HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Services Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (13) Citing Articles via Google Scholar Google Scholar Articles by Carinci, F. Articles by Volinia, S. Search for Related Content PubMed PubMed Citation Articles by Carinci, F. Articles by Volinia, S.

Genetic Expression Profiling of Six Odontogenic Tumors

¹ Department of Maxillofacial Surgery, University of Ferrara;
² Department of Morphology and Embryology, University of Ferrara, Via Fossato di Mortara 64/b, 44100 Ferrara, Italy;
³ Dental Clinic, University of Chieti;
⁴ Department of Pathology, University of Ancona;
⁵ Histology, University of Ferrara;
⁶ Institute of Histology and Centre of Molecular Genetic, CARISBO Foundation, University of Bologna;
⁷ Institute of Histology and General Embryology, University of Bologna; and
⁸ TIGEM-Telethon Institute of Genetics and Medicine, Via P. Castellino, Napoli;

*corresponding author, enr{at}unife.it

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS & METHODS RESULTS DISCUSSION REFERENCES

 
Odontogenic tumors are rare neoplasms arising from the odontogenic apparatus. We aimed to identify molecular characteristics associated with odontogenic tumorigenesis and malignancy. To this end, we investigated the expression level of human genes by using, for the first time in odontogenic tumors, the technique of expression profiling. Gene expression alterations common to all six odontogenic tumors were identified by the use of cDNA microarrays containing 19,000 human cDNAs. Statistical analysis on a subset of 4974 cDNAs present in the biopsies identified 506 distinct genes associated with the tumors (p-value < 0.01). Gene ontology analysis of the cellular processes which were differentially regulated in odontogenic tumors was accomplished by the use of a subset of 1409 annotated genes. Finally, 43 cDNAs differentiated the three malignant odontogenic tumors (ameloblastic carcinoma, clear cell odontogenic tumor, granular cell odontogenic tumor) from the three benign ameloblastoma biopsies (p < 0.01). The identified genes might help us better classify borderline odontogenic tumors.

KEY WORDS: expression profiling • DNA microarray • odontogenic tumors

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS & METHODS RESULTS DISCUSSION REFERENCES

 
The classification of rare neoplasms arising from the odontogenic apparatus has undergone many changes since the first classification by the World Health Organization (Pindborg et al., 1972) and is sometimes controversial. Ameloblastoma is the most frequent benign tumor, is locally aggressive, and, for about 80%, is located in the mandible, more often in the molar area. Histologically, the ameloblastoma consists of proliferating odontogenic epithelium that forms sheets, islands, and cords in which the peripheral layer is composed of columnar or cuboid cells resembling ameloblasts, and the central mass is formed by stellate cells resembling the stellate reticulum of the enamel organ (Neville et al., 2002). Malignant odontogenic tumors are very rare. Among these entities, ameloblastic carcinoma (AC) represents an aggressive odontogenic neoplasm of the jaws, more common in the mandible, in which both the microscopic pattern of ameloblastoma and cytological features of malignancy are shown by the primary tumor and/or by any metastasis (Infante-Cossio et al., 1998). It may eventually arise as a result of malignant changes in a pre-existing ameloblastoma. In the last WHO classification, clear cell odontogenic tumor (CCOT) was classified as a benign locally invasive lesion (Kramer et al., 1992). However, it has now become evident that CCOT is capable of locally destructive growth, recurrences, and both nodal and distant metastases, and should be designated as a carcinoma (Li et al., 2001;Iezzi et al., 2002). The anterior portions of the jaws, especially the mandible, are most frequently affected. The histopathological pattern is characterized by large and irregular islands and sheets of clear cells, cells with eosinophilic cytoplasm, and squamoid cells separated by thin collagenous stroma. Granular cell odontogenic tumor (GCOT) is extremely rare and is considered a benign neoplasm. Histologically, it consists of densely packed granular cells immersed in a fibroblastic stroma, with interspersed strands of apparently inactive odontogenic epithelium (Ardekian et al., 1998). In the latest WHO histological typing of odontogenic tumors, lesions with a predominant granular cell component have been considered among ameloblastic fibroma (Kramer et al., 1992). More recently, tumors predominantly composed of granular cells and showing ameloblastic and/or odontogenic features have been considered a distinct entity and named granular cell odontogenic tumors (GCOT) (Gesek et al., 1995;Yih et al., 1995;Ardekian et al., 1998). At present, only one case of malignant GCOT has been reported (Carinci et al., 2003). The pathogenesis for odontogenic tumor is unknown, and studies are currently under way in many laboratories with the goal of detailing perturbations in cell cycling, proto-oncogene expression, and tumor suppressor mutations. However, these studies are complicated by the diseases‘ rarity, and often knowledge is derived from case reports or small series reports.

DNA microarrays are emerging as a technology that allows for the identification of novel cancer-related genes (Golub et al., 1999) and the fine molecular classification of human cancer (Francioso et al., 2002). They facilitate a comparative screening of gene expression at the genome-wide level, i.e., many thousands of genes are investigated in parallel, rather than with the traditional serial approach. In the present study, we planned to define the genetic expression profile of odontogenic tumors. By applying this technology for the first time to odontogenic tumors, we compared three cases of ameloblastoma with three cases of malignant tumors (i.e., one ameloblastic carcinoma, one clear cell odontogenic tumor, and one granular cell odontogenic tumor). The aims of the research were to detect alterations in gene expression and to identify molecular characteristics associated with odontogenic tumorigenesis, as well as to identify differences in gene expression patterns between benign and malignant tumors.

	MATERIALS & METHODS

TOP ABSTRACT INTRODUCTION MATERIALS & METHODS RESULTS DISCUSSION REFERENCES

 
Patients and Tissue Specimens
Samples were derived from surgically resected odontogenic tumors from three patients with ameloblastoma and three with malignant odontogenic tumors (i.e., one AC, one CCOT, and one GCOT). Informed consent to a protocol reviewed and approved by the Arcispedale S. Anna (Ferrara) Institutional Ethical Committee was obtained from patients to use their tissue specimens for research purposes. Part of each sample was frozen in liquid nitrogen and stored at -80°C until RNA extraction, and the remaining part was used for histopathological analysis.

Microarray Assays
DNA microarrays allow for the genome-wide analysis of gene expression in qualitative and quantitative fashion. They detect the relative expression levels of the genes in the tissue under investigation and in a reference tissue. Technically, reference RNA and tumor RNA are retro-transcribed and labeled with different fluorescent dyes (Cy3 for the reference and Cy5 for the tumor), and finally hybridized to a robotically printed cDNA microarray. The slides are then scanned with a high-resolution laser-scanning device, and false color images are generated for each hybridization. Genes up-regulated in the tumor are visualized in red, whereas those with decreased expression appear green.

Total RNA was extracted from the biopsies by means of RNAzol. Ten micrograms of total RNA were used for each labeling reaction. cDNA was synthesized by Superscript II (Life Technologies, Carlsbad, CA, USA) and amino-allyl dUTP (Sigma, Milwaukee, WI, USA). Mono-reactive Cy3 and Cy5 esters (Amersham Pharmacia, Piscataway, NJ, USA) were used for indirect cDNA labeling. A pool of 8 normal gum mucosa RNA derived from healthy patients was labeled with Cy3 and used as reference tissue against the Cy5-labeled tumor cDNA. Human 19.2 K DNA microarrays, containing 15,448 annotated genes and 3684 anonymous expressed tag sequences (EST), were used (Ontario Cancer Institute, Toronto, ON, Canada, http://www.uhnres.utoronto.ca/services/microarray/). One hundred microliters of the sample and control cDNA in DIG Easy hybridization solution (Roche, Basel, Switzerland) were used in a sandwich hybridization of the two slides constituting the 19.2 K set at 37°C overnight. Washing was performed three times for 10 min with 1 x SSC, 0.1% SDS at 42°C, and three times for 5 min with 0.1 x SSC at room temperature. Slides were dried by centrifugation for 2 min at 2000 rpm. Hybridized slides were scanned by means of the GenePix 4000A scanner (Axon Instruments, Foster City, CA, USA). Maximum intensity is at 65,500 in both channels, although we scanned the arrays to ensure that less than 1% of the total number of spots measured on the slide had median intensities close to maximal. Arabidopsis RNA was used as a reference for RNA labeling.

Images were analyzed by means of GENEPIX PRO 3.0 (Axon Instruments, Foster City, CA, USA). Spots showing no signal or obvious defects were flagged accordingly by GENEPIX PRO and excluded from the analysis. Local background was subtracted from the remaining spots, and the ratios of net fluorescence from the Cy5-specific channel to the net fluorescence from the Cy3-specific channel were calculated for each spot, representing tumor mRNA expression relative to the reference tissue. In each single DNA array, spots with absolute expression levels, after the removal of local background, of less than 3 x SD above background were flagged as absent, since ratios might not be reliable below that detection level. A normalization factor was estimated for each different array by the use of GP3 Perl script adapted to Linux (GP3 z-score normalization; Fielden et al., 2002). The z-score normalization process uses a subset of the log2 background-corrected signals for each channel. The script removes the outliers (5% off each end) at the extremes of the data distribution and uses the remaining 90% of original data points to calculate means and standard deviations.

DNA spots which passed the abovementioned measurements in at least 75% of the arrays were input into the expression table (4974 ESTs) as log2 of the ratio of Cy5/Cy3 background-corrected and normalized signals. To select the DNA spots (genes or ESTs) which distinguish between two sample groups (e.g., ameloblastoma vs. malignant odontogenic tumors), we used a program performing t test analysis. Multiple t testing correction was applied by the Benjamini and Hochberg procedure (Francioso et al., 2002), which controls the false-positive discovery rate (pFDR), defined as the proportion of spots expected to be identified by chance (assuming that spots are independent) relative to the total number of spots called significant. This procedure provides a good balance between discovery of significant spots and protection against false-positives, since occurrence of the latter is held to a small proportion of the list. A balanced bootstrap analysis with 10,000 permutations was therefore applied to the tests to define pFDR.

Gene ontology analysis was performed by the GOAL application (Volinia et al., submitted). The goal of the Gene Ontology^TM (GO) Consortium is to produce a dynamic controlled vocabulary that can be applied to all organisms even as knowledge of gene and protein roles in cells is accumulating and changing (Asburner et al., 2000). GO provides controlled vocabularies for the description of the molecular function, biological process, and cellular component of gene products. The GOAL application identifies GO terms which are differentially expressed in two different biopsy groups by calculation of average t scores. Bootstrap analysis is performed to associate p-values to t scores.

We used principal component analysis (PCA) function in Jexpress v.2.1 (www.molmine.com) to visualize the separation of biopsies by expression profiles in a bidimensional space. PCA involves a mathematical procedure that transforms several (possibly) correlated variables into a (smaller) number of uncorrelated variables called "principal components". The first principal component accounts for as much of the variability in the data as possible, and each succeeding component accounts for as much of the remaining variability as possible.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS & METHODS RESULTS DISCUSSION REFERENCES

 
We compared the expression profiles of benign and malignant odontogenic tumors by hybridization to microarrays containing 19,200 cDNAs. Subsequent statistical analysis was performed with the 4974 selected genes and ESTs present in 75% of the arrays, as detailed in MATERIALS & METHODS. By using the t test, we identified gene expression alterations common to all six of the odontogenic tumors analyzed. cDNAs with p-value < 0.01 (506 distinct) separated into two groups by principal component analysis (Fig.). In parallel, we evaluated the cellular processes which are differentially regulated in all six odontogenic tumors when compared with the pool of eight non-tumor biopsies. To this end, Gene Ontology (GO) analysis was performed by means of the GOAL application (Volinia et al., submitted), which calculates the average t score for every GO term associated with all the annotated ESTs in the expression table (1409 out of the total 4974). The ESTs which are not GO-annotated are excluded by the analysis. An average t score was calculated for each GO term (molecular functions and cellular processes), and only those present at least three times in the expression table were selected. The over-expressed and under-regulated GO terms corresponding to the best 5% are displayed in Table 1. We next selected a subset of cDNAs that most strongly defined the division between malignant and benign lesions (Table 2). By again using a t test, we identified 43 cDNAs differentiating malignant odontogenic tumors from benign ameloblastoma (p-value < 0.01, as determined by bootstrap distribution). Additional tables can be found at the http://biotecnologie.unife.it/microarrays/odontogenic_tumors Web site.

View larger version (71K): [in this window] [in a new window]   Figure. Principal component analysis of cDNAs differentially regulated in odontogenic tumors vs. non-tumor tissues as selected by t test analysis (P < 0.01). On the right are the genes and ESTs which are overexpressed in the tumors, while on the left are those which are underexpressed.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS & METHODS RESULTS DISCUSSION REFERENCES

We performed Gene Ontology analysis by using GOAL to identify modulated functions and processes in the biopsies. Although a large percentage of the genes in the human genome is still not associated with GO terms (in jargon, "annotation"), we used GO to dissect the cellular functions automatically in odontogenic tumors. The procedure yielded useful results, albeit, due to technical limitations, not all genes sampled were included in the analysis. It was in fact possible to detect modulation of several pathways and mechanisms within the tumor biopsies, whereas a complete overview of the human cell physiology is far from complete.

Notable genes over-expressed in all six odontogenic tumors and selected by GO analysis (Table 1) are those that encode for proteins involved in intercellular adhesion and for receptors used by cells to bind to the extracellular matrix, such as integrins (alpha 3, 5, 6, 11; beta 2 and integrin-associated PTK2), genes involved in the signaling pathways from the integrin receptor, and others encoding for enzymes of collagen biosynthesis. Important genes with elevated expression participate in many aspects of the signaling of cell growth and differentiation (ARHI, RPIP8, VAV3, SIAH2), including receptor protein tyrosine phosphatases (PTPR R, H, G, D), cell cycle regulator (CDC37, WEE1, CDKN1A, 1B, BTG1), and nucleic acid binding (PPARD, NRD1, NR2F6). Up-regulated genes are also involved in oxidative metabolism, specifically in aerobic respiration and in biogenesis and organization of peroxisomes. Cytochrome b5 reductase, a central component of the plasma membrane anti-oxidant system, has, as a consequence, the inhibition of apoptosis (Villalba and Navas, 2000).

Relevant under-expressed genes in all six analyzed odontogenic tumors and selected by GO analysis encode for DNA repair enzymes, protein synthesis, and basement membrane. Extracellular matrix was similarly affected, with proteoglycans and type V collagen, a key determinant in the assembly of tissue-specific matrices. Finally, connexins CX26, CX32, and CX43, that form gap junctions and provide long-range cell signaling within epithelia, appear down-regulated as much as cyclin-dependent protein kinases: CDK4, required for cell cycle progression, phosphorylates Rb and inactivates its repressor function; CDK2 associates with cyclin A and cyclin E and is involved in promoting DNA synthesis and cell cycle progression; and CDC2-related kinase, a member of the cyclin-dependent protein kinase family, probably phosphorylates retinoblastoma (Rb) protein.

We then compared the expression profiles of three malignant odontogenic tumors with those of the three benign ameloblastomas. Up-regulated genes in malignant biopsies include coronin, a ubiquitous actin-binding protein essential for organizing the normal actin cytoskeleton that plays a significant role in cell division (Fukui et al., 1999), and MYD88, a protein containing death domains with a role in negative growth control, cell cycle arrest, and apoptosis (Liebermann and Hoffman, 2002). On the other hand, repressed genes in malignant tumors correspond to: STK19, a Ser/Thr nuclear protein kinase; ABT1, a transcription co-activator; and CTBP2, a transcriptional repressor which acts as a tumor suppressor and plays an important role in oncogenesis (Chinnadurai, 2002). Down-regulated genes also include RFP, a DNA-binding protein associated with the nuclear matrix and initially identified as an oncogene product, and TXK, a member of the Tec non-receptor tyrosine kinase family, expressed in T-cells and other hemopoietic cell types, that specifically regulates IFN-gamma gene expression (Takeba et al., 2002).

In conclusion, we have identified genes differentially expressed in all analyzed odontogenic tumors, and genes associated with three malignant tumors (AC, CCOT, GCOT). This gene subset can potentially improve the classification of borderline odontogenic tumors and eventually might contain candidates for disease-specific cancer therapy targets. However, a similar study design should be applied to a larger and more varied group of odontogenic tumors to confirm the present results, and further comparisons need to be made with other malignant tumors so that unique odontogenic features may be found.

	ACKNOWLEDGMENTS

 
We thank Hillary Siddons for manuscript revision. This work was supported by grants from UNIFE 60% (F.C., S.V.) and Fondazione CARISBO (P.C.) and CARIFE (F.C.).

Received July 1, 2002; Last revision March 31, 2003; Accepted April 4, 2003

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS & METHODS RESULTS DISCUSSION REFERENCES

 
Ardekian L, Manor R, Gaspar R, Laufer D (1998). Central granular cell odontogenic tumor: case report and review of literature. J Oral Maxillofac Surg 56:1343–1345.[Medline]

Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. (2000). Gene Ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 25:25–29.[ISI][Medline]

Carinci F, Francioso F, Rubini C, Fioroni M, Tosi L, Pezzetti F, et al. (2003). Genetic portrait of malignant granular cell odontogenic tumour. Oral Oncol 39:69–77.[Medline]

Chinnadurai G (2002). CtBP, an unconventional transcriptional corepressor in development and oncogenesis. Mol Cell 9:213–224.[ISI][Medline]

Fielden MR, Halgren RG, Dere E, Zacharewski TR (2002). GP3: GenePix post-processing program for automated analysis of raw microarray data. Bioinformatics 18:771–773.[Abstract/Free Full Text]

Francioso F, Carinci F, Tosi L, Scapoli L, Pezzetti F, Passerella E, et al. (2002). Identification of differentially expressed genes in human salivary gland tumors by DNA microarrays. Mol Cancer Ther 1:533–538.[Abstract/Free Full Text]

Fukui Y, Engler S, Inoue S, de Hostos EL (1999). Architectural dynamics and gene replacement of coronin suggest its role in cytokinesis. Cell Motil Cytoskeleton 42:204–217.[ISI][Medline]

Gesek DJ Jr, Adrian JC, Reid EN (1995). Central granular cell odontogenic tumor: a case report including light microscopy, immunohistochemistry, and literature review. J Oral Maxillofac Surg 53:945–949.[Medline]

Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M, Merisov JP (1999). Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 286:531–537.[Abstract/Free Full Text]

Iezzi G, Rubini C, Fioroni M, Piattelli A (2002). Clear cell odontogenic carcinoma. Oral Oncol 38:209–213.[Medline]

Infante-Cossio P, Hernandez-Guisado JM, Fernandez-Machin P, Garcia-Perla A, Rollon-Mayordomo A, Gutierrez-Perez JL (1998). Ameloblastic carcinoma of the maxilla: a report of 3 cases. J Craniomaxillofac Surg 26:159–162.[Medline]

Kramer JRH, Pindborg JJ, Shear M (1992). WHO histological classification of tumours. Histological typing of odontogenic tumours. 2nd ed. Berlin: Springer-Verlag.

Li TJ, Yu SF, Gao Y, Wang EB (2001). Clear cell odontogenic carcinoma: a clinicopathologic and immunocytochemical study of 5 cases. Arch Pathol Lab Med 125:1566–1571.[Medline]

Liebermann DA, Hoffman B (2002). Myeloid differentiation (MyD)/growth arrest DNA damage (GADD) genes in tumor suppression, immunity and inflammation. Leukemia 16:527–541.[ISI][Medline]

Neville BW, Damm DD, Allen CM, Bouquot JE (2002). Oral and maxillo-facial pathology. 2nd ed. Philadelphia: W.B. Saunders, pp. 611-616.

Pindborg JJ, Kramer IRH, Torloni H (1972). Histologic typing of the odontogenic tumors, jaw, cysts, and allied lesions. Geneva: World Health Organization, pp. 35-36.

Takeba Y, Nagafuchi H, Takeno M, Kashiwakura J, Suzuki N (2002). Txk, a member of nonreceptor tyrosine kinase of Tec family, acts as a Th1 cell-specific transcription factor and regulates IFN-gamma gene transcription. J Immunol 168:2365–2370.[Abstract/Free Full Text]

Villalba JM, Navas P (2000). Plasma membrane redox system in the control of stress-induced apoptosis. Antioxid Redox Signal 2:213–230.[Medline]

Yih WY, Thompson C, Meshul CK, Bartley MH (1995). Central odontogenic granular cell tumor of the jaw: report of case and immunohistochemical and electron microscopic study. J Oral Maxillofac Surg 53:453–459.[Medline]

This Article Abstract Figures Only Full Text (PDF) Services Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (13) Citing Articles via Google Scholar Google Scholar Articles by Carinci, F. Articles by Volinia, S. Search for Related Content PubMed PubMed Citation Articles by Carinci, F. Articles by Volinia, S.