Oncogene (2011) 30, 3513–3521 & 2011 Macmillan Publishers Limited All rights reserved 0950-9232/11 www.nature.com/onc ORIGINAL ARTICLE An siRNA screen identifies RSK1 as a key modulator of metastasis

R Lara1,7, FA Mauri2, H Taylor3, R Derua4, A Shia3, C Gray5, A Nicols5, RJ Shiner2, E Schofield6, PA Bates6, E Waelkens4, M Dallman3, J Lamb3, D Zicha5, J Downward7, MJ Seckl1 and OE Pardo1

1Department of Oncology, Hammersmith Campus, Cyclotron Building, London, UK; 2Histopathology Imperial College London, Hammersmith Campus, London, UK; 3Division of and Molecular Biology, Department of Life Sciences, Faculty of Natural Sciences, Imperial College London, London, UK; 4Labo Proteı¨ne Fosforylatie en Proteomics, Katholieke Universiteit Leuven, Leuven, Belgium; 5Light Microscopy Department, London Research Institute, London, UK; 6Biomolecular Modelling Laboratory, London Research Institute, London, UK and 7Signal Transduction Laboratory, Cancer Research UK, London Research Institute, London, UK

We performed a kinome-wide siRNA screen and identified Introduction 70 kinases altering cell migration in A549 lung cancer cells. In particular, ribosomal S6 kinase 1 (RSK1) Lung cancer is the most common cancer killer with silencing increased, whereas RSK2 and RSK4 down- a 5-year survival rate o5%. Non-small cell lung cancer regulation inhibited cell motility. In a secondary collagen- (NSCLC) accounts for 80% of cases of which adeno- based three-dimensional invasion screen, 38 of our hits carcinoma represents the majority. Most patients cross-validated, including RSK1 and RSK4. In two further present with metastatic lesions and are incurable. Hence, lung cancer cell lines, RSK1 but not RSK4 silencing a better understanding of the biological processes showed identical modulation of cell motility. We therefore underlying lung cancer cell motility, invasion and selected RSK1 for further investigation. Bioinformatic metastasis is needed to improve patient survival. analysis followed by co-immunoprecipitation-based vali- The 90-kDa ribosomal S6 kinase (RSK) family is dation revealed that the actin regulators VASP and Mena activated downstream of the Ras/MEK/Erk signalling interact with RSK1. Moreover, RSK1 phosphorylated pathway (Blenis et al., 1991). Four human isoforms VASP on T278, a site regulating its binding to actin. In (RSK1–4) exist (Anjum and Blenis 2008). RSKs are addition, silencing of RSK1 enhanced the metastatic characterised by the existence of two kinase domains potential of these cells in vivo using a zebrafish model. that come into close proximity following activating Finally, we investigated the relevance of this finding in phosphorylation events. Their downstream substrates human lung cancer samples. In isogenically matched include a number of cytoplasmic and nuclear targets tissue, RSK1 was reduced in metastatic versus primary (CREB, Fos, Jun, TSC2 and filamin A (Anjum and lung cancer lesions. Moreover, patients with RSK1- Blenis 2008)) that explain their involvement in diverse negative lung tumours showed increased number of cellular processes, such as cell proliferation and survival. metastases. Our results suggest that the findings of our Increased expression of RSKs was shown in breast high-throughput in vitro screen can reliably identify (Smith et al., 2005) and prostate cancer (Clark et al., relevant clinical targets and as a proof of principle, 2005), whereas RSK2 activity has been linked to cell RSK1 may provide a biomarker for metastasis in lung transformation (Cho et al., 2007; Kang et al., 2007). cancer patients. RSKs have been shown to phosphorylate filamin A Oncogene (2011) 30, 3513–3521; doi:10.1038/onc.2011.61; (Woo et al., 2004) and p27Kip (Larrea et al., 2009), with published online 21 March 2011 RSK1 being implicated in promoting melanoma cell migration in vitro (Larrea et al., 2009). However, the Keywords: ribosomal S6 kinase 1; metastasis; lung relevance of these findings to cancer metastasis in vivo or cancer; siRNA screen; VASP in patients has not been established. The Ena/VASP family of actin-binding proteins is not known to mediate RSK1 effects, but is involved in various processes, including cell migration (Krause et al., 2003). Three human family members exist; Mena, VASP and EVL. Their overlapping function is modu- Correspondence: Professor MJ Seckl and Dr OE Pardo, Department of lated by both homo- (Zimmermann et al., 2002) or Oncology, Cancer Medicine, Imperial College London, Hammersmith heterotetramerisation (Gertler et al., 1996) and phos- Campus, Cyclotron Building, Du Cane Road, London, W12 0NN, phorylation events (Benz et al., 2009) that control UK. E-mail: [email protected] or [email protected] their effect on actin polymerisation. Increased expres- Received 5 July 2010; revised 29 November 2010; accepted 2 February sion of Ena/VASP family members enhances cancer 2011; published online 21 March 2011 cell invasiveness in vitro and in vivo (Han et al., 2008; RSK1 in lung cancer cell metastasis R Lara et al 3514 Philippar et al., 2008) and correlates with tumour 779 Targets 1.8 progression in several malignancies (Gurzu et al., Motility screen 1.6 2008; Hu et al., 2008; Toyoda et al., 2009), including 1.4 RSK1 Scramble lung adenocarcinoma (Dertsiz et al., 2005). 160 Hits 1.2 ERBB2 Here, we performed an siRNA kinome library screen Toxicity screen 1 0.8 RSK2 0.6 in A549 human lung adenocarcinoma cells and demon- 70 Hits SRC RSK4 0.4 strated that RSK1 silencing increased migration and Mean speed ratio MST1R Publication 0.2 invasion. This could potentially be explained by VASP 0 phosphorylation on T278, a site regulating binding 57 Hits Individual Targets to actin (Blume et al., 2007). Knockdown of RSK1 enhanced the metastatic potential of A549 cells in vivo. Sc RSK1 RSK2 RSK4 Moreover, immunohistochemical staining revealed that 2 2 2 2 1 1 1 1 RSK1 was downregulated in metastatic lung cancer 0 0 0 0

samples compared with isogenically matched primary Units) -1 -1 -1 -1 Distamce (Arbitrary -2 -2 -2 -2 tumours. Thus, in lung cancer, our results suggest that -2 -10 1 2 -2-10 1 2 -2-1 0 1 2 -2 -1 0 1 2 RSK1 expression inhibits metastasis and might be a predictive biomarker for disease dissemination. RSK1 RSK2 RSK4 8 *** 7 2 2 6 ** 5 *** 1.5 1.5 * Results ** * * * 4 * * ** ** Identification of novel regulators of cell motility in lung 3 1 1 cancer cells 2 ** ** Speed (Arbitrary Units) Using the well-characterised A549 NSCLC cell line, we 1 0.5 0.5 established a high-throughput random-walk motility screen in which 60 cells per condition were automatically tracked ScP 1 2 3 4 ScPSc41 2 3 4 P 1 2 3 using Metamorph software (Molecular Devices, Chicago, RT RSK1 RSK2 RSK4 IL, USA). The resulting trajectory plots were analysed with a Mathematica notebook (Katso et al., 2006) and analysis GAPDH GAPDH GAPDH of variance performed to assess the significance of the changes observed with Po0.05 used as a cutoff. Our Figure 1 RSK family members regulate cell migration. screen identified 160 proteins that when downregulated, (a) Overview of the migration screen. (b) Distribution of the mean speed ratio to scramble for individual targets. (c) Trajectory plots significantly altered A549 cell migration (Figures 1a of representative fields of view (n ¼ 15) following transfection and b). To exclude cellular toxicity as a cause for with RSK1, 2, 4 or non-targeting (Sc) SmartPool siRNAs. decreased cell motility, we used data from our pre- (d) Representative validation experiment for RSK1, 2 and 4. P; viously published toxicity screen (Swanton et al., 2007) pool, 1–4; deconvoluted single oligonucleotides. The average random migration speed distribution of 60 cells per condition is performed with the same cells and RNAi library. represented as a box plot. Box; 75% and side bar; 12.5% of the cell Cell motility hits that showed 410% toxicity were population. Transverse bar; median speed. Analysis of variance discarded, thereby reducing the number of our candi- with Sc as reference: *Po0.05, **Po0.005, ***Po0.001. Repre- dates to 70 proteins (Figure 1a). Reassuringly, several of sentative reverse transcriptase–PCR (RT) for the target mRNAs these were already known modulators of cell migration, with GAPDH as a loading control. including MAP4K4 (Collins et al., 2006), AKT2 (Irie et al., 2005), EphB6 (Fox and Kandpal, 2009) and CDC42 (Nobes and Hall, 1995). However, 57 of our hits As metastasis involves both enhanced motility and were novel modulators of migration (Figure 1a and invasiveness, we next performed a secondary three- Supplementary Figure 1 for a full hit list). Four kinase dimensional high-throughput collagen-based invasion families were found, with several members capable of screen on our 70 motility hits. Thirty-eight of these altering migration as follows (Supplementary Figure 2): cross-validated in the two screens (Supplementary the ribosomal protein S6 kinase , casein kinase , CDC- Figure 4). Figure 2a shows side views of representative like kinase and Arf-domain containing kinase families. stacks in which knockdown of 4.1B was used as a To validate our hits, a second toxicity assay and positive control (Cavanna et al., 2007). We decided to SmartPool (Dharmacon, Lafayette, CO, USA) decon- focus on the RSK family as a paradigm to validate the volution were performed. We confirmed that changes in importance of our screen and because of our previous cell viability did not account for the observed motility interest in S6 kinases (Pardo et al., 2001, 2006). effects (Supplementary Figure 2) and that 14 of the 17 In keeping with our migration data, silencing of RSK1 targets validated with at least 2 of the 4 deconvoluted enhanced, whereas RSK4 knockdown suppressed inva- individual sequences (Supplementary Figure 2 and sion (Figures 2a and b). However, unlike the motility Figure 1d). These effects correlated with target down- screen, RSK2 silencing failed to suppress invasion. regulation at the protein and/or mRNA levels (Figure 1d Hence, silencing of RSK1 and RSK4 alter invasion in a and Supplementary Figure 3). similar way to their effects on migration in A549 cells.

Oncogene RSK1 in lung cancer cell metastasis R Lara et al 3515 We next investigated whether the effects of RSK1 and but not RSK4 validated in both NME35 and H23 RSK4 silencing on motility could be generalised to other NSCLC cells. Consequently, we focused on RSK1 and NSCLC cell lines. Figure 2c demonstrates that RSK1 further validated this hit in our three cell lines using four independent siRNA sequences from an alternative provider (Figure 2d). All four sequences were able to enhance the migration in the three NSCLC cell lines siRNA: examined. In contrast, overexpression of wild-type Sc RSK1 in A549 cells inhibited cell migration (Supple- mentary Figures 9A–C). This effect was even more 4.1B pronounced upon overexpression of a kinase-active mutant of this protein (Supplementary Figure 9C). RSK1 Hence, overexpression and silencing of RSK1 have opposing effects on A549 cell migration, and the role of RSK2 this kinase on cell motility might be mediated through its kinase activity. RSK4

RSK1 binds and phosphorylates VASP * 12 To identity new interacting partners of RSK1 that 4 8 might explain the effects of this protein on cell migration and invasion, we performed bioinformatic analysis. 3 A human protein–protein interaction database was 8 assembled that integrated two previously published n.s. collections containing both experimentally verified 4 2 or in-silico predicted interactions (Jonsson and Bates 4 2006; Linding et al., 2007). Subsequent analysis revealed Invasion (fold) 1 several known modulators of cell migration (for example, * VASP(Krause et al.,2003),VIM(Ivaskaet al.,2007)) 1 1 that might interact with RSK1 (Figure 3a). The changes 0 0 0 seen in the actin cytoskeleton upon downregulation of Sc Sc Sc Sc Sc Sc RSK1 (Figure 3b) suggested a potential link with RSK2 RSK4 RSK1 modulators of actin dynamics. As VASP is known to +EGF +EGF +EGF mediate such effects (Trichet et al.,2008),wetested whether the predicted interaction between VASP NME35 H23 1.8 * 6 and RSK1 could be validated. Figure 3c (upper panel) * shows that immunoprecipitation of RSK1 pulled down endogenous VASP in A549 cells. Similarly, RSK1 was 1.6 5 detected in VASP immunoprecipitates (Figure 3c, lower panel). As VASP is known to heterotetramerize with 1.4 Mena (Gertler et al., 1996), another protein regulating 4 actin dynamics (Kwiatkowski et al., 2003), we next tested 1.2 the potential for Mena to associate with RSK1. Co-immunoprecipitation experiments confirmed that 1 3 RSK1 associates with Mena in A549 cells (Figure 3d). Speed (Arbitrary Units) To investigate the functional relevance of the RSK1/ 0.8 VASP interaction and determine whether these proteins 2 directly interact, we performed an in vitro kinase assay Sc Sc using purified proteins. Figure 3e shows that RSK1 RSK1 RSK2 RSK4 RSK2 RSK4 RSK1

A549 H23 NME35 ** ** ** Figure 2 Validation of RSK1 in additional cell lines and by * 5 ** * 4 * invasion assay. (a) Side views of representative confocal stacks of 4 A549 cells invading upward into a collagen matrix. Arrows point to ** * * 4 * the bottom of the wells. (b) Quantification of cell invasion in A549 3 * * ** * 3 treated with RSK1, 2, 4 or Sc siRNAs. The combined invasion distribution for 27 fields of view per condition is represented as a 2 3 2 box-plot. Box; 75% and side bar; 12.5% of the cell population. Transverse bar; median. Analysis of variance with epidermal growth 1 2 factor-treated Sc as reference: *Po0.05, **Po0.005. (c) RSK1 1 regulates cell migration in NME35 and H23 cell lines. (d) Validation Speed (Arbitrary Units) P 1234 P1234 P1234 of the effects of RSK1 silencing using four alternate siRNA oligo- Sc Sc Sc nucleotides in A549, NME35 and H23 cells. (c, d) Random migration RSK1 RSK1 RSK1 speed distributions are shown as in Figure 1.

Oncogene RSK1 in lung cancer cell metastasis R Lara et al 3516 polymerisation (Blume et al., et al., 2007). We therefore tested whether RSK1 might modulate phosphorylation of T278 in intact cells. Figure 3f (left panel) demon- strates that RSK1 silencing prevented phosphorylation of VASP on T278 in A549 cells growing in serum. Knockdown of VASP confirmed that the T278 phos- phosite detected was specific for VASP. We also examined the phosphorylation of S157 and S239, two other sites involved in the binding of VASP to actin. Unlike T278, basal phosphorylation of these sites was low and was unaffected by RSK1 downregulation, but as expected, could be induced by forskolin (Figure 3f, WB: Ly B IP right panel). Thus, RSK1 can directly associate and RSK1 - + - + selectively phosphorylate VASP on T278 in A549 RSK1 VASP + + -- S6 --+ + NSCLC cells. VASP As VASP T278 phosphorylation is known to correlate IP: RSK1 VASP with an inhibitory effect on lamellipodia formation RSK1 and cell migration, one might predict that silencing of VASP could reverse the promigratory effects of RSK1 VASP S6 IP: VASP knockdown. In addition, because VASP heterotetra- merises with Mena we investigated the potential role WB: Ly B IP of Mena in controlling RSK1 effects. Figure 4a shows

Loading that knockdown of either VASP or Mena decreased RSK1 A549 cell migration. Moreover, knockdown of VASP

IP: RSK1 siRNA or Mena partially reversed, whereas combined down- Mena regulation completely prevented the effects of RSK1 silencing. Furthermore, although RSK1 silencing siRNA WB: Sc RSK1 DMSO Forskolin induced a 10-fold increase in lamellipodia, combined VASP pS157 VASP and Mena knockdown dramatically reduced this RSK1 VASP WB: Sc effect (Figures 4b and c). We next investigated the effect VASP pS239 VASP pT278 of downregulating VASP or Mena on the proinvasive phenotype of RSK1-silenced A549 cells. VASP, but not Total VASP Total VASP Mena silencing, slightly decreased the ability of RSK1 siRNAs to promote A549 invasiveness (Supplementary Figure 3 VASP and Mena interact with RSK1. (a) Bioinformatic Figure 8). Taken together, these data suggest that VASP analysis showing proteins involved in cell migration interacting and/or Mena mediate the effects of RSK1 on A549 cell with RSK1. The link with VASP is shown in red. (b) Represen- migration. However, it seems likely that in addition to tative actin cytoskeleton staining of A549 cells treated with Sc and VASP, one or more other molecules are involved in RSK1 siRNAs. (c) Co-immunoprecipitation (IP) experiments validate interactions between RSK1 and VASP. (d) Co-IP identify RSK1-mediated invasion. interactions between RSK1 and Mena. (c, d) Beads without antibody (b) were used as control. (e) VASP is phosphorylated by RSK1 in vitro. VASP immunoprecipitates were subject to RSK1 modulates tumour cell dissemination in vivo in vitro kinase assay using purified recombinant RSK1 and [g-32P]ATP. Ribosomal S6 was used as positive control. Upper Our preceding results indicated that RSK1 downregula- panel: autoradiogram; black bars indicate molecular weight tion enhanced cell motility and invasion in A549 cells markers. Lower panel: Coomassie stain. (f) VASP T278 phosphor- in vitro. To assess the relevance of these findings in vivo, ylation depends on RSK1. A549 cells were treated with RSK1, we established a cancer metastasis zebrafish model. VASP or Scramble (Sc) siRNAs. Untreated cells were stimulated A549 cells expressing a histone 2B GFP fusion protein with forskolin 10 mM for 30 min as a positive control. Dimethyl sulfoxide; diluent-alone control. (c, d and f) Samples analysed by were injected into the cell mass of zebrafish eggs 4 h post SDS-polyacrylamide gel electrophoresis/western blotting (WB) fertilisation. Figure 5a shows that these cells could easily using antibodies against the indicated targets. be visualised using wide-field fluorescent microscopy. Whole-fish imaging after 48 h demonstrated that the injected cells formed a tight tumour mass in the head of could directly phosphorylate VASP more efficiently all injected fish with only very limited dissemination into than its classical substrate, S6 protein. To determine the the fish body (data not shown). Using this model, we nature of the phosphorylation sites, the phosphorylated hypothesized that downregulation of RSK1 might recombinant VASP from our in vitro experiment was enhance dissemination of A549 cells away from this sent for mass spectrometry. Results in Supplementary mass. Indeed, injection of A549 cells silenced for RSK1 Figure 5 showed that T278 was the only site specifically resulted in a more diffuse tumour mass in the fish head, phosphorylated by RSK1. Indeed, neither recombinant accompanied by single cell invasion throughout the RSK2 nor RSK4 induced phosphorylation of this site, fish body, including the tail (Figure 5b). In contrast, fish which is known to be involved in controlling actin injected with A549 cells transfected using a scrambled

Oncogene Lamellipodia / Cell Speed (arbitrary units) 0.2 0.4 0.6 0.8 1.2 1 2 3 4 0 1 RNAi RNAi Sc Sc *** ** *** VASP VASP *** ***

*** Mena Mena

VASP + *** VASP +Mena Mena *** RSK1 RSK1 *** ** *** VASP VASP + RSK1 *** + RSK1 Mena ** Mena VASP + VASP +Mena Mena S1i ugcne elmetastasis Lara cell R cancer lung in RSK1 ettv s netdwt 2-F 59clsiae tthe at invading imaged single of cells details dividing. show A549 cell panels H2B-GFP Lower points. with time indicated injected fish sentative uorcl nte‘ed,‘ide r‘al oprmn sindi- as compartment ‘Tail’ ( or ‘Middle’ in ‘Head’, cated the in cell tumour ( ( in as cells. injected invading single to rSrml S)RA eeijce t4hf ersnaiefish Representative hpf. 4 at injected were ( (from embryos. RNAi hpf (Sc) 4 Scramble of mass or cell the in cells h ( 4 (hpf). embryos zebrafish fertilisation into post injected were cells A549 GFP-expressing 5 Figure ihS srfrne ** ( ( reference: as s.e.m. in Sc represent were with as view bars of view Error field the excluded. of outside partially fields Cells observers. independent 36 box from The cell box-plot. ( shown. Spot; a are population. condition as cell the represented of ( 12.5% speed. is bar median side condition each and per 75% contains cells with 60 treated of were ( cells siRNAs. A549 indicated cytoskeleton. the actin the and migration 4 Figure Transmitted tal et Merged n GFP 30 ¼ 2 e odto)aesona 8hf e rospoint arrows Red hpf. 48 at shown are condition) per 120 b S1slnigpooe eatssi erfih H2B- zebrafish. in metastasis promotes silencing RSK1 APadMn ouaeteefcso Sso cell on RSKs of effects the modulate Mena and VASP ). n b ¼ b .Preto ae eest rsneo bec of absence or presence to refers cases of Percent ). ersnaiepcue fatnsann o each for staining actin of pictures Representative ) 2 e odto.* condition. per 120 Hours post-injection c

Percent of cases 100 siRNA 60 a h vrg ubro aelppi per lamellipopdia of number average The ) P a 50 h vrg irto pe distribution speed migration average The ) ersnaiepcuesoiginjected showing picture Representative ) o 0 c RSK1 Sc unicto fivso rmfish from invasion of Quantification ) .0,*** 0.005, Head Head b P eedtrie ytwo by determined were ) P o b 84 a o el rae ihRSK1 with treated Cells ) , Middle 0.001. .5( 0.05 c nlsso variance of Analysis ) Middle * t ts) ( -test). RSK1 Scramble 90 d Tail Repre- ) Tail Oncogene 3517 RSK1 in lung cancer cell metastasis R Lara et al 3518 siRNA displayed tight tumours in the head with control RSK1 Score -5 levels of invasion and total absence of dissemination to p=5.4 x 10 80 the fish tail (Figure 5b). This difference in phenotype 17% could be quantified by scoring for the presence or 70 37% absence of tumour cells within three defined regions of 60 the fish body; head, middle and tail (Figure 5c, n ¼ 120 50 46% fish per condition). Importantly, the cells that had 40 migrated to the tail retained the ability to proliferate, a 30 condition required for the formation of metastatic 20

I neg/II neg % of RSK1 score tumours (Figure 5d). Thus, RSK1 knockdown enhances 10 I >II the metastatic potential of A549 cells in vivo. 0 I

secondary lesion. Examples of pictures of adenocarci- RSK1 noma samples with increasing RSK1 staining scores are shown in Figure 6d. Taken together, these results correlate with our previous experimental findings and suggest that RSK1 negativity in primary tumours may be a clinical biomarker for metastasis and poor prognosis. A much larger cohort of patients is now required to examine the predictive power of RSK1 for Positive Score 200 Positive Score 300 metastasis in the clinic. Figure 6 RSK1 expression decreases during metastasis in lung adenocarcinoma patients. (a) and (b) Histoimmunopathological analysis of a human lung tissue microarray containing isogenically matched primary and metastatic lung adenocarcinoma lesions Discussion (n ¼ 100 patients). (a) Comparison of RSK1 staining score between primary and secondary sites. ‘I & II neg’; both primary and secondary lesions are negative for RSK1. ‘I>II’; staining higher in Lung cancer like other common neoplasms is fatal for primary than secondary sites. ‘I o II’; staining lower in primary several reasons, including the dissemination of tumour than secondary sites. (b) Percentage of RSK1 scores in the primary to multiple sites. Therefore, understanding the molecu- (‘I’) and the secondary (‘II’) tumours of cases with positive RSK1 lar mechanisms responsible for metastasis should result staining. Error bars are s.e.m. P value from Student’s t-test. (c) Number of secondary lesions per patients depending on RSK1 in new therapeutic approaches to prevent, and biomar- positive or negative staining in the primary tumour. Percent of kers to predict, this process. Recently, an siRNA screen cases are indicated. (d) Examples of scoring for RSK1 staining in for new proteins and kinases that regulate migration of adenocarcinoma tissue samples. RSK1 staining was scored on the the normal breast epithelial cell line MCF10A was basis of both the percentage of positive cells (0–100) and the intensity of the staining (0–3, corresponding to negative, weak, published (Simpson et al., 2008). Although the results moderate and strong positivity). The range of possible scores was generated were informative for normal cell migration, thus 0–300. the value of these findings for neoplastic cells and their metastatic potential is less clear. To gain new insight into cancer metastasis, we have revealed, including the RSKs. Downregulation of RSK1 performed an siRNA library screen and identified over dramatically increased cell migration, wheras silencing 50 kinases previously unknown to regulate the migra- of RSK2 and RSK4 impaired this process. These effects tion of lung cancer cells. Several families of kinases were correlated with changes in the actin cytoskeleton

Oncogene RSK1 in lung cancer cell metastasis R Lara et al 3519 (Figure 3b). However, only RSK1 downregulation Potential reasons for this discrepancy might include cross-validated in both a secondary invasion screen differences between cell types and their genetic back- and multiple NSCLC cell lines (Figure 2 and Supple- ground, as well as the experimental approaches used. mentary Figure 4). However, our results clearly demonstrate that at least So how might RSK1 link to the motility machinery? for NSCLC, downregulation of RSK1 enhances meta- A combination of bioinformatic and biochemical static potential in vitro and in vivo and correlates with analysis (Figure 3) revealed that RSK1 interacts with increased disease dissemination in patients. two well-known regulators of actin dynamics, VASP and Mena (Krause et al., 2003). We show that silencing of either VASP or Mena decreased cell migration speed Materials and methods in A549 cells. Moreover, our results demonstrated that co-silencing of these proteins with RSK1 abolished the promigratory and cytoskeletal effects of selective A549 cells were cultured in Dulbecco’s modied Eagle’s medium RSK1 downregulation (Figures 4a–c). It is tempting to (DMEM)/10% foetal bovine serum and LNM35 and speculate that RSK1 might modulate the function of H23 NSCLC cells in RPMI/10% foetal bovine serum at VASP and/or Mena by directly or indirectly phosphory- 37 1C, 10% CO2. lating them. Indeed, using purified proteins, we show that RSK1 phosphorylates VASP. Three principal phosphor- RNAi transfection ylation sites (S157, S239 and T278) have been described In all, 1 Â 103 A549 cells were plated in 96-well plates for 24 h on VASP that regulate its activity and for which before transfection with the relevant siRNA (35 nM)using antibodies are available (Benz et al., 2009). Amongst Oligofectamine (Invitrogen, Carlsbad, CA, USA). The siRNA these, S239 and T278 lie within an AGC kinase consensus library was from Dharmacon (Lafayette, CO, USA). Each protein sequence and could potentially be targeted by RSKs. So was targeted with a mix of four sequences (SmartPool). Cells were then incubated at 37 1C/10% CO for 48 h for target silencing. how does RSK1 effect migration? Two lines of evidence 2 suggest that T278 may be important as follows: (1) our Plasmid DNA transfection MS results from in vitro kinase assays show that RSK1 5 but not RSK2 or RSK4 induces phosphorylation of In all, 1 Â 10 A549 cells were plated in 6-well plates for 24 h before transfection with the relevant plasmids using Lipofec- this site (Supplementary Figure 5); (2) RSK1 silencing tamine Plus (Invitrogen). Wild-type and kinase-active human only reduces phosphorylation on T278 in intact cells RSK1 constructs were kindly provided by the laboratory of stimulated by serum (Figure 3f). In keeping with this Dr JM Slingerland (University of Miami Sylvester Compre- hypothesis, the phosphorylation of T278 was shown to hensive Cancer Center, Miami, FL, USA). These were impair VASP-driven actin polymerisation (Benz et al., co-transfected with a GFP-H2B encoding vector to allow 2009). One prediction of these results is that VASP identification of transfected cells under fluorescent microscopy. and possibly Mena might mediate the enhanced migra- At 24 h following transfection, individual wells were either tion seen with RSK1 knockdown. Indeed, our results lysed for western blotting of RSK1 or used for cell motility support this notion as VASP knockdown partially assays (see below). reversed, whereas combined silencing of VASP and Mena completely abolished the effects of RSK1 silencing on cell Cell motility assay migration (Figure 4). Additional experimental work is In the case of siRNA-based experiments, 48 h post siRNA now warranted to test the importance of the T278 transfection, cells were fluorescently stained using cell tracker phosphosite and the role of Mena in the migratory effects green (Molecular Probe, Eugene, OR, USA) in phenol red-free controlled by RSK1. DMEM/10% foetal calf serum. In the case of overexpression experiments, 24 h post transfection with the mix of the relevant So does RSK1 mediate metastasis in vivo? Zebrafish vector together with the GFP-H2B tracking vector, transfec- models have been used in the past as a versatile system tion efficiency was controlled under fluorescent microscopy. to study the metastatic process (Lee et al., 2009; Time-lapse imaging was performed for 18 h (1 image/10 min) Marques et al., 2009; Weiss et al., 2009). In the present using a Nikon microscope driven by Metamorph (Molecular report, we have established a zebrafish model for human Devices, Chicago, IL, USA). Each condition was acquired in lung cancer cell invasion and dissemination. In this quadruplicate. Sixty cells per condition for siRNA experiments system A549 cells form a tight tumour mass in the head or all green fluorescent cells in the case of overexpression were of the developing zebrafish and only slowly invade and tracked using Metamorph and tracks analysed using a disseminate down the body towards the tail. Down- previously published Mathematica (Wolfram Research, Cham- regulation of RSK1 markedly increased this process paign, IL, USA) notebook (Katso et al., 2006). (Figure 5). Moreover, comparison of RSK1 protein levels in primary human lung cancers and isogenically Actin cytoskeleton staining matched metastases revealed that decreased RSK1 Cells in 96-well plates were fixed with a 4% buffered para- expression correlated with increased numbers of sec- formaldehyde solution for 20 min, permeabilized with 0.1% Triton X-100 in phosphate-buffered saline for 5 min, and ondary lesions (Figure 6). blocked using 3% bovin serum albumin in phosphate-buffered It is noteworthy that a previous report in melanoma saline. Actin was stained with Alexa Fluor 488-Phalloidin cells indicates that overexpression of kinase-active (Molecular Probe) and nuclear DNA revealed using DAPI RSK1 promotes in vitro migration (Larrea et al., (Molecular Probe). Images were acquired for 36 fields per well 2009). These findings contrast with the present study. using an ImageXpress high-content imager (Molecular Devices).

Oncogene RSK1 in lung cancer cell metastasis R Lara et al 3520 Reverse transcriptase–PCR validation of target downregulation were fixed in 4% paraformaldehyde and cellular actin revealed Total cellular mRNA was extracted using RNAeasy (Qiagen, using TRITC-phalloidin (Sigma, Steinhelm, Germany). Nine Hilden, Germany) and used for reverse transcriptase–PCR stacks of 150 mm(4mm intervals) were acquired per well with a using Qiagen’s one-step reverse transcriptase–PCR kit. Reverse Zeiss invert confocal microscope. Stacks were analysed with transcriptase–PCR for glyceraldehyde-3-phosphate dehydrogenase Mathematica (Wolfram Research). siRNAs targeting 4.1B was used as internal control. For sequences used for each target, were used as positive control (Cavanna et al., 2007). see Supplementary Figure 7. Zebrafish metastatic model Western blotting and co-immunoprecipitation Zebrafish embryos at 4 h post fertilisation were microinjected Cellular proteins were extracted using a lysis buffer (LB ¼ 0.5% under stereomicroscopic observation with Histone 2B-GFP Triton X-100, 150 mM NaCl, 2 mM EDTA, 10% glycerol expressing A549 cells into the cell mass using a 12 mm guage supplemented with protease inhibitors cocktail tablets (Roche borosilicate pipettes mounted on a Narishige microinjector. Diagnostics, Basel, Switzerland), 10 mM b-glycerophosphate, The embryos were then imaged under anaesthetic by wide field 1mM sodium orthovanadate, 10 mM sodium fluoride). Equal fluorescent microscopy at 48 h post fertilisation and the protein amounts were analysed by SDS–polyacrylamide gel presence of A549 cells in three different compartments (head, electrophoresis/western blotting using the relevant antibodies. middle and tail), recorded. H2B-GFP expression had no effect on baseline motility or invasiveness of A549 cells (data not Bioinformatic analysis shown). A human protein–protein interaction database was compiled from two previously published databases (Jonsson and Bates Immunohistochemistry (IHC) 2006; Linding et al., 2007) containing both predicted and For the syngenic primary tumour/metastasis tissue microarray, demonstrated interactions. The database was queried using 100 lung cancer cases (Ethic Ref: 06/Q0406/154) were Python programming and Cytoscape for first–node inter- considered suitable on the basis of histotype confirmation actions (http://www.cytoscape.org). and immunostaining for pan-cytokeratin MNF116 and Vimentin. Tissue microarrays were prepared by obtaining In vitro kinase assay three 1 mm cores from the most representative and better- In all, 1 mg recombinant VASP (Origene, Rockville, MD, USA) preserved areas of the tumours and re-embedded in microarray was incubated on ice in 5 mM 3-(N-morpholino)propanesulfonic blocks. Immunohistochemistry was performed using an RSK1 acid (pH 7.2), 2.5 mM b-glycerophosphate, 1 mM ethylene glycol antibody (sc-231, Santa Cruz, Santa Cruz, CA, USA) at 1/1200 tetraacetic acid, 0.4 mM EDTA, 5 mM MgCl2 and 0.05 mM dithio- dilution. A semiquantitative immunohistochemical score was threitol in the presence or absence of 10 ng of recombinant active used, as previously described (Herberger et al., 2007). Signal GST-RSK1 (Cell Signaling, Danvers, MA, USA). The reaction specificity was determined using A549 cells treated with and was started by adding an ATP-mix resulting in a final concentra- without RSK1 siRNA and embedded in paraffin (Supplemen- tion of 100 mM ATP with or without 10 nCi/ml; [g-32P]ATP and tary Figure 6A). For representative staining see Supplementary incubated for 30 min at 30 1C. As a positive control, 1 mgrecom- Figure 6B. For further details see Supplementary Data. binant GST-S6 was used as a substrate for RSK1. Reactions were terminated in SDS-sample buffer and analysed by SDS- polyacrylamide gel electrophoresis and autoradiography. Conflict of interest

Invasion assay The authors declare no conflicts of interest. In all, 3 Â 103 A549 cells were plated in 96-well plates for 24 h before RNAi transfection. These were covered with a final concentration of 3 mg/ml of rat tail Collagen I (BD Bioscience, Franklin Lakes, NJ, USA) in DMEM/1% foetal calf serum Acknowledgements 24 h later. After 4 h, 25 ml of DMEM/1% foetal calf serum were placed above the gel. 25 ml of DMEM/1% foetal calf This work was supported by grants from Cancer Treatment serum with or without 200 ng/ml epidermal growth factor were and Research Trust, Cancer Research UK, Department of added 24 h later. Cells were allowed to invade for 48 h. The gels Health and our Experimental Cancer Medicine Centre.

References

Anjum R, Blenis J. (2008). The RSK family of kinases: emerging roles skeleton assembly by phosphorylating vasodilator-stimulated phos- in cellular signalling. Nat Rev Mol Cell Biol 9: 747–758. phoprotein. J Biol Chem 282: 4601–4612. Benz PM, Blume C, Seifert S, Wilhelm S, Waschke J, Schuh K et al. Cavanna T, Pokorna E, Vesely P, Gray C, Zicha D. (2007). Evidence (2009). Differential VASP phosphorylation controls remodeling of for protein 4.1B acting as a metastasis suppressor. J Cell Sci 120: the actin cytoskeleton. J Cell Sci 122: 3954–3965. 606–616. Blenis J, Chung J, Erikson E, Alcorta DA, Erikson RL. (1991). Cho YY, Yao K, Kim HG, Kang BS, Zheng D, Bode AM et al. (2007). Distinct mechanisms for the activation of the RSK kinases/ Ribosomal S6 kinase 2 is a key regulator in tumor promoter induced MAP2 kinase/pp90rsk and pp70-S6 kinase signaling systems are cell transformation. Cancer Res 67: 8104–8112. indicated by inhibition of protein synthesis. Cell Growth Differ Clark DE, Errington TM, Smith JA, Frierson Jr HF, Weber MJ, 2: 279–285. Lannigan DA. (2005). The serine/threonine protein kinase, p90 Blume C, Benz PM, Walter U, Ha J, Kemp BE, Renne T. (2007). ribosomal S6 kinase, is an important regulator of prostate cancer AMP-activated protein kinase impairs endothelial actin cyto- cell proliferation. Cancer Res 65: 3108–3116.

Oncogene RSK1 in lung cancer cell metastasis R Lara et al 3521 Collins CS, Hong J, Sapinoso L, Zhou Y, Liu Z, Micklash K et al. Lee SL, Rouhi P, Dahl Jensen L, Zhang D, Ji H, Hauptmann G et al. (2006). A small interfering RNA screen for modulators of tumor cell (2009). Hypoxia-induced pathological angiogenesis mediates tumor motility identifies MAP4K4 as a promigratory kinase. Proc Natl cell dissemination, invasion, and metastasis in a zebrafish tumor Acad Sci U S A 103: 3775–3780. model. Proc Natl Acad Sci U S A 106: 19485–19490. Dertsiz L, Ozbilim G, Kayisli Y, Gokhan GA, Demircan A, Kayisli Linding R, Jensen LJ, Ostheimer GJ, van Vugt MA, Jorgensen C, UA. (2005). Differential expression of VASP in normal lung tissue Miron IM et al. (2007). Systematic discovery of in vivo phosphory- and lung adenocarcinomas. Thorax 60: 576–581. lation networks. Cell 129: 1415–1426. Fox BP, Kandpal RP. (2009). EphB6 receptor significantly alters Marques IJ, Weiss FU, Vlecken DH, Nitsche C, Bakkers J, Lagendijk invasiveness and other phenotypic characteristics of human breast AK et al. (2009). Metastatic behaviour of primary human tumours carcinoma cells. Oncogene 28: 1706–1713. in a zebrafish model. BMC Cancer 9: 128. Gertler FB, Niebuhr K, Reinhard M, Wehland J, Soriano P. (1996). Nobes CD, Hall A. (1995). Rho, rac and cdc42 GTPases: regulators of Mena, a relative of VASP and Drosophila Enabled, is implicated actin structures, cell adhesion and motility. Biochem Soc Trans 23: in the control of microfilament dynamics. Cell 87: 227–239. 456–459. Gurzu S, Jung I, Prantner I, Ember I, Pavai Z, Mezei T. (2008). The Pardo OE, Arcaro A, Salerno G, Tetley TD, Valovka T, Gout I et al. expression of cytoskeleton regulatory protein Mena in colorectal (2001). Novel cross talk between MEK and S6K2 in FGF-2 induced lesions. Rom J Morphol Embryol 49: 345–349. proliferation of SCLC cells. Oncogene 20: 7658–7667. Han G, Fan B, Zhang Y, Zhou X, Wang Y, Dong H et al. (2008). Pardo OE, Wellbrock C, Khanzada UK, Aubert M, Arozarena I, Positive regulation of migration and invasion by vasodilator- Davidson S et al. (2006). FGF-2 protects small cell lung cancer cells stimulated phosphoprotein via Rac1 pathway in human breast from apoptosis through a complex involving PKCepsilon, B-Raf cancer cells. Oncol Rep 20: 929–939. and S6K2. Embo J 25: 3078–3088. HerbergerB,PuhallaH,LehnertM,WrbaF,NovaleS,BrandstetterA Philippar U, Roussos ET, Oser M, Yamaguchi H, Kim HD, et al. (2007). Activated mammalian target of rapamycin is an Giampieri S et al. (2008). A Mena invasion isoform potentiates adverse prognostic factor in patients with biliary tract adenocarcino- EGF-induced carcinoma cell invasion and metastasis. Dev Cell 15: ma. Clin Cancer Res 13: 4795–4799. 813–828. Hu LD, Zou HF, Zhan SX, Cao KM. (2008). EVL (Ena/VASP-like) Simpson KJ, Selfors LM, Bui J, Reynolds A, Leake D, Khvorova A expression is up-regulated in human breast cancer and its relative et al. (2008). Identification of genes that regulate epithelial cell expression level is correlated with clinical stages. Oncol Rep 19: migration using an siRNA screening approach. Nat Cell Biol 10: 1015–1020. 1027–1038. Irie HY, Pearline RV, Grueneberg D, Hsia M, Ravichandran P, Smith JA, Poteet-Smith CE, Xu Y, Errington TM, Hecht SM, Kothari N et al. (2005). Distinct roles of Akt1 and Akt2 in Lannigan DA. (2005). Identification of the first specific inhibitor of regulating cell migration and epithelial-mesenchymal transition. p90 ribosomal S6 kinase (RSK) reveals an unexpected role for RSK J Cell Biol 171: 1023–1034. in cancer cell proliferation. Cancer Res 65: 1027–1034. Ivaska J, Pallari HM, Nevo J, Eriksson JE. (2007). Novel functions of Swanton C, Marani M, Pardo O, Warne PH, Kelly G, Sahai E et al. vimentin in cell adhesion, migration, and signaling. Exp Cell Res (2007). Regulators of mitotic arrest and ceramide metabolism are 313: 2050–2062. determinants of sensitivity to and other chemotherapeutic Jonsson PF, Bates PA. (2006). Global topological features of cancer drugs. Cancer Cell 11: 498–512. proteins in the human interactome. Bioinformatics 22: 2291–2297. Toyoda A, Kawana H, Azuhata K, Yu J, Omata A, Kishi H et al. Kang S, Dong S, Gu TL, Guo A, Cohen MS, Lonial S et al. (2007). (2009). Aberrant expression of human ortholog of mammalian FGFR3 activates RSK2 to mediate hematopoietic transformation enabled (hMena) in human colorectal carcinomas: implications for through tyrosine phosphorylation of RSK2 and activation of the its role in tumor progression. Int J Oncol 34: 53–60. MEK/ERK pathway. Cancer Cell 12: 201–214. Trichet L, Sykes C, Plastino J. (2008). Relaxing the actin cytoskeleton Katso RM, Pardo OE, Palamidessi A, Franz CM, Marinov M, for adhesion and movement with Ena/VASP. J Cell Biol 181: De Laurentiis A et al. (2006). Phosphoinositide 3-Kinase 19–25. C2beta regulates cytoskeletal organization and cell migration via Weiss FU, Marques IJ, Woltering JM, Vlecken DH, Aghdassi A, Rac-dependent mechanisms. Mol Biol Cell 17: 3729–3744. Partecke LI et al. (2009). Retinoic acid receptor antagonists inhibit Krause M, Dent EW, Bear JE, Loureiro JJ, Gertler FB. (2003). Ena/ miR-10a expression and block metastatic behavior of pancreatic VASP proteins: regulators of the actin cytoskeleton and cell cancer. Gastroenterology 137: 2136–2145 e2131–2137. migration. Annu Rev Cell Dev Biol 19: 541–564. Woo MS, Ohta Y, Rabinovitz I, Stossel TP, Blenis J. (2004). Kwiatkowski AV, Gertler FB, Loureiro JJ. (2003). Function and Ribosomal S6 kinase (RSK) regulates phosphorylation of filamin regulation of Ena/VASP proteins. Trends Cell Biol 13: 386–392. A on an important regulatory site. Mol Cell Biol 24: 3025–3035. Larrea MD, Hong F, Wander SA, da Silva TG, Helfman D, Lannigan Zimmermann J, Labudde D, Jarchau T, Walter U, Oschkinat H, D et al. (2009). RSK1 drives p27Kip1 phosphorylation at T198 to Ball LJ. (2002). Relaxation, equilibrium oligomerization, and promote RhoA inhibition and increase cell motility. Proc Natl Acad molecular symmetry of the VASP (336–380) EVH2 tetramer. Sci U S A 106: 9268–9273. Biochemistry 41: 11143–11151.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene