Gene expression analysis after receptor tyrosine kinase activation reveals new potential melanoma proteins
© Teutschbein et al. 2010
Received: 13 April 2010
Accepted: 21 July 2010
Published: 21 July 2010
Skip to main content
© Teutschbein et al. 2010
Received: 13 April 2010
Accepted: 21 July 2010
Published: 21 July 2010
Melanoma is an aggressive tumor with increasing incidence. To develop accurate prognostic markers and targeted therapies, changes leading to malignant transformation of melanocytes need to be understood. In the Xiphophorus melanoma model system, a mutated version of the EGF receptor Xmrk (Xiphophorus melanoma receptor kinase) triggers melanomagenesis. Cellular events downstream of Xmrk, such as the activation of Akt, Ras, B-Raf or Stat5, were also shown to play a role in human melanomagenesis. This makes the elucidation of Xmrk downstream targets a useful method for identifying processes involved in melanoma formation.
Here, we analyzed Xmrk-induced gene expression using a microarray approach. Several highly expressed genes were confirmed by realtime PCR, and pathways responsible for their induction were revealed using small molecule inhibitors. The expression of these genes was also monitored in human melanoma cell lines, and the target gene FOSL1 was knocked down by siRNA. Proliferation and migration of siRNA-treated melanoma cell lines were then investigated.
Genes with the strongest upregulation after receptor activation were FOS-like antigen 1 (Fosl1), early growth response 1 (Egr1), osteopontin (Opn), insulin-like growth factor binding protein 3 (Igfbp3), dual-specificity phosphatase 4 (Dusp4), and tumor-associated antigen L6 (Taal6). Interestingly, most genes were blocked in presence of a SRC kinase inhibitor. Importantly, we found that FOSL1, OPN, IGFBP3, DUSP4, and TAAL6 also exhibited increased expression levels in human melanoma cell lines compared to human melanocytes. Knockdown of FOSL1 in human melanoma cell lines reduced their proliferation and migration.
Altogether, the data show that the receptor tyrosine kinase Xmrk is a useful tool in the identification of target genes that are commonly expressed in Xmrk-transgenic melanocytes and melanoma cell lines. The identified molecules constitute new possible molecular players in melanoma development. Specifically, a role of FOSL1 in melanomagenic processes is demonstrated. These data are the basis for future detailed analyses of the investigated target genes.
Melanoma development is a complex process based on many epigenetic and genetic factors. The most frequent genetic changes in human melanoma are activating mutations in either BRAF or NRAS. This is often combined with inactivating mutations in phosphatase and tensin homologue (PTEN) or cyclin-dependent kinase inhibitor 2 a (CDKN2A) .
The search for other characteristics shared between human melanoma from different individuals has revealed the importance of several proteins influencing melanoma cell cycle progression, apoptosis, cell adhesion, and angiogenesis. Examples are cyclin-dependent kinase 4 (CDK4), AKT, β-catenin, melanoma inhibitory activity protein (MIA), and Ephrin-A1 (EFNA1) [1, 2]. Still, the search for further melanoma-relevant genes is a promising concept with potential therapeutic value, and several recent studies applying high-throughput gene expression profiling have associated previously unknown candidate genes with melanoma progression [3–5]. However, the comparability among different studies is low due to the variability of human tumor biopsies and the cultivation-dependent changes in melanoma-derived cell lines.
By contrast, animal models represent genetic systems with well defined genetic background where tumorigenesis is initiated by known molecular events. In the Xiphophorus fish melanoma model, a single oncogenic epidermal growth factor receptor, termed Xiphophorus melanoma receptor kinase (xmrk) is responsible for spontaneously developing melanoma .
Xmrk uses several signaling cascades that are also involved in human melanomagenesis, e.g. the phosphoinositide-3 kinase (PI3K) pathway and the RAS-RAF-MAPK cascade. Other molecules, e.g. the signal transducer and activator of transcription 5 (STAT5) and osteopontin (OPN), were first identified as essential mediators of Xmrk signaling and were subsequently shown to be relevant in human melanomagenesis as well [7–10]. These findings prompted us to search for novel Xmrk-regulated genes that may potentially play a role in human melanoma development.
It was shown many times that Xmrk signaling is highly comparable between its natural host cells - pigment cells from Xiphophorus - and mammalian cells that ectopically express the receptor [11–14]. Xmrk is permanently active due to its dimerization. However, to be able to differentiate between inactive and active receptor signaling, we are using the melanocytes cell line melan-a stably expressing a chimeric protein consisting of the extracellular part of EGFR ("HER") and the cytoplasmic part of Xmrk ("mrk") (melan-a "HERmrk" or shortly "Hm"). Melan-a cells lack endogenous EGFR, and the stimulation of Hm cells with EGF results in specific induction of Xmrk-dependent signaling pathways and tumorigenic transformation.
Here, we have analyzed gene expression profiles of stimulated versus unstimulated cells using a microarray approach. The genes with the strongest regulation in response to activated HERmrk were FOS-like antigen 1 (Fosl1), early growth response 1 (Egr1), osteopontin (Opn), insulin-like growth factor binding protein 3 (Igfbp3), dual-specificity phosphatase 4 (Dusp4), and tumor-associated antigen L6 (Taal6). We investigated the pathways regulating these genes and analyzed their expression in human melanoma cell lines. We furthermore found that the knockdown of FOSL1 reduced proliferation and migration of human melanoma cell lines. Thus, this study reveals FOSL1 as new potential molecular player in melanomagenesis by using the Xmrk melanoma model.
Cells were starved for 72 h with DMEM containing 2.5% dialyzed FCS. After stimulation with 100 ng/ml human EGF (hEGF) (tebu-bio, Le Perray en Yvelines, France) for indicated times, RNA was extracted from the cells using the RNeasy kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. Only RNA samples with an A260/A280 ratio > 1.8 were used for microarray hybridization.
Transcriptional profiling was done on a microarray containing 21,168 DNA spots from the mouse cDNA library NIA 15 k and 7.4 k Mouse cDNA Clone Set (National Institute on Aging, Bethesda, MD). Total RNA was purified with RNeasy spin columns (Qiagen). After mRNA amplification with MessageAmp II aRNA Kit (Ambion, Austin, TX USA), Cy3- and Cy5-labeled cDNA probes were generated using the CyScribe cDNA Post-Labelling Kit (GE Healthcare, Amersham Place, Little Chalfont England). The labeled probes were purified with QIAquick PCR Purification Kit (Qiagen, Hilden, Germany), combined in hybridization buffer (0.2% SDS, 4.5× SSC) and hybridized on the microarray for 16 h at 55°C. Finally, the chips were washed at a stringency of 0.1 × SSC/0.1% SDS, dried by centrifugation, scanned and quantified using Scan Array Express (Perkin Elmer, Waltham, MA USA).
Each experiment was performed as sandwich hybridization, i.e. instead of a coverslip, a second microarray slide was used. This provides a replicated measurement for each hybridization that can be used for quality control and that reduces technical variability. For each spot, median signal and background intensities for both channels were obtained. To account for spot differences, the background-corrected ratio of the two channels was calculated and log2 transformed. To balance the fluorescence intensities for Cy3 and Cy5 as well as to allow for comparison of expression levels across experiments, the raw data were standardized. We used the print-tip-LOWESS normalization to correct for inherent bias on each chip. Expression data and gene annotations were stored in Array Express http://www.ebi.ac.uk/arrayexpress/ (accession: E-MEXP-1311), which complies with MIAME (minimal information about a microarray experiment) guidelines. The R environment software http://www.r-project.org/ was used for data analysis. To find differently expressed genes, changes in mRNA expression levels in stimulated versus unstimulated cells were calculated for each gene. The normalized data were filtered due to strict quality criteria and analyzed using Microsoft Excel. For experimental comparisons, genes showing at least a twofold change were chosen.
Mouse melanocytes transfected with HERmrk [melan-a Hm cells, ] or with human EGF receptor (melan-a HER) were cultured as described previously . The human immortalized melanocyte cell line Hermes 3a (kindly provided by D. Bennett) was grown in RPMI supplemented with penicillin (400 U/ml), streptomycin (50 μg/ml), L-glutamine (300 μg/ml), TPA (200 nM), cholera toxin (2 pM), hSCF (10 ng/ml), endothelin (10 nM), and 10% FCS, as previously described . Human melanoma cell lines Mel Im, Mel Wei, Mel Juso, and SK-Mel-3 (kind gifts by A. Bosserhoff) as well as A375, A375M, DX-3, LT5.1, and SK-Mel-28 were maintained in DMEM supplemented with penicillin (400 U/ml), streptomycin (50 μg/ml), L-glutamine (300 μg/ml) and 10% FCS. Normal human epidermal melanocytes (NHEM) derived from foreskin were obtained from PromoCell (Heidelberg, Germany) and grown in melanocyte growth medium MGM (PromoCell) under a humidified atmosphere of 5% CO2 at 37°C. NHEM cells were used between passages 3 and 6.
Cells were starved as described and subsequently stimulated with 100 ng/ml hEGF for indicated times. RNA extraction from stimulated melan-a Hm cells and human cell lines was done using Total RNA Isolation Reagent (ABgene, Epsom, UK) as recommended by the manufacturer. For the identification of pathways regulating expression of candidate genes, the small molecule inhibitors AG1478 (20 μM), PP2 (20 μM), LY294002 (10 μM), or U0126 (10 μM), respectively, were applied one hour prior to hEGF stimulation. Cells without inhibitor treatment received the equivalent amount of DMSO. Primary data of these experiments are available on request. cDNA was prepared from total RNA using the RevertAid kit with random hexamer primers (Fermentas, Burlington, Canada). PCR primers were designed using Primer3 software version 0.4.0 http://frodo.wi.mit.edu/primer3/. PCR was carried out using the iCycler IQ (Bio-Rad, Hercules, CA). Values for each gene were normalized to expression levels of β-actin (mouse cell lines) or ribosomal protein S14 (human cell lines). Primer sequences are available on request.
Cells were lyzed and blotted as described . Before probing with the reference antibody, the blots were stripped as described. Monoclonal anti-phosphotyrosine (PY20) was from BD Biosciences (San José, CA). Phospho-p44/42 MAPK (Thr202/Tyr204) antibody was purchased from Cell Signaling Technology (Danvers, MA). Polyclonal antibodies against FOSL1 (N-17, C-12), EGR1 (C-19), OPN (P-18 and K-20), IGFBP3 (H-98), MKP2 (H-67), and ERK2 (C-14) were from Santa Cruz Biotechnology (Santa Cruz, CA). The MYC-tag antibody (9B11) was from Cell Signaling Technology (Danvers, MA). Secondary antibodies were conjugated with horseradish peroxidase and were directed against mouse (Pierce, Rockford, IL), rabbit (Bio-Rad) or goat (Abcam, Cambridge, UK). TAAL6 mouse IgM antibody (10H6) was a kind gift from S. Roffler. Secondary antibody against mouse IgM was purchased from Rockland (Gilbertsville, PA).
One day before siRNA transfection, melanoma cells were seeded at a density of 3 × 104 cells per well of a 12-well plate. For human melanoma cells, commercially available siRNA against human FOSL1 as well as control siRNA (Smart Pool On Target Plus, Thermo Scientific) were used. siRNA was transfected using X-treme gene transfection reagent (Roche), according to the manufacturer's recommendations. Downregulation was monitored after 48 h by realtime PCR and Western blot analysis.
Melanoma cell lines A375 and Mel Juso were plated in triplicate (2.5 × 103 cells per well of a 96-well plate) in DMEM containing 10% FCS. 48 h after siRNA treatment, cells were incubated with 10 μM BrdU for 6 h. BrdU incorporation was then quantified using a colorimetric BrdU cell proliferation ELISA, as recommended by the manufacturer (Roche).
2 × 104 A375 or Mel Juso cells were transfected with control- or FOSL1-specific siRNA. One day later, they were serum-starved in DMEM containing 1% dialyzed FCS for 24 h and applied to the upper chamber of a transwell inlay (polycarbonate, 10 mm diameter, 8 μm pores, Nunc). Migration was measured as described before . The migration rate of the siFOSL1-treated cells was determined relative to the migration of control cells.
To obtain a detailed picture of the time course of Xmrk-dependent gene expression, we compared unstimulated melan-a HERmrk cells to those stimulated for 15 minutes, 1 h, 2 h, 4 h, 8 h or 24 h with EGF.
Successful receptor activation was controlled by Western blot (Additional file 1, Figure S1a) and realtime PCR (Additional file 1, Figure S1b). Both assays revealed successful stimulation, visible by receptor- and MAPK phosphorylation as well as induction of the Xmrk target gene Opn.
Subsequent microarray analysis revealed the regulation of 1,273 genes at one or more time-points after HERmrk stimulation. A gene was considered to be regulated when changed twofold and more. The complete list of expression data and gene annotations is available at http://www.ebi.ac.uk/arrayexpress/ (accession: E-MEXP-1311).
Regulated genes were categorized with respect to their molecular functions and biological processes according to the Gene Ontology (GO) terminology. Using the expression analysis systematic explorer (EASE) software , overrepresentation of gene ontology terms in the 1,273 regulated genes compared to the total number of genes assayed (21,168 spots on the chip) was calculated. Significantly enriched categories (EASE score < 0.05) are listed in Additional file 2, Table S1. Among the biological processes, protein metabolism and protein modification were particularly enriched, indicating a high metabolic activity as expected from growth factor stimulated cells, and enhancement of signal transduction processes. The most overrepresented molecular function was nucleic acid binding, encompassing transcription factors and factors regulating nucleic acid stability.
Altogether, in all cases except Igfbp3, transcript induction as well as its time course was confirmed by realtime PCR.
Previous analyses have revealed that Xmrk transfers signals via STAT5, the PI3K pathway, the RAS-RAF-MAPK cascade, and the cytoplasmic SRC kinase family member FYN . To identify the pathways responsible for modulation of the chosen ten genes, we blocked known Xmrk-induced pathways using the small molecule inhibitors AG1478, U0126, PP2, or LY294002. Target gene expression in presence or absence of the inhibitors was analyzed by realtime PCR (Additional file 3, Table S2 and, for a schematic overview, Addditional file 4, Figure S2).
As expected, the inducing effects of EGF on all genes were abrogated when HERmrk was inhibited by AG1478. While regulation of Emp1, Fosl1, and Opn was both MEK- and SRC family kinase-dependent, induction of Sos1, Ube2I, Dusp4, and Taal6 was only restrained by inhibiting SRC-family kinases with PP2. Egr1 expression could only be decreased after MEK inhibition, and Cyr61 transcription was dependent on MEK and PI3K. Finally, expression of Igfbp3 was decreased after application of each of the inhibitors.
For further analysis, we focused on six significantly regulated genes from four different functional groups: the transcription factors FOSL1 and EGR1, the secreted proteins OPN and IGFBP3, the phosphatase DUSP4, and the membrane protein TAAL6. We monitored their expression levels in human melanoma cell lines compared to normal human epidermal melanocytes (NHEM). For this analysis we chose eight different cell lines, containing either activating N-RAS- (DX-3, LT5.1 and Mel Juso) or B-RAF mutations (A375, A375M, Mel Wei, Mel Im, SK-Mel-3 and SK-Mel-28).
Here we describe the regulation of potential novel melanoma candidate genes using an inducible version of the EGFR orthologue Xmrk.
Ten of the most strongly regulated genes were examined in presence of pathway inhibitors to determine the respective signal transduction chain responsible for their regulation. Eight out of ten genes were at least partly controlled by members of the SRC kinase family, while six genes were induced by the MAPK pathway and only two genes by the PI3 kinase pathway. Usually, gene regulation was reduced by inhibition of more than one pathway, which indicates an interplay between the simultaneously activated pathways. Although we could only perform the inhibitor analysis with a small group of genes, which is not representative of the overall mode of gene regulation, it was surprising that such a high gene number was regulated by SRC kinases. From previous studies it is known that FYN is the only SRC family kinase member that is directly activated by Xmrk . FYN prevents inactivation of MAPK by reducing MAPK phosphatase 1 (MKP-1) expression, but also stabilizes the complex between Xmrk and the p85 subunit of PI3K [13, 26]. This might explain why in our study many genes are not only affected by SRC family kinase inhibition, but also by blocking MEK or PI3K. Although the effect of SRC kinases on gene expression has not yet been investigated in human melanoma, accumulating data indicate that this pathway plays a vital role for the malignant cells. Specifically, it was shown by kinase activity profiling that SRC is activated in primary human melanoma and its inhibition leads to reduced growth . Activated SRC was also detected in cutaneous, mucosal and metastatic melanoma, and its inhibition by dasatinib or bosutinib blocked the growth of several melanoma cell lines [28, 29]. Furthermore, an active SRC family kinase-focal adhesion kinase complex, responsible for migration and metastatic spread, was described both for the Xmrk model and for human melanoma [14, 30]. Together with our own data, this depicts an important melanomagenic role for SRC kinases in general and raises the necessity to further scrutinize SRC family kinase-dependent gene regulation in human melanoma.
Among the identified genes, expression of Egr1, Dusp4, Igfbp3, Fosl1, Opn, and Taal6 were investigated more closely. Importantly, five of these genes have been found to be upregulated in several different melanoma microarray studies. Recently, B-RAFV600E-dependent genes were analysed by MEK inhibition or BRAF siRNA in human melanoma cell lines, revealing regulation of the transcription factors FOSL1 and EGR1 by this pathway . Interestingly, EGR1 belongs to a distinct group of salivary marker genes expressed in melanoma-bearing mice . IGFBP3 was more strongly expressed in metastatic and cutaneous melanoma compared to melanocytes . Its function seems to depend on the cellular context, as both anti- and pro-tumorigenic roles were attributed to this protein [34, 35].
DUSP4 and OPN were upregulated in cutaneous melanoma in comparison to normal skin or benign nevi . DUSP4 is a dual phosphatase capable of dephosphorylating p38, JNK1 and ERK1/2, though JNK1 seems to be the preferred target in vitro . Its upregulation may simply reflect counterregulation of the ERK1/2 pathway in melanoma, but has not yet been investigated more closely. OPN, on the other hand, is a well-known melanoma marker [37, 38] and served as internal control in our studies.
Finally, in a murine melanoma model with xenografted A375 cells, EGR1 and OPN expression was enhanced in cells derived from metastatic melanoma compared to the parental cell line as well as in metastatic melanoma versus primary melanoma . The original data of all above-mentioned microarray studies are available at http://www.oncomine.org.
Due to the important role of the AP-1 complex in human melanoma, the lack of information on FOSL1 function in this tumor type, and the MAPK pathway dependent induction of FOSL1 in melanoma cell lines, we investigated the effect of FOSL1 knockdown on the proliferation of two melanoma cell lines and found a significant inhibition of cells entering the S phase, which was not due to the induction of apoptosis (data not shown). Also, migration of Mel Juso cells was decreased after FOSL1 knockdown. Its function in melanoma was not described previously, but pro-tumorigenic roles of FOSL1 were reported for other solid cancers [40, 41]. Oncogenic EGFRvIII in glioblastoma cells induces FOSL1 , and it modulates the malignant features of glioma cells, so it was suggested as target for therapeutic interventions against malignant gliomas . An oral DNA vaccine directed against FOSL1 has been demonstrated to effectively suppress tumor growth, angiogenesis and metastasis in mice injected with breast carcinoma cells .
In summary, we used the high overlap between pathways downstream of Xmrk and established human melanoma pathways for the search of new melanoma-relevant target genes. Our gene and protein expression results indicate that Xmrk serves as a suitable model oncogene for this purpose. As a proof of principle, we investigated the AP-1 complex component FOSL1 in more detail. We found that the gene is similarly regulated in a MAPK-dependent manner by Xmrk and by human melanoma oncogenes. Importantly, we also could demonstrate a pro-tumorigenic role of FOSL1 in human melanoma cell lines, thus confirming the Xmrk oncogene as instrumental in the search of new melanoma players.
Here we show the receptor tyrosine kinase Xmrk is a valuable tool to identify target genes and proteins that are commonly shared between Xmrk-transgenic melanocytes and human melanoma cell lines. Interestingly, of the ten most strongly upregulated genes, the majority was regulated by SRC kinases, followed by the MAPK pathway. The transcription factor component gene FOSL1 also belongs to this group of genes. By knocking down FOSL1, we could demonstrate a pro-proliferative and pro-migratory function of this protein in melanoma cell lines. The presented data reveal new potential melanoma-relevant genes that can now be investigated for their melanomagenic function in detail.
v-akt murine thymoma viral oncogene homolog
activator protein 1
v-raf murine sarcoma viral oncogene homolog B1
cyclin dependent kinase inhibitor 2A
cysteine-rich, angiogenic inducer 61
dual specificity phosphatase 4
epidermal growth factor
epidermal growth factor receptor
early growth response 1
epithelial membrane protein 1
extracellular regulated MAP kinase
FOS-like antigen 1
FYN oncogene related to SRC, FGR, YES
human epidermal growth factor receptor
chimera of HER and Xmrk
human stem cell factor
insulin-like growth factor binding protein 3
JUN N-terminal Kinase 1
mitogen-activated protein kinase
MAK kinase phosphatase 1
normal human epidermal melanocyres
neuroblastoma RAS viral (v-ras) oncogene homolog
phosphatase and tensin homolog
small inhibitory RNA
son of sevenless homolog 1
v-src sarcoma (Schmidt-Ruppin A-2) viral oncogene homolog
signal transducer and activator of transcription 5
tumor-associated antigen L6
ubiquitin-conjugating enzyme E2I
Xiphophorus melanoma receptor kinase
We thank Anita Hufnagel for excellent technical assistance. Futhermore, we gratefully acknowledge the following people: Jochen Wilhelm kindly provided us the R package for microarray data analysis. The TAAL6 antibody (10H6) was donated by Steve Roffler. Melanoma cell lines and useful information on the cell lines were provided by courtesy of Anja Bosserhoff.
This work was supported by Transregio 17 ("Ras-dependent pathways in human cancer"), and Deutscher Melanomverbund (Deutsche Krebshilfe).
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.