p21-Activated Kinases Are Required for Transformation in a Cell-Based Model of Neurofibromatosis Type 2

Background NF2 is an autosomal dominant disease characterized by development of bilateral vestibular schwannomas and other benign tumors in central nervous system. Loss of the NF2 gene product, Merlin, leads to aberrant Schwann cell proliferation, motility, and survival, but the mechanisms by which this tumor suppressor functions remain unclear. One well-defined target of Merlin is the group I family of p21-activated kinases, which are allosterically inhibited by Merlin and which, when activated, stimulate cell cycle progression, motility, and increased survival. Here, we examine the effect of Pak inhibition on cells with diminished Merlin function. Methodology/Principal Findings Using a specific peptide inhibitor of group I Paks, we show that loss of Pak activity restores normal cell movement in cells lacking Merlin function. In addition, xenografts of such cells form fewer and smaller tumors than do cells without Pak inhibition. However, in tumors, loss of Pak activity does not reduce Erk or Akt activity, two signaling proteins that are thought to mediate Pak function in growth factor pathways. Conclusions/Significance These results suggest that Pak functions in novel signaling pathways in NF2, and may serve as a useful therapeutic target in this disease.


Introduction
Neurofibromatosis type 2 (NF2) is an autosomal dominant disorder characterized by the development of bilateral vestibular schwannomas and other benign tumors in central nervous system [1,2]. While several mitogenic pathways are known to be upregulated in NF2-mutant cells, despite considerable effort, there is as yet no consensus as to how loss of the NF2 tumor suppressor gene leads to schwannoma growth, nor are there effective medical therapies for this disorder.
The protein encoded by the NF2 gene, Merlin, exhibits significant homology to Ezrin-Radixin-Moesin (ERM) proteins, sharing a FERM (Four-point one, Ezrin, Radixin, and Moesin) domain at the N-terminus followed by an alpha-helical segment. Merlin has a unique C-terminal domain lacking a binding region for F-actin that exists in all other ERM proteins [3]. Within the FERM domain, a seven amino-acid conserved sequence (termed the ''Blue Box''), is important for Merlin functions. In Drosophila, deletion of this sequence (DBB) or substitution of polyalanine within this region (BBA) results in a dominant-negative form of the protein [4], most likely by disrupting intramolecular association between the N-and C-termini of Merlin [5]. This self-interaction can also be disrupted by phosphorylation of Merlin at residue serine 518, leading to a functionally inactive ''open state'' [6]. Merlin phosphorylation at this site is stimulated by Rac1 and Cdc42 GTPases via activation of their downstream effectors, p21activated kinases (Paks) [7,8].
Merlin is known to play an inhibitory role in Rac-mediated signaling [6]. NF2-deficient Schwannoma cells display aberrant membrane ruffling and concomitant hyperactivation of Rac and Pak1 [9,10,11]. Fibroblasts and keratinocytes lacking Merlin lose contact inhibition and Nf2-null neuroendocrine cells are defective in assembly of tight junctions and adherens junctions [12], processes disrupted by activated Rac and Pak [13,14,15]. Merlin affects Rac trafficking to the plasma membrane and is a direct inhibitor of Pak1, suggesting a negative feed-forward loop between Rac/Pak and Merlin [16]. For these reasons, The Rac/Pak signaling axis has garnered increasing attention as a possible therapeutic target in NF2.
As downstream mediators of Rac function, Paks have been implicated in regulating cell morphology, motility proliferation, and survival [17,18,19]. Group I Paks (Pak1, -2, and -3) affect a wide variety of central signaling pathways, including positively regulating mitogen activated kinases (MAPKs), Akt, and NFkB [20,21]. In addition, group I Paks influence the G2/M transition by activating Aurora-A and Polo-like kinase 1 (Plk-1) [22,23]. In Ras-transformed cells, expression of dominant-negative Pak1 blocks transformation and prevents full activation of Erk and Jnk, and it is thought such MAPKs represent important signaling targets for Pak in cancer [24,25]. In the Erk pathway, Pak has been shown to phosphorylate c-Raf at S338 and Mek1 at S298 [26,27,28,29]. These phosphorylations are thought to be necessary, but not sufficient, for full activation of c-Raf and Mek by Ras.
In this study, we show that inhibition of group I Paks in Nf2mutant fibroblasts reduces proliferation, restores normal morphology, and reduces invasiveness of cells lacking Merlin function, as well as the tumorigenicity of such cells in nude mice. However, these changes are not associated with reductions in Erk or Akt activity; in fact, the activity of these pathways is augmented, suggesting that the beneficial effects of inhibiting Pak in these cells involves novel signal transduction pathways.

Establishing Merlin-deficient cells that express a Pak inhibitor
To study the influence of Pak inhibition in cells lacking Merlin function, we established a stable NIH-3T3 fibroblast cell line overexpressing Merlin DBB, a mutant form of the tumor suppressor that lacks the ''Blue Box'' motif and which, in Drosophila, acts in a dominant negative fashion, in effect mimicking loss of the NF2 gene ( Fig. 1A) [30]. Another ''Blue Box'' mutant, Merlin BBA, has been shown to be tumorigenic in mammals [31,32], though Merlin DBB and Merlin BBA are reported to show some differences with respect to effects on cell shape and adhesiveness [30,31]. The expression level of Merlin DBB in such infected cells was about 5-fold higher than endogenous Merlin ( Fig. 1B and Fig. S1). Note that, despite deletion of the Blue Box motif, exogenous Merlin DBB migrates at the same position as endogenous Merlin in SDS/PAGE. NIH-3T3 cells expressing dominant negative Merlin (DBB cells) were then infected with recombinant retrovirus encoding GST-tagged Pak inhibitor domain (PID) or a nonfunctional inhibitor control, PID L107F (PID LF) [27]. The DBB cells infected with the PID or PID LF retrovirus displayed readily detectable signals in anti-GST immunoblots (Fig. 1B).

Effects of Pak inhibition on invasiveness in cells expressing dominant negative Merlin
We first examined the effect of Merlin on cell morphology and invasiveness and asked if such effects could be reversed by PID expression. Cells expressing DBB became fusiform as compared to normal NIH-3T3 cells ( Fig. 2A). Expression of the PID, but not the control PID LF, showed substantial restoration of normal morphology. For invasion studies, cells were plated in a chamber above a layer of Matrigel and assessed for their ability to penetrate through this layer, indicating invasiveness. As shown in Fig. 2B and Table S1, the invasiveness of DBB cells was almost twice that of control NIH-3T3 cells. Assuming linear invasion kinetics, expression of PID, but not PID LF, significantly inhibited the invasive capacity of Merlin DBB cells. Thus, a Pak inhibitor substantially reduced invasiveness in cells expressing a dominantnegative form of Merlin. In control NIH-3T3 cells, PID expression had a slight effect on morphology, and reduced invasiveness by about 10% (Fig. 3 and Table S2).

Inhibition of tumor formation by Pak inhibition
To assess the effects of Pak inhibition on NF2-related tumorigenicity, we injected nude mice only with DBB, DBB/ PID, or DBB/PID LF cells. DBB cells developed substantial tumors by 6 weeks post-injection (average volume .200 mm 3 ), similar to previous studies in which the Merlin BBA mutant was used [31,32]. Tumors derived from DBB/PID cells were much smaller in size, with an average volume of 38 mm 3 (Fig. 4). Interestingly, mice injected with DBB/PID LF cells developed tumors even larger (average volume ,450 mm 3 ) than those injected with DBB cells. Similar experiments were also carried out in NIH-3T3 cells lacking the DBB transgene (Fig. 5). In this latter case, only very small tumors were noted after 6 weeks (compare Fig. 4 with Fig. 5), but, as with the DBB xenograft, these masses were larger in cells expressing PID LF (Fig. 5).
Lastly, we examined signaling activity in tumor lysates from these xenografts. Pak activity was inhibited in DBB/PID cells, indicating that this suppressor remained effective in vivo. Surprisingly, while Akt and Erk were not active in tumors derived from animals injected with DBB or DBB/PID LF cells, both Akt and Erk were activated in lysates from the small tumors that developed in mice injected with DBB/PID (Fig. 6). It is also of interest that, in the tumors from DBB/PID mice, Merlin expression (presumably exogenous Merlin DBB) was substantially elevated. These data show that PID expression strongly inhibited tumor formation in the NF2 xenograft model, suggesting that Group I Paks are required for transformation in cells that have lost Merlin function. Despite this requirement, loss of Pak function did not reduce Akt or Erk activity; on the contrary, it activated them. In the small tumors observed in NIH-3T3 xenografts, the effects of the PID on signaling were consistent with prior literature: PID expression decreased Pak activity, as well as the Mek phosphorylation at S298, Mek activity, and ERK activity (Fig. 7). In these small tumors, Merlin expression was not affected in a consistent way, nor was activation of Akt altered.

Discussion
In this report, we have shown that the effects of a dominant acting Merlin protein can be largely reversed by inhibition of group I Paks. As has been noted previously, expression of a different ''Blue Box'' mutant of Merlin alters the actin cytoskeleton, promotes invasiveness, and confers the ability to form tumors in nude mice [31]. Given that Pak is activated in cells lacking functional Merlin [32] and that some of these aberrant phenotypes resemble those shown in cells with hyperactivation of Pak1, blockade of Pak might represent a reasonable therapeutic strategy in NF2. The involvement of Akt and Erk in NF2 pathobiology is controversial. Some studies report elevated Akt and Erk in NF2deficient cells [33,34,35,36], whereas others report no correlation [37,38]. Our results are consistent with these latter studies, in that cells lacking Merlin, whether grown in vitro or when recovered from xenograft tumors, displayed low basal Akt and Erk activity. Curiously, in the few small tumors that developed from DBB/PID xenografts, Akt and Erk activities were elevated (Fig. 6). The same was true for DBB/PID cells grown in vitro (data not shown). These activations may reflect an altered signaling strategy in the tumor cells, necessary to overcome loss of Pak activity due to PID expression.
Our studies also show that the DBB mutant of NF2, like another commonly studied ''Blue Box'' mutant, Merlin BBA, is in fact tumorigenic in mice. Whether this mutant acts in precisely the same manner as the better-studied BBA mutant is unclear, as these two mutants are reported to have different effects on cell adhesiveness and morphology. Despite these issues, it is clear that the DBB mutant has major effects on mouse cell morphology, invasiveness, and tumorigenicity, and that these changes are not accompanied by marked upregulation of Erk or Akt.
The data reported here are in general agreement with a previous study conducted by Yi et al., in which Pak inhibition by shRNAs reduced tumor volume of dominant-negative NF2expressing cells [32], but also have some intriguing differences. Interestingly, in the study by Yi et al., escape from Pak knock-down occurred when expression of the Pak1 shRNA hairpin was silenced by methylation of the H1 promoter in the NF2 tumor cells. In our studies, which employed a peptide inhibitor of Pak rather than shRNA, we noted that the few, small tumors that arose in animals injected with DBB/PID cells did not shut off the PID transgene but did show significant increases in Merlin expression (Fig. 5). Because endogenous Merlin and the DBB Merlin mutant have similar mobility on SDS/PAGE, it was not possible to determine if the increased expression is due to endogenous or exogenous (DBB) Merlin, but we favor the idea that it represents DBB Merlin, especially in light of the lack of consistent increase in endogenous Merlin levels in xenograft tissues from cells expressing PID alone (Fig. 7). Therefore, we believe that the most likely explanation for these findings is that increased Merlin DBB expression was required for the tumor cells to escape inhibition consequent to diminished Pak function. Thus, both the results of our study and that of Yi et al. suggest that Pak activity is required for efficient tumorigenesis in cells that have lost Merlin function. In addition, our study shows that down-regulating Pak expression per se is not required for inhibiting tumor formation; inhibiting the catalytic activity of endogenous Pak is sufficient for these beneficial effects.
A number of peptide based reagents, such as PID and cellpenetrating peptides based on the Nck or PIX binding regions of Pak, have been used to effectively block Pak function in cells and in vivo [39,40]. One note of caution raised by our studies is that the putative negative control for the PID, PID LF, appears to have gain-of-function effects in a variety of cell types. The PID LF mutant has been thought to represent a functionless, inert control for the PID, incapable of inhibiting Pak or binding to its partners such as the Fragile X protein [41]. Our results here should inject a note of caution in the use of this construct. Instead, one may consider small molecule inhibitors of Pak.
Recently, a few specific small molecule inhibitors of group I Paks have also been described, including OSU03012, lambdaFL172 [42], and IPA-3 [43]. The latter of these three compounds has been shown to inhibit membrane ruffling and cell spreading in NF2 2/2 schwannoma cells [44]. Interestingly, in these experiments, blockade of Pak with IPA-3 reduced Rac activity, suggesting that Pak acts upstream of Rac in these cells. Though Rac also activates Pak, such an idea is consistent with models which link Pak to further Rac activation via the Pak-bound guanine-nucleotide exchange factor, PIX [37,45]. These results, like ours, support the notion that the Rac/Pak signaling axis is activated as a consequence of Merlin loss, and that these enzymes might provide useful targets for therapy in NF2.

Materials and Methods
Cell culture conditions and retroviral transductions NIH-3T3 (obtained from American Type Culture Collection) cells were cultured in high-glucose DMEM supplemented with 10% fetal calf serum, 2 mM L-glutamine and 100 U/ml penicillin/streptomycin at 37uC in a humidified 5% CO 2 incubator. The wNX packaging cell line (Orbigen) was transfected using Lipofectamine 2000 (Invitrogen). Viral supernatants were harvested 48 hr post-transfection and filtered. Cells were incubated with retroviral supernatant supplemented with 4 mg/ml polybrene for 4 hr at 37uC, and then were cultured in growth media for 48 hr for viral integration. Infected cells were selected with 2 mg/ml of puromycin or by flow cytometry for cells with green fluorescent protein (GFP).

Cell invasion assay
Matrigel invasion chamber (BD Biosciences) were rehydrated in serum-free DMEM medium for 2 hr and then placed in 0.75 ml of DMEM medium supplemented with 5% fetal calf serum. Cells at a density of 1.5610 4 suspended in 0.5 ml of DMEM, and seeded onto Matrigel chambers. Cells were allowed to invade for 12 or 24 hr and cells on the upper surface were gently removed with a cotton bud, and cells that had migrated through the 8 mm pores were fixed with 4% paraformaldehyde for 15 min and stained with 0.1% crystal violet for 15 min. Membranes were washed, removed and mounted on a glass slide, and the level of invasion was quantified by visual counting using a microscope with a 20X objective.

Cell cycle analysis
Cell cycle profiles were analyzed using flow cytometry with propidium iodide staining. Cells were trypsinized and washed with PBS and fixed in 70% ethanol at 220uC. After fixation, cells were washed once with PBS and resuspended in 0.5 ml of a solution containing 1 M Tris-HCl pH 8.0, 0.1% Nonidet P-40, 10 mM NaCl, 50 mg/ml propidium iodide and 70 Kunitz units/ml RNase A. FACS analysis was performed using CELLQuest TM software (Becton Dickinson). A minimum of 10,000 events was collected per sample.

Immunofluorescence
Cells were plated on glass coverslips in 6-well culture plates and fixed in 4% paraformaldehyde for 10 min, permeabilized with 0.2% Triton X-100 for 5 min and blocked with 1% BSA in PBS for 30 min. Filamentous actin and nuclei were visualized by staining with Rhodamine-phalloidin (Molecular Probe) and DAPI (Sigma), respectively. Images were observed and captured on an inverted phase/fluorescence microscope (Nikon TE300).

Tumorigenicity assays
Cells were trypsinized, washed with PBS and resuspended at 10 7 cells per ml in PBS. 2610 6 cells were injected subcutaneously into flanks of 5-week old nude mice (BALB/c nu/nu). Mice were monitored and tumor diameters were measured every 3 days using caliper for 6 weeks following injection. Tumor volume was calculated by the following formula: volume = 0.56 (length) 6 (width) 2 . The mice were sacrificed and the tumors were resected for histological examination and immunoblot analysis. All animals Figure 6. Activity of signaling pathways in tumor samples from xenografts. Tumor from six mice were resected and immediately lysed in a buffer containing 10 mM PBS pH 7.2, 1% Triton X-100, 0.1% SDS, 20% glycerol, plus complete protease and phosphatase inhibitor tablets (Roche). Lysates were separated by SDS-PAGE and incubated with the indicated antibodies. doi:10.1371/journal.pone.0013791.g006 were handled in strict accordance with good animal practice as defined by the relevant national and/or local animal welfare bodies, and all animal work was approved by the FCCC IACUC committee (Protocol 96-10), active 08/09 -08/10.

Data analysis
All experiments were performed at least three times. Results are reported as means 6 SD. The significance of the data was determined by two-tailed, unpaired Student's t-test, and differences were considered statistically significant at P,0.05.