Background

Over the last decades the incidence rate of thyroid cancer increases worldwide [1]. In Saudi Arabia, thyroid carcinoma (TC) is considered the second most common cancer in young women [2]. About 80% of all TCs are PTCs. The majority of PTCs are histologically classified as conventional PTCs. The follicular variant of PTC (FVPTC) represents the largest subtype and accounts for about 30% of all PTCs [3]. Minor and rare subtypes include Hurthle cell variant PTC and insular PTC which bears an aggressive clinical behavior [4]. Conventional PTCs are characterized on the molecular level by a moderate to high frequency of BRAF mutations (40 % - 70 %) that distinguishes them from FVPTCs (10 % - 20 %) [5].

BRAF is a cytoplasmic receptor serine/threonine kinase and a key molecule in the mitogen activated protein kinase (MAPK) pathway. BRAF is mutated in diverse human malignancies although frequency and clinical presentation varies considerably between different types of cancers [6]. Over 90 % of all BRAF mutations are a valine by glycine substitution at codon 600 (V600E) in exon 15. Other BRAF mutations affect commonly codons adjunct to codon 600. Although the impact of BRAF mutations in PTC is controversially discussed, many studies found an association of BRAFmut PTCs with unfavorable clinical features including larger tumor size, advanced tumor stage, vessel invasion, capsular invasion, tumor extension, higher risk for lymph node (LN) involvement, distant metastasis, and poor prognosis [7–10]. Consistent with this, patients with a BRAFmut PTC are considerably older than those with a BRAFwt PTC [5, 10]. Whereas a BRAF mutation represents are valuable target for molecular therapy in advanced solid tumors including PTCs, molecular profiles of BRAFwt PTCs are less known and genetic screening for valuable target genes is primarily limited to research studies [11]. The major deregulated key genes in the BRAFwt group are RET and RAS. RAS consists of the highly related genes for HRAS, KRAS, and NRAS. Within the MAPK pathway the RAS molecules transmit signals to the downstream target BRAF. The most common RET/PTC fusions are paracentric fusions with the gene entitled coiled-coil domain containing 6 (CCDC6) contributing to ~80 % and with the nuclear receptor coactivator 4 (NCOA4) contributing to ~10 % of all known RET/PTC rearrangements. The frequency of RAS mutations and RET/PTC rearrangements differs between the populations studied and depends in part on the inclusion/exclusion criteria for the different histological PTC subtypes [12]. RAS mutations and RET rearrangements are unlikely to act as molecular drivers for onset of malignancy as they are already present in benign thyroid neoplasms [13, 14]. This distinguishes them from BRAF mutations which are virtually absent in precursor lesions of PTCs.

Until now only a few studies compared expression profiles between BRAFwt and BRAFmut PTCs using array technologies [15–18]. The relevance of expressional screening in PTC according to their BRAF mutational status is in part related to the different clinical behavior of both PTC groups with the necessity to identify appropriate biomarkers and in part related to the different tumor biology of both groups which is not thoroughly understood [7]. In our study we took advantage of current state-of-the-art technology to screen and analyze a case series of BRAFwt and BRAFmut PTCs for detecting new biomarkers which could become useful to distinguish both groups on the molecular level. We did not include FVPTCs and other smaller histological subtypes of PTC in our screening to minimize expressional bias which might be related to a different histology.

Methods

Thyroid samples

We examined 25 specimens from PTCs and seven normal thyroid samples (TN) from patients who were treated surgically in the period between November 2008 and February 2013 at the King Abdulaziz University Hospital, Jeddah, and the King Faisal Specialist Hospital & Research Center (KFSH&RC), Jeddah. In two BRAFmut PTCs, specimens were derived from a recurrence or a local metastasis and in one BRAFwt PTC from an LN metastasis. Normal thyroid specimens were derived from histopathologically unaffected normal thyroid tissue in the course of lobo- or thyroidetomies of thyroid lesions (4 goiters, 1 hyperthyroiditis, 1 PTC, and 1 FVPTC). Histopathological diagnosis and staging of thyroid lesions was performed by an experienced oncologic pathologist (JM) according to established criteria [19, 20]. Five BRAFwt PTCs were classified as micro PTCs (≤ 1 cm). This study was approved by the Research Ethics Committee of the King Abdulaziz University, Faculty of Medicine, #358-10, and the Institutional Review Board of the KFSH&RC, #IRB2010-07, and included written informed consent provided by the participants.

DNA extraction and BRAFmutational screening

Genomic DNA from fresh-frozen samples was extracted using the QIAmp DNA tissue kit (Qiagen, Hilden, Germany). DNA concentration was measured with the Nanodrop ND-1000 spectrophotometer (Thermo Scientific, Wilmington, DE). Screening of the BRAF mutational hotspot region of exon 15 was performed as described earlier involving direct sequencing of PCR products spanning the region [5].

RNA sample and array processing

Total RNA was extracted from freshly preserved thyroid tissue specimens using the Qiagen RNeasy Mini Kit (Qiagen, Hilden, Germany) including an on-column DNAse treatment according to manufacturer’s recommendations. Quality of the purified RNA was verified on an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). RNA integrity number for all evaluated samples was at least 5.0. RNA concentrations were determined using a NanoDrop ND-1000 spectrophotometer. Samples containing each 250 ng of RNA were processed using the Ambion WT Expression Kit (Life Technologies, Austin, TX) and the GeneChip WT Terminal Labeling and Controls Kit (Affymetrix, Santa Clara, CA) according to the manufacturers` recommendations. Fragmentation and endlabeling of samples were monitored by electrophoresis on 3 % agarose gels. Affymetrix GeneChip hybridization, wash and stain kits were utilized in subsequent processing steps. Hybridization mixtures containing each 5500 ng of cDNA were hybridized at 45°C for 17 hrs and 60 rpm to Affymetrix Human Gene 1.0 ST GeneChip arrays. Subsequent to wash and staining at the GeneChip Fluidics Station 450, the arrays were scanned with the GeneChip Scanner 3000 7G. Probe cell intensity data (CEL files) were generated using the GeneChip Command Console Software (AGCC). Human Gene 1.0 ST GeneChip arrays interrogate in total with a set of 764,885 probes 36.079 reference sequences (NCBI build 36).

Gene Expression Analysis

CEL files were imported to Partek Genomics Suite version 6.6 (Partek Inc., MO) and a log-transformed data set of robust multi-array averaged (RMA), background-adjusted, and normalized values was generated. Non-annotated genes and multiple transcripts generated from the same gene were excluded from further analysis. Principal component analysis (PCA) was performed to assess quality as well as overall variance in gene expression between groups of samples. Analysis of Variance (ANOVA) was applied to generate a list of differentially expressed genes using a p-value with a false discovery rate (FDR) (Step up method) < 0.05 and a fold change > 2.0. Two dimensional average linkage hierarchical clustering was performed using Spearman’s correlation as a similarity matrix. The generated array data set complies with MIAME [21] and was submitted to NCBI’s Gene Expression Omnibus (GEO), accession number GSE54958.

Functional network and pathway analysis

To define molecular networks and canonical pathways in differentially regulated gene sets, pathway analyses were performed by using Ingenuity Pathways Analysis (IPA) software (Ingenuity Systems, Redwood City, CA). For this purpose, statistically differentially expressed genes and their corresponding probe set ID, gene symbol as clone identifier, p-value and fold change were imported into IPA. The program identifies with its functional algorithms those biological functions, interacting drugs and/or diseases that are most significantly related to a differentially expressed gene set. The canonical pathway analysis identifies pathways that are most significantly related to the data set.

Results

Demographic data of patients and histopathological criteria of BRAFwt and BRAFmut PTC are listed in Table 1. A gender shift towards females was observed in the BRAFwt group and mean age was considerable lower in the BRAFwt than in the BRAFmut group (30.9 years vs. 45.9 years). Histopathological criteria including tumor size, LN involvement, tumor focality and tumor stage were comparably more unfavorable in BRAFmut PTCs (Table 1).

Table 1 Demographic and clinicopathological features of BRAFwt and BRAFmut PTCs
Full size table

Expression BRAFwtvs. BRAFmutPTCs

Employing whole-transcript microarrays (HuGene 1.0 ST) we compared expression profiles of 11 BRAFwt PTCs with 14 BRAFmut PTCs. Seven TN specimens were used as a reference for normal thyroid tissue expression which allowed us to identify differentially expressed genes between both PTC groups and TN samples. Three-D presentation of the PCA displays clustering of BRAFwt PTCs, BRAFmut PTCs and TN samples (Figure 1). We identified, after excluding non-annotated genes and multiple transcriptional isoforms, 237 annotated genes that were significantly differentially expressed (p-value with FDR <0.05 and a fold change >2) between BRAFwt and BRAFmut PTCs (Additional file 1). Of these genes, 127 were up-, and 110 were downregulated in BRAFwt compared to BRAFmut PTCs. The most significantly upregulated genes in BRAFwt were inositol 1,4,5-triphosphate receptor, type 1 (ITPR1), hepatic leukemia factor (HLF), potassium voltage-gated channel, shaker-related subfamily, beta member 1 (KCNAB1), engulfment and cell motility 1 (ELMO1), Rho GTPase activating protein 24 (ARHGAP24), thyroid peroxidase (TPO) and solute carrier family 4, sodium bicarbonate cotransporter, member 4 (SLC4A4). The most significantly downregulated genes in BRAFwt PTCs were dendrocyte expressed seven transmembrane protein (DCSTAMP), ladinin 1 (LAD1), keratin 19 (KRT19), chromosome 19 open reading frame 33 (C19orf33), poliovirus receptor-related 4 (PVRL4), EPH receptor A10 (EPHA10), and TBC1 domain family, member 2 (TBC1D2). The list of 237 genes contained one microRNA (MIR492) that was downregulated and one microRNA (MIR32) that was upregulated in BRAFwtvs. BRAFmut PTCs. Additional file 1 contains also the cellular/extracellular location, function of gene products as well as a selection of drugs known to interact with a gene product. Hierarchical cluster analysis of the 237 genes separates both PTC groups according to their differential gene expression (Figure 2). TN samples were included in this data set to provide normal expression values.

Figure 1
figure 1

PCA scatter plot wherein each dot represents a sample with a group specific color. Distance between dots is a dimensional measure for the similarity of the respective expression profiles of the samples. Ellipsoids are a measure to visualize distance of relationships between samples of a group. Red, normal thyroid; blue, BRAFwt PTCs, green, BRAFmut PTCs.

Full size image
Figure 2
figure 2

Hierarchical cluster analysis of 237 genes which were differentially expressed between BRAFwt and BRAFmut PTCs. Gene expression of TN samples is included in analysis. Color scheme for expression levels: red for comparably higher and blue for comparably lower expression. Color scheme for groups in the left sided branches: green for TN samples, purple for BRAFwt, and red for BRAFmut PTCs. Stars mark samples from recurrence or local metastasis, and LN metastasis.

Full size image

Expression PTCs vs. TN samples

Comparison of genes differentially expressed between BRAFwt and TN samples revealed a set of 8249 genes and between BRAFmut and TN samples a set of 8836 genes. To identify genes of interests, e.g. conceivable immunohistochemistry markers, we selected those genes which are differentially expressed between all three groups (Table 2). In sum, 32 genes met the statistical criteria and the vast majority revealed the comparatively highest expression in BRAFmut PTCs.

Table 2 Differentially expressed genes between BRAFwt and BRAFmut PTCs and TN samples
Full size table

Gene networks and canonical pathways

Most significant network functions identified by IPA algorithms and associated with the set of 237 differentially expressed genes were involved in cell signaling, cancer, and cellular development (Figure 3). This comprehensive network includes 12 molecules which are overexpressed and 16 molecules which are underexpressed in BRAFwt compared to BRAFmut PTCs. The most significantly associated canonical pathway related to the differentially expressed set of 237 genes is the proteinkinase A (PKA) signaling cascade (Additional file 2). The PKA pathway is involved in second messenger signaling and it is stimulated by upstream cascades including the G protein-coupled receptor pathway (Figure 4). The most significant pathways with thyroid specific functions were the thyroid hormone metabolic and thyronamine/iodothyronamine metabolic pathways.

Figure 3
figure 3

A molecular network including 12 overexpressed (red) and 16 underexpressed (green) molecules in BRAFwt compared to BRAFmut PTCs. The relationships between molecules were conveyed by IPA functions on basis of the set 237 differentially expressed genes in BRAFwtvs. BRAFmut PTCs. Gene names in Additional file 1.

Full size image
Figure 4
figure 4

The G protein-coupled receptor pathway is involved in transmembrane signaling and diverse physiological functions including hormone signaling, regulation of cell contraction, cell migration, growth and differentiation. Members of the G protein-coupled receptors, G protein alpha q, G protein alpha i/o, and PKA are upregulated (red) whereas members of MAPK phosphatases (MKP), phosphodiesterases (PDE) and adenylate cyclases (AC) are downregulated (green) in BRAFwt compared to BRAFmut PTCs.

Full size image

Discussion

We performed one of the first studies using whole transcript microarrays to compare expression profiles solely in conventional BRAFwt and BRAFmut PTCs. Comparison to studies including different histological subtypes of PTCs may result in detecting a lower number of common genes. An expression array study including histological variants of PTC identified, on basis of enhanced stringent threshold criteria, over 80 up- and downregulated genes in the BRAFmut group in comparison to PTCs with either a RAS mutation or a RET/PTC rearrangement [15] and the 40 most up- and downregulated genes have an overlap of ~40 % to our list of 237 differentially expressed genes.

Selected differentially expressed molecules

DCSTAMP, also known as TM7SF4, has been previously identified as one of the most overexpressed genes in BRAFmut PTC compared to BRAFwt PTC as well in PTC with undetermined mutational status compared to normal thyroid tissue [15, 22, 23]. It has been supposed that DCSTAMP expression is an immune response related to BRAFmut tumors [15]. DCSTAMP contains signature motifs owned by the family of transmembrane serine proteases and it exhibits degradation activity against extracellular matrix proteins. One of the functions of the hepatic leukemia factor (HLF) as a transcription factor is to mediate thyroid hormone activation from the thyroid hormone receptor/retinoid X receptor heterodimer to the hypoxia-inducible factor (HIF-1alpha) [24]. The function of C19orf33, also known as H2RSP (Hepatocyte growth factor activator inhibitor type 2-related small protein) is virtually unknown. An enhanced expression with higher levels in LN positive tumors has been observed for C19orf33 in colorectal adenocarcinoma cells at the invasive front [25]. The voltage gated channel molecule KCNAB1 exhibits diverse functions including electrolyte transport, and insulin secretion. Downregulation of KCNAB1 expression has been identified in follicular thyroid carcinomas compared to benign follicular adenomas [26]. ELMO1 functionally interacts with dedicator of cytokinesis 1 (DOCK1) that promotes Rac guanine exchange factor (GEF) activity for Rac proteins of the Rho GTPases family. GTP-loaded Rac proteins initiate downstream pathways that promote cell elongation, migration, and cytoskeleton remodeling [27, 28]. The active ELMO1/DOCK1 complex is anchored via phosphoinositides to the membrane. In our study, the DOCK1 related DOCK5 gene was identified as a 2-fold upregulated molecule in BRAFwt compared to BRAFmut PTCs. The leucine-rich repeat and immunoglobulin-like domain-containing nogo receptor-interacting protein 2 (LINGO2) encodes a single-pass type I membrane protein that is primarily expressed during development in cells adjunct to the epithelial lining of the olfactory pit and in adult brain [29]. In our study, LINGO2 was downregulated in BRAFwt PTCs and even more in BRAFmut PTCs compared to TN samples which may imply a tumor suppressor function for this molecule. Higher expression levels of the ERK1/2-specific cytoplasmic dual specificity phosphatase 6 (DUSP6) in comparison to benign and normal thyroid cells has been previously associated with PTC, especially with advanced thyroid carcinomas [30, 31]. In our study, DUSP6 was 2.1 times higher expressed in BRAFmut than in BRAFwt PTCs assuming that the ERK1/2 related pathway is frequently more utilized in the BRAFmut group. MET overexpression in thyroid cancer has been identified in a number of studies and this molecule was 2.3-fold higher expressed in our study in BRAFmut compared to BRAFwt PTCs which is in accordance with another survey [15] (Table 2).

Second messenger molecules

ITPR1 is an intracellular receptor for inositol 1,4,5-trisphosphate (IP3) and implicated in the thyroid hormone synthesis pathway. The receptor mediates calcium release from the endoplasmic reticulum upon stimulation by IP3. Downregulation of ITPR1 has been demonstrated in thyroid cancer in comparison to non-malignant thyroid tissue in a number of studies [32–35]. Among the members of the phospholipase C (PLC) family, PLCH1 was upregulated in BRAFwt PTCs whereas PLCD3 was downregulated. PLC molecules hydrolyze phosphatidylinositol 4,5-bisphosphate to generate the second messengers diacylglycerol and IP3.

Thyroid hormone pathway molecules

LAD1 has been identified as an overexpressed gene in BRAFmut thyroid carcinomas compared to those with a RET/PTC rearrangement [18]. Lad1 expression is regulated by the glucocorticoid receptor (GR) and requires for induction the GR coactivators thyroid hormone receptor associated protein 220 (MED1) and the thyroid hormone receptor associated protein 170 (MED14) [36]. Downregulation of TPO has been considered in a number of studies comparing thyroid carcinomas with benign thyroid tumors or normal thyroid samples [32, 33, 37, 38]. However, downregulation of TPO and the sodium iodine symporter genes as been previously associated with BRAFmut PTCs in comparison to PTCs with a RET/PTC rearrangement [39]. This is in line with observations that BRAFmut tumors are refractory for radioactive iodine ablation due to downregulation of thyroid hormone biosynthesis pathways [40]. TPO was in our study 13-fold downregulated in BRAFmut compared to BRAFwt PTCs. In addition, other genes involved in thyroid hormone biosynthesis pathway including the solute carrier family-5 member-8 (SLC5A8), solute carrier family 26 (anion exchanger), member 4 (SLC26A4), deiodinase, iodothyronine, type I (DIO1) and deiodinase, iodothyronine, type II (DIO2) were downregulated in our set of BRAFmut PTCs. Other members of the SLC family that were downregulated in BRAFmut PTCs were SLC1A3, SLC4A4, SLC16A2, and SLC26A7 whereas SLC22A31, SLC30A2, and SLC34A2 were downregulated in BRAFwt PTCs.

ERBB3 and ERRB4genes

Two of the four structurally related receptor tyrosine kinases ErbB, namely ERBB3 and ERBB4, were differentially expressed in the two PTC groups, i.e. ERBB3 was downregulated and ERBB4 upregulated in BRAFwt PTCs (Additional file 1). Of notice, ERBB4 was also lower expressed in BRAFmut PTCs compared to TN samples (p = 0.0035) (data not shown). Oncogenic functions of ERBB3 and ERBB4, which can form heterodimers and signal through the PI3K/AKT signaling pathway have not been elucidated yet in detail in relation to the BRAF mutational status in thyroid cancer [41–43]; however, decreased expression of ERBB4 in PTC vs. normal thyroid tissue has been demonstrated in a RT-PCR study whereas ERBB2 and ERBB3 expression was shown to be increased [44]. On the protein level, a tissue microarray study in proliferative thyroid lesions found a correlation of ERBB3 expression with LN metastasis whereas ERBB4 expression correlated with lower tumor stage [45]. A possible link of ERBB3 to apoptosis can be deduced from a functional in vivo study wherein deletion of ERBB3 in mouse intestinal epithelium induced tumor-specific cell death [46].

MicroRNAs

The most significantly downregulated microRNA in BRAFwt compared to BRAFmut PTCs was mir492 and the most upregulated one was mir32. Overexpression of mir492 has been previously linked to progressive hepatoblastoma and tumorigenesis of retinoblastoma [47, 48]. Of notice, mir492 is processed from the KRT19 gene [47] and both are higher expressed. A possible target for miR492 is the 3`UTR of KRT19 [49] which is 4.4-fold higher expressed in BRAFmut PTCs compared to BRAFwt PTCs suggesting an accumulation of the KRT19 mRNA. Another possible target of miR492 is the thyroid hormone receptor-associated protein 3 (THRAP3) mRNA that harbors two predicted target sites for miR492 [49] and that is significantly downregulated in our case series by approximately 3-fold in both, BRAFwt and BRAFmut PTCs compared to TN samples. Upregulation of mir32 in thyroid cancer vs. benign thyroid tumors has been detected in a microarray study; however functional implication of this microRNA in PTC is not known yet [50].

In summary, our microarray expression study provides a detailed overview of differentially expressed genes, networks, and pathways between BRAFwt and BRAFmut PTCs that gain interest for basic molecular genetics and translational studies in PTCs.