Fra-1 governs cell migration via modulation of CD44 expression in human mesotheliomas

We first focused on whether heterogeneous pathways of Fra-1 regulation occurred in human MM cells using inhibition of upstream signaling cascades. In Figure 1A, we observed that the PI3K inhibitor, LY294002 (LY 20 μM), caused significant reduction of Fra-1 mRNA levels in 3 of the 4 MM lines initially examined, whereas addition of AG1478 (AG 10 μM), an inhibitor of EGFR phosphorylation, had no effects on Fra-1 expression. Two of the 4 MM lines showed inhibition of Fra-1 expression after pretreatment with PP2 (10 μM) or PD98059 (PD, 30 μM), respectively, suggesting additional pathways of Fra-1 modulation in some MMs. In MM3 cells in which Fra-1 mRNA levels were diminished significantly only after inhibition of the PI3K pathway, transactivation of Fra-1 dependent gene expression (Figure 1B) and protein levels (Figure 1C) were also inhibited selectively by LY294002 (20 μM). In accordance with data presented in Figure 1A, pre-addition of the EGFR kinase inhibitor, AG1478 (10 μM) did not affect Fra-1 protein levels (Figure 1C). In Figure 1D, we also show using an EMSA super-shift assay that inhibition of the PI3K pathway by LY294002 (10 and 20 μM) causes reduced expression of Fra-1 in the AP-1 complexes of these cells in a dose-related fashion.

Western blot analyses were performed on a panel of 7 MM cell lines at near confluency to determine possible correlations between expression of these proteins in control cells and with addition of serum (+ lanes) for 4 h (Figure 2). Fra-1 was inducible in all MMs after addition of serum. In the 3 SV40 + cell lines, both Fra-1 and CD44 protein were markedly reduced in comparison to SV40- cell lines in the presence or absence of serum.

Here we examined levels of endogenous CD44 in SV40+ and SV40- MM lines after addition of LY294002 (Figure 3). These studies revealed that the PI3K inhibitor LY294002 diminished CD44 protein in a dose-related fashion in the SV40+ line, but not in the SV40- MM lines. The Src inhibitor, PP2, did not affect CD44 expression in the SV40+ MM line, but decreased CD44 expression in the SV40- MM line (Figure 3A). CD44 protein was decreased after addition of the PD98059 MEK1 inhibitor and in a SV40- cell line transformed with a dn-Fra-1 construct (Figure 3B). However, CD44 depletion was not increased synergistically after PD98059 or LY294002 was added to the dnFra-1 stable cell line (Figure 3B). These results show the complexity of Fra-1 regulation in MM and suggest that Fra-1, Src, ERK1/2-and PI3K are also modulators of CD44 expression in different MM lines. CD44 promoter analysis does not suggest the direct participation of Fra-1 in CD44 expression (data not shown).

To determine a role of Fra-1 in CD44 expression, we next developed a shFra-1 construct and verified in both an SV40+ and SV40- MM line that CD44 expression was reduced (Figure 4A). We further showed that shFra-1 in the SV40- line reduced CD44 expression (red) using immunochemistry and CSLM, whereas the SV40+ line had low basal levels of CD44 (Figure 4B). The efficiency of the shFra-1 construct and a shCD44 construct for inhibition of CD44 expression in functional assays below is depicted in Figure 4C using semi-quantitative PCR in a SV40- MM line.

Two Fra-1 and CD44 expressing SV40- MM lines (MM1, MM2) and one SV40+ MM cell line (MM3) were used to demonstrate that Fra-1 and CD44 were causally linked to cell migration. In Figure 5A, migration was compared in stable MM1 lines transfected with shFra-1, shCD44, or the empty vector (EV) control. These studies showed that after addition of serum for 24 h, numbers of migrating cells in the shFra-1 or shCD44 cell lines were significantly decreased (p < .05) in comparison to the MM1 (EV) cells. These results were confirmed using shFra-1 transformed MM2 and MM2 (EV) stable cell lines (Figure 5B). In Figure 5C, no migration was detected in stable MM3 lines transfected with shFra- 1 or the empty vector (EV).

CD44 is known to bind hyaluronic acid (HA), and is also thought to contribute to HA internalization. The binding of HA to CD44 stimulates cytoskeleton-mediated tumor cell migration [29]. In Figure 6 we show that the uptake of BODIPY FL hyaluronic acid (green) is decreased in SV40- MM cells transfected with shFra-1 (Fra-1 expression is shown in red) compared to MM cells transfected with empty vector (EV). Moreover, BODIPY FL hyaluronic acid did not get internalized in the SV40+ MM line.

We have shown previously that endogenous levels of Fra-1 are increased in human and rat MM cell lines as compared to normal rat mesothelial cells [15]. To confirm this in human MMs, tissue arrays containing normal lung (Figure 7A), and MMs (Figure 7B) were examined by multi-fluorescence approaches using antibodies for Fra-1 (red), phosphorylated (p)-ERK1/2 (blue) and SYTOXgreen (green) to detect nuclei. The images in Figure 7A and 7B are split into an upper portion showing Fra-1 expression, and a lower portion showing triple fluorescence studies for detection of p-ERK1/2 (blue), Fra-1 (red) and SYTOXgreen (green). In comparison to normal lungs (negative control), 33 out of 34 MMs stained positively for Fra-1 with different intensities. On a scale of 0–3 (0 = no expression, 1 = low expression, 2 = medium expression and 3 = high expression) using a blind coding system, ~3% MMs showed no Fra-1 immunoreactivity (score of 0), 29% were scored 1, 27% were scored 2 and 41% were scored 3. Fra-1 immunolocalization in MMs was either nuclear (41%), mainly nuclear with some cytoplasmic staining (32%), entirely cytoplasmic (6%), mainly cytoplasmic with some nuclear staining (18%) or negative (~3%). As shown in Figure 7 (lower panel), increased p-ERK1/2 immunoreactivity was observed in approximately 30% of tumors, and its localization was mainly cytoplasmic.

Using a similar tissue microarray set as used for Fra-1 expression above, we performed dual fluorescence studies on MMs using antibodies for SV40 T-antigen (blue/nuclear) and CD44 (red) (Figure 8). Of the 34 MMs, 17 (50%) were SV40+ and 8 (24%) were CD44+. Moreover, all SV40+ tumors were CD44-. Of the 17 SV40- tumors (50%), 9 (26%) were CD44-.

A side by side comparison of the Fra-1 array with the SV40/CD44 array showed that Fra-1 localization was primarily nuclear for all tumors with some cytoplasmic localization in CD44- tumors (Table 1).

Human pleural mesothelioma cell lines were isolated from patients at autopsy (from Dr. Michele Carbone, University of Hawaii, Honolulu, HI (MM3, MM4, MM7), and Dr. Luciano Mutti (Maugeri Foundation, Pavia, Italy) (MM1, MM6), or from surgical debulking of MMs from Dr. Harvey Pass (New York University, NY, NY) (MM2). The cell line MM5 (#CRL-2081) was obtained from the ATCC (Manassas, Virginia). All MM cell lines were tested for the mRNA expression of large and small T/t-antigen by PCR before each experiment. The MM1, MM2, MM5, and MM6 lines were negative for SV40 large T-antigen (SV40-), whereas MM3, MM4 and MM7 were SV40+. Cells were maintained in frozen stocks and propagated in DMEM/F12 medium (GIBCO BRL, NY) containing 10% fetal bovine serum (FBS), hydrocortisone (100 ng/ml), insulin (2.5 μg/ml), transferrin (2.5 μg/ml), and selenium (2.5 ng/ml) (Sigma, St Louis, MO).

Stock solutions of all inhibitors were diluted in dimethyl sulfoxide (DMSO) and used at effective nontoxic concentrations as reported previously: The MEK1/2 inhibitor, PD98059, at 30 μM [15]; the EGFR inhibitor, AG1478, at 10 μM [13]; the PI3K inhibitor, LY294002, at 10 and 20 μM [59]; and the Src inhibitor, PP2, at 10 μM [44]. All chemicals were obtained from Calbiochem (La Jolla, CA). Control groups of cells also received DMSO (0.1%) in medium.

Nearly confluent MM cells were washed 3 × with cold phosphate-buffered saline (PBS), scraped from culture plates, and collected by centrifugation at 14,000 rpm for 1 min. The pellet was resuspended in lysis buffer [20 mM Tris (pH 7.4), 1% Triton X-100, 10% glycerol, 137 mM NaCl, 2 mM EDTA, 25 mM β-glycerophosphate, 1 mM Na₃VO₄, 2 mM pyrophosphate, 1 mM PMSF, 10 μg/ml leupeptin, 1 mM DTT, 10 mM NaF, 1% aprotinin], incubated at 4°C for 15 min, and centrifuged at 14,000 rpm for 20 min. Protein concentrations were determined using a Bio-Rad assay (Bio-Rad, Hercules, CA). Twenty μg of protein in sample buffer [62.5 mM Tris-HCl (pH 6.8), 2% sodium dodecyl sulfate (SDS), 10% glycerol, 50 mM dithiothreitol, 0.1% w/v bromophenol blue] was resolved by electrophoresis in 10% SDS-polyacrylamide gels, and transferred to nitrocellulose using a semi-dry transfer apparatus (Ellard Instrumentation, Ltd., Seattle, WA). Blots were incubated in blocking buffer [Tris-buffered saline (TBS) containing 5% nonfat dry milk plus 0.1% Tween-20 (Sigma)] for 1 h, washed 3 × for 5 min each in TBS/0.1% Tween-20, and incubated at 4°C overnight with antibodies specific to CD44 and Fra-1 at a dilution of 1:500 (Santa Cruz Biotechnology Inc., Santa Cruz, CA). Blots were then washed 3 × with TBS/0.1% Tween-20 and incubated with a specific peroxidase-conjugated secondary antibody at a dilution of 1:5,000 (Amersham Pharmacia Biotech, Piscataway, NJ) for 1 h. After washing blots 3 × in TBS/0.1% Tween-20, protein bands were visualized with the LumiGlo enhanced chemiluminescence detection system (Kirkgaard and Perry Laboratories, Gaithersburg, MD) and quantitated by densitometry [44]. Blots were reprobed with an antibody to α-Tubulin in a dilution 1:1,000 (Santa Cruz Biotechnology, Santa Cruz, CA) or β-Actin at a dilution of 1:5,000 (Abcam Inc, Cambridge, MA) to validate equal loading between lanes [15].

Electrophoretic gel mobility shift assays (EMSA) were used to assess binding of AP-1 to DNA and composition of AP-1 complexes. Nuclear extracts were prepared and analyzed as described by Ramos et al.[15]. The amount of protein in each sample was determined using the Bio-Rad protein assay (Bio-Rad, Hercules, CA). For supershift assays, nuclear extracts were incubated with antibodies to Fra-1 (Santa Cruz, CA) for 15 min at room temperature prior to addition of labeled oligonucleotide. Gels were quantitated using a Bio-Rad phosphoimager (Bio-Rad, Hercules, CA).

MM cells (5 × 10⁷) were starved in medium containing 0.5% FBS overnight and then stimulated with 10% FBS for 3 h. ChIP was performed using a commercially available kit (AVIVA Systems biology, San Diego, CA). Briefly, chromatin was crosslinked by adding formaldehyde (1%) to culture medium for 15 min and sonicated. A fraction of the soluble chromatin was saved for measurement of total chromatin input. The soluble chromatin was precleared and then was immunoprecipitated with 3 μg of Fra-1 antibodies (Santa Cruz Biotechnologies), 18 h at 4°C, and the immune complexes were absorbed with protein A/G beads. Immunoprecipitated purified DNA was analyzed by semi-quantitative PCR and densitometry was used to quantify the PCR results. The CD44 promoter was analyzed using primers: forward 5'-tttacagcctcagcagagc-3' and reverse 5'-ggaagttggctgcagttttt-3", which yield a 184-bp DNA product.

Total RNA (1 μg) was reverse-transcribed with random primers using the Promega AMV Reverse Transcriptase kit (Promega, Madison WI USA.) according to the recommendations of the manufacturer. To quantify gene expression, we amplified the cDNA by TaqMan Real Time Q-PCR using the 7700 Sequence Detector (Perkin Elmer Applied Biosystems, Foster, CA). Reactions contained 1 × TaqMan Universal PCR Master Mix, 900 nM of forward and reverse primers and 200 nM of TaqMan-probes. Thermal cycling was performed using 40 cycles of 95°C for 15 s and 60°C for 1 min. Original input RNA amounts were calculated with relative standard curves for both the mRNAs of interest and the hypoxanthine phosphoribosyl transferase (HPRT) control. Duplicate assays were performed with RNA samples isolated from at least 2 independent experiments. The values obtained from cDNAs and HPRT controls provided relative gene expression levels for the gene loci investigated. The primers and probe sequences used are presented in Table 2.

RK7-Fra-1Δzip, a dominant negative Fra-1 (dnFra-1) construct with deletion of the leucine zipper responsible for the dimerization function of Fra-1, was obtained from Dr. M. Busslinger (Research Institute of Molecular Pathology, Vienna, Austria) and cloned into pcDNA3 (InVitrogen, San Diego, CA) before transfection using electroporation. Briefly, cells were grown to 80–90% confluence, trypsinized, counted, and resuspended at 3 × 10⁶ cells/ml at room temperature. An aliquot of the cell suspension (400 μl) was mixed with 10 μg of plasmid DNA (expression or control plasmids) and electroporated at 280 V and 850 μF capacitance. Cells were immediately plated in fresh growth medium in 35 mm culture dishes and allowed to recover overnight. Following an overnight recovery, cells were selected for neomycin resistance using 200 μg/ml G418 (Sigma). Colonies surviving G418 selection were expanded and tested for the presence of Fra-1 as an indicator of plasmid activity. The Fra-1 and CD44 RNA interference (RNAi) duplexes were constructed from sequence information about mature mRNA extracted from the NIH genetic sequence database (GenBank^®). The open frame region from the cDNA sequence of exon 2 of the Fra-1 gene, and exon 5 (common exon to all splice forms) of the CD44 gene. The siRNA sequences targeting Fra-1 corresponded to the 230–250 coding region relative to the first nucleotide of the start codon, and the sequence targeting CD44, corresponded to the 480–500 coding region relative to the first nucleotide of the start codon. The sequences were BLAST-searched (NCBI database) against EST libraries to ensure the specificity of the siRNA molecule. The designed oligonucleotides were inserted and structured as follows: BamHI-sense-loop-antisense-HindIII small hairpin RNA (shRNA) in the expression vector pSilencer 3.1 H1-neo siRNA (Ambion). The vector was transfected using Lipofectamine 2000 (Invitrogen) as recommended by the manufacturer. Cells were immediately plated in fresh growth medium in 35 mm culture dishes and allowed to recover overnight. Following an overnight recovery, cells were selected for neomycin resistance using 200 μg/ml G418 (Sigma St. Louis, MO). Colonies surviving G418 selection were expanded and tested for the presence of mRNA levels of Fra-1 and CD44 using 27 cycles of PCR at an annealing temperature of 57°C (Fra- 1 forward primer: agtcaggagctgcagtgga and reverse primer: ctgctgctactcttgcgatg; CD44 forward primer: aagacatctaccccagcaac and reverse primer: ccaagatgatcagccattctgg and GAPDH control forward primer: cgggaagcttgtgatcaatgg and reverse primer: ggcagtgatggcatggactg).

Twelve-well plastic plates were coated with 1–2 mm isotonically prepared type I/II collagen (Vitrogen 100. Cohesion, PaloAlto, CA) as recommended by the manufacturer, incubated at 37°C for 1 h to promote gelation, and dried in a laminar flow hood overnight. The dry film was then rinsed with sterile water to remove salts and re-hydrate the film. Cells were seeded on the collagen-coated plates with complete medium as described above, and grown until confluent. Cells were maintained in serumless medium for 48 h before each assay. Control experiments were performed where cell replication was inhibited 2 h prior to treatment with 1 μg/ml aphidicolin to account for cell movement due to cell growth. After treatment for 1 h with the different inhibitors, a wound was made on the coverslips using a 100 μl plastic tip, and a small 2-mm² area was marked with a template for further observations using phase contrast microscopy (Olympus M081, Olympus Industrial America Inc.) After complete medium was added, the migration of cells then was examined at 8, 24, and 48 h. A final count of cells that moved into the marked area that exhibited a spreading morphology (thin, long axis) was performed at 24 h, and the relative cell motility was estimated as a % of the area covered by cells over the total area.

Receptor-mediated internalization of BODIPY FL dye-labeled hyaluronic acid derivative was assayed as described previously [60]. Briefly, cells were grown in glass coverslips, and then treated with BODIPY FL HA conjugate probe (Molecular Probes, Carlsbad, CA, now Invitrogen) at 100 μg/ml for 2 h. Unbound probe was removed by washing 3 × with phosphate buffered saline (PBS). Cells were fixed with 3% paraformaldehyde for 10 min at room temperature and washed again with PBS. Then the slides were nuclear counterstained with DAPI (5 μg/ml solution) (Molecular Probes, Carlsbad, CA). Coverslips were mounted onto slides with AquaPolyMount (Polysciences Inc., Warrington, PA). The sections were viewed with a BioRad MRC 1024 Confocal Scanning Laser Microscope (BioRad, Hercules, CA), and images were captured in sequential mode using Lasersharp 2000 software.

Human MM tissue arrays consisting of 2 mm representative areas of resected MM (N = 34) and normal lung tissue (N = 2) were obtained from Dr. Pass (New York University, NY, NY). Arrays and other samples containing formalin-fixed, paraffin-embedded samples were deparaffinized in xylene (2 × 15 min) and rehydrated in a graded ethanol series (95% to 50%). Slides were rinsed in water and placed in 1 × DAKO antigen retrieval solution (DakoCytomation, Glustrup, Denmark) and incubated at 95–99°C for 40 min. Slides then were washed 2 × for 5 min in 1 × PBS, and blocked with 50 μl of normal goat serum (Jackson ImmunoResearch Laboratories Inc., West Grove, PA) diluted in 950 μl PBS for 30 min in a humidified chamber at room temperature. A cocktail of polyclonal anti-rabbit Fra-1 antibody (R-20) (Santa Cruz Biotechnology Inc., Santa Cruz CA) diluted 1:100 and anti-mouse phosphorylated-p42/p44 (p-ERK1/2) (Cell Signaling Technology, Beverly, MA) diluted 1:100 in 1% BSA in PBS was applied to each slide and incubated overnight in a humidified chamber at 4°C. Sections then were washed in PBS and incubated for 30 min at room temperature in the dark with a secondary antibody cocktail [AlexaFluor goat anti-rabbit 568 (red), and AlexaFluor goat anti-mouse 647 (far red = blue) (Molecular Probes, Eugene, OR)]. Following a final wash in PBS, sections were counter-stained with SYTOX green (1:1,000 in PBS) or DAPI (5 μg/ml) (Molecular Probes, Eugene, OR), to detect nuclei, washed 1× in PBS, and mounted on glass slides using AquaPoly/Mount (Polysciences Inc., Warrington, PA). A negative control tissue omitting incubation with the primary antibody was included in each run.

A second set of arrays were deparaffinized as described above and double-stained with antibodies to SV40 T-antigen (Pab 101)(Santa Cruz Biotechnology Inc., Santa Cruz CA) and CD44 (Zymed/Invitrogen, Carlsbad, CA) as described previously [61]. The sections were viewed with a BioRad MRC 1024 Confocal Scanning Laser Microscope (BioRad, Hercules, CA), and images were captured in sequential mode using Lasersharp 2000 software. In addition to qualitative observations for staining, localization and extent, slides were semi-quantitatively scored for intensity on a scale of 0 to 3 using a blind coding system (data not shown).

In all experiments, duplicate or triplicate determinations were conducted for each group per time point. Experiments were performed in duplicate. Results were evaluated by one-way analysis of variance using the Student-Newman-Keuls procedure for adjustment of multiple pairwise comparisons between treatment groups. Differences with p values ≤ .05 were considered statistically significant.