Dissertation committee:
Τόσιος Κωνσταντίνος, Αναπληρωτής Καθηγητής, Τμήμα Οδοντιατρικής, Σχολή Επιστημών Υγείας, ΕΚΠΑ
Σκλαβούνου Αλεξάνδρα, Καθηγήτρια, Τμήμα Οδοντιατρικής, Σχολή Επιστημών Υγείας, ΕΚΠΑ
Πετσίνης Βασίλειος, Επίκουρος Καθηγητής, Τμήμα Οδοντιατρικής, Σχολή Επιστημών Υγείας, ΕΚΠΑ
Βασταρδή Ελένη, Αναπληρώτρια Καθηγήτρια, Τμήμα Οδοντιατρικής, Σχολή Επιστημών Υγείας, ΕΚΠΑ
Βλαχοδημητρόπουλος Δημήτριος, Αναπληρωτής Καθηγητής, Τμήμα Ιατρικής, Σχολή Επιστημών Υγείας, ΕΚΠΑ
Θεοχάρης Σταμάτιος, Καθηγητής, Τμήμα Ιατρικής, Σχολή Επιστημών Υγείας, ΕΚΠΑ
Κιτράκη Ευθυμία, Καθηγήτρια, Τμήμα Οδοντιατρικής, Σχολή Επιστημών Υγείας, ΕΚΠΑ
Summary:
Aim: Odontogenic keratocyst (OKC) has great research and clinical interest, due to its high frequency and aggressive behavior, i.e., growth potential within jaw bones and high recurrence rate. OKC may appear as a single lesion (sporadic OKC) or as a manifestation of Nevoid Basal Cell Carcinoma Syndrome (syndromic OKC). The aim of the present study is to perform a) transcriptomics analysis of the whole OKC via RNA-sequencing (RNA-seq), b) comparison of the transcriptome of primary cases and recurrences, c) comparison of the transcriptome of sporadic and syndromic OKC, d) validation of the RNA-seq results via real-time Polymerase Chain Reaction (qPCR) and immunohistochemistry, and e) intersection of the results of the present study with those obtained by a previous microarray-based study (GSE38494), and f) comparison of the immunohistochemical profile of syndromic and sporadic OKC.
Methods: The study material consisted of formalin-fixed paraffin-embedded tissues of 21 OKC cases and 6 dental follicles of impacted teeth. The samples had been fixed in formalin for up to 48 hours and had been stored for up to 20 years. Total RNA was extracted, followed by RNase H-based rRNA depletion. The library preparation protocol utilized random hexamers for first strand priming. For 27 samples, 150-bp paired-end RNA-sequencing was implemented using an Illumina Hiseq4000 sequencer (Illumina Inc., San Diego, CA). A second run of 150-bp paired-end RNA-sequencing was applied for 7 samples using NextSeq 500 sequencer (Illumina Inc., San Diego, CA). %Q30 Phred score was calculated for the quality control of sequencing results. Principal Component Analysis and Euclidean Distance Analysis were performed to identify outliers. Finally, RNA-seq data from 18 cases were included in downstream bioinformatics analysis. Differential gene expression was evaluated using the DEsSeq2 algorithm, setting the significance threshold as i) False Discovery Rate (FDR)-adjusted p-value<0.05, ii) |log2[fold-change (FC)]|>1, iii) base mean of DESeq2's normalized counts>20. Three generations of enrichment analysis were performed to identify Gene Ontologies and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways that were overrepresented among the upregulated and downregulated genes. Real-time Polymerase Chain Reaction was applied as a validation screen method of the induction levels of ALDH3A1, ATP12A, EPCAM, FETUB, GRHL3, OVOL1, SERPINB3, TACSTD2, and TP63. Immunohistochemical analysis was performed to verify the RNA-seq results at the protein level and to spatially map the expression of p63, SOX2, KLF4, OVOL1, IRF6, TACSTD2/TROP2 and E-cadherin. The GEO2R online platform was implemented, with default settings, to extract the differentially expressed genes (DEGs) of the previous microarray-based transcriptomics study (GSE38494). Next, we searched MEDLINE/Pubmed, Web of Science, EMBASE via OVID and grey literature for publications in the English language that compared the immunohistochemical profile of the sporadic and syndromic OKC subtype. The studies were qualitatively assessed using the Critical Appraisal Tool for Case Series proposed by Joanna Briggs Institute. Sensitivity and Specificity, Positive and Negative Likelihood Ratio, Diagnostic Odds Ratio and Area Under the Curve and pooled estimates were calculated, using a random effects model.
Results: On average, RNA-seq generated 19.3 million paired-end reads, of which a median of 73.5% of reads per sample was uniquely mapped to the human genome. The count of total reads, mapped reads and uniquely mapped reads did not differ significantly between samples stored for up to 36 months in comparison with samples stored for more than 82 months. 2,654 and 2,427 DEGs were captured to characterize the transcriptome of sporadic and syndromic OKCs, respectively. Gene ontologies related to “epidermis/skin development” and “keratinocyte/epidermal cell differentiation” were enriched among the upregulated genes (KRT10, NCCRP1, TP63, GRHL3, SOX21), while “extracellular matrix (ECM) organization” (ITGA5, LOXL2) and “odontogenesis” (MSX1, LHX8) Gene Ontologies were overrepresented among the downregulated genes in OKC. Moreover, several metabolic pathways were overrepresented among the upregulated genes, e.g., “arachidonic acid metabolism” (ALOX12B, PLA2G4E) and “glutathione metabolism” (GGT6, RRM1), whereas genes related to cell-matrix adhesion were downregulated (PECAM1, VCAM1). Interestingly, upregulation of various embryonic stem cells (ESCs) markers (EPHA1, SCNN1A) and genes committed in cellular reprogramming (SOX2, KLF4, OVOL1, IRF6, TACSTD2, CDH1) was found in OKC. These findings were highly shared between sporadic and syndromic OKCs. qPCR confirmed the induction of ALDH3A1, ATP12A, EPCAM, FETUB, GRHL3, OVOL1, SERPINB3, TACSTD2, and TP63 in OKC. Immunohistochemistry verified SOX2, KLF4, OVOL1, IRF6, TACSTD2/TROP2, CDH1/E-cadherin, and p63 expression predominantly in the OKC suprabasal epithelial layers. In addition, we observed a considerable overlap by intersecting the DEGs dataset obtained from this study with the DEG dataset captured by a previous microarray-based study (GSE38494) and noticed the enrichment of Gene Ontologies related to epidermis development and keratinocyte differentiation among the overlapping upregulated genes. The qualitative analysis phase of the systematic review included 71 studies reporting data about the immunoexpression of 93 markers. Risk of bias was high in 34 studies, moderate in 22 studies, and low in 15 studies. 25 studies reported differential expression of 29 markers in the form of higher number of positive cells or/and greater staining intensity in syndromic OKCs. Meta-analysis for bcl-2, Cyclin D1, CD56, CK18, p53 and PCNA showed that none of those markers is distinguishable between syndromic and sporadic OKCs, in a 95% confidence interval.
Conclusions: The results of the present study are in accordance with the pertinent literature regarding the utility of formalin-fixed paraffin-embedded tissues, stored for up to 20 years, in RNA-seq-based experiments. Our findings indicate a highly shared gene expression program between the sporadic (either primary cases or recurrences) and the syndromic OKCs. The OKC transcriptome is characterized by a prominent epidermal and dental epithelial fate, a repressed dental mesenchyme fate, combined with deregulation of the ECM organization, and enhanced stemness gene signatures. Moreover, significant upregulation of several metabolic pathways was observed in the transcriptome of OKC and could be linked to its locally aggressive behavior. On the other hand, the downregulation of cell-matrix adhesion mediators in OKC is in alignment with the microscopically noticed detachment of its epithelium from the underlying cystic wall. Core pluripotency factors (SOX2, KLF4), genes committed in cellular reprogramming (OVOL1, IRF6) or induced during early (CDH1/E-cadherin) or intermediate (TACSTD2/TROP2) reprogramming stages, were predominantly expressed in the suprabasal cells of OKC epithelium. Finally, there are no published immunohistochemical markers that can discriminate between syndromic and sporadic OKCs.