The RET/PTC-RAS-BRAF linear signaling cascade mediates the motile and mitogenic phenotype of thyroid cancer cells

In papillary thyroid carcinomas (PTCs), rearrangements of the RET receptor (RET/PTC) and activating mutations in the BRAF or RAS oncogenes are mutually exclusive. Here we show that the 3 proteins function along a linear oncogenic signaling cascade in which RET/PTC induces RAS-dependent BRAF activation and RAS- and BRAF-dependent ERK activation. Adoptive activation of the RET/PTC-RAS-BRAF axis induced cell proliferation and Matrigel invasion of thyroid follicular cells. Gene expression profiling revealed that the 3 oncogenes activate a common transcriptional program in thyroid cells that includes upregulation of the CXCL1 and CXCL10 chemokines, which in turn stimulate proliferation and invasion. Thus, motile and mitogenic properties are intrinsic to transformed thyroid cells and are governed by an epistatic oncogenic signaling cascade.

Papillary thyroid carcinoma (PTC) is the most prevalent endocrine malignancy in humans (1). Four genetic lesions, at the somatic level, are associated with PTC: chromosomal alterations that affect the RET or TRKA tyrosine kinase receptors and oncogenic activation of the RAS or BRAF genes. RET encodes the tyrosine kinase receptor of growth factors belonging to the glial-derived neurotrophic factor (GDNF) family (2). RET gene rearrangements occur in up to 30% of PTCs (3). They cause the recombination of the intracellular kinase–encoding domain of RET with heterologous genes, thereby generating RET/PTC chimeric oncogenes. RET/PTC1 (the H4-RET fusion) and RET/PTC3 (the RET-fused gene–RET [RFG-RET] fusion) are the most prevalent variants. RET/PTC3 is frequently found in radiation-associated PTC (4). The finding that RET/PTC-transgenic mice develop PTC demonstrates that RET/PTC oncogenes can initiate thyroid carcinogenesis (5, 6). RET/PTC oncogenes are frequent in clinically silent small PTCs and are thus an early event of thyroid tumorigenesis (7). Similar rearrangements of TRKA, e.g., the high-affinity receptor for the nerve growth factor (NGF), can also be found, at a low frequency, in human PTC (8). Activating point mutations in RAS small GTPases occur in about 10% of PTCs, mainly in the follicular variant (9). Finally, point mutations in BRAF are the most common genetic lesions found in PTCs (up to 50% of cases) (10). BRAF belongs to the RAF family of serine/threonine kinases that includes c-RAF and ARAF. RAF proteins are components of the RAF–MAPK kinase–ERK (RAF-MEK-ERK) pathway, which is a highly conserved signaling module in eukaryotes. They are activated through binding to RAS in its GTP-bound state. Once activated, RAF kinases can phosphorylate MEK, which in turn phosphorylates and activates ERKs (11). As occurs in melanomas (12), a V600E substitution (formerly designated V599E), in the activation segment accounts for more than 90% of BRAF mutations in PTC (10, 13–15). This mutation enhances BRAF activity by disrupting the autoinhibited state of the kinase (15).

Thanks to their intrinsic kinase activity, receptor tyrosine kinases (RTKs) activate many intracellular signaling pathways. Upon binding to ligand, RTKs dimerize and autophosphorylate various cytoplasmic tyrosines. The phosphorylated tyrosines thus become binding sites for intracellular molecules containing phosphotyrosine-binding motifs, thereby initiating a diverse array of signaling pathways (16). In RET/PTC rearrangements, fusion with protein partners possessing protein-protein interaction motifs provides RET/PTC kinases with dimerizing interfaces, which results in ligand-independent autophosphorylation. The RET intracellular domain contains at least 12 autophosphorylation sites, 11 of which are maintained in RET/PTC proteins (17). Tyrosine 905 (Y905) is a binding site for Grb7/10 adaptors (18), Y1015 for phospholipase Cγ (19), and Y981 for c-Src (20). Tyrosine 1062 is the binding site for several proteins, including the Shc proteins, insulin receptor substrate–1/2 (IRS-1/2), FGFR substrate 2 (FRS2), downstream of kinase 1/4/5 (DOK1/4/5), and Enigma, which, in turn, lead to the activation of many signaling pathways (2, 21). Binding to Shc and FRS2 mediates recruitment of Grb2-SOS complexes, which thus leads to GTP exchange on RAS and RAS/ERK stimulation (22).

In human PTC, RET/PTC, RAS, and BRAF genetic alterations are mutually exclusive, which suggests that mutations at more than 1 of these sites are unlikely to provide an additional biological advantage (10, 13, 14). This is what would be expected if the 3 proteins function in tandem along a common signaling cascade. To verify whether this is indeed the case, we examined the link among the 3 oncogenic proteins in a thyroid cell culture model. RET/PTC triggered the RAS-BRAF-ERK signaling cascade in a Y1062-dependent fashion. Analysis of the transcriptional profile by oligonucleotide microarrays revealed that the RET/PTC3, HRAS, and BRAF oncogenes induced changes in the expression of widely overlapping sets of genes. The RET/PTC3-RAS-BRAF axis triggered upregulation of CXC chemokines and their receptors, which in turn stimulated the mitogenic and invasive capacity of thyroid cancer cells.

A biochemical cascade linking RET/PTC to the activation of RAS, BRAF, and ERK. We previously showed that oncogenic RET/PTC proteins activate GTP loading on RAS (23). Here, we transiently transfected HEK293 cells with myc-tagged BRAF and with the constructs shown in Figure 1A. We examined BRAF activity in an immunocomplex kinase assay, with the oncogenic BRAF(V600E) and the kinase-dead BRAF(K^–) mutants as positive and negative controls, respectively. As shown in Figure 1B, BRAF activation was induced by the expression of the RET/PTC3 and HRAS(V12) oncogenes. Activation of BRAF depended on RET/PTC3 kinase activity and on the integrity of tyrosine 1062. Indeed, neither the kinase-dead RET/PTC3(K^–) mutant nor a RET/PTC3 mutant carrying a tyrosine to phenylalanine (Y → F) mutation at Y1062 activated BRAF. In contrast, the Y → F mutation at tyrosine 1015 had virtually no effect on BRAF activation. By Western blot analysis with phospho-specific anti-RET antibodies, we demonstrated that the tyrosine 1062 substitution does not affect RET/PTC3 autophosphorylation levels overall or Y1015 and Y905 phosphorylation (Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI200522758DS1). Expression of the dominant-interfering HRAS(N17) mutant blocked RET/PTC3-mediated BRAF activation, which indicates that in this context, BRAF activation requires RAS (Figure 1B). We explored whether the RET/PTC3(Y1062)-RAS-BRAF cascade resulted in ERK1/2 stimulation. RET/PTC3 stimulated ERK activation, as revealed by immunoblot with a phospho-specific antibody (Figure 1C). This activation did not depend on Y1015 (data not shown); instead, it was dependent on the integrity of Y1062 and on the activity of RAS and BRAF, being obstructed by the expression of HRAS(N17) and BRAF(K^–) dominant negative mutants (Figure 1C). The BRAF(K^–) dominant negative mutant probably acts as a pan-RAF inhibitor in that it blocks the activity of the various RAF proteins. To specifically block BRAF, we used small interfering duplex RNA (siRNA) oligonucleotides that target endogenous BRAF. This duplex affected only BRAF and no other member of the RAF family (Figure 1D and data not shown). Silencing of BRAF in RET/PTC3-transfected HEK293 cells inhibited ERK activation. This effect was specific, since a control scrambled siRNA did not alter ERK activity (Figure 1D). Taken together, these findings demonstrate that phosphorylation of RET/PTC tyrosine 1062 triggers RAS-dependent stimulation of BRAF signaling.

Figure 1

RET/PTC-mediated ERK activation is dependent on RAS and BRAF. (A) Schematic representation of the RET, HRAS, and BRAF constructs. The RET/PTC breakpoint and RET tyrosines 1015 and 1062 are indicated. Residues V600 and K483 are mutated to E and M, respectively, in the BRAF(V600E) and BRAF(K^–) plasmids. Residues G12 and S17 are mutated to V and N in the HRAS(V12) and HRAS(N17) plasmids. C, CAAX tail; CR1–3, conserved BRAF regions 1–3; Cys, cysteine-rich; EC, extracellular domain; ED, HRAS effector domain; H, heterogeneous region; JX, juxtamembrane; SP, RET signal peptide; TK, tyrosine kinase; TM, transmembrane. (B) Protein lysates (500 μg) extracted from HEK293 cells transfected with the indicated plasmids underwent immunoprecipitation with anti-tag (myc) antibody. Kinase assay was performed with GST-MEK as a substrate. BRAF and RET/PTC3 were detected by Western blotting (W.B.) with anti-myc and anti-RET antibodies, respectively. RAS expression was detected with an anti-RAS monoclonal antibody that also recognizes the endogenous protein. (C) HEK293 cells transfected with the indicated plasmids were harvested, and protein extracts were subjected to immunoblotting with anti–phospho-p44/p42 ERK (pERK) antibodies. The blot was reprobed with anti-p44/p42 antibodies for normalization. RET/PTC3 and RAS were detected by Western blotting with specific antibodies. These experiments are representative of at least 3 independent assays. (D) Transient BRAF suppression was achieved by RNA interference. Whole cell lysates were prepared 48 hours after transfection and analyzed for protein expression by Western blotting with the indicated antibodies. siRNA(SCR), scrambled siRNA.

Generation of the cellular model system. PC Cl 3 (referred to hereafter as PC) is a continuous line of follicular thyroid cells, derived from Fischer rats, that constitutes a model system with which to study differentiation and growth regulation in an epithelial thyroid cell setting. PC cells express the thyroid-specific gene thyroglobulin (TG) and the thyroid-specific transcription factors TTF-1 and PAX-8. They require a mixture of 6 hormones (6H), including thyrotropic hormone (TSH) for proliferation (24). RET/PTC1 expression causes hormone-independent proliferation of PC cells (25–27); hence, the system lends itself to the study of RET/PTC-mediated effects. We generated marker-selected mass populations of several PC cell clones stably transfected with RET/PTC3, HRAS(V12), or BRAF(V600E). The corresponding proteins were correctly synthesized (Figure 2A); ERK activity was stimulated by each of the 3 oncoproteins (Figure 2C). The oncogene-transfected cell populations showed a similar, albeit not identical, morphologically transformed phenotype, i.e., spindle-shaped and refractile cells scattered on the surface of the culture dish (Figure 2B). RET/PTC3, HRAS(V12), and BRAF(V600E) oncogenes abolished the dependency of PC cell proliferation on the 6H (Figure 3A). Efficiency of colony formation was 35% ± 5% in the case of HRAS(V12), perhaps due to the reported proapoptotic role of acute overexpression of the oncoproteins (23, 27). Oncogene-transfected cells maintained increased levels of G₁ cyclin D1 under conditions of 6H deprivation (Figure 3A, inset). Moreover, oncogene-transfected PC cells exerted an invasive phenotype through Matrigel (Figure 3B) and lost the differentiated phenotype (data not shown). In the case of RET/PTC3, all these features, as well as BRAF-ERK activation, depended on tyrosine 1062 but not on tyrosine 1015 (Figures 2 and 3). Since tyrosine 1062 is a multi-docking site in RET and is required to activate several signaling pathways, we used RNA interference to specifically downregulate BRAF in PC cells transfected with RET/PTC3 (PC RET/PTC3 cells). As shown in Figure 2D, transfection with BRAF siRNA, but not with the control scrambled siRNA, diminished BRAF protein levels. ERK activity was downregulated by BRAF knockdown (Figure 2D). Importantly, the transient silencing of BRAF inhibited the invasive activity and growth of PC RET/PTC3 cells, whereas the negative control siRNA was ineffective (Figure 3, B and C). Downregulation of ERK activity achieved with U0126, a specific MEK inhibitor, also inhibited PC RET/PTC3 proliferation (data not shown) and Matrigel invasion (Figure 3B).

Figure 2

Generation of thyroid cell cultures. (A) Protein lysates (50 μg) extracted from the indicated cell lines underwent Western blotting with anti-RET, -tag (myc), or -RAS antibodies. Equal protein loading was ascertained by anti-tubulin immunoblot. (B) Mass populations of PC cells transfected with the indicated plasmids were photographed using a phase-contrast light microscope (magnification, ×150). RET/PTC3-, RET/PTC3(Y1015F)-, BRAF(V600E)-, and HRAS(V12)-expressing cells displayed a transformed morphology, whereas RET/PTC3(Y1062F)-expressing cells were virtually indistinguishable from parental cells. (C) Protein extracts from the indicated cell lines were subjected to immunoblotting with anti–phospho-p44/42 ERK antibodies. The blot was reprobed with anti-p44/42 antibodies for normalization. (D) BRAF targeting by siRNA but not by scrambled siRNA induced BRAF downregulation and ERK inhibition, as shown by Western blotting with specific antibodies.

Figure 3

The transformed phenotype of RET/PTC3 thyroid cells requires the integrity of the Y1062-RAS-BRAF-ERK pathway. (A) PC cells were transfected with the indicated plasmids or the empty vector and either selected in neomycin-containing medium or left in the absence of 6H. Three weeks later, colonies were stained with crystal violet and counted. The ratio of neomycin-resistant clones to 6H-independent colonies was calculated. The results of 3 independent experiments performed in duplicate ± SD are reported [the number of 6H-independent colonies induced by HRAS(V12) was set at 100]. Inset: Protein lysates (50 μg) underwent Western blotting with the indicated antibodies. cyc D1, cyclin D1. (B) Matrigel invasion of parental and transformed PC cells. Where indicated, in PC RET/PTC3 cells, suppression of endogenous BRAF was obtained by transfection of RNA interference against BRAF, and ERK inhibition was achieved by U0126 treatment. Cells were seeded in the upper chamber of 8-μM-pore transwells and incubated for 24 hours. Thereafter, filters were fixed and stained. The upper surface was wiped clean and cells on the lower surface photographed (top) and then solubilized. Absorbance at 570 nm was measured with a microplate reader. Cell migration is expressed as percentage with respect to RET/PTC3 cells, whose migration was arbitrarily set at 100. Each column represents the average ± SD of 3 independent experiments (bottom). (C) Suppression of endogenous BRAF in PC RET/PTC3 cells was obtained by transfection of RNA interference against BRAF. Cells were counted at different time points, and the average results of 3 independent experiments are reported.

A gene expression signature of the RET/PTC3-RAS-BRAF axis in thyroid cells. We explored gene expression changes after PC cell transfection with RET/PTC3, HRAS(V12), or BRAF(V600E). We also used cells expressing RET/PTC3(Y1062F) and RET/PTC3(Y1015F) mutants. To study the expression profiles, we used oligonucleotide-based DNA microarrays (GeneChip; Affymetrix) containing oligonucleotide probe pairs corresponding to more than 16,000 known genes and expressed sequence tag (EST) clusters. RNA was extracted from mass populations of cells (pool no. 3 for each oncogene), converted into fluorescently labeled complementary RNA (cRNA), and hybridized to arrays. Each chip was analyzed with Affymetrix Microarray Suite 5.0 Software to generate raw expression data. Fold change (signal log ratio [SLR]) was calculated by pairwise comparison of probe pairs from the experiment (cells expressing the different oncogenes) with baseline (untransfected parental cells). In order to sort the genes with changes in mRNA abundance in response to the different oncogenes, we defined a filter query that “passed” only data sets that were denoted “increased” or “decreased” by the software and that showed a fold change of 4 or more (SLR ≥ 2 for upregulated genes; SLR ≤ –2 for downregulated genes). The entire data set is available as Supplemental Table 1.

We first examined genes whose expression was expected to be modified by thyroid cell transformation. Consistent with the above-described cell phenotype, microarray screening demonstrated upregulation of G₁ cyclins (D and E cyclins) and downregulation of thyroid differentiation markers (including TG, TTF-1, and PAX-8) (Supplemental Table 1). Then we examined the global expression profile. Some genes were regulated by only 1 or 2 oncoproteins. However, cross-comparison of the results revealed a group of sequences that were regulated in a similar fashion in RET/PTC3, HRAS(V12), and BRAF(V600E) cells (Figure 4 and Supplemental Figures 2 and 3). Whereas the former probably reflect the specific biological activity of each oncoprotein, the common targets represent a transcriptional signature of the expression of the 3 oncogenes in thyrocytes. Of the 786 oligonucleotide probe pairs whose expression was increased in RET/PTC3 cells (Supplemental Figure 2A), 146 showed an SLR of 2 or more (average SLR, 3.2; Figure 4A). Tyrosine 1062 played a pivotal role in these expression changes. In fact, upregulation of most sequences (131 of 146, 90%) required the integrity of Y1062; the remaining 15 sequences were still induced after the Y1062F mutation, albeit to a lesser extent (average SLR, 1.8) (Figure 4A). In contrast, upregulation of a few of them (less than 15%) depended on tyrosine 1015; most of them depended also on Y1062 (data not shown). Many of the sequences induced by RET/PTC3 were also induced by HRAS(V12) (80 of 146, 54%) and BRAF(V600E) (70 of 146, 48%) cells with a similar average SLR (Figure 4A). Overall, 60 (41%) of the oligonucleotide probe pairs induced in RET/PTC3 cells were induced in both HRAS(V12) and BRAF(V600E) cells, and as many as 87% (52 of 60) were Y1062 dependent (Figure 4A). These upregulated oligonucleotide probe pairs corresponded to the 32 genes and 19 ESTs listed in Table 1. Similarly, sorting of HRAS(V12)- and BRAF(V600E)-induced sequences showed that RET/PTC3 cells upregulated a large set of them in a Y1062-dependent fashion (Supplemental Figure 3). Of the 2,517 oligonucleotide probe pairs whose expression was decreased in RET/PTC3 cells (Supplemental Figure 2B), 338 were downregulated 4-fold or more (average SLR, –2.96). Again, a large fraction of these sequences was also downregulated in HRAS(V12) (148 of 338, 44%; average SLR, 3.2) and in BRAF(V600E) (121 of 338, 36%; average SLR, 2.5) cells (Figure 4B). Overall, 104 (31%) sequences were downregulated by all 3 oncogenes, with 66 of them (64%) being dependent on the integrity of Y1062 (Figure 4B). These downregulated oligonucleotide probe pairs corresponded to the 50 genes and 41 ESTs listed in Table 2.

Figure 4

Graphic representation of the gene expression changes (fold change of at least 4) in RET/PTC3, BRAF(V600E), and HRAS(V12) cells versus baseline (PC): upregulated genes (A) and downregulated genes (B). The numbers of upregulated and downregulated genes are represented on the y axis. Thick arrows indicate the Y1062-dependent genes; thin arrows indicate the targets common to RET/PTC3, BRAF(V599E), and HRAS(V12). The different groups of genes are highlighted. Average SLRs (ASLR) are indicated.

Table 1

Genes upregulated 4-fold or more in RET/PTC3 cells^A

Table 2

Genes downregulated 4-fold or more in PC RET/PTC3 cells^A

Validation of the expression data in cultured cells and in human PTCs. We used quantitative RT-PCR (Q-RT-PCR) to evaluate the expression of 50 genes (31 of the 32 upregulated and 19 of the 50 downregulated genes listed in Tables 1 and 2). To verify that gene expression changes were not due to a clonal effect during selection, we analyzed gene expression levels in the mass population (pool no. 3) and in 2 individual clones (Cl1 and Cl2) for each cell line. We then statistically analyzed the expression changes induced by each oncogene. In most cases (37 of the 50 genes), RET/PTC3, HRAS(V12), and BRAF(V600E) activity altered the expression of a particular gene identified by the microarray method. The results of these experiments and the relative statistical analysis are reported in Table 3. In 5 cases (marked in bold in the table), Q-RT-PCR data were not statistically significant (P > 0.05); however, in these cases, expression changes showed a trend consistent with that in microarray data.

Table 3

Gene expression levels in transformed PC cells by Q-RT-PCR

All the genes examined were affected by the Y1062F mutation, their expression changes being negligible in PC RET/PTC(Y1062F) cells (Tables 1 and 2 and data not shown). Moreover, using PC cells transfected by RET/PTC1, another RET/PTC variant, we also demonstrated that gene changes were not restricted to the isoform (RET/PTC3) used in the array screen (data not shown). To determine whether activation of the RAS-BRAF-ERK pathway was required for gene expression regulation induced by RET/PTC3, we used Q-RT-PCR to measure the expression levels of the 37 genes listed in Table 3 in RET/PTC3 cells in which the pathway was transiently obstructed, either by BRAF siRNA or by the MEK inhibitor U0126. As shown in Table 4, expression of 22 of the 37 genes was affected by both BRAF silencing and U0126 treatment. With the sole exception of the Ca2 gene, which was obstructed by U0126 but not by BRAF silencing, there was complete concordance between the results obtained with siRNA and U0126 treatment (Table 4). Scrambled siRNA, used as a negative control, did not affect expression levels (Table 4). These findings support the notion that the BRAF-MEK cascade is essential for the gene expression signature of RET/PTC-transformed cells, but they do not necessarily prove that it is also sufficient. It might well be that additional pathways (perhaps triggered by the same Y1062) contribute to the establishment of the phenotype.

Table 4

BRAF/MEK dependence of gene expression levels in PC RET/PTC3 cells

We selected 4 genes upregulated more than 4-fold and 3 genes (Arpc1b, CD44, RUNX1) upregulated less than 4-fold in the microarray screen. Using Q-RT-PCR, we compared the expression levels of these genes in a set of primary human PTC tissue specimens (n = 13) characterized for oncogene activation (BRAF or RET/PTC) (28) to those in 5 samples of normal thyroid tissue obtained from different donors. Successfully amplified samples are reported in Figure 5A. All 7 analyzed genes were found to be upregulated in most of the tumor samples, and for 5 of them, P was 0.05 or less. We then examined genes coding for the chemokines CXCL1/growth-related oncogene–α (CXCL1/GRO-α) and CXCL10/interferon-γ–inducible protein 10 (CXCL10/IP-10), using a larger set of tumor samples (n = 18) in this case. Both chemokine-encoding genes were significantly upregulated in PTC samples with respect to normal thyroid tissue (Figure 5B).

Figure 5

Expression levels of selected genes (A) and chemokines (B) in human PTC samples versus 5 normal thyroid tissues as determined by Q-RT-PCR. The PTC samples were characterized for the presence of either a RET/PTC rearrangement or a BRAF(V600E) mutation. For each target (indicated on the x axis), the expression levels values of tumors (y axis) were calculated relative to the average expression level in normal tissues (TN). All the experiments were performed in triplicate; SDs were smaller than 25% in all cases (data not shown). P values were calculated by the Mann-Whitney U test.

Autocrine loops that sustain mitogenesis and motility of thyroid cancer cells. Chemokines are small chemotactic cytokines that are subdivided into 2 main families (α or CXC and β or CC chemokines) on the basis of the relative position of cysteine residues (29). Chemokines bind to 7-transmembrane receptors present in the cell surface that are coupled to Gα_i class G proteins and are therefore inhibited by Bordetella pertussis toxin (PTX). Chemokine receptor activation leads to a cascade of cellular events: generation of diacylglycerol and inositol triphosphate, release of intracellular calcium, inhibition of adenylyl cyclase, and activation of several signaling proteins, including JAK/STATs, PKC, phospholipase C, PI3K, and small GTPases of the Ras and Rho families. This cascade results in activation of AKT and ERK and in cell polarization, adhesion, and migration (30, 31).

According to the microarray screen, CXCL1/GRO-α and CCL2/monocyte chemotactic protein–1 (CCL2/MCP-1) were upregulated more than 4-fold in transformed thyrocytes (Table 1). CXCL10/IP-10 was also upregulated, albeit to a lesser extent (Supplemental Table 1). CCL2 is known to be abundantly produced in human PTC (32). However, its receptor, CCR7, was not detectable in thyroid follicular cells (R.M. Melillo et al., unpublished observations). On the other hand, our Q-RT-PCR data demonstrated that CXCL1 and CXCL10 were upregulated in human PTCs (Figure 5B). According to the microarray data, CXCR2 (the CXCL1 receptor) and CXCR3 (the CXCL10 receptor) were expressed in parental and transformed PC cells (Supplemental Table 1).

First, an ELISA assay demonstrated that CXCL1 and CXCL10 were more abundantly secreted by human PTC cell lines spontaneously carrying the RET/PTC1 rearrangement (TPC1, FB2, BHP2-7) or the BRAF(V600E) mutation (BCPAP, BHP5-16) than by P5, a primary culture of normal human thyroid follicular cells (Figure 6A). Moreover, the mRNA for CXCR2 and CXCR3 was upregulated in all the PTC cell lines compared with normal P5 cells (Figure 6B). Accordingly, both receptors were found to be expressed on the cell surface of the 5 PTC cell lines by flow cytometric analysis (Figure 6C and data not shown).

Figure 6

Chemokines and chemokine receptors are expressed in human PTC-derived cell lines. (A) CXCL1 and CXCL10 secretion in human PTC cells was evaluated by ELISA. Experiments were performed in triplicate, and the average value of the results ± SD was plotted. Normal thyroid cells (P5) were used as a negative control. (B) Expression levels of CXCR2 and CXCR3 in PTC cell lines were evaluated by Q-RT-PCR. Expression values were calculated relative to the expression level in normal P5 cells. Experiments were performed in triplicate, and the average value of the results ± SD was plotted. (C) Flow cytometric analysis of surface expression of CXCR2 and CXCR3 in TPC1 cells. (D) TPC1 cells were transfected by BRAF or scrambled siRNA and harvested 72 or 96 hours later. Protein lysates were subjected to immunoblotting with anti-BRAF and anti–phospho-p44/p42 ERK antibodies. (E) BRAF siRNA interference in TPC1 cells affected chemokine production as determined by ELISA.

Treatment of TPC1 cells with BRAF siRNA caused a decrease in endogenous BRAF levels and a parallel reduction of ERK activity (Figure 6D). In parallel, BRAF siRNA, but not scrambled siRNA, remarkably attenuated CXCL1 and CXCL10 chemokine secretion (Figure 6E).

TPC1 cells were stimulated with recombinant CXCL1 or CXCL10; cells were harvested at different time points, and activation of ERK and AKT was analyzed with phospho-specific antibodies. Both chemokines stimulated potent ERK and AKT phosphorylation starting as early as 1 minute after treatment; ERK activation occurred earlier than AKT activation (Figure 7A). To measure DNA synthesis, we counted BrdU-positive cells upon a 1-hour BrdU pulse. The average results of 3 independent experiments are reported in Figure 7B. Treatment with recombinant CXCL1 and CXCL10 stimulated (about 2- to 3-fold) DNA synthesis of TPC1 cells. In contrast, treatment with selective receptor antagonists or blocking antibodies, but not with nonspecific IgG, impaired basal DNA synthesis (about 2-fold). Finally, we asked whether chemokines stimulated cell invasiveness through Matrigel. Representative micrographs are shown in Figure 7C (left) together with the average results of 3 independent assays (right). TPC1 cells had basal levels of invasiveness; exogenous CXCL1 or CXCL10 further induced a strong migratory response. Basal TPC1 migration through Matrigel was inhibited by treatment with PTX, with CXCR2- or CXCR3-blocking antibodies, and with CXCL1- or CXCL10-blocking antibodies but not with nonspecific IgG. Cell motility was also blocked by selective CXCR2 (SB225002) and CXCR3 (TAK-779) antagonists (33, 34). Finally, treatment of TPC1 cells with BRAF siRNA and U0126 inhibited basal migration in Matrigel. This effect was not observed when a scrambled control was used (Figure 7D).

Figure 7

Functional activities of CXCL1 and CXCL10 in human PTC cell lines. (A) Stimulation with CXCL1 and CXCL10 (100 ng/ml) induced time-dependent ERK and AKT activation in TPC1 cells. Cell lysates were harvested at the indicated time points; Western blots were probed with the indicated antibodies. (B) BrdU incorporation was measured to evaluate S-phase entry upon treatment with CXCL1 or CXCL10 or the indicated inhibitors. The average of the results of 3 independent experiments ± SD is indicated. (C) TPC1 cells were allowed to migrate for 24 hours toward serum-free medium or, where indicated, a gradient of CXCL1 or CXCL10. Where indicated, cells were preincubated with blocking antibodies, control antibodies, chemical inhibitors, or PTX. Cells were treated with either BRAF siRNA or U0126. Representative micrographs (left) and absorbance at 570 nm (average ± SD of 3 experiments; right) are shown.

A RET/PTC-RAS-BRAF pathway drives thyroid cancer initiation. Here we show that the oncogenic proteins involved in the initiation of PTC work along the same signaling cascade. This pathway starts at the level of RET tyrosine 1062 and sequentially triggers RAS, BRAF, and ERK stimulation. These conclusions are supported by a body of experimental evidence. First, the transient block of RAS or BRAF activity in HEK293 cells, either by dominant negative mutants or by RNA interference, inhibited RET/PTC-mediated ERK activation. Second, in a thyroid cell setting, RET/PTC3-induced ERK activity depended on BRAF. Finally, the biological effects mediated by RET/PTC3, i.e., cell proliferation and Matrigel invasion, depended on the integrity of the RET/PTC3-RAS-BRAF pathway.

According to such a model, in PC thyroid cells, the 3 activated oncoproteins stimulated largely overlapping gene expression signatures. Most of the common targets were under the control of RET/PTC(Y1062), and the use of BRAF siRNA and chemical blockade of MEK demonstrated that 22 of the 37 genes tested were under the control of the RET/PTC3-RAS-BRAF-ERK cascade. Consistently, some genes had been previously shown to be regulated via the MAPK cascade. One example is DUSP6, a dual-specificity protein phosphatase, which binds to and inactivates ERK1/2 (35). It is likely that the gene expression signature shared by cells transformed by RET/PTC, RAS, or BRAF characterizes a vast number of PTCs. Accordingly, our analysis of a small sample set of primary tumors showed that this gene expression signature may be exploited as a clinical diagnostic test. A larger number of samples will be required to refine the test and to determine whether it is sufficiently robust for clinical implementation.

One corollary of these observations is that therapeutic approaches can be tailored to target the proteins that act downstream from this cascade. Although selective RET kinase inhibitors (36) are effective only in tumors featuring RET rearrangements, and not in those carrying activating mutations in genes functioning downstream from the cascade, chemical blockade of BRAF (37) might be beneficial in carcinomas characterized by oncogenic activation of 1 of the 3 proteins. Moreover, our findings imply that other proteins coupled, at different levels, to the RET/PTC-RAS-BRAF cascade could be involved in PTC samples negative for mutations in the 3 oncoproteins. For instance, it is likely that oncogenic versions of the TRKA receptor, which occur in a small fraction of PTCs (8), substitute RET/PTC upstream from the cascade. It is also conceivable that proteins that modulate the transmission of signals among RET, RAS, BRAF, and ERK can modify the phenotype of thyroid tumors carrying mutations at different levels of the cascade.

Differences in RET/PTC-, RAS-, and BRAF-driven transcriptional profiles. The analysis of transcriptional profiles indicated that the 3 oncoproteins are not completely equivalent. Indeed, in addition to targets common to RET/PTC3, HRAS, and BRAF, there were relatively large sets of genes specifically modulated by only 1 or 2 of the 3 oncogenes. Overall, the similarity between transcriptional profiles of HRAS(V12) and BRAF(V600E) thyroid cells was greater than that between RET/PTC3 and HRAS or RET/PTC3 and BRAF cells. For instance, HRAS and BRAF cells (but not RET/PTC cells) were characterized by overexpression of genes such as IL-6, CXCR1, PDGF-α, SPARC, MCM6, RAF1, and Akt and by downregulation of PLC-β1. This was not unexpected; although the 3 oncoproteins work together along a single cascade, they are biochemically different and therefore able to trigger specific signals in addition to the common ones. Molecular genetics evidence nicely supports this concept. For instance, BRAF mutations are frequently associated with aggressive thyroid carcinomas, such as poorly differentiated and anaplastic carcinomas, that rarely, if ever, harbor RET/PTC rearrangements (28). On the other hand, RAS point mutations are rare in classic PTCs and, instead, characterize PTCs of the follicular variant that are rarely affected by mutations in BRAF or RET (9). Detailed analysis of the gene expression signatures in a large set of samples may help to identify molecular profiles that characterize each oncoprotein.

Function of the genes modulated by RET/PTC3, HRAS(V12) and BRAF(V600E). Many of the commonly modulated genes were involved in hallmarks of neoplastic transformation, e.g., altered cell morphology, uncontrolled growth, loss of differentiation, and apoptosis. Induced genes included MMPs (MMP3, -10, -12, -13) (38). Another category of genes that was greatly overexpressed in transformed PC cells were those encoding adhesion/structure-associated proteins, such as collagen 1 (Col1α1), thymosin β4, and galectin-3, previously reported to be overexpressed in thyroid carcinomas (39–42). The dual-specificity phosphatase (DUSP6) was also previously identified as a gene upregulated in human PTC samples (41). Finally, upregulated genes included those coding for S100A4 (p9KA) and S100A6 (calcyclin); their elevated levels have been associated with cancer and metastasis (43). Consistent with the loss of the differentiated phenotype of transformed cells, genes coding for differentiation markers were downregulated. Interestingly, the gene Hhex, which encodes for a proline-rich homeodomain transcription factor and is regulated by thyroid-specific transcription factors, such as TTF-1 and PAX8 (44), was also downregulated. Another downregulated gene was mitogen-inducible gene–6 (Mig6). This protein is a negative-feedback regulator of the epidermal growth factor receptor and a potential tumor suppressor (45).

Chemokine autocrine loops in transformed thyroid cells. PTC is associated with a striking chronic inflammatory reaction in about 30% of cases (1, 32). Our data provide a molecular explanation for this phenomenon by showing that the RET/PTC-RAS-BRAF signaling cascade that drives PTC initiation stimulates the overexpression of several chemokines. In turn, chemokines secreted by tumor cells can recruit leukocytes (macrophages, dendritic cells, T cells, and natural killer cells) to tumor sites (30, 31). An interesting aspect of the relationship between inflammation and cancer is that several tumors use molecules of the innate immune system not only to recruit leukocytes, but also for growth, survival, and metastasis. It can be envisaged that a mixture of chemokines is produced in the tumor microenvironment and act in an autocrine/paracrine fashion between neighboring homotypic cells (46, 47). Expression of CXCR2 and CXCR3, the receptors for CXCL1 and CXCL10, respectively, was detected in transformed thyroid cells. Therefore, 2 autocrine/paracrine loops may mediate biological activities in PTCs that are relevant for the establishment of the neoplastic phenotype, e.g., autonomous proliferation and motility.

In conclusion, we propose that PTCs are initiated by a set of transforming events that target proteins that act along a linear signaling cascade. Activation of this signaling cascade results in upregulation of chemokines and their receptors that are relevant for sustained proliferation and cell motility of tumor cells. We suggest that these genes may act cooperatively to commit transformed thyroid cells to a malignant invasive phenotype.

View Supplemental data

View Supplemental table

We are grateful to C.J. Marshall for the BRAF expression plasmid; to F. Curcio for the P5 primary thyroid culture; to J.M. Hershman for the BHP2-7 and BHP5-16 cell lines; and to L. Vitiello and L. Racioppi for FACS analysis of thyroid cell cultures. We would like to thank J. Gilder for editing the text. This study was supported by the Associazione Italiana per la Ricerca sul Cancro (AIRC), the Progetto Strategico Oncologia of the CNR/Italian Ministero per l’Istruzione, Università e Ricerca Scientifica (MIUR), Biotecnologia e Genetica Molecolare nel Mezzogiorno d’Italia (BioGeM) s.c.ar.l., the Italian Ministero della Salute, and the Centro Regionale di Competenza GEAR (Genomic for Applied Research). V. Guarino and M.D. Castellone were recipients of BioGeM s.c.ar.l. fellowships.

Nonstandard abbreviations used: EST, expressed sequence tag; GDNF, glial-derived neurotrophic factor; GRO-α, growth-related oncogene–α; 6H, 6 hormones; IP-10, interferon-γ–inducible protein 10; MCP-1, monocyte chemotactic protein–1; MEK, MAPK kinase; NGF, nerve growth factor; PTC, papillary thyroid carcinoma; PTX, Bordetella pertussis toxin; Q-RT-PCR, quantitative RT-PCR; RTK, receptor tyrosine kinase; siRNA, small interfering duplex RNA; SLR, signal log ratio; TG, thyroglobulin; TSH, thyrotropic hormone.

Conflict of interest: The authors have declared that no conflict of interest exists.

1. Rosai, J., Carcangiu, M.L., and DeLellis, R.A. 1992. Atlas of tumor pathology: tumors of the thyroid gland. 3rd series. Armed Forces Institute of Pathology. Washington, D.C., USA. 343 pp.
2. Manie, S, Santoro, M, Fusco, A, Billaud, M. The RET receptor: function in development and dysfunction in congenital malformation. Trends Genet. 2001. 17:580-589.
   View this article via: PubMed CrossRef Google Scholar
3. Santoro, M, et al. Ret oncogene activation in human thyroid neoplasms is restricted to the papillary cancer subtype. J. Clin. Invest. 1992. 89:1517-1522.
   View this article via: JCI PubMed CrossRef Google Scholar
4. Fagin, JA. Perspective: lessons learned from molecular genetic studies of thyroid cancer--insights into pathogenesis and tumor-specific therapeutic targets. Endocrinology. 2002. 143:2025-2028.
   View this article via: PubMed CrossRef Google Scholar
5. Santoro, M, et al. Development of thyroid papillary carcinomas secondary to tissue-specific expression of the RET/PTC1 oncogene in transgenic mice. Oncogene. 1996. 12:1821-1826.
   View this article via: PubMed Google Scholar
6. Buckwalter, TL, et al. The roles of phosphotyrosines-294, -404, and -451 in RET/PTC1-induced thyroid tumor formation. Oncogene. 2002. 21:8166-8172.
   View this article via: PubMed CrossRef Google Scholar
7. Fusco, A, et al. Assessment of RET/PTC oncogene activation and clonality in thyroid nodules with incomplete morphological evidence of papillary carcinoma: a search for the early precursors of papillary cancer. Am. J. Pathol. 2002. 160:2157-2167.
   View this article via: PubMed Google Scholar
8. Pierotti, MA. Chromosomal rearrangements in thyroid carcinomas: a recombination or death dilemma. Cancer Lett. 2001. 166:1-7.
   View this article via: PubMed CrossRef Google Scholar
9. Zhu, Z, Gandhi, M, Nikiforova, MN, Fischer, AH, Nikiforov, YE. Molecular profile and clinical-pathologic features of the follicular variant of papillary thyroid carcinoma. An unusually high prevalence of ras mutations. Am. J. Clin. Pathol. 2003. 120:71-77.
   View this article via: PubMed CrossRef Google Scholar
10. Kimura, ET, et al. 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. 63:1454-1457.
    View this article via: PubMed Google Scholar
11. Malumbres, M, Barbacid, M. RAS oncogenes: the first 30 years. Nat. Rev. Cancer. 2003. 3:459-465.
    View this article via: PubMed CrossRef Google Scholar
12. Davies, H, et al. Mutations of the BRAF gene in human cancer. Nature. 2002. 417:949-954.
    View this article via: PubMed CrossRef Google Scholar
13. Cohen, Y, et al. BRAF mutation in papillary thyroid carcinoma. J. Natl. Cancer Inst. 2003. 95:625-627.
    View this article via: PubMed Google Scholar
14. Soares, P, et al. BRAF mutations and RET/PTC rearrangements are alternative events in the etiopathogenesis of PTC. Oncogene. 2003. 22:4578-4580.
    View this article via: PubMed CrossRef Google Scholar
15. Fagin, JA. How thyroid tumors start and why it matters: kinase mutants as targets for solid cancer pharmacotherapy. J. Endocrinol. 2004. 183:249-256.
    View this article via: PubMed CrossRef Google Scholar
16. Schlessinger, J, Lemmon, MA. SH2 and PTB domains in tyrosine kinase signaling [review]. Sci. STKE. 2003. 191:RE12.
17. Kawamoto, Y, et al. Identification of RET autophosphorylation sites by mass spectrometry. J. Biol. Chem. 2004. 279:14213-14224.
    View this article via: PubMed CrossRef Google Scholar
18. Pandey, A, et al. Direct association between the Ret receptor tyrosine kinase and the Src homology 2-containing adapter protein Grb7. J. Biol. Chem. 1996. 271:10607-10610.
    View this article via: PubMed CrossRef Google Scholar
19. Borrello, MG, et al. The full oncogenic activity of Ret/ptc2 depends on tyrosine 539, a docking site for phospholipase Cgamma. Mol. Cell. Biol. 1996. 16:2151-2163.
    View this article via: PubMed Google Scholar
20. Encinas, M, Crowder, RJ, Milbrandt, J, Johnson (Jr), EM. Tyrosine 981, a novel Ret autophosphorylation site, binds c-Src to mediate neuronal survival. J. Biol. Chem. 2004. 279:18262-18269.
    View this article via: PubMed CrossRef Google Scholar
21. Ichihara, M, Murakumo, Y, Takahashi, M. RET and neuroendocrine tumors. Cancer Lett. 2004. 204:197-211.
    View this article via: PubMed CrossRef Google Scholar
22. Melillo, RM, et al. Docking protein FRS2 links the protein tyrosine kinase RET and its oncogenic forms with the mitogen-activated protein kinase signaling cascade. Mol. Cell. Biol. 2001. 21:4177-4187.
    View this article via: PubMed CrossRef Google Scholar
23. Castellone, MD, et al. Ras-mediated apoptosis of PC CL 3 rat thyroid cells induced by RET/PTC oncogenes. Oncogene. 2003. 22:246-255.
    View this article via: PubMed CrossRef Google Scholar
24. Fusco, A, et al. One- and two-step transformations of rat thyroid epithelial cells by retroviral oncogenes. Mol. Cell. Biol. 1987. 9:3365-3370.
25. Santoro, M, et al. The TRK and RET tyrosine kinase oncogenes cooperate with ras in the neoplastic transformation of a rat thyroid epithelial cell line. Cell Growth Differ. 1993. 4:77-84.
    View this article via: PubMed Google Scholar
26. Knauf, JA, Kuroda, H, Basu, S, Fagin, JA. RET/PTC-induced dedifferentiation of thyroid cells is mediated through Y1062 signaling through SHC-RAS-MAP kinase. Oncogene. 2003. 22:4406-4412.
    View this article via: PubMed CrossRef Google Scholar
27. Shirokawa, JM, et al. Conditional apoptosis induced by oncogenic ras in thyroid cells. Mol. Endocrinol. 2000. 14:1725-1738.
    View this article via: PubMed CrossRef Google Scholar
28. Nikiforova, MN, et al. BRAF mutations in thyroid tumors are restricted to papillary carcinomas and anaplastic or poorly differentiated carcinomas arising from papillary carcinomas. J. Clin. Endocrinol. Metab. 2003. 88:5399-5404.
    View this article via: PubMed CrossRef Google Scholar
29. Dhawan, P, Richmond, A. Role of CXCL1 in tumorigenesis of melanoma. J. Leukoc. Biol. 2002. 72:9-18.
    View this article via: PubMed Google Scholar
30. Luster, AD. Chemokines--chemotactic cytokines that mediate inflammation. N. Engl. J. Med. 1998. 338:436-445.
    View this article via: PubMed CrossRef Google Scholar
31. Mellado, M, Rodriguez-Frade, JM, Manes, S, Martinez-A, C. Chemokine signaling and functional responses: the role of receptor dimerization and TK pathway activation. Annu. Rev. Immunol. 2001. 19:397-421.
    View this article via: PubMed CrossRef Google Scholar
32. Scarpino, S, et al. Papillary carcinoma of the thyroid: hepatocyte growth factor (HGF) stimulates tumor cells to release chemokines active in recruiting dendritic cells. Am. J. Pathol. 2000. 156:831-837.
    View this article via: PubMed Google Scholar
33. White, JR, et al. Identification of a potent, selective non-peptide CXCR2 antagonist that inhibits interleukin-8-induced neutrophil migration. J. Biol. Chem. 1998. 273:10095-10098.
    View this article via: PubMed CrossRef Google Scholar
34. Gao, P, et al. The unique target specificity of a nonpeptide chemokine receptor antagonist: selective blockade of two Th1 chemokine receptors CCR5 and CXCR3. J. Leukoc. Biol. 2003. 73:273-280.
    View this article via: PubMed CrossRef Google Scholar
35. Eblaghie, MC, et al. Negative feedback regulation of FGF signaling levels by Pyst1/MKP3 in chick embryos. Curr. Biol. 2003. 13:1009-1018.
    View this article via: PubMed CrossRef Google Scholar
36. Carlomagno, F, et al. ZD6474, an orally available inhibitor of KDR tyrosine kinase activity, efficiently blocks oncogenic RET kinases. Cancer Res. 2002. 62:7284-7290.
    View this article via: PubMed Google Scholar
37. Lyons, JF, Wilhelm, S, Hibner, B, Bollag, G. Discovery of a novel Raf kinase inhibitor. Endocr. Relat. Cancer. 2001. 8:219-225.
    View this article via: PubMed CrossRef Google Scholar
38. Liotta, LA, Kohn, EC. The microenvironment of the tumour-host interface. Nature. 2001. 411:375-379.
    View this article via: PubMed CrossRef Google Scholar
39. Dahlman, T, et al. Collagen type I expression in experimental anaplastic thyroid carcinoma: regulation and relevance for tumorigenicity. Int. J. Cancer. 2002. 98:186-192.
    View this article via: PubMed CrossRef Google Scholar
40. Califano, D, et al. Thymosin beta-10 gene overexpression correlated with the highly malignant neoplastic phenotype of transformed thyroid cells in vivo and in vitro. Cancer Res. 1998. 58:823-828.
    View this article via: PubMed Google Scholar
41. Huang, Y, et al. Gene expression in papillary thyroid carcinoma reveals highly consistent profiles. Proc. Natl. Acad. Sci. U. S. A. 2001. 98:15044-15049.
    View this article via: PubMed CrossRef Google Scholar
42. Bartolazzi, A, et al. Application of an immunodiagnostic method for improving preoperative diagnosis of nodular thyroid lesions. Lancet. 2001. 357:1644-1650.
    View this article via: PubMed CrossRef Google Scholar
43. El-Rifai, W, et al. Gastric cancers overexpress S100A calcium-binding proteins. Cancer Res. 2002. 62:6823-6826.
    View this article via: PubMed Google Scholar
44. Elsalini, OA, von Gartzen, J, Cramer, M, Rohr, KB. Zebrafish hhex, nk2.1a, and pax2.1 regulate thyroid growth and differentiation downstream of Nodal-dependent transcription factors. Dev. Biol. 2003. 263:67-80.
    View this article via: PubMed CrossRef Google Scholar
45. Hackel, PO, Gishizky, M, Ullrich, A. Mig-6 is a negative regulator of the epidermal growth factor receptor signal. Biol. Chem. 2001. 382:1649-1662.
    View this article via: PubMed CrossRef Google Scholar
46. Balkwill, F, Mantovani, A. Inflammation and cancer: back to Virchow? Lancet. 2001. 357:539-545.
    View this article via: PubMed CrossRef Google Scholar
47. Coussens, LM, Werb, Z. Inflammation and cancer. Nature. 2002. 420:860-867.
    View this article via: PubMed CrossRef Google Scholar
48. Curcio, F, Ambesi-Impiombato, FS, Perrella, G, Coon, HG. Long-term culture and functional characterization of follicular cells from adult normal human thyroids. Proc. Natl. Acad. Sci. U. S. A. 1994. 91:9004-9008.
    View this article via: PubMed CrossRef Google Scholar
49. Hingorani, SR, Jacobetz, MA, Robertson, GP, Herlyn, M, Tuveson, DA. Suppression of BRAF(V599E) in human melanoma abrogates transformation. Cancer Res. 2003. 63:5198-5202.
    View this article via: PubMed Google Scholar
50. Yuan, B, Latek, R, Hossbach, M, Tuschl, T, Lewitter, F. siRNA Selection Server: an automated siRNA oligonucleotide prediction server. Nucleic Acids. Res. 2004. 32:W130-W134.
    View this article via: PubMed CrossRef Google Scholar
51. Castellone, MD, et al. Autocrine stimulation by osteopontin plays a pivotal role in the expression of the mitogenic and invasive phenotype of RET/PTC-transformed thyroid cells. Oncogene. 2004. 23:2188-2196.
    View this article via: PubMed CrossRef Google Scholar