Signal Pathways in Up-regulation of Chemokines by Tyrosine Kinase MER/NYK in Prostate Cancer Cells

INTRODUCTION

AXL tyrosine kinase family members are characterized by an N-CAM–like extracellular domain and a common ligand, GAS6 [growth arrest-specific protein 6 (1 , 2)] . The family consists of three members, AXL (UFO, Ark, and Tyro7), Tyro3 (SKY, Rse, Brt, Tif, Dtk, and Etk-2), and MER [Eyk, N-CAM–related kinase (NYK), and Tyro12]. The MER tyrosine kinase was first described by Graham et al. (3) , and our laboratory independently cloned it as NYK (4) . The AXL family kinases are transcriptionally regulated by extracellular stimuli, showing significant variation in expression among different tissues and different stages of diseases (5, 6, 7) . Several functions including cell adhesion, migration, phagocytosis, and apoptosis regulation have been ascribed to this family of tyrosine kinases (8, 9, 10, 11, 12, 13, 14) . MER/NYK, in particular, plays a significant role in spermatogenesis, apoptotic clearance, and retinal development (11 , 15, 16, 17) . There is also a large body of evidence linking this family of tyrosine kinases to carcinogenesis. AXL was initially identified as an NIH3T3 transforming gene (18) , and MER is the human ortholog of the transforming retroviral oncogene, v-eyk (19) . Overexpression of AXL, MER, and Tyro3 has been reported in various cancer types (7 , 20 , 21) . A recent report showed that the overexpression of any two of the three kinases is a predictor of poor prognosis for gastric cancer, suggesting these kinases synergize with each other in gastric carcinogenesis (22) .

How the AXL family tyrosine kinases function to effect cellular transformation is not entirely clear. They appear to connect to different nodes of transforming pathways. In some cell types, they induce cell proliferation, but in most others, they are antiapoptotic and mediate cell adhesion and migration (8 , 10 , 14 , 23) . Activation of phosphoinositide 3′-kinase (PI3K)/AKT and extracellular signal-regulated kinase (ERK) pathways is often associated with activation of this family of kinases. We showed previously (4) that in fibroblasts overexpressing MER/NYK, PI3K and ERK are activated. The activation of signal transducer and activator of transcription 3 by v-eyk, the transforming ortholog of MER/NYK in chicken, and its involvement in transformation have also been reported previously (24) . Although overexpression of AXL in prostate cancer has been reported (25) , there has been no study of the signal pathways involved.

Interleukin (IL)-8 is a proinflammatory chemokine with multiple functions ranging from chemotaxis, angiogenesis, and mitogenesis to induction of cell motility. The role of IL-8 in cancer progression has been supported by abundant literature, including a number of studies on the overexpression of IL-8 in advanced prostate cancers (26, 27, 28, 29, 30) . Two in vivo models further suggested the possible involvement of IL-8 in prostate cancer. First, the expression level of IL-8 in PC3 prostate cancer cells correlates with induction of angiogenesis, tumorigenicity, and metastasis in nude mice (31) . Second, selection of metastatic variants of LNCaP prostate cancer cells leads to the selection of IL-8–overexpressing cells (32) . IL-8 expression can be induced by a variety of ligands, notably cytokines tumor necrosis factor α and IL-1. Signal pathways such as nuclear factor (NF)-κB, ERK, and p38 mitogen-activated protein kinase (MAPK) have been implicated in IL-8 induction in a cell content-dependent manner (33 , 34) . Whereas these pathways are often associated with the activation of tyrosine kinases, there have been few studies concerning the induction of IL-8 by tyrosine kinases.

In the present study, we report overexpression of the tyrosine kinase MER/NYK in a subset of prostate cancer cells and characterization of the biological consequences of the activation of this tyrosine kinase in the prostate cancer cell line DU145. Activation of NYK does not lead to increased proliferation; instead, it leads to a differentiation phenotype. We analyzed the gene expression profile and the downstream pathways engaged by an activated NYK. The results showed that NYK is a strong inducer of endocrine factors including chemokines such as IL-8. IL-8 is induced at both the transcriptional and posttranscriptional levels. The signal pathway leading to the transcriptional induction of IL-8 is attributed principally to the mitogen-activated protein kinase kinase (MEK)/ERK/Jun/Fos pathway. In contrast, IL-1α–induced IL-8 transcription in the same cell relies on the NF-κB pathway. This study thus provides a basic understanding not only of the gene expression and signal pathways of AXL family tyrosine kinases, but also of the various mechanisms involved in IL-8 induction in prostate cancer cells.

MATERIALS AND METHODS

Cell Culture and Reagents.

Normal human prostate epithelial cells (PrEC) and prostate stromal cells (PrSC) were from Cambrex/Clonetics (East Rutherford, NJ) and maintained in the recommended media. The human normal prostate epithelial cell line MLCSV40, which has been immortalized by an origin-defective SV40 genome, was maintained in keratinocyte serum-free medium with 5 ng/mL human recombinant epidermal growth factor and 0.05 mg/mL bovine pituitary extract (Invitrogen, Carlsbad, CA). RWPE-1, PWR-1E, and MDA-PCa-2b were from American Type Culture Collection (Manassas, VA) and maintained in the recommended media. The prostate carcinoma cell lines LNCaP, CWR22, PC3, and DU145 were from American Type Culture Collection and maintained in RPMI 1640 supplemented with 10% fetal bovine serum (Invitrogen). Recombinant human colony-stimulating factor 1 (M-CSF) was from PeproTech (Rocky Hill, NJ). Recombinant human IL-1α was from R&D Systems Inc. (Minneapolis, MN). The anti–phospho-ERK1/2, anti–phospho-p38 MAPK, anti–phospho-c-Jun NH₂-terminal kinase (JNK), anti–phospho-Raf, anti–phospho-MEK1/2, anti–phospho-p90RSK, and anti–phospho-Akt antibodies were from Cell Signaling Technology (Beverly, MA).

Tyrosine Kinase Displays of Prostate Normal/Cancer Cells.

Total RNA was isolated from 10 human prostate cell lines and 5 manually microdissected tumor/normal matched pairs of prostate cancer samples using TRIzol reagent (Invitrogen). The expression profiles of tyrosine kinases were then analyzed with a method developed in our laboratory (35 , 36) . Basically, cDNAs were synthesized, and polymerase chain reactions (PCRs) were performed with a pair of degenerate primers derived from two conserved motifs in the catalytic domains of various tyrosine kinases. To quantify kinase expression, one of the PCR primers was labeled with [γ-³³P]ATP at the 5′ end. Aliquots of reverse transcription-PCR (RT-PCR) products with the same radioactive counts were digested with various restriction enzymes, resolved on DNA sequencing gels, and subjected to phosphorimaging (Molecular Imager FX; Bio-Rad, Hercules, CA). The restriction fragments representing unique kinases were quantified by the Quantity One program (Bio-Rad).

Reverse Transcription-Polymerase Chain Reaction Analysis.

Total RNAs from eight matched tumor/normal pairs of prostate cancer samples were isolated, and first-strand cDNAs were synthesized. Subsequent PCR analysis was performed with MER/NYK gene-specific primers as follows: forward, 5′-TAGCAAGCACGACTGAAGGAGC-3′; and reverse, 5′-AGCCAAAGATGATGAGCACAGG-3′. β-Actin (forward, 5′-GGCATCCACGAAACTACCTTCAAC-3′; reverse, 5′-ACTGCTGTCACCTTCACCGTTC-3′) was used as an internal standard. For analysis of MER/NYK and β-actin expression, cycles (MER/NYK, 32 cycles; β-actin, 24 cycles) of amplification were performed; each cycle consisted of 50 seconds at 93°C, 1 minute at 55°C, and 45 seconds at 72°C followed by a final step of 10 minutes at 72°C. PCR products were analyzed by agarose gel electrophoresis.

Construction of the FMS-NYK Chimera and Generation of FMS-NYK/DU145 Cells.

The FMS-NYK chimera was constructed by joining DNA fragments encoding amino acids 1 to 512 (extracellular domain) of human M-CSF receptor (FMS) and amino acids 502 to 999 (transmembrane and intracellular domains) of human NYK. The chimeric construct was cloned into the pcDNA3.1 vector (Invitrogen) and transfected into DU145 for stable expression. After selection with G418 (Invitrogen), cells stably expressing FMS-NYK chimeric proteins were confirmed by immunoblotting with antibodies specific for the extracellular domain of FMS (Upstate Biotechnology, Lake Placid, NY). The growth rates of parental DU145, pcDNA/DU145 control, and two independently selected FMS-NYK/DU145 clones (clones 2 and 6) were measured in a 0.5% fetal bovine serum-containing medium with or without M-CSF (50 ng/mL) using WST-1 assay (Roche, Indianapolis, IN).

Immunoprecipitation and Immunoblotting Assays.

Before stimulation with human M-CSF, DU145 and FMS-NYK/DU145 cells were serum starved for 24 to 48 hours. After treatment with 100 ng/mL M-CSF for various times, cells were lysed in lysis buffer (58 mmol/L Na₂HPO₄, 17 mmol/L NaH₂PO4, 68 mmol/L NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 1 mmol/L sodium orthovanadate, 1 mmol/L sodium fluoride, 1 mmol/L phenylmethylsulfonyl fluoride, and 10 μg/mL each of leupeptin, aprotinin, chymostatin, and pepstatin). Immunoprecipitation was performed by incubating 500 μg of total lysates of each sample with anti-FMS antibody and protein A-agarose beads (Upstate Biotechnology). Total cell lysates or immune complexes from immunoprecipitation were resolved by SDS-polyacrylamide gels, transferred to polyvinylidene difluoride membranes (Millipore, Billerica, MA), and detected with anti–phospho-tyrosine antibody (clone 4G10; Upstate Biotechnology) or other appropriate primary antibodies. The blots were subsequently incubated with secondary antibodies conjugated to horseradish peroxidase, and images were developed using the enhanced chemiluminescence system (Amersham Pharmacia Biotech, Piscataway, NJ). To detect the activation of endogenous AXL and SKY kinases, DU145 cells were serum starved for 24 to 48 hours before treatment with GAS6 (100 ng/mL) for 10 and 30 minutes. Cell lysates were immunoprecipitated with anti-AXL or anti-SKY antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) and analyzed by Western blotting using anti–phospho-tyrosine antibody.

Microarray Analysis and Real-Time Polymerase Chain Reaction.

FMS-NYK/DU145-6 cells were grown to 60% to 70% confluence, serum starved for 48 hours, and stimulated with 100 ng/mL M-CSF for 0, 1, 4, and 16 hours, and total RNAs were extracted. The cRNA for each sample was in vitro transcribed and labeled according to the protocol recommended by Affymetrix (Santa Clara, CA). Fragmented cRNAs were hybridized to HG-U95Av2 GeneChips containing ∼12,000 oligonucleotide probe sets. Fluorescence intensities were obtained with a laser scanner and analyzed using MicroArray Suite 5.0 software (Affymetrix). To find transcripts that were up-regulated by NYK activation, we selected for mRNAs that were classified as increasing from a detectable signal, with a difference of >2-fold on M-CSF treatment relative to the unstimulated control. Transcripts were included in the final lists only if they fulfilled these selection criteria for at least two of the three tested time points. Quantitative real-time PCR analysis was performed with the iCycler System (Bio-Rad) using TaqMan gene expression assays (Applied Biosystems, Foster City, CA). In all experiments, samples were run in triplicate, and mean threshold cycle (C_T) values were calculated. Quantification of a given gene, expressed as fold induction over control (untreated sample), was calculated after normalization to 18S rRNA, using the ΔΔC_T formula as described in ABI Prism Sequence Detection System User Bulletin 2: ΔC_T (sample) = C_T (target gene) − C_T (18S rRNA), ΔΔC_T = ΔC_T (time course) −ΔC_T (0 hours), and relative expression = 2^−ΔΔCT.

Construction of Interleukin-8 Promoter-Luciferase Reporter Plasmids and Analysis of Interleukin-8 Promoter Activities.

The roles of NYK and IL-1α signal pathways in regulating IL-8 transcription were examined using transient transfection with IL-8 promoter-luciferase reporter constructs. A part of the human IL-8 promoter, ranging from nucleotides −416 to +44, was PCR amplified and ligated into the firefly luciferase reporter vector pGL3-Basic (Promega, Madison, WI), yielding the wild-type reporter construct pIL8(−416)-Luc. Three additional constructs [pNF-κB-mut, pC/EBPβ-mut, and pAP-1-mut, with mutations in the NF-κB, CCAAT/enhancer binding protein β(C/EBPβ), and activator protein (AP)-1 elements, respectively, in the context of pIL8(−416)-Luc construct] were created by site-directed mutagenesis (QuikChange; Stratagene, La Jolla, CA). The sequences of the mutants constructed are shown in Fig. 6 ⇓ . FMS-NYK/DU145 cells were cotransfected with pIL8(−416)-Luc or the cognate-mutated constructs along with an internal control plasmid pRL-TK, using FuGENE 6 (Roche). Cells were incubated in the transfection medium for 5 hours before treatment with 100 ng/mL M-CSF or 10 ng/mL IL-1α for 16 hours. Cells were harvested in passive lysis buffer (Dual-Luciferase Reporter Assay System; Promega), and luciferase activities were measured on a MicroLumat Plus LB96V luminometer (Berthold Technologies, Bad Wildbad, Germany).

Analysis of Gene Expression.

Total cellular RNA was extracted, electrophoresed in 0.9% agarose gel containing 6.5% formaldehyde, and transferred onto nylon membranes (Pall Corp., East Hills, NY). The cDNA fragments of BMP2, CTGF, IL-8, PLAB, PTHLH, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were obtained by PCR; radiolabeled with [α-³²P]dATP by random priming (StripEZ DNA); and hybridized to the membranes in ULTRAhyb solution (Ambion, Austin, TX). The blots were washed to high stringency with 0.5× SSC/0.25% SDS at 52°C, analyzed by phosphorimager, and quantified by the Quantity One program. Reagents for chemiluminescent IL-8 enzyme-linked immunosorbent assay (ELISA) were obtained from R&D Systems Inc. FMS-NYK/DU145 and parental DU145 cells were starved for 48 hours before M-CSF stimulation at a concentration of 100 ng/mL for 1, 4, 12, and 24 hours. To test the effects of signaling inhibitors on IL-8 production, cells were pretreated with MEK1/2 inhibitor U0126 (10 μg/mL) or PI3K inhibitor LY294002 (10 μg/mL) for 30 minutes before the addition of M-CSF. Supernatants were collected after treatments, and each sample was tested in triplicate by ELISA. After development, the light intensity was measured by a luminometer, converted to picograms of IL-8 per milliliter based on a standard curve of the same assay, and normalized with cell numbers for each sample. The data are presented as the mean of the triplicates.

Analysis of Transcription Factor Activities.

The transcription factors activated in response to NYK signals were investigated using an ELISA-based detection kit (Mercury TransFactor Kit; BD Biosciences Clontech, Palo Alto, CA). FMS-NYK/DU145 or parental DU145 cells were serum starved for 48 hours and stimulated with M-CSF (100 ng/mL) for 1, 4, and 16 hours. Nuclear extracts were tested in a 96-well format plate with oligonucleotides containing the consensus binding sequences for each transcription factor coated on the well. Competitor oligonucleotides were added in control wells to demonstrate the specificity of the DNA–protein interactions. Bound transcription factors were detected by specific primary antibodies and horseradish peroxidase-conjugated secondary antibodies. The absorbance of enzyme product was measured with a microplate reader.

Construction and Expression of GAS6.

A cDNA fragment encoding GAS6 was subcloned into pcDNA3.1 vector and V5-(His)₆ tagged at the COOH terminus. 293 cells were transfected with the GAS6-V5-(His)₆ plasmid, and supernatants were collected at different time points. The expression of GAS6 in supernatants was detected by anti–V5-Tag antibody (Invitrogen). The supernatant collected 48 hours after transfection was applied to an Immobilized Metal Affinity Chromatography column (BD TALON Metal Affinity Resins; BD Biosciences Clontech). The purified GAS6 was eluted with imidazole, dialyzed, quantified, and used to stimulate serum-starved DU145 cells at 100 ng/mL. IL-8 secretion after GAS6 treatment was detected by ELISA as described above.

RESULTS

Expression of NYK in Prostate Cancer Cells.

Our laboratory has been studying the role of tyrosine kinases in prostate cancer progression and developed an effective tyrosine kinase display method to determine the content and quantity of tyrosine kinases expressed in a given cell (35 , 36) . This approach displays the radiolabeled RT-PCR products of different tyrosine kinases according to their restriction fragment sizes, and the quantities are reflected by the intensities of the bands. Using this approach, we have profiled the tyrosine kinases in prostate cancer cells and tumor specimens and found that the transcriptional levels of most of the tyrosine kinases do not vary significantly. 1 The tyrosine kinase NYK is one of the exceptions: it is generally not expressed in normal prostate cell lines (Fig. 1A ⇓ , PWR-1E, RWPE-1, MLCSV40, PrEC, and PrSC), but it is up-regulated in all of the prostate cancer cell lines tested (LNCaP, CWR22, MDA-PCa-2b, PC3, and DU145). The expression level varies; CWR22 has the highest level of NYK expression, and DU145 and PC3 have the lowest level of NYK expression. The identical expression of CSK and SRC tyrosine kinases in these cell lines serves as internal control for this assay. We then extended this analysis to five matched normal tissue/tumor specimens (Gleason grade > 3 + 3; 80% tumor cells) from patients with radical prostatectomies. As shown in Fig. 1B ⇓ , four of the five samples demonstrated an elevated expression level of NYK. To further confirm the elevated expression pattern of NYK in prostate tumors, we analyzed another eight matched normal tissue/tumor specimens (Gleason grade > 3 + 3) by RT-PCR analysis using a pair of gene-specific primers for NYK. Six of the eight samples showed an increased expression level of NYK (Fig. 1C) ⇓ . Thus, NYK seems to be consistently expressed at a low level in normal prostate epithelial cells and has an elevated expression in a subset of prostate tumor samples. The lack of a monospecific NYK antibody precludes an immunohistochemical analysis of clinical samples.

Inducible Activation of NYK in DU145 Cells.

To study the signal pathways and the biological effects of NYK, we chose DU145, a cell line that expresses a moderate level of NYK, as an expression system. As such, DU145 provides a functionally relevant environment without a high basal level of NYK kinase activity that may interfere with the study. Kinases of the AXL family share the same ligand, GAS6. Because DU145 expresses all three kinases of this gene family, application of GAS6 would activate the three kinases and complicate the analysis of NYK-specific signals. We therefore adopted an approach whereby the extracellular domain of FMS is fused to the transmembrane and intracellular domains of NYK to generate a FMS-NYK chimeric receptor (Fig. 2A) ⇓ . The chimera approach was successfully used by us (4) and others (14) to study NYK signaling in other systems. FMS is not expressed in DU145, providing a clean background for this analysis.

Stable cell lines were screened for expression of the FMS-NYK chimeric receptor (Fig. 2B) ⇓ . Clones 2 and 6 both express FMS-NYK and were used in the analyses. Unless otherwise indicated, representative data of clone 6 (FMS-NYK/DU145-6) are shown. M-CSF, the ligand for FMS, stimulates the tyrosine kinase activity of FMS-NYK (as evaluated by tyrosine phosphorylation) in a dose-dependent manner (Fig. 2C) ⇓ . The activation is transient and peaks around 10 minutes (Fig. 2D) ⇓ , consistent with the mode of activation of NYK (4) .

Because this is the first study of NYK signaling in a prostate cancer cell line, we were curious about the overall biological effects of NYK on DU145. The growth rates of parental DU145, pcDNA/DU145 (vector control), FMS-NYK/DU145-2, and FMS-NYK/DU145-6 were measured in a medium containing 50 ng/mL M-CSF. No significant difference in the growth kinetics after M-CSF treatment was observed for the parental and vector-transfected DU145 cells. Treatment of FMS-NYK/DU145 clones with M-CSF also does not increase the growth rate; indeed, if anything, it slightly inhibits the growth (data not shown). Interestingly, the M-CSF–treated FMS-NYK/DU145 cells exhibit significant morphologic changes; they round up after ligand stimulation for 16 to 24 hours and display a scattered growth pattern, more akin to a differentiation phenotype.

The Transcriptional Profiles of NYK Activation.

To gain further understanding of the molecular changes after NYK activation, we applied microarray analysis using Affymetrix Hu95Av2 gene chips. The FMS-NYK/DU145-6 clone was treated with M-CSF for 1, 4, and 16 hours, and RNA samples were analyzed. There are significant changes in gene expression on NYK activation at both early (1 and 4 hours) and late time points (16 hours). A total of >140 genes are up-regulated in their expression by >2-fold, in at least two time points and with a confidence level of P < 0.05. For this report, we will focus on the two classes of genes, secreted proteins and transcription factors (the entire microarray data will be published elsewhere 2 ). Most remarkably, a great number of the genes up-regulated by NYK are secreted chemokines and peptide hormones (Table 1) ⇓ . These include CXCL1/GROα, CXCL2/GROβ, CXCL6/GCP2, CXCL8/IL-8, CCL20/MIP3A, PTHLH/PTHrP, CTGF/IGFBP8, HB-EGF/DTR, vascular endothelial growth factor (VEGF), PDGFβ, PLAB/PDF, BMP2/BMP2A, and LIF. Four of the seven known ELR-containing CXCL chemokines (37) , CXCL1, CXCL2, CXCL6, and CXCL8, are significantly up-regulated by NYK (4–247 fold). Two of the transforming growth factor β family ligands, PDF and BMP2, which were implicated in bone morphogenesis (38 , 39) , are also up-regulated. To confirm the results obtained from the microarray analysis, we performed Northern analysis with five genes (BMP2, CTGF, IL-8, PLAB, and PTHLH) and real-time PCR analysis with four genes (CXCL1, CXCL2, VEGF, and LIF). As shown in Fig. 3 ⇓ and Tables 1 ⇓ and 2 ⇓ , expression patterns for all of the genes tested correlate strongly with the microarray data. These results suggest that NYK activation bestows on DU145 cells a facility to produce endocrine factors, which may play a significant role in prostate cancer progression.

A second class of genes relevant to our discussion later is transcription factors (Table 1) ⇓ . NYK activation leads to the up-regulation of JunB, c-Jun, c-Fos, MafF, EGR3, EGR4, CRE-BPa, Hox-A1, and COPEB/KLF6/Zf9. Most of these are known early response genes. Interestingly, KLF6, a Kruppel-like zinc finger protein, was found to be involved in epithelial differentiation (40) and may be related to the differentiation phenotype induced by NYK activation.

The Transcriptional Activation of Interleukin-8.

Among the genes up-regulated by NYK activation in the microarray analysis, the most dramatic is IL-8, which was up-regulated 247-fold 4 hours after M-CSF treatment (Table 1) ⇓ . Northern analysis similarly showed a nearly 50-fold increase in IL-8 mRNA (Fig. 3B) ⇓ . The induction of IL-8 expression is a specific response to NYK activation because parental DU145, which lacks the inducible NYK, shows a complete absence of IL-8 induction. The secretion of IL-8 protein on NYK activation was measured by ELISA, using supernatants from M-CSF–treated FMS-NYK/DU145 cells (Fig. 3C) ⇓ . A steady increase of IL-8 release beginning 1 hour after NYK activation was observed in both NYK-expressing stable clones. The increase at 24 hours after stimulation is 43-fold for clone 2 and 63-fold for clone 6, which is in good agreement with the Northern blot data.

The Signal Pathways Involved in NYK-Induced Interleukin-8 Production.

To explore the mechanism underlying the transcriptional activation of IL-8, we studied the signal pathways activated by NYK in DU145 cells. IL-8 can be activated by a variety of proinflammatory cytokines such as tumor necrosis factor α and IL-1 via transcription factors including NF-κB and AP-1 (33) . We therefore focused on the PI3K and MAPK pathways that activate these transcription factors. As shown in Fig. 4A and B ⇓ , NYK activation leads to phosphorylation/activation of ERK1/2 and AKT, but not p38 and JNK, as measured by Western blotting. Consistent with the ERK1/2 activation is the activation of Raf, MEK1/2, and p90RSK, kinases upstream and downstream in the pathway (Fig. 4C) ⇓ . These data are in good agreement with our previous finding in fibroblasts, in which we showed that NYK is a strong activator of both the ERK and PI3K pathways (4) , and the report by Katagiri et al. (41) , in which Tyro3 was found to activate ERK1/2, but not p38 or JNK, in osteoclasts.

To further narrow down the transcription factors involved in IL-8 induction by NYK activation, we used a quantitative ELISA-based DNA binding assay, in which oligonucleotides corresponding to the response elements for individual transcription factors were incubated with nuclear extracts prepared from DU145 or FMS-NYK/DU145 cells with or without NYK activation. The binding of each family member is distinguished from the binding of others by a transcription factor-specific antibody. As shown in Fig. 5 ⇓ , NYK activation leads to increased binding activity of c-Fos (1 hour and 4 hours) and c-Jun (4 hours), but not CREB-1, ATF2, cRel, NFκBp65, and NFκBp50. The validity of this assay was confirmed by comparison with parental DU145 cells treated with M-CSF, in which no such activations were observed. These data are in good agreement with the up-regulation of c-Jun and c-Fos observed in microarray analysis (Table 1) ⇓ .

The Transcriptional Activation of Interleukin-8 Promoter.

To more precisely define the transcription factors involved in the transcriptional activation of IL-8, we isolated a segment of the human IL-8 promoter (nucleotides −416 to +44 relative to the start of transcription; Fig. 6A ⇓ ) and fused it to a luciferase reporter. We first showed that this promoter construct responds to NYK activation with a 3-fold increase in the luciferase activity (Fig. 6C) ⇓ . Within this promoter, several motifs contributing to IL-8 activation have been identified, including response elements for AP-1, C/EBPβ, and NF-κB. We have mutated the AP-1, C/EBPβ, and NF-κB sites, respectively (Fig. 6B) ⇓ , and the mutant promoter constructs were used to assess the importance of these factors contribution to the NYK-mediated activation of IL-8 promoter. As shown in Fig. 6C ⇓ , the AP-1 site mutant had a substantially reduced inducibility, whereas the inducibility of the C/EBPβ binding site mutant remained unaffected. The NF-κB binding site mutant had a modest reduction in promoter activity. These results confirm the above-mentioned data that AP-1 transcription factors such as c-Jun and c-Fos are the major factors activated by NYK signaling pathways, which participate in the transcriptional activation of IL-8 promoter. As a control, we studied the mode of activation of these mutants by IL-1α, a well-recognized activator of IL-8 promoter. In this case, NF-κB and C/EBPβ seem to be much more important than AP-1 in the transcriptional activation of IL-8 (Fig. 6D) ⇓ . These data are in total agreement with previous reports (34 , 42) , and, more importantly, point to the different activation pathways used by IL-1α and NYK.

The Posttranscriptional Regulation of Interleukin-8.

The increased level of IL-8 transcript could be attributed to increased transcriptional initiation, increased transcript stability, or a combination of both. Indeed, in response to NYK activation, the induction of IL-8 transcript as measured by microarray and Northern analyses is much higher than that obtained by the promoter assay, suggesting possible posttranscriptional regulation. We therefore studied whether the NYK signal affects IL-8 transcript stability. Transcriptional inhibitor 5,6-dichloro-1-β-d-ribofuranosylbenzimidazole was used to inhibit the nascent RNA synthesis, and the decay of IL-8 transcript through the time course was measured. NYK activation significantly extends the half-life of IL-8 from 1 hour in the absence of ligand M-CSF to >6 hours in the presence of M-CSF (Fig. 7A) ⇓ .

The Involvement of the MEK/ERK Pathway in Interleukin-8 Activation.

To further delineate whether the MEK/ERK or PI3K/AKT pathway is involved in the regulation of IL-8 expression, we treated FMS-NYK/DU145 cells in combination with ligand and inhibitors of MEK and PI3K and analyzed the IL-8 production. As shown in Fig. 7B and C ⇓ , treatment with the MEK1/2 inhibitor U0126 results in an attenuated ERK1/2 activity and reduced IL-8 production, whereas treatment with the PI3K inhibitor LY294002 does not. The results are consistent with the above finding that AP-1 is the key activator of IL-8 transcription and the published report that the ERK pathway is involved in stabilization of IL-8 transcript (43) . The increased IL-8 production in the presence of LY294002 was curious and will be discussed later.

GAS6 Induces Interleukin-8 Production in DU145.

Having demonstrated the effect of activated chimeric NYK on IL-8 production, we asked whether GAS6, the potential natural ligand for NYK family kinases, is also able to stimulate IL-8 production through the endogenous receptors. We generated affinity-purified GAS6 by overexpressing a His-tagged human GAS6 clone in 293 cells. Parental DU145 cells, which express AXL, NYK, and SKY, were treated with either GAS6 or buffer, and the activation of receptors was examined. As shown in Fig. 8A ⇓ , the tyrosine kinase activities of endogenous AXL and SKY kinases were stimulated by GAS6. The low level of NYK expression in the parental DU145 cells (Fig. 1A) ⇓ , coupled with the absence of a suitable antibody for immunoprecipitation of NYK, precludes similar analysis of endogenous NYK. The amount of IL-8 released from DU145 cells after GAS6 stimulation was analyzed by ELISA. As shown in Fig. 8B ⇓ , IL-8 production substantially increased on GAS6 treatment (5–6 fold). These results demonstrate that the activation of IL-8 by NYK kinase is not an artifact of the chimeric receptor but an intrinsic property of this family of tyrosine kinases. To study the functions of each individual AXL family kinase, chimeric constructs of FMS/AXL and FMS/SKY were also generated, and the activation of these individual chimeric receptors by M-CSF in DU145 cells also led to the elevated production of IL-8 as measured by Northern analysis (data not shown).

DISCUSSION

Here, we report the characterization of the signal pathways and transcriptional profile of genes induced by the activation of MER/NYK, a member of the AXL family of tyrosine kinases. Based on published reports and our tyrosine kinase profiling of various cancers, it appears that this family of tyrosine kinases is most prone to transcriptional modulation. For instance, AXL is overexpressed in gastric cancer cells (22) , and MER/NYK is overexpressed in mantle cell lymphomas (44) and induced by androgen in the LNCaP prostate cancer cell line (45) . We show here that NYK is generally, but not universally, overexpressed in prostate cancer cells.

The mechanisms whereby NYK or other family members contribute to carcinogenesis are not clear and are likely to be cell type specific. In general, however, the actions of this family of kinases are related more to cell adhesion, migration, and cell survival processes than cell proliferation (8, 9, 10 , 13 , 14) . The natural role of MER/NYK during development, as revealed by targeted disruption of the loci in mice and germ-line mutations of patients with retinal dystrophy, also seems to be more related to differentiation functions such as phagocytosis and chemotaxis (11 , 12 , 15 , 16) . Here, we present data for potential functions of MER/NYK in prostate carcinogenesis.

To explore the role of NYK in prostate carcinogenesis, we developed a chimeric receptor construct combining the extracellular domain of FMS and the transmembrane and intracellular domains of NYK. The chimeric approach is necessary for the AXL family tyrosine kinases. Because they share a common ligand, GAS6, it would have been difficult to discern the specific signals emanating from GAS6-activated NYK in cells such as DU145, which expresses all three family members. We chose the FMS extracellular domain as the fusion partner because this receptor is not expressed in DU145 and M-CSF does not induce any effects in DU145. In contrast, M-CSF activates the FMS-NYK chimeric kinase in a dose- and time-dependent manner, providing a sensitive and kinase-specific system to study NYK signals. The validity of this system was further confirmed by the experiment in which GAS6 was used to stimulate parental DU145, yielding results in agreement with that of the chimeric NYK.

A striking feature of the gene expression profile altered by NYK activation is the increased transcription of cytokines and chemokines, and the induction of IL-8 was the most dramatic. This increase in transcription is accompanied by an increased release of IL-8 protein. We found that NYK activation results in phosphorylation of AKT and ERK, but not JNK and p38. This result generally agrees with the published reports, although in some cell types, p38 activation by NYK/MER was observed (4 , 14 , 46) . When the nuclear activities of some potential transcription factors were analyzed, c-Jun and c-Fos were found to increase on NYK activation. This is consistent with MEK/ERK activation of Elk-1, which transcriptionally activates c-Fos and increases the c-Fos/c-Jun dimer binding to the AP-1 response element (47 , 48) . Nuclear NF-κB (p65 and p50) activity did not show much increase on NYK activation, and a mutant IL-8 promoter construct that lacks NF-κB binding ability still responded to NYK activation. In contrast, mutation of the AP-1 site significantly diminished NYK-induced promoter activity. This suggests that activation of c-Jun/c-Fos or AP-1, but not NF-κB, is largely responsible for the increased activity of the IL-8 promoter by NYK. Interestingly, this mode of activation is different from the induction of IL-8 by IL-1α in the same cell type, where all three response elements (NF-κB, C/EBPβ, and AP-1) contribute to activation, with NF-κB being the most critical.

The argument that AP-1, but not NF-κB, is responsible for NYK activation of IL-8 promoter is further supported by inhibitor studies. Treatment of ligand-stimulated FMS-NYK/DU145 with U0126, a MEK1/2 inhibitor, inhibited the activation of IL-8, whereas treatment with LY294002, the PI3K inhibitor, not only failed to inhibit the IL-8 production but actually increased it. The increase of IL-8 production by the PI3K inhibitor is not unprecedented. The activation of IL-8 by the tyrosine kinase RET was found to be augmented by PI3K inhibition (49) . This is likely due to the opposing effects of AKT and ERK (50) . It was reported that activated AKT is able to phosphorylate either c-Raf or B-Raf and negatively regulates Raf kinase activity, which in turn has a negative effect on signaling through Ras/MEK/ERK to AP-1 (51 , 52) . Diminishing AKT activity by a PI3K inhibitor thus may enhance AP-1 activity, leading to a higher IL-8 production. Our data suggest that AP-1 activation is a major contributor to the NYK tyrosine kinase-induced IL-8 expression.

Our studies were prompted by the observation that NYK is overexpressed in some prostate cancer tissues when compared with their normal counterparts. What is the potential role of NYK in prostate cancer progression? As described here, NYK activation does not seem to induce cell proliferation but instead converts DU145 into a rich source of endocrine factors known to be involved in angiogenesis and bone metastasis. NYK activation induces the expression of four angiogenic ELR-CXC chemokines, but none of the angiostatic non–ELR-CXC chemokines (37 , 53) . In addition, VEGF is also induced. It has been shown that DU145 is a highly tumorigenic and angiogenic cell line in vivo and the tumorigenic potential of DU145 correlates with IL-8 expression and depends on CXCL1 production (54) . The significant increase in IL-8, CXCL1, CXCL2, and CXCL6 after NYK activation should further enhance its tumorigenicity in a paracrine fashion. It is interesting to point out that DU145 expresses neither CXCR1 nor CXCR2 receptors and should not be affected by these chemokines. In addition to angiogenic chemokines, other factors induced by NYK are also noteworthy. PDF has a highly restricted tissue distribution in primarily placenta and prostate and is androgen regulated (39 , 55) . It is also highly expressed in advanced prostate tumor specimens and in androgen-independent variants of LNCaP (as compared with the androgen-dependent parental line; ref. 55 ). CTGF, another growth peptide induced by NYK, is implicated in angiogenesis, osteoblast and chondrocyte proliferation/differentiation, and is a gene whose overexpression correlates with selections of bone metastasis variants of breast cancer cells (56) . Likewise, parathyroid hormone-related protein (PTHrP or PTHLH), which is also induced by NYK, is overexpressed in prostate cancer cells and known to play a role in osteoclast activation (57) . Recent work further showed that parathyroid hormone-related protein enhances IL-8 production by an intracrine mechanism (58) . These data suggest that NYK activation may be linked to bone metastasis of prostate cancer.

In summary, we have shown here that the receptor tyrosine kinase NYK is able to activate IL-8 principally through activation of the ERK pathway and the AP-1 transcription factors. The modulation is through both transcriptional and posttranscriptional mechanisms. Our studies have direct implications on the role of the tyrosine kinase NYK in prostate cancer progression and also, more generally, on the role of tyrosine kinases in chemokine production.