bioinformatic analysis of the signature revealed potential roles of several metabolic pathways in the oncogenic action of PPFP.


   Materials and Methods
Top
 Abstract
 Materials and Methods
Results
 Discussion
 References
 
Tumors, histopathology, and RNA isolation. A total of 93 unique thyroid samples consisting of 4 normal thyroids and 89 thyroid tumors [7 follicular carcinomas with the PAX8-PPARG translocation, 6 follicular carcinomas without the translocation (including Thy203, described below), 10 follicular adenomas, 8 oncocytic carcinomas, 7 oncocytic adenomas, and 51 papillary carcinoma] were used to generate the gene expression profiles. Cases were derived from the University of Michigan, the University of Cincinnati Medical Center, and the Cooperative Human Tissue Network. The 51 papillary carcinomas and the four normal thyroids were described previously (3). Institutional review board approval was obtained. All tumors were diagnosed using accepted morphologic criteria. Frozen section slides and original permanent sections were reviewed, when available, to confirm the diagnoses and ensure research tissues were in agreement with the final pathologic diagnosis. All tissues were processed and RNAs were extracted similarly, as previously described (13).

Microarray analysis. DNA microarray analysis was done using commercially available oligonucleotide DNA microarrays containing 22,283 probe sets (U133A GeneChip, Affymetrix, Santa Clara, CA) as reported (3, 14). cRNA preparation and hybridization, and scanning and image analysis of the arrays were done according to protocols of the manufacturer and as previously described (13), as was probe set intensity estimation and normalization. Our procedures gave average probe set intensities of ~1,500 units, which we log-transformed using log[max(x + 50,0) + 50]. Estimates of fold changes between groups are the antilogarithms of the differences in means of the log-transformed data. Normalized and raw versions of the data are publicly available at http://dot.ped.med.umich.edu:2000/pub/PPARG/index.html.

Quantitative reverse transcription-PCR and sequencing. Reverse transcription real-time PCR was done as previously described (15). The probes and labels, shown in Table 1 , were designed using Primer Express (ABI, Foster City, CA) and were obtained from Biosearch Technologies (Novato, CA). PCR conditions for each primer-probe combination were optimized for time, temperature, and magnesium concentration and done using a SmartCycler (Cepheid, Sunnyvale, CA). PCR products were sequenced in both directions by the University of Michigan DNA Sequencing Core.


View this table:
[in this window]
[in a new window]
  		Table 1. Primers, probes, and labels used in quantitative reverse transcription-PCR studies

 
Tissue array and immunohistochemistry. A thyroid tissue array was constructed for validation by immunohistochemisty studies. The four PPFP(+) follicular carcinomas from the University of Michigan were used along with two PPFP(–) follicular carcinomas, four papillary carcinomas, two follicular adenomas, and four normal thyroids. One-millimeter-diameter cores were arrayed in duplicate. Immunohistochemistry was done using a robotic autostainer (DAKO, Carpinteria, CA) and standard procedures using the Envision detection system (DAKO). The following antibodies and conditions were used: PPAR{gamma} (Santa Cruz Biotechnology, Santa Cruz, CA), 1:100 dilution, high-pH Tris antigen retrieval, 60 minutes room temperature incubation; enolase 3 (ENO3; BD Transduction Laboratories, San Jose, CA), 1:50 dilution, citrate buffer antigen retrieval, 60 minutes room temperature incubation; and aquaporin 7 (AQP7; Abcam, Cambridge, MA), 1:800 dilution, citrate buffer antigen retrieval, 30 minutes room temperature incubation.

Cell culture and transfection assays. All cells were maintained at 37°C with 5% CO2. JEG-3 human choriocarcinoma cells were cultured in Eagle's MEM with 10% fetal bovine serum and penicillin/streptomycin. N2a mouse preneuronal cells were cultured in DMEM with 10% fetal bovine serum and penicillin/streptomycin. Rat FRTL-5 thyroid cells were cultured in F12 Coon's media with 5% fetal bovine serum, six hormone combination (1 mU/mL bovine TSH, 4 ng/mL insulin, 10 ng/mL somatostatin, 5 µg/mL apotransferrin, 4 mg/mL hydrocortisone, and 10 ng/mL glycyl-L-histidyl-L-lysine acetate; Sigma, St. Louis, MO) and penicillin/streptomycin.

Whole thyroid glands were removed from dogs that had been previously anesthetized and exsanguinated as part of an unrelated, institutionally approved study. Thyroid glands were removed within 10 minutes of exsanguination. Glands were trimmed, minced, and primary cultures of thyrocytes were obtained following the method of Uyttersprot et al. (16).

The promoters of four genes that, according to our microarray data, were induced specifically in PPFP(+) follicular carcinomas were selected for analysis by transfection. The PCR was used with Accuprime Pfx polymerase (Invitrogen, Carlsbad, CA) to amplify human AQP7 bp –2,359 to +90 (the transcription start site is +1), angiopoietin-like protein 4 (ANGPTL4) bp –2,565 to +77, placental growth factor (PGF) bp –2,372 to +34, and ENO3 bp –2,808 to +56. The respective templates for these reactions were human genomic DNA and bacterial artificial chromosomes RP11-886P16, RP11-104F2, and RP5-1050D4. The 5' PCR primers contained an Mlu1 restriction enzyme site and the 3' primers contained either an Xho1 or Sal1 site. The PCR products were digested with the appropriate enzymes and ligated into the Mlu1 and Xho1 sites of pGL3-basic (Promega, Madison, WI). All constructs were confirmed by sequencing.

For transfection, cells were plated into 24-well clusters. The day before transfection, the medium was replaced to include charcoal-stripped serum. Transfections were done with LipofectAMINE and Plus reagents according to the protocol of the manufacturer (Invitrogen) in serum-free medium, and included 100 ng of the above-described pGL3-based firefly luciferase reporter plasmids, 100 ng transcription factor expression plasmid (PAX8, PPAR{gamma}, PPFP, or empty vector pCDNA3.1+; Invitrogen), and 0.5 to 1 ng of the internal control Renilla luciferase plasmid pRL-SV40 (Promega). After 3 hours of transfection, an equal volume of culture medium containing 20% charcoal-stripped FCS and penicillin/streptomycin was added to the wells. The next day, the culture medium was replaced with medium containing either 10 µmol/L PPAR{gamma} agonist ciglitazone (17) or vehicle ethanol (again with 10% stripped serum) for an additional 24 hours. The cells were lysed and analyzed for firefly and Renilla luciferase activities using the Promega dual luciferase reagents and protocol.

Enriched feature tests. We tested a selected set of 977 probe sets for overrepresentation of any Gene Ontology terms, GenMAPP maps using probe set annotation from Affymetrix (http://www.affymetrix.com/analysis/index.affx, version of May 31, 2005), as well as pathways defined in the Kyoto Encyclopedia of Genes and Genomes (http://www.genome.jp/kegg/) using methods similar to those previously reported (Thy203 was omitted in this analysis; ref. 18). The 22,283 U133A GeneChip probe sets were collapsed to 12,44 distinct genes with unambiguous Entrez gene numbers, which reduced the 977 probe sets to 761 genes (460 up, 301 down). Overrepresentation of each annotation term (e.g., membership in a particular pathway) in this set of genes was tested using one-sided Fisher's exact tests. To estimate the false discovery rates for the most significantly enriched terms, the resulting P values were compared with P values obtained from 100 data sets in which the 761 genes were randomly selected.

Bioinformatic analysis using oncomine. The Oncomine data mining platform (19) was used to compare the PPFP(+) and PPFP(–) follicular carcinoma gene expression profiles (including Thy203). Genbank accession IDs corresponding to Affymetrix probe set IDs were downloaded from Netaffx (www.affymetrix.com). Genbank IDs were mapped to Unigene Build 185. A map from Unigene to Entrez Gene ID was downloaded from Entrez Gene (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db = gene). The data set was base 2 log transformed (negative intensity values were removed) and median centered per array, and the SDs were normalized to one per array. Each gene was assessed for differential expression with Student's t test, done using the R statistical computing package. Tests were conducted both as two-sided for differential expression analysis and one-sided for overexpression analysis. To account for multiple hypothesis testing, Q values (estimated false discovery rates) were calculated as follows: Formula, where P is P value, N is the total number of genes analyzed, and R is the sorted rank of P value.

Gene set collection. All identifiers were mapped to Entrez Gene IDs for analysis. The 22,283 probe sets were collapsed to 13,046 distinct Entrez gene IDs. In the case of multiple probe sets per Entrez gene ID, the probe set with the minimum P value was kept. Sets of biologically related genes were collected or derived from a number of external resources; those relevant to the data presented here are as follows: chromosome arm mappings were downloaded from the National Center for Biotechnology Information Map Viewer (http://www.ncbi.nlm.nih.gov/mapview/), protein-protein interaction sets were downloaded from the Human Protein Reference Database (http://www.hprd.org/), and predicted micro-RNA (miRNA) target genes were downloaded from PicTar (http://pictar.bio.nyu.edu/; ref. 20).

Gene set analysis. Oncomine gene expression signatures were defined as the top 20% of Entrez gene IDs with enough nonnegative values to perform a t test, rank-ordered by their P values in each differential expression analysis. This constitutes 12,078 distinct genes, giving 2,415 in the top 20%. The association of a gene expression signature and the gene set was assessed with Fisher's exact test. The false discovery rate was again estimated using Q values, calculated as follows: Formula, where N is the number of gene sets of a given type tested against each gene expression signature and R is the ascending order rank of the respective P value.

Interactome. Approximately 16,000 known protein-protein interactions were downloaded from the Human Protein Reference Database (http://www.hprd.org; ref. 21), a manually curated database of pairs of proteins that have experimental evidence for physical interaction. Oncomine reports pairs of differentially expressed genes that encode proteins with documented protein-protein interactions. Oncomine generates interactome maps for the top 10% of genes rank-ordered by their P values in each differential expression analysis.


   Results
Top
 Abstract
 Materials and Methods
 Results
Discussion
 References
 
Gene expression profiling identifies follicular carcinomas with the PAX8-PPARG translocation. The main focus of this work was to identify the transcriptional changes that are specific to follicular carcinomas that contain the PAX8-PPARG translocation. For this purpose, we obtained gene expression profiles on 93 thyroid samples consisting of 4 normal thyroids and 89 thyroid tumors (13 follicular carcinomas, 10 follicular adenomas, 8 oncocytic carcinomas, 7 oncocytic adenomas, and 51 papillary carcinomas). It was possible to identify cases with the PAX8-PPARG translocation by examining the microarray data for increased expression of PPAR{gamma} (Fig. 1 ). High PPAR{gamma} transcript levels, compared with the other thyroid tumors, were present in seven of the follicular carcinomas. All the follicular patterned tumors (follicular carcinomas, follicular adenomas, oncocytic carcinomas, and oncocytic adenomas) were analyzed by reverse transcription-PCR for the presence of the fusion transcript (data not shown). The fusion transcript was detected in all seven follicular carcinomas with high PPAR{gamma} expression and in only one other sample, a follicular carcinoma (Thy203) that expressed very low levels of PPAR{gamma} by microarray (Fig. 1). By reverse transcription real-time PCR, the threshold for detection of the fusion transcript occurred 10 cycles later for Thy203 than for the seven follicular carcinomas with high PPAR{gamma} expression, indicating that Thy203 expresses the fusion transcript at ~0.1% the level of those seven follicular carcinomas. Further, real-time PCR for the 3' end of PPAR{gamma} showed an undetectable level of expression after 40 cycles of amplification (Table 2 ). Therefore, for analysis, we grouped Thy203 with the PPFP(–) follicular carcinomas. As expected from this extremely low-level expression, the microarray profile of Thy203 was similar to those of the five other PPFP(–) follicular carcinomas.


Figure 1
View larger version (10K):
[in this window]
[in a new window]
  		Fig. 1. Microarray analysis of PPAR{gamma} expression in benign and malignant thyroid samples. High expression of PPAR{gamma} correlates with PPFP expression, as analyzed by reverse transcription real-time PCR (not shown). Arrow, follicular carcinoma Thy203.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Quantitative reverse transcription-PCR validation of selected microarray data

 
Characterization of transcript fusions in the PPFP(+) follicular carcinomas and Thy203. PPFP transcripts have been reported to contain PAX8 exons 7, 8, or 9 fused to PPAR{gamma}1 exon 1 (11). Reverse transcription-PCR using a forward primer in PAX8 exon 7 and a reverse primer in PPAR{gamma}1 exon 1 followed by sequencing revealed that six of our seven PPFP(+) follicular carcinomas had transcripts with PAX8 exon 8 fused to PPAR{gamma}1 exon 1, and one (Thy150) had PAX8 exon 7 fused to PPAR{gamma}1 exon 1. Thy203 also showed fusion of PAX8 exon 8 to PPAR{gamma}1 exon 1.

Gene expression among follicular patterned lesions is a function of the PAX8-PPARG translocation. Principal component analysis was done to examine global differences in gene expression between samples. Principal component analysis of all follicular neoplasms (23 follicular carcinomas and follicular adenomas) revealed significant separation of the PPFP(+) follicular carcinomas from the PPFP(–) follicular carcinomas and the follicular adenomas (Fig. 2 ). This result indicates that the PAX8-PPARG translocation is the predominant source of the gene expression variation within this set of tumors. Thy203 plotted among the other PPFP(–) follicular carcinomas, providing further support for its inclusion in the PPFP(–) follicular carcinoma group.


Figure 2
View larger version (10K):
[in this window]
[in a new window]
  		Fig. 2. Principle component analysis of log-transformed data for all probe sets for PPFP(+) follicular carcinomas (PPFP + FC), PPFP(–) follicular carcinomas (PPFP – FC), follicular adenomas (FA), and normal thyroids. The first two principal components are plotted. The seven PPFP(+) follicular carcinomas are distinctly separated from the other thyroid follicular patterned tumors and the normal thyroid samples.

 
Gene expression profile of follicular carcinomas with the PAX8-PPARG translocation. The most direct way to define the expression profile of follicular carcinomas with the PAX8-PPARG translocation would be comparison of a large number of follicular carcinomas with and without the translocation. However, in general, follicular carcinomas are relatively rare thyroid tumors and microarray analysis requires frozen tissue. Thus, only 13 follicular carcinomas were available for analysis. Therefore, we used all of the data from the various tumor types to identify genes with larger (or smaller) mRNA levels in the seven PPFP(+) follicular carcinomas compared with the five PPFP(–) follicular carcinomas without this translocation (Thy203 was omitted), which also were increased (or decreased) compared with non-follicular carcinoma tumor samples and normal tissue. We asked that two-sample t tests give P < 0.01 for the comparison of PPFP(+) follicular carcinomas to PPFP(–) follicular carcinomas, as well as for the comparison of PPFP(+) follicular carcinomas to the set of nonfollicular carcinoma samples. We further asked that the fold difference between PPFP(+) follicular carcinomas and each of the six groups individually be at least 1.5 and be in the same direction. This selected a set of 322 probe sets, 239 of which had increased values in the PPFP(+) follicular carcinomas. To estimate the false discovery rate for this gene list, we permuted the sample labels 1,000 times, and on average obtained only 3.85 qualifying probe sets in the 1,000 resulting data sets, so that we estimate the false discovery rate to be ~1.2%. These 322 probe sets are identified in our Supplementary Data. When performing statistical tests for enriched features among sets of genes, below, we desired a somewhat larger list of genes, and for this used a weaker selection criterion that asked that the P values be <0.05 and the fold changes be at least 1.2. This selected 977 probe sets with an estimated false discovery rate of 11.4%.

We show a smaller subset in Fig. 3 that qualified under a similar but more stringent selection criteria that required the two P values to be <0.001 and the fold changes be at least 2.0. This selected 80 probe sets (67 up, 13 down) representing 68 distinct genes (55 up, 13 down), and gave an estimated false discovery rate of 0.07% using 1,000 permuted data sets. Note that in our data, PPARG is the most differentially expressed gene, but this reflects the expression of PPFP in tumors with the PAX8-PPARG translocation. We also examined the expression of several thyrocyte differentiation markers genes (SLC5A5, TG, TPO, and TSHR) between the PPFP(+) follicular carcinomas and the other follicular cohorts and found few significant changes.


Figure 3
View larger version (116K):
[in this window]
[in a new window]
  		Fig. 3. Gene expression signature of follicular carcinomas with the PAX8-PPARG translocation. Differentially expressed genes were identified by comparing follicular carcinomas with high levels of PPAR{gamma} to normal thyroid and the other types of thyroid tumors according selection criteria described in Results. This yielded 55 genes whose expression is greater in the PPFP(+) follicular carcinoma group and 13 genes whose expression is down in the PPFP(+) follicular carcinoma group. Fold change from the average of the two means for PPFP(+) follicular carcinomas and all other samples combined. FC(+), PPFP(+) follicular carcinoma; FC(–), PPFP(–) follicular carcinoma; FA, follicular adenoma; OA, oncocytic adenoma; OC, oncocytic carcinoma; PC, papillary carcinoma.

 
Validation of select genes by reverse transcription real-time PCR. To validate the microarray data, reverse transcription real-time PCR was done using RNA from a set of tumors that partially overlapped with the set used for DNA microarray analysis. PPARG and four additional genes with increased expression in the PPFP(+) follicular carcinomas were selected for validation (ANGPTL4, AQP7, ENO3, and PGF). The results, reported as the number of cycles needed to reach threshold (cycle to threshold, CT), are shown in Table 2 listed by histologic type and translocation status. Overall, the PCR results validate the microarray data, including classification of Thy203 as PPFP(–).

Validation of select proteins by immunohistochemistry. To validate the microarray data at the protein level, immunohistochemistry for PPAR{gamma} and two proteins (ENO3 and AQP7) identified in the PPFP(+) signature was done using a thyroid tissue array that contained four PPFP(+) follicular carcinomas as well as a 10 other thyroid tumors [including two PPFP(–) follicular carcinomas] and four normal thyroids. The results confirmed increased protein expression in PPFP(+) follicular carcinomas of PPAR{gamma} (four of four, 100%), ENO3 (three of four, 75%), and aquaporin (three of four, 75%) compared with normal thyroid and the other thyroid tumors (Fig. 4 ).


Figure 4
View larger version (111K):
[in this window]
[in a new window]
  		Fig. 4. Immunohistochemistry for PPAR{gamma}, ENO3, and AQP7. Representative results are shown for PPFP(+) follicular carcinoma, PPFP(–) follicular carcinoma, and papillary carcinoma. Note the nuclear staining for PPAR{gamma}, the cytoplasmic staining for ENO3, and the membranous staining for AQP7, as well as the absence of staining of intratumoral endothelial cells. (Original magnification, x 200).

 
Functional validation of the gene expression signature by transient transfection assays. Two of the genes most strongly induced specifically in the PPFP(+) follicular carcinomas, AQP7 and ANGPTL4, are induced by PPAR{gamma} in other tissues (22, 23). This suggests that PPFP might be inducing these genes in a PPAR{gamma}-like manner, which runs counter to the conventional view that PPFP blocks PPAR{gamma} action (11). Therefore, we used transient transfection to compare the abilities of PPFP, PPAR{gamma}, and PAX8 to regulate the AQP7 and ANGPTL4 promoters. The promoters from two additional genes induced specifically in the PPFP(+) follicular carcinomas, PGF and ENO3, were also studied. We transfected three different cell lines and primary cultures of dog thyrocytes to assess whether cell type–specific factors might regulate the response.

Preliminary studies were done to show the functional capacity of the primary dog thyrocyte cultures. The thyrocytes were transfected with a reporter plasmid in which the rat sodium iodide symporter gene upstream enhancer element and 2 kbp proximal promoter direct firefly luciferase expression (NIS-luc), together with a cytomegalovirus-Renilla luciferase internal control plasmid. Exposure to 15 mIU/mL TSH for 24 hours induced NIS-luc 2.3 ± 0.09-fold (n = 3), indicating that the cells are responsive to TSH. Separate immunohistochemical experiments showed uniformly positive thyroglobulin staining (data not shown). Therefore, the dog thyrocytes were deemed appropriate for study.

The AQP7 promoter was strongly induced by PPAR{gamma} and PPFP, but not by PAX8, in all four cell types (Fig. 5 ). In general, PPAR{gamma} and PPFP showed similar levels of induction in the presence of the PPAR{gamma} agonist ciglitazone, although PPFP tended to have stronger ligand-independent activity. For example, in JEG-3 cells, PPAR{gamma} induced luciferase 2.1-fold in the absence and 14-fold in the presence of ciglitazone, whereas the inductions with PPFP were 5.9- and 14-fold. Similarly, in primary cultures of dog thyrocytes, PPAR{gamma} induced luciferase 8.8-fold in the absence and 29-fold in the presence of ciglitazone, whereas the inductions with PPFP were 23- and 46-fold.


Figure 5
View larger version (15K):
[in this window]
[in a new window]
  		Fig. 5. Regulation of the AQP7 promoter. JEG-3 cells, N2a cells, FRTL-5 cells, or primary cultures of dog thyrocytes were cotransfected with an AQP7-luciferase plasmid, an internal control Renilla luciferase plasmid, and an expression vector for PAX8, PPAR{gamma}, PPFP, or empty vector. The cells were cultured ±ciglitazone for 2 days before determining reporter gene activities. Luciferase is normalized to Renilla luciferase, and for each experiment the normalized activity with empty vector –ciglitazone is defined as 1. Columns, mean for four to six experiments; bars, SE.

 
The ANGPTL4 promoter was less responsive than AQP7 and the data showed some cell type specificity, but the overall trend was similar with PPFP being at least as active as PPAR{gamma}, and PAX8 having no activity (Fig. 6 ). For example, in JEG-3 cells, PPAR{gamma} induced luciferase 1.2-fold in the absence and 2.4-fold in the presence of ciglitazone, whereas the inductions with PPFP were 1.9- and 3.4-fold. The ANGPTL4 promoter was not induced by either PPAR{gamma} or PPFP in FRTL-5 cells, but in dog thyrocytes PPFP expression resulted in a 3.5-fold induction in the absence and a 5.2-fold induction in the presence of ciglitazone. The response in N2a cells was qualitatively similar to that in dog thyrocytes, with PPAR{gamma} not inducing this promoter but PPFP resulting in inductions of 2.2- and 3.1-fold in the absence and presence of ciglitazone.


Figure 6
View larger version (16K):
[in this window]
[in a new window]
  		Fig. 6. Regulation of the ANGPTL4 promoter. JEG-3 cells, N2a cells, FRTL-5 cells, or primary cultures of dog thyrocytes were cotransfected with an ANGPTL4-luciferase plasmid, an internal control Renilla luciferase plasmid, and an expression vector for PAX8, PPAR{gamma}, PPFP, or empty vector. The cells were cultured ±ciglitazone for 2 days before determining reporter gene activities. Luciferase is normalized to Renilla luciferase, and for each experiment the normalized activity with empty vector –ciglitazone is defined as 1. Columns, mean for four to six experiments, except FRTL-5 (two experiments, each in triplicate); bars, SE.

 
In JEG-3 cells, the ENO3 promoter was also induced more strongly by PPFP (2.9-fold minus ciglitazone and 6.9-fold plus ciglitazone) than by PPAR{gamma} (1.3- and 3.8-fold; Fig. 7A ). However, this promoter was not induced by PPAR{gamma}, PPFP, or PAX8 in N2a cells, FRTL-5 cells, or dog thyrocytes (data not shown).


Figure 7
View larger version (9K):
[in this window]
[in a new window]
  		Fig. 7. Regulation of the ENO3 and PGF promoters. A, JEG-3 cells were cotransfected with an ENO3-luciferase plasmid, an internal control Renilla luciferase plasmid, and an expression vector for PAX8, PPAR{gamma}, PPFP, or empty vector. The cells were cultured ±ciglitazone for 2 days before determining reporter gene activities. Luciferase is normalized to Renilla luciferase, and for each experiment the normalized activity with empty vector –ciglitazone is defined as 1. Columns, mean for four experiments; bars, SE. B, similar to (A), except the reporter plasmid was driven by the PGF promoter and the transfection was in primary cultures of dog thyrocytes. Columns, mean for six experiments; bars, SE.

 
The PGF promoter was not induced by PPAR{gamma}, PPFP, or PAX8 in any of the cell lines (data not shown). However, in dog thyrocytes, PPFP caused inductions of 6.4-fold minus ciglitazone and 10-fold plus ciglitazone, compared with no induction by PPAR{gamma} and a modest ~2.5-fold induction by PAX8 (Fig. 7B).

Pathway analysis of the PAX8-PPARG signature genes. We analyzed the larger set of 977 probe sets found to be altered with the PAX8-PPARG translocation for enriched Gene Ontology terms, Kyoto Encyclopedia of Genes and Genomes pathways, and GenMAPP maps (Table 3 ). The most substantial enrichment was observed for pathways related to fatty acid metabolism. Induced genes in these pathways include several acyl-CoA dehydrogenases (ACADL, ACADM, ACADS), acetyl-CoA acyltransferases (ACAA1, ACAA2), and hydroxyacyl-CoA dehydrogenases (HADHA, HADHSC), all of which participate in fatty acid ß-oxidation. Other metabolic pathways also were enriched, such as Kyoto Encyclopedia of Genes and Genomes pathways valine, leucine, and isoleucine degradation, and glycolysis/gluconeogenesis. These results are striking because PPAR{gamma} regulates adipogenesis and glucose metabolism.


View this table:
[in this window]
[in a new window]
  		Table 3. Pathways enriched in follicular carcinomas with the PAX8-PPARG translocation

 
Bioinformatic analysis using Oncomine. The Oncomine data mining platform (www.oncomine.org/; ref. 19) was used to compare the PPFP(+) and PPFP(–) follicular carcinoma gene expression profiles (Thy203 included), as a means of exploring for differences of potential biological significance between these groups of follicular carcinomas. Genes located on chromosome 3p were found to be overrepresented, with 95 of 341 measured genes on 3p being in the top 20% of the PPFP(+) up-regulated profile (P = 5.1E–5, Q = 0.002; all other chromosome arms had Q values of at least 0.2). Presumably, this is a consequence of the t(2;3)(q13;p25) chromosomal translocation and may reflect strong PAX8 regulatory sequences from chromosome 2 exerting effects on chromosome 3p genes or other chromosome structural effects. Interestingly, the genes on 3p that are induced include two genes that are directly involved in fatty acid metabolism—carnitine/acylacrnitine translocase (SLC25A20), which transfers fatty acylcarnitines into mitochondria, and acetyl-CoA acyltransferase 1 (ACAA1), which participates in peroxisomal fatty acid ß oxidation. Thus, although these genes seem to be functionally related to PPAR{gamma} and it would be plausible to postulate that they are induced directly by PPFP, their membership in the set of induced genes from chromosome 3p suggests they may be induced secondary to the translocation itself.

Recently, it has become clear that miRNAs down-regulate the expression of a large number of genes posttranscriptionally by binding to short sequences in mRNA 3' untranslated regions. Each miRNA may regulate multiple mRNAs, and one mRNA may be regulated by multiple miRNAs. Oncomine uses PicTar (20) to analyze for miRNA target genes. Putative target genes for four miRNAs are strongly overrepresented among the up-regulated genes in PPFP(+) follicular carcinomas: miR-101 [104 of 329 measured target genes are in the top 20% of the PPFP(+) profile, P = 2.1E–7, Q = 3.6E–5], miR-30A-3P (55 of 160 measured target genes, P = 1.1E–5, Q = 9.3E–4), miR-200A (81 of 262 measured target genes, P = 1.2E–5, Q = 6.7E–4), and miR-199A (92 of 309 measured target genes, P = 1.7E–5, Q = 7.1E–4). Twenty-one up-regulated genes are putative targets for at least three of these four miRNAs, suggesting coordinate regulation. Included in this list are the oncogenes RUNX1/AML1 and SS18; PUM2, which encodes a protein thought to be involved in stem cell proliferation and self renewal; and NRP2, which encodes the vascular endothelial growth factor/PGF receptor neuropilin 2.

The Oncomine "Interactome" identifies known physically interacting proteins (based up the Human Protein Reference Database; ref. 21) among the differentially expressed genes. This analysis revealed correlations between the expression of PPAR{gamma} (which also measures PPFP) and two proteins that can function as PPAR{gamma} coactivators, GADD45G (r2 = 0.85) and NCOA4/ARA70 (r2 = 0.48). This suggests that, in PPFP(+) follicular carcinoma, the PPAR{gamma}-like transcriptional activity of PPFP may be magnified by increased expression of these proteins.

Interactome analysis also revealed that the set of genes with increased transcript expression in PPFP(+) follicular carcinomas includes the epidermal growth factor receptor (EGFR) and several EGFR-interacting proteins: BRAF, which is activated by EGFR; CRK, an adapter protein that participates in EGFR-mediated BRAF activation; VAV2, an oncogene that is phosphorylated by EGFR; STATs 1 and 5B, which also are activated by EGFR; the ERBB3 oncogene, which dimerizes with EGFR and is amplified in numerous cancers; PTK2, a tyrosine kinase that binds to and helps transmit motility signals from the EGFR; and HBEGF, which binds and activates EGFR with greater potency than EGF. The Interactome analysis also allows one to visualize overall networks of interactions by drawing an interaction map. This reveals that the EGFR is a central node that connects to numerous other up-regulated genes, including the oncogenes BRAF, PTK2, and EPHA2 (Fig. 8 ).


Figure 8
View larger version (17K):
[in this window]
[in a new window]
  		Fig. 8. EGFR-related protein interaction network for up-regulated genes in PPFP(+) versus PPFP(–) follicular carcinomas. A protein interaction network was generated by Oncomine illustrating physical interactions between the EGFR and other up-regulated genes in PPFP(+) follicular carcinomas, based up on the Human Protein Reference Database (21). Proteins with direct physical interactions are represented as nodes connected by lines. EPHA2, PTK2, EGFR, and BRAF are indicated by enlarged nodes to signify that they are known oncoproteins.

 

   Discussion
Top
 Abstract
 Materials and Methods
 Results
 Discussion
References
 
In their original description of the PPFP fusion protein, Kroll et al. (11) used transiently transfected U2OS (osteosarcoma) cells to show that PPFP does not activate PPAR-responsive promoters. Instead, PPFP functioned as a dominant-negative inhibitor of PPAR{gamma}-induced reporter gene activation. This led to the hypothesis that inhibition of endogenous PPAR{gamma} is an important mechanism by which PPFP causes follicular carcinomas to develop or progress. A number of other observations would also be consistent with this hypothesis. There are no thyroid cancer cell lines that express PPFP; however, a PPAR{gamma} agonist caused growth inhibition of thyroid cancer cell lines that express PPAR{gamma} (24–26), and forced overexpression of PPAR{gamma} reduced cell growth in a thyroid cancer cell line that does not express endogenous PPAR{gamma} (24). Stable transfection of the SV40 T antigen-transformed human thyroid-derived cell line Nthy-ori with PPFP led to increased growth in soft agar, and treatment of Nthy-ori cells with a PPAR{gamma} antagonist had a similar effect (12).

Comparing the gene expression profiles of follicular carcinomas that express PPFP versus those that lack the translocation should provide insight into the mechanism of action of this fusion protein. Although the data summarized above would suggest that PPFP functions by inhibiting thyroid PPAR{gamma}, two of the genes most strongly induced in PPFP(+) follicular carcinomas, AQP7 and ANGPTL4, are induced by PPAR{gamma} in other tissues (22, 23) and a PPAR response element has been identified in the mouse AQP7 promoter. Thus, we considered the hypothesis that PPFP might function in a PPAR{gamma}-like manner, at least on some target genes. To test this hypothesis, we did a reporter gene analysis of AQP7 and ANGPTL4 promoter-luciferase constructs cotransfected with PPFP, PPAR{gamma}, PAX8, or empty vector. The promoters of two additional genes induced in the PPFP(+) follicular carcinomas, ENO3 and PGF, were also evaluated. The major conclusions from these experiments are that PPFP can indeed function in a PPAR{gamma}-like manner, although it also has transcriptional properties distinct from either PAX8 or PPAR{gamma}, and the effects of this fusion protein can be cell type dependent. Thus, the concept of PPFP contributing to follicular carcinoma largely by antagonizing endogenous PPAR{gamma} needs reevaluation. Endogenous PPAR{gamma} is expressed at very low levels in thyrocytes and has no known function in these cells. Furthermore, there are other ways of blocking endogenous PPAR{gamma}, yet follicular carcinomas have not been reported to develop these mechanisms. For example, germ line dominant-negative mutations in the ligand-binding domain of PPAR{gamma} cause severe insulin resistance (27), yet similar mutations have not been described in follicular carcinomas.

Because PPFP is a transcription factor, it likely regulates a number of genes that contribute to the phenotype of follicular carcinoma. PGF and ANGPTL4 are angiogenic factors that are overexpressed in nonthyroid malignancies (28–30), and hence could be important in the development or progression of PPFP(+) follicular carcinoma. PGF also is induced in the thyroid glands of patients with Graves' disease and rats treated with thiouracil, suggesting that it participates in the physiologic vascular response that accompanies goiter formation (31). The modest induction of the PGF promoter by PAX8 in our transfection studies may in fact be part of the mechanism of this physiologic response, and suggests that PPFP also may have PAX8-like activity on some target genes. AQP7 is a glycerol channel (32, 33) and could provide energy for follicular carcinomas, especially because our microarray data indicate follicular carcinomas express glycerol kinase (as does normal thyroid; ref. 34).

The gene expression signature of PPFP(+) follicular carcinomas was most significantly enriched in pathways related to fatty acid ß oxidation and metabolism, with significant enrichments also in pathways related to amino acid and carbohydrate metabolism (Table 2). This is striking considering the known roles of PPAR{gamma} in lipid and carbohydrate metabolism. Interestingly, however, it is primarily PPAR{alpha} that stimulates fatty acid ß oxidation (35), suggesting that PPFP also has PPAR{alpha}-like activity. PPARs {alpha} and {gamma} bind to the same response elements, and the factors that dictate their distinct activities are not fully defined.

Oncomine and PicTar revealed a striking enrichment of putative miRNA target genes among the up-regulated genes in PPFP(+) follicular carcinomas. Included among these are genes plausibly connected to cancer, such as SS18, RUNX1, PUM2, and NRP2. Because miRNAs down-regulate gene expression, these data suggest that certain miRNAs are under expressed in PPFP(+) follicular carcinoma, which would be consistent with the observed decrease in miRNA expression in other cancers (36). Recently, miRNA expression profiles have been shown to accurately predict subtypes of acute lymphoblastic leukemia (36). Our analysis suggests that miRNA profiling also might distinguish subtypes of follicular thyroid cancer.

Interactome analysis revealed that the PPFP(+) follicular carcinoma gene signature includes EGFR and numerous proteins that interact directly with EGFR. In addition, EGFR is a central node in a large network of interacting proteins with increased transcript expression in PPFP(+) follicular carcinomas (Fig. 8). This suggests that inhibition of EGFR could have multiple beneficial downstream effects, making it a potentially attractive target in PPFP(+) follicular carcinomas. Inhibitors of the EGFR tyrosine kinase currently are in use to treat other cancers (37).

The potential value of EGFR inhibitors is but one link to clinical oncologic practice provided by our data. We also show increased expression of the angiogenic factors ANGPTL4 and PGF, suggesting that antiangiogenic agents may be therapeutic. Diagnostically, increased expression of PPAR{gamma} would almost always indicate follicular thyroid cancer because we found that high expression of PPAR{gamma} uniformly signifies expression of PPFP, which is rarely found in benign lesions.

Two recent reports have examined gene expression in PPFP(+) versus PPFP(–) follicular carcinomas, although neither study provided functional validation of their findings. Lui et al. (38) did Affymetrix profiling of four follicular carcinomas with the rearrangement and five without. For unknown reasons, there is virtually no overlap between their PPFP-specific gene signature and ours. While our manuscript was in preparation, Lacroix et al. (39) published gene expression profiling of four follicular tumors containing PPFP (one of which was benign) and eight follicular tumors without the fusion protein (three of which were benign). Our PPFP signature overlaps substantially with theirs, including all four genes that we studied by transfection (AQP7, ANGPTL4, ENO3, and PGF), as well as the enrichments in pathways related to lipid, glucose, and amino acid metabolism. In addition, Lacroix et al. used computational methods to identify putative PPAR binding sites in the promoters of many genes in their PPFP(+) profile. The concordance of our findings, which were done on different platforms, substantiates our conclusions but does not provide insight into the discrepant results of Lui et al. It remains to be determined which PPFP target genes are important in the development and progression of follicular carcinoma.


   Acknowledgments
 
We thank the many current and former technicians of the Tissue Procurement Service of the University of Michigan Comprehensive Cancer Center, Barbara Lamb for technical assistance with the DNA microarray analysis, Nancy McAnsh for immunohistochemical expertise, Mary Hong for organizational assistance, and Dr. Sissy M. Jhiang (Department of Physiology and Cell Biology, Ohio State University, Columbus, Ohio) for providing NIS-luc.


   Footnotes
 
Grant support: Marilynn Collins Thyroid Cancer Fund and the Millie Schembechler Endocrine Cancer Program of the University of Michigan Comprehensive Cancer; the Michigan National Institutes of Diabetes, Digestive, and Kidney Diseases Biotechnology Center (NIH DK58771); the Cell and Molecular Biology Core of the Michigan Diabetes Research and Training Center (NIH DK20572); NIH (NIH CA88041 and NIH 5P30CA46592); National Cancer Institute Director's Challenge Program at the University of Michigan (NIH CA84952); and the University of Michigan Comprehensive Cancer Center Bioinformatics Core (NIH 5P30CA46592).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

Received 9/19/05; revised 1/18/06; accepted 1/20/06.


   References
Top
 Abstract
 Materials and Methods
 Results
 Discussion
 References
 

  1. Kimura ET, Nikiforova MN, Zhu Z, Knauf JA, Nikiforov YE, Fagin JA. High prevalence of BRAF mutations in thyroid cancer: genetic evidence for constitutive activation of the RET/PTC-RAS-BRAF signaling pathway in papillary thyroid carcinoma. Cancer Res 2003;63:1454–7.[Abstract/Free Full Text]
  2. Melillo RM, Castellone MD, Guarino V, et al. The RET/PTC-RAS-BRAF linear signaling cascade mediates the motile and mitogenic phenotype of thyroid cancer cells. J Clin Invest 2005;115:1068–81.[CrossRef][Medline]
  3. Giordano TJ, Kuick R, Thomas DG, et al. Molecular classification of papillary thyroid carcinoma: distinct BRAF, RAS, and RET/PTC mutation-specific gene expression profiles discovered by DNA microarray analysis. Oncogene 2005;24:6646–56.[CrossRef][Medline]
  4. Nikiforova MN, Lynch RA, Biddinger PW, et al. 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;88:2318–26.[Abstract/Free Full Text]
  5. Kroll TG, Sarraf P, Pecciarini L, et al. PAX8-PPAR {gamma} 1 fusion in oncogene human thyroid carcinoma. Science 2000;289:1357–60.[Abstract/Free Full Text]
  6. Maximo V, Botelho T, Capela J, et al. Somatic and germline mutation in GRIM-19, a dual function gene involved in mitochondrial metabolism and cell death, is linked to mitochondrion-rich (Hurthle cell) tumours of the thyroid. Br J Cancer 2005;92:1892–8.[CrossRef][Medline]
  7. Brun RP, Spiegelman BM. PPAR {gamma} and the molecular control of adipogenesis. J Endocrinol 1997;155:217–8.[CrossRef][Medline]
  8. Chawla A, Schwarz EJ, Dimaculangan DD, Lazar MA. Peroxisome proliferator-activated receptor (PPAR) {gamma}: adipose-predominant expression and induction early in adipocyte differentiation. Endocrinology 1994;135:798–800.[Abstract]
  9. Lazar MA. PPAR {gamma}, 10 years later. Biochimie 2005;87:9–13.[Medline]
  10. Picard F, Auwerx J. PPARg and glucose homeostasis. Annu Rev Nutr 2002;22:167–97.[CrossRef][Medline]
  11. Kroll TG, Sarraf P, Pecciarini L, et al. PAX8-1 fusion oncogene in human thyroid carcinoma [corrected]. Science 2000;289:1357–60.[Abstract/Free Full Text]
  12. Powell JG, Wang X, Allard BL, et al. The PAX8/PPAR{gamma} fusion oncoprotein transforms immortalized human thyrocytes through a mechanism probably involving wild-type PPAR{gamma} inhibition. Oncogene 2004;23:3634–41.[CrossRef][Medline]
  13. Giordano TJ, Shedden KA, Schwartz DR, et al. Organ-specific molecular classification of primary lung, colon, and ovarian adenocarcinomas using gene expression profiles. Am J Pathol 2001;159:1231–8.[Abstract/Free Full Text]
  14. Shedden K, Chen W, Kuick R, et al. Comparison of seven methods for producing Affymetrix expression scores based on false discovery rates in disease profiling data. BMC Bioinformatics 2005;6:26.[CrossRef][Medline]
  15. Thomas DG, Giordano TJ, Sanders D, et al. Expression of receptor tyrosine kinases epidermal growth factor receptor and HER-2/neu in synovial sarcoma. Cancer 2005;103:830–8.[CrossRef][Medline]
  16. Uyttersprot N, Costagliola S, Miot F. A new tool for efficient transfection of dog and human thyrocytes in primary culture. Mol Cell Endocrinol 1998;142:35–9.[CrossRef][Medline]
  17. Willson TM, Cobb JE, Cowan DJ, et al. The structure-activity relationship between peroxisome proliferator-activated receptor {gamma} agonism and the antihyperglycemic activity of thiazolidinediones. J Med Chem 1996;39:665–8.[CrossRef][Medline]
  18. Creighton C, Kuick R, Misek DE, et al. Profiling of pathway-specific changes in gene expression following growth of human cancer cell lines transplanted into mice. Genome Biol 2003;4:R46.[CrossRef][Medline]
  19. Rhodes DR, Yu J, Shanker K, et al. ONCOMINE: a cancer microarray database and integrated data-mining platform. Neoplasia 2004;6:1–6.[Medline]
  20. Krek A, Grun D, Poy MN, et al. Combinatorial microRNA target predictions. Nat Genet 2005;37:495–500.[CrossRef][Medline]
  21. Peri S, Navarro JD, Amanchy R, et al. Development of human protein reference database as an initial platform for approaching systems biology in humans. Genome Res 2003;13:2363–71.[Abstract/Free Full Text]
  22. Kishida K, Shimomura I, Nishizawa H, et al. Enhancement of the aquaporin adipose gene expression by a peroxisome proliferator-activated receptor {gamma}. J Biol Chem 2001;276:48572–9.[Abstract/Free Full Text]
  23. Yoon JC, Chickering TW, Rosen ED, et al. Peroxisome proliferator-activated receptor {gamma} target gene encoding a novel angiopoietin-related protein associated with adipose differentiation. Mol Cell Biol 2000;20:5343–9.[Abstract/Free Full Text]
  24. Martelli ML, Iuliano R, Le Pera I, et al. Inhibitory effects of peroxisome poliferator-activated receptor {gamma} on thyroid carcinoma cell growth. J Clin Endocrinol Metab 2002;87:4728–35.[Abstract/Free Full Text]
  25. Park JW, Zarnegar R, Kanauchi H, et al. Troglitazone, the peroxisome proliferator-activated receptor-{gamma} agonist, induces antiproliferation and redifferentiation in human thyroid cancer cell lines. Thyroid 2005;15:222–31.[CrossRef][Medline]
  26. Klopper JP, Hays WR, Sharma V, Baumbusch MA, Hershman JM, Haugen BR. Retinoid X receptor-{gamma} and peroxisome proliferator-activated receptor-{gamma} expression predicts thyroid carcinoma cell response to retinoid and thiazolidinedione treatment. Mol Cancer Ther 2004;3:1011–20.[Abstract/Free Full Text]
  27. Barroso I, Gurnell M, Crowley VEF, et al. Dominant negative mutations in human PPAR {gamma} associated with severe insulin resistance, diabetes mellitus and hypertension. Nature 1999;402:880–3.[Medline]
  28. Chen CN, Hsieh FJ, Cheng YM, et al. The significance of placenta growth factor in angiogenesis and clinical outcome of human gastric cancer. Cancer Lett 2004;213:73–82.[CrossRef][Medline]
  29. Donnini S, Machein MR, Plate KH, Weich HA. Expression and localization of placenta growth factor and PlGF receptors in human meningiomas. J Pathol 1999;189:66–71.[CrossRef][Medline]
  30. Le Jan S, Amy C, Cazes A, et al. Angiopoietin-like 4 is a proangiogenic factor produced during ischemia and in conventional renal cell carcinoma. Am J Pathol 2003;162:1521–8.[Abstract/Free Full Text]
  31. Viglietto G, Romano A, Manzo G, et al. Upregulation of the angiogenic factors PlGF, VEGF and their receptors (Flt-1, Flk-1/KDR) by TSH in cultured thyrocytes and in the thyroid gland of thiouracil-fed rats suggest a TSH-dependent paracrine mechanism for goiter hypervascularization. Oncogene 1997;15:2687–98.[CrossRef][Medline]
  32. Maeda N, Funahashi T, Hibuse T, et al. Adaptation to fasting by glycerol transport through aquaporin 7 in adipose tissue. Proc Natl Acad Sci U S A 2004;101:17801–6.[Abstract/Free Full Text]
  33. Hara-Chikuma M, Sohara E, Rai T, et al. Progressive adipocyte hypertrophy in aquaporin-7-deficient mice: adipocyte glycerol permeability as a novel regulator of fat accumulation. J Biol Chem 2005;280:15493–6.[Abstract/Free Full Text]
  34. Su AI, Cooke MP, Ching KA, et al. Large-scale analysis of the human and mouse transcriptomes. Proc Natl Acad Sci U S A 2002;99:4465–70.[Abstract/Free Full Text]
  35. Lee SS, Pineau T, Drago J, et al. Targeted disruption of the {alpha} isoform of the peroxisome proliferator-activated receptor gene in mice results in abolishment of the pleiotropic effects of peroxisome proliferators. Mol Cell Biol 1995;15:3012–22.[Abstract]
  36. Lu J, Getz G, Miska EA, et al. MicroRNA expression profiles classify human cancers. Nature 2005;435:834–8.[CrossRef][Medline]
  37. Baselga J, Arteaga CL. Critical update and emerging trends in epidermal growth factor receptor targeting in cancer. J Clin Oncol 2005;23:2445–59.[Abstract/Free Full Text]
  38. Lui WO, Foukakis T, Liden J, et al. Expression profiling reveals a distinct transcription signature in follicular thyroid carcinomas with a PAX8-PPAR{gamma} fusion oncogene. Oncogene 2005;24:1467–76.[CrossRef][Medline]
  39. Lacroix L, Lazar V, Michiels S, et al. Follicular thyroid tumors with the PAX8-1 rearrangement display characteristic genetic alterations. Am J Pathol 2005;167:223–31.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 Endocr. Rev.Home page
M. Eszlinger, K. Krohn, A. Kukulska, B. Jarzab, and R. Paschke
Perspectives and Limitations of Microarray-Based Gene Expression Profiling of Thyroid Tumors
Endocr. Rev., May 1, 2007; 28(3): 322 - 338.
[Abstract] [Full Text] [PDF]

Home page
 EndocrinologyHome page
H. V. Reddi, B. McIver, S. K. G. Grebe, and N. L. Eberhardt
The Paired Box-8/Peroxisome Proliferator-Activated Receptor-{gamma} Oncogene in Thyroid Tumorigenesis
Endocrinology, March 1, 2007; 148(3): 932 - 935.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Giordano, T. J.
Right arrow Articles by Koenig, R. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Giordano, T. J.
Right arrow Articles by Koenig, R. J.