Differential roles and regulation of the protein kinases PAK4, PAK5 and PAK6 in melanoma cells

The protein kinases PAK4, PAK5 and PAK6 comprise a family of ohnologues. In multiple cancers including melanomas PAK5 most frequently carries non-synonymous mutations; PAK6 and PAK4 have fewer; and PAK4 is often amplified. To help interpret these genomic data, initially we compared the cellular regulation of the sister kinases and their roles in melanoma cells. In common with many ohnologue protein kinases, PAK4, PAK5 and PAK6 each have two 14-3-3-binding phosphosites of which phosphoSer99 is conserved. PAK4 localises to the leading edge of cells in response to phorbol ester-stimulated binding of 14-3-3 to phosphoSer99 and phosphoSer181, which are phosphorylated by two different PKCs or PKDs. These phosphorylations of PAK4 are essential for its phorbol ester-stimulated phosphorylation of downstream substrates. In contrast, 14-3-3 interacts with PAK5 in response to phorbol ester-stimulated phosphorylation of Ser99 and epidermal growth factor-stimulated phosphorylation of Ser288; whereas PAK6 docks onto 14-3-3 and is prevented from localising to cell–cell junctions when Ser133 is phosphorylated in response to cAMP-elevating agents via PKA and insulin-like growth factor 1 via PKB/Akt. Silencing of PAK4 impairs viability, migration and invasive behaviour of melanoma cells carrying BRAFV600E or NRASQ61K mutations. These defects are rescued by ectopic expression of PAK4, more so by a 14-3-3-binding deficient PAK4, and barely by PAK5 or PAK6. Together these genomic, biochemical and cellular data suggest that the oncogenic properties of PAK4 are regulated by PKC–PKD signalling in melanoma, while PAK5 and PAK6 are dispensable in this cancer.


Figure S2
Although the quantitation and linearity range of 14-3-3 overlays have not been explored, and these should be considered as semi-quantitative, here ImageJ analyses (Schneider CA, Rasband WS, Eliceiri KW. (2012) NIH Image to ImageJ: 25 years of image analysis. Nat Methods. 9, 671-675) of the following results are displayed, each given for n=3 and indicating the standard deviation. A. Image J analyses of data represented in Fig 2A. Gö6976 inhibits pSer181 phosphorylation and 14-3-3 binding to PAK4-GFP, whereas Gö6983 primarily inhibits phosphorylation of pSer99. B. Image J analyses of data represented in Fig 2E. Mutation of Ser288 inhibits binding of PAK5-GFP to 14-3-3, assessed by 14-3-3 overlays. C. ImageJ analyses of data represented in Fig

Figure S3
A. GFP-tagged PAK4 K350M S474A was immunoprecipitated from cells treated with kinase activators and inhibitors as indicated and binding to 14-3-3s was analysed. Phosphorylation state of Ser99 and Ser181 was analysed using phosphospecific antibodies. Cell lysates (30 µg) were blotted for pThr202/Tyr204 ERK, pSer473 PKB and pS157 VASP. B. SKMEL13 and SBcl2 melanoma cells were serum starved overnight and treated with MAPK and PKC/PKD inhibitors as indicated. Cell lysates from each condition were pre-cleared, incubated with 14-3-3 Sepharose beads overnight and eluted in sample buffer. The eluates were subjected to SDS-PAGE and blotted for PAK4. The phosphorylation state of PAK4 was analysed using pSer99 and pSer181 antibodies. Cell lysates (30 µg) were analysed with antibodies that recognise PAK4, pThr202/Tyr204 ERK and GAPDH.

Figure S4
A. PAK5-GFP was immunoprecipitated from lysates of cells serum starved overnight and treated with the combination of kinase activators and inhibitors as indicated. The immunoprecipitates were tested for binding to 14-3-3s. B. Cell lysates from HEK293 cells over expressing PAK5 wildtype and mutant for predicted 14-3-3 binding sites were immunoprecipitated and tested for their ability to bind directly to 14-3-3s in Far-Western overlay assay and by co-immunoprecipitation of endogenous 14-3-3s

Figure S5
Map of the 14-3-3 binding sites of group II PAK kinases showing the stimuli and kinases regulating phosphorylation of these sites.

Figure S6
U2OS cells were transfected with PAK4, PAK5 and PAK6 kinase-dead mutants. The cells were seeded onto coverslips 16 h post-transfection and allowed to attach overnight. Cells were fixed in paraformaldehyde, stained for actin (Alexafluor 594 Phalloidin), nucleus (DAPI) and mounted onto slides. The samples were imaged using LSM710 confocal microscopy; scale bar = 20 µm.

Figure S8
PAK4 and PAK6 knockdown SBcl2 cells were seeded in 96-well plates and cell proliferation was monitored over a period of 96 h using CellTiter AQueous non-radioactive cell proliferation assay kit.

Figure S9
Lysates from SKMEL13 cells, serum-starved overnight and pre-treated with Gö6976 and PF-3758309 followed by PMA, were analysed for phosphorylation of GEF H1, LIMK1 and cofilin with β-tubulin as the loading control.

Figure S10
The mRNA expression profile of group II PAK kinases in various melanoma cell lines carrying B-RAF V600E (SKMEL13, UACC62 and G361) and N-RAS Q61K (SBcl2, WM1361, MM485) driver mutations relative to 18S rRNA levels are represented. The mRNA from SH-SY5Y cell line (neuroblastoma) was used as the positive control for PAK5 expression.