Oncogene (2007) 26, 6071–6081 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE Protein WNK2 inhibits cell proliferation by negatively modulating the activation of MEK1/ERK1/2

S Moniz1, F Verı´ ssimo1, P Matos1, R Braza˜ o1, E Silva1, L Kotevelets2,3, E Chastre2,3, C Gespach4,5 and P Jordan1

1Centre of Human Genetics, Instituto Nacional de Sau´de ‘Dr Ricardo Jorge’, Lisbon, Portugal; 2INSERM U 773, Centre de Recherche Biome´dicale Bichat Beaujon CRB3, Paris, France; 3Universite´ Paris 7 Denis Diderot, site Bichat, Paris, France; 4INSERM U 673, Paris, France and 5Universite´ Pierre et Marie Curie-Paris 6, Faculte´ de Me´decine, Laboratory of Molecular and Clinical Oncology of Solid Tumors, Paris, France

The recently identified subfamily of WNK protein Introduction is characterized by a unique sequence variation in the catalytic domain and four related human WNK Protein kinases are critical components of signal were identified. Here, we describe the cloning and transduction pathways and form a superfamily of functional analysis of the human family member WNK2. around 518 genes in humans (Kostich et al., 2002; We show that the depletion of endogenous WNK2 Manning et al., 2002). A recently recognized subfamily, expression by RNA interference in human cervical HeLa the WNK (with no K ¼ lysine) protein kinases, has cancer cells led to the activation of the extracellular attracted much attention, as germline mutations in the signal–regulated kinase (ERK)1/2 mitogen-activated pro- WNK1 and WNK4 genes were found to cause pseudo- tein kinases but, in contrast to the depletion of WNK1, hypoaldosteronism type II, a hereditary form of had no effect on ERK5. Furthermore, expression of a hypertension (Wilson et al., 2001). WNK kinases are kinase-dead WNK2-K207M mutant also activated ERK1/ found in multicellular organisms and the 2 suggesting that WNK2 catalytic activity is required. contains four related WNK genes (Verı´ ssimo and Depletion of WNK2 expression increased G1/S progres- Jordan, 2001). They are unique among protein kinases sion and potentiated the cellular response to low epidermal in lacking the highly conserved catalytic lysine in growth factor concentrations. The molecular mechanism subdomain II (Hanks and Hunter, 1995), important of ERK1/2 activation in WNK2-depleted cells lies down- for the correct position of (ATP) stream of the Raf kinases and involves MEK1 phosphory- within the catalytic domain. An alternative lysine lation at serine 298 in both HeLa and HT29 colon cancer residue from subdomain I functionally reconstitutes cells. This modification is linked to the upregulation of the in WNKs (Xu et al., 2000; Min et al., MEK1 activity toward ERK1/2. Together, these results 2004), and is part of a characteristic WNK sequence provide evidence that WNK2 is involved in the modulation signature (Verı´ ssimo and Jordan, 2001). of growth factor–induced cancer cell proliferation through WNK4 is expressed in the distal and the MEK1/ERK1/2 pathway. The data identify WNK2 mutations that cause type II as a candidate tumor suppressor and suggest a affect the regulation of paracellular chloride (ClÀ) coordinated activity of WNK kinases in the regulation of permeability as well as the traffic of epithelial ion cell proliferation. channels such as the sodium chloride (Na þ -ClÀ) Oncogene (2007) 26, 6071–6081; doi:10.1038/sj.onc.1210706; cotransporter (NCCT) and the renal potassium channel published online 30 July 2007 ROMK (reviewed in Kahle et al., 2004; Gamba, 2005). WNK1 has been found to phosphorylate synaptotag- Keywords: cell cycle progression; EGF; ERK MAP kinase min-2, a protein involved in calcium-dependent mem- pathway; MEK1; WNK protein kinases brane fusion during exocytosis and endocytosis (Lee et al., 2004), and to activate the protein kinases SGK1 (Xu et al., 2005), SPAK and OSR1 (Moriguchi et al., 2005; Vitari et al., 2005; Gagnon et al., 2006), all of which regulate the activation of various renal ion channels, including the epithelial sodium channel. WNK kinases were further reported to participate in signal transduction pathways related to cell growth and Correspondence: Dr P Jordan, Centro de Gene´ tica Humana, Instituto survival. WNK1 was shown to interact with MEKK2 Nacional de Sau´ de ‘Dr Ricardo Jorge’, Avenida, Padre Cruz, Lisboa and MEKK3 and act as an upstream activator of the 1649-016, Portugal. E-mail: [email protected] extracellular signal–regulated kinase (ERK)-5 mitogen- Received 8 May 2007; revised 3 July 2007; accepted 4 July 2007; activated protein (MAP) kinase pathway (Xu et al., published online 30 July 2007 2004). WNK1 also serves as a substrate for protein WNK2 negatively modulates MEK1/ERK1/2 activation S Moniz et al 6072 kinase B/Akt upon the activation of phosphatidyl Results inositol (PI)-3-kinase (Vitari et al., 2004) and was found to negatively regulate insulin-stimulated mito- Cloning and expression of human WNK2 genesis in adipocytes (Jiang et al., 2005). WNK3 was We subcloned the human WNK2 cDNA 6894 bp shown to interact with procaspase-3 and its expression encoding a protein of 2297 amino acids and 243 kDa levels can modulate the apoptotic/cell survival balance calculated molecular weight (GenBank accession num- (Verı´ ssimo et al., 2006). ber AJ242724). Further details on the cloning proce- Here, we describe a functional analysis of human dure, on the exon–intron organization and alternative WNK2, the yet uncharacterized fourth family member, splicing of the WNK2 gene, as well as its homology with and demonstrate its role in counteracting ERK1/2 the other three WNK kinases are included as Supple- activation and epidermal growth factor (EGF)- mentary Information S1–S4. The WNK2 sequence mediated G1/S progression in HeLa human cervical contains the catalytic at the N- cancer cells. terminal end between residues 195 and 453 (Figure 1a),

Figure 1 Expression and activity of human WNK2. (a) Schematic representation of the WNK2 coding sequence containing 2297 amino acids. Depicted are the protein kinase domain, a domain (CC) and three regions of high homology shared with WNK1, WNK3 and WNK4. (b) Northern blot analysis of human WNK2 expression in various human tissues, showing strongest expression in heart and muscle followed by brain and small intestine. (c) Expression of endogenous WNK2 in various human cell lines analysed by reverse transcription–PCR (RT–PCR) (top panels) and western blot (middle panels). For RT–PCR, the respective cloned cDNA (‘pcDNA3-WNK2’) provided the positive control and reactions without cDNA (‘no cDNA’) the negative control. For western blots, transfection of Myc-WNK2 into HEK293 cells served as positive control. As loading controls, endogenous RNA polymerase II transcripts (pol II) amplified by RT–PCR or a-tubulin protein levels detected by western blot were used. A western blot analysis of various human tissue extracts revealed WNK2 in brain and muscle (bottom panel). (d) Specificity of the protein band detected by the anti-WNK2 serum. HT29 cells were transfected with either a control small interfering RNA (siGFP) or one of two different WNK2- specific small interfering RNAs (siRNAs) (a and b). At 24 h post-transfection, the expression of WNK2 was analysed by western blot (top panels) and RT–PCR (bottom panels). Note that the detected protein band as well as the WNK2 transcripts was specifically reduced in the presence of both WNK2-specific siRNAs. As internal controls, the amplification of RNA polymerase II transcripts (pol II) or the protein level of a-tubulin was used. (e) In vitro protein kinase activity of immunoprecipitated WNK2. Endogenous WNK2 was immunoprecipitated from HT29 cells (left panels), whereas ectopic WNK2 was isolated from HEK293 cells transfected with either Myc-WNK2, or kinase-dead mutant Myc-WNK2-K207M, or Myc empty vector (EV) (right panels). Cell lysates were prepared, immunoprecipitated with either anti-WNK2 (left panels) or anti-Myc (right panels) antibodies and assayed in the presence of radioactive adenosine triphosphate for autophosphorylation. Proteins were separated by SDS–PAGE, transferred to polyvinylidene fluoride membranes, exposed to an X-ray film and subsequently labelled with either anti-WNK2 (left panel, bottom) or anti-Myc (right panel, bottom) antibodies to document the levels of immunoprecipitated WNK2. (f) Subcellular localization of WNK2. HeLa cells were transfected with GFP-tagged WNK2 and 20 h later analysed by fluorescence microscopy.

Oncogene WNK2 negatively modulates MEK1/ERK1/2 activation S Moniz et al 6073 including the typical WNK kinase signature motif protein (GFP)-tagged WNK2 was predominantly (Verı´ ssimo and Jordan, 2001). Except for a C-terminal cytoplasmic (Figure 1f). coiled-coil domain (residues 1931–1959), the sequence does not reveal any homology to other functional WNK2 expression modulates activation of the ERK1/2 protein domains. The full-length WNK2 sequence MAP kinase also allows assignment of two previously reported A of the WNK2 catalytic domain cDNAs as incomplete fragments of WNK2, namely, with other human protein kinase domains revealed the the ‘serologically-defined colon cancer antigen 430 highest homology scores for MEKK3 (54%) and B-Raf or NY-CO3 (Scanlan et al., 1998) and the T-cell (50%), both involved in the MAP kinase activation. recognized pancreatic cancer cell antigen P/OKcl13 Moreover, WNK1 has previously been shown to (Ito et al., 2001). activate the ERK5 MAP kinase pathway (Xu et al., Northern blot analysis revealed a strong expression of 2004). We, therefore, compared the effect of WNK2 WNK2 in heart and skeletal muscle and a weaker signal and WNK1 on the activation of MAP kinases. The in brain and small intestine (Figure 1b). This is in expression of WNK2 and WNK1 in HeLa cells was agreement with the previous data showing that brain, suppressed by specific siRNAs and the levels of heart, small intestine and colon are the main WNK2- activated phosphorylated MAP kinases were analysed positive tissues by reverse transcription–PCR (RT– by western blot using specific antibodies. As shown in PCR) (Verı´ ssimo and Jordan, 2001). To identify the Figure 2a, the WNK1-specific siRNA did not affect relevant experimental models, WNK2 expression was WNK2 transcript levels, but reduced WNK1 expression further analysed by RT–PCR in various human cancer by 55% and inhibited the EGF-induced activation cell lines. WNK2 transcript and protein were detected in of ERK5, as previously reported (Xu et al., 2004). In HT29 (colon), HeLa (cervix carcinoma) and HEK293 contrast, the specific suppression of WNK2 had no (human embryonic ) cells, but were absent in effect on ERK5 activation (Figure 2a). Interestingly, the SW480 colorectal cells (Figure 1c). A WNK2-specific WNK2-specific siRNA remarkably increased the phos- antibody was raised against a peptide and did not cross- pho-ERK1/2 signal, whereas the depletion of WNK1 react with the remaining three WNK family members had no significant effect on ERK1/2 activation (see in Supplementary Information S5). This anti- (Figure 2b). In addition, two different WNK2-specific WNK2 serum detected ectopic expression of Myc- siRNAs reduced WNK2 expression on average by 59% WNK2 protein with an apparent molecular weight (n ¼ 14) and this increased ERK1/2 activation twofold, above 250 kDa in the transfected HEK293 cells. The with no change in total ERK1/2 (Figure 2c). In the same Myc-WNK2 band co-migrated with an endogenous cell lysates, the depletion of WNK2 expression had only protein in HT29, HeLa and HEK293 cells (Figure 1c; a marginal effect on the activation of Jun N-terminal middle panels), which could also be detected in human kinase (JNK) or p38 MAPK under the same growth tissue extracts from brain and muscle (Figure 1c; conditions (data not shown). bottom). The endogenous protein band in HT29 cells These results suggested that WNK2 and WNK1 are was specifically reduced after the transfection with involved in the regulation of two distinct MAP kinase two different WNK2-specific small interfering RNAs pathways. Whereas WNK1 is required for the activation (siRNAs), which also depleted the endogenous WNK2 of ERK5, WNK2 exerts an inhibitory role on the transcripts (Figure 1d). The protein kinase activity of activation of ERK1/2. transfected or endogenous WNK2 was determined by in vitro kinase assays following the immunoprecipitation. These experiments revealed the autophosphorylation of WNK2 silencing potentiates the activation of ERK1/2 by endogenous WNK2 isolated from HT29 cells as well as EGF in HeLa cells of overexpressed Myc-WNK2 (Figure 1e). A kinase- To further analyse the effect of WNK2 on the ERK1/2 dead Myc-WNK2 K207M protein was engineered by pathway, the kinase-dead K207M mutant was over- the mutation of the catalytic lysine residue that expressed in HeLa cells. An increase in the activation of characterizes the WNK kinase family (Xu et al., 2000; ERK1/2 was observed upon the expression of Myc- Verı´ ssimo and Jordan, 2001). Following the transfection WNK2-K207M under the standard growth conditions into HEK293 cells, the immunoprecipitated mutant (Figure 2d), suggesting that the catalytic activity of protein revealed a weak but consistent autophosphory- WNK2 is required for the negative regulation of ERK1/ lation, indicating that endogenous WNK2 or another 2 activation. Interestingly, WNK2-K207M had no protein kinase co-precipitated with the WNK2 mutant activating effect on ERK1/2 in serum-starved cells. protein (Figure 1e). In agreement with the previous The transfection of wild-type WNK2 had no effect reports showing that myelin basic protein (MBP) is a under the standard growth conditions but inhibited poor substrate for immunoprecipitated full-length ERK1/2 activation following the stimulation with EGF WNK1 kinase (Vitari et al., 2004; Verı´ ssimo et al., at low concentration (Figure 2e). 2006), we also observed that immunoprecipitated Myc- To test whether WNK2 affected ERK1/2 activation WNK2 had no significant kinase activity toward MBP directly, cells were incubated under the conditions that (data not shown), and a suitable WNK2 substrate inhibit the ERK cascade. As shown in Figure 3a, growth remained to be identified. The subcellular localization factor depletion due to serum starvation abolished of WNK2 in HeLa cells expressing green fluorescent ERK1/2 activation in the presence of WNK2 siRNAs.

Oncogene WNK2 negatively modulates MEK1/ERK1/2 activation S Moniz et al 6074

Figure 2 Depletion of endogenous WNK2 activates the extracellular signal-regulated kinase (ERK)-1/2 mitogen-activated protein kinase. (a) HeLa cells were transfected with control (siGFP), WNK2-specific (siWNK2) or WNK1-specifc (siWNK1) small interfering RNAs (siRNAs) and 24 h later analysed by western blot for the activation of ERK5 following the stimulation of cells with 1 ng/ml epidermal growth factor (EGF) for 15 min (top panels). Note that upon suppression of WNK1, but not of WNK2, the phosphorylation of ERK5 by EGF is inhibited. The reverse transcription–PCR (RT–PCR) analysis (bottom panels) shows specificity of the two WNK siRNAs. (b) HeLa cells were transfected with control (siGFP), WNK2-specific (siWNK2) or WNK1-specifc (siWNK1) siRNAs and 24 h later analysed by western blot for the activation of ERK1/2. Note the activation of ERK1/2 upon depletion of WNK2, but not of WNK1 transcripts. (c) Quantification of the activation of ERK1/2 in WNK2-depleted cells. A representative western blot comparing the total ERK1/2 levels with the levels of phosphoT183/Y185 ERK1/2 is shown (top panel). Transfection with either WNK2-specific siWNK2-a or siWNK2-b revealed a mean ERK1/2 activation of about 2.1-fold compared to the control siRNA (middle panel), where the mean residual expression of WNK2 was 41% (bottom panel). Data are presented as the mean7s.d. of seven independent experiments for each siRNA (n ¼ 14). (d) The activation of ERK1/2 was further analysed by comparing the total ERK1/2 with the phospho-T183/Y185 ERK1/2 levels in HeLa cells previously transfected with either the kinase-dead Myc-WNK2 K207M mutant or a control empty vector (EV). Under the standard growth conditions in the presence of serum, WNK2-K207M increased ERK1/2 activation but had no stimulatory effect on serum-starved cells. (e) The activation of ERK1/2 was analysed as described above in HeLa cells overexpressing wild-type Myc-WNK2 or the corresponding control empty vector. In the presence of serum, the overexpression of WNK2 had no effect on ERK1/2 activation but following the stimulation of cells with 0.1 ng/ml EGF for 5 min ERK1/2, activation was inhibited by the presence of Myc-WNK2.

In addition, the two MEK1/2 inhibitors PD98059 and WNK2 modulates DNA synthesis and cell proliferation U0126 prevented any increase in ERK activation A well-described biological effect of increased ERK1/2 following the depletion of WNK2. Our data indicate activation is cell cycle progression. To determine that WNK2 is not acting directly upon ERK1/2 but whether the depletion of WNK2 and its consequent rather modulates its upstream stimulus-dependent acti- impact on ERK1/2 activation had indeed a phenotypic vation. To further sustain this interpretation, the dose– relevance, HeLa cells were transfected with control or response of ERK1/2 activation by EGF was analysed in WNK2-specific siRNAs, grown for 24 h and then pulse- relation to WNK2 expression levels. Figure 3b shows labelled for 30 min with bromodeoxyuridine (BrdU) to that depletion of WNK2 in serum-starved cells poten- detect cells in S-phase by immunofluorescence micro- tiated about 10-fold the concentration-dependent scopy (Figure 4a). As shown in Figure 4b, the average ERK1/2 activation following EGF stimulation. At number of cells in S-phase was about 26% in cells higher concentrations of EGF (100 ng/ml), the effect of transfected with control siRNA. Notably, this percen- WNK2 became less pronounced (data not shown). tage increased 1.6-fold to 42.5% in cells treated with Taken together, our data demonstrate that WNK2 is a WNK2-specific siRNAs. This significant increase in negative modulator of the EGF-stimulated ERK1/2 S-phase cells (P ¼ 0.0003) was observed when WNK2 MAP kinase pathway. expression was reduced by a mean value of 50% (n ¼ 3)

Oncogene WNK2 negatively modulates MEK1/ERK1/2 activation S Moniz et al 6075

Figure 3 WNK2 silencing upregulates serum and epidermal growth factor (EGF)-induced extracellular signal-regulated kinase (ERK)1/2 activation. (a) HeLa cells were transfected with control (siGFP) or WNK2-specific (siWNK2) small interfering RNAs (siRNAs) and ERK1/2 activation was compared 24 h later in the presence and absence of either the serum or the MEK inhibitors U0126 and PD98059. A representative western blots for activated and total ERK1/2 as well as a reverse transcription–PCR (RT–PCR) control of the WNK2 depletion (right panel) are shown. (b) Depletion of WNK2 potentiates EGF-induced ERK activation. HeLa cells were transfected as above, starved before the treatment with the indicated concentrations of EGF and the resulting ERK1/2 activation was compared by western blot. Note that in WNK2-depleted cells, EGF was 10 times more potent to produce the same level of ERK1/2 activation as compared to control siGFP-treated cells.

(Figure 4c). Under these conditions, western blot cells and assayed in vitro for their kinase activity using analysis revealed an increase in the protein level of recombinant MEK as substrate. Again, WNK2 deple- cyclin D1 (Figure 4b) and the total cell number tion had no influence on control or EGF-induced increased by 25% (P ¼ 0.027). These data indicate that stimulation of B-Raf and Raf-1 (Figures 5b and c). WNK2-depleted cells passed the G1 restriction point, Importantly and in accordance with our previous results entered S-phase and then completed the mitosis. The (Figures 2 and 3), a clear increase in ERK1/2 or MEK1/ results confirm an inhibitory role of WNK2 on the 2 activity was detected in the same cell lysates as a result ERK1/2 MAP kinase activation and downstream cell of WNK2 depletion (Figures 5a–c). proliferation. Next, the activation of MEK1/2 was analysed using phospho-specific antibodies that recognize phosphoryla- tion of serines 218 and/or 222 in the activation loop as a WNK2 depletion activates the ERK/MAP kinase surrogate measure of MEK1/2 activity (Alessi et al., pathway via phosphorylation of MEK1 at serine 298 1994; Zheng and Guan, 1994). In four independent To determine which step of the signalling cascade is experiments, we observed that two different siRNAs regulated by WNK2, HeLa cells depleted of WNK2 (WNK2-a and -b) reduced endogenous WNK2 levels by were analysed for EGF receptor (EGFR) internalization on average 50% and upregulated about twofold the and activation of its downstream components, the Ras activation status of MEK1/2 (Figure 5d) and its GTPase and the kinases B-Raf, Raf-1 and MEK1/2. immediate downstream target ERK1/2. This indicated Following the addition of EGF to cultured HeLa that WNK2 affected MEK1/2 activation in a Raf- cells, the internalization of the EGF–EGFR complex independent manner. Such a mechanism was previously was followed in a time-course experiment at 371Cby shown to originate from the phosphorylation of MEK1 immunofluorescence microscopy. No difference in at Ser-298. Accordingly, both HeLa and HT29 cells EGF internalization was observed between the cells showed increased levels of MEK1 phosphorylation at transfected with control or WNK2-specific siRNAs (see Ser-298 after WNK2 depletion (Figure 5e). As expected, Supplementary Information S6). Next, the guanosine the MEK1 inhibitor U0126 abolished ERK1/2 activa- triphosphate (GTP)-bound, activated pool of Ras was tion in siWNK2-treated cells, but the phosphorylation analysed. Whereas the stimulation of cells with EGF led status of MEK at Ser-298 was still observed (not to a clear increase in the level of activated Ras, the shown). Since HT29 colon cancer cells carry an depletion of WNK2 had no stimulating effect on the activating B-Raf-V600E mutation (Davies et al., 2002), cells neither under the standard growth conditions nor the observed increase in MEK1 Ser-298 induced by after the EGF stimulation. Then, endogenous B-Raf or WNK2 depletion in this model confirms our interpreta- Raf-1 were immunoprecipitated from siRNA-treated tion that the mechanism involved lies downstream of

Oncogene WNK2 negatively modulates MEK1/ERK1/2 activation S Moniz et al 6076 B-Raf and involves the signalling component MEK1, upstream of ERK1/2.

Discussion

The subfamily of WNK protein kinases was recently identified and the human genome encodes four distinct WNK genes. This paper describes the cloning and functional analysis of the yet uncharacterized family member WNK2. As a major finding, we demonstrate that WNK2 negatively regulates the EGF-induced activation of the ERK/MAPK-pathway and the down- stream cell cycle progression. The EGF-induced Ras-Raf-MEK-ERK MAP kinase pathway is evolutionary conserved and controls multiple pathophysiological cellular processes such as prolifera- tion, survival or differentiation. For the regulation of these processes the level of ERK activation in cells plays a critical role and determines the signalling responses to several growth factors and oncogenic pathways. For example, quiescent fibroblasts begin to proliferate following sustained, but not transient, ERK activation (Dobrowolski et al., 1994; Balmanno and Cook, 1999; Murphy et al., 2004). Moreover, the amplitude of ERK1/2 signalling must be correctly modulated, be- cause if it is too strong, cells will stop cycling and will differentiate or senesce (Marshall, 1995; Sewing et al., 1997; Woods et al., 1997; Kerkhoff and Rapp, 1998). Therefore, the signalling output depends on a fine tuning of ERK1/2 activity in normal versus transformed cells (Ebisuya et al., 2005), generating regulatory flexibility of the pathway. For example, scaffold proteins such as Sur-8 (Li et al., 2000) or kinase suppressor of Ras (KSR) recruit cascade components and facilitate signal propagation, thereby potentiating ERK activation (Morrison and Davis, 2003). In contrast, the depletion of the impedes mitogenic signal propagation (IMP) scaffold protein by siRNAs was found to result in increased and stimulus-dependent MEK activation (Matheny et al., 2004). Such a thresh- Figure 4 Depletion of endogenous WNK2 increases the percen- old modulation was also observed in our experiments, tage of HeLa cells in S-phase. Cells were transfected with either control (siGFP) or one of the two WNK2-specific siWNK2-a or indicating that WNK2 participates in the fine-tuning of siWNK2-b small interfering RNAs (siRNAs) and 24 h later pulse- ERK1/2 activation as a negative modulator. labelled for 30 min with BrdU, allowing its incorporation into In particular, we found that overexpression of WNK2 replicating DNA. (a) Representative fluorescence microscopy raised the threshold for ERK1/2 activation in EGF- images showing HeLa cells in S-phase as BrdU-positive nuclei stimulated cells (Figure 2e), whereas WNK2 depletion and the total number of cells as 4,6-diamidino-2-phenylindole (DAPI)-stained nuclei. (b) The percentage of BrdU-positive cells rendered cells more sensitive to low concentrations of was counted in seven randomly chosen fields and the mean7s.d. EGF and increased ERK1/2 activation with subsequent values obtained in three independent RNAi experiments (n ¼ 3) cell cycle progression (Figures 3b and 4). This effect were plotted. At least 1000 cells were counted for each siRNA requires an active WNK2 catalytic domain because the treatment. Below the graph, a western blot analysis shows the expression levels of cyclin D1 and a-tubulin (loading control). Note kinase-inactive WNK2 K207M mutant also increased the increase in S-phase cells from 26 to 42.5% following the ERK1/2 signalling (Figure 2d). We show that several depletion of WNK2 using either siWNK2-a or siWNK2-b. (c) The signalling components upstream of MEK1 are not efficiency of WNK2 depletion was quantified by reverse transcrip- involved in the WNK2 molecular mechanism. These tion–PCR (RT–PCR) and an average residual expression of 51% include the endocytosis of the EGF receptor, the was observed (n ¼ 3). activation status of GTP-bound Ras and the kinase activity of B-Raf and Raf-1. In contrast, WNK2 depletion increased the phosphorylation levels of MEK1/2 at Ser-218/222, within the conventional

Oncogene WNK2 negatively modulates MEK1/ERK1/2 activation S Moniz et al 6077

Figure 5 The mechanism of extracellular signal-regulated kinase (ERK)1/2 activation in WNK2-depleted cells involves the phosphorylation of MEK1. HeLa cells were transfected with control (siGFP) or WNK2-specific (siWNK2-a and -b) small interfering RNAs (siRNAs) and analysed after 24 h. (a) The level of active guanosine triphosphate (GTP)-Ras was determined in pull-down assays with or without cell stimulation by epidermal growth factor (EGF). HeLa cells transfected with a constitutively active Myc-Ras G12V mutant served as a positive control. Note that the GTP-Ras levels were not further increased following the depletion of WNK2, although an increase in ERK activation was observed in the cells, as expected. (b) The kinase activity of B-Raf in HeLa cells was determined in the presence or absence of EGF using immunoprecipitated endogenous B-Raf and recombinant kinase-inactive MEK1 as substrate in an in vitro kinase assay. HeLa cells transfected with a constitutively active GFP-B-Raf V600E mutant served as a positive control. Note that the B-Raf kinase activity remains unaffected following the depletion of WNK2, although an increase in MEK1/2 activation is observed in the cells. (c) The kinase activity of Raf-1 was determined in the presence or absence of EGF by an in vitro kinase assay using immunoprecipitated endogenous Raf-1 and recombinant kinase-inactive MEK1 as substrate. HeLa cells transfected with a constitutively active Myc-Ras G12V mutant served as a positive control. Note that the Raf-1 kinase activity remains unaffected following the suppression of WNK2, although an increase in MEK1/2 activation is observed in the lysates. (d) Depletion of WNK2 in HeLa cells increases MEK1/2 activation. A representative western blot shows the effect of siWNK2 on the phosphorylation of MEK1/2 at Ser-218/222, a surrogate marker for MEK activity (top panels). The increase in MEK activation (middle) as well as the achieved inhibition of WNK2 expression (bottom) was determined in four independent experiments and the respective mean7s.d. values were graphically displayed (n ¼ 4). (e) Depletion of WNK2 in HeLa (top panels) or HT29 cells (bottom panels) results in the phosphorylation of MEK1 at Ser-298, a modification known to upregulate the activity of MEK1 toward ERK1/2.

MEK1/2 activation loop sites (Alessi et al., 1994; Zheng extracellular matrix proteins (Slack-Davis et al., 2003). and Guan, 1994). In addition, an increase in phosphory- This mechanism was shown to potentiate both the lation of MEK1 at Ser-298 was observed following the amplitude and duration of ERK1/2 activation in WNK2 silencing. response to EGF (Frost et al., 1997), and was proposed This Ser-298 phosphorylation site in MEK1 is an to integrate cell matrix and growth factor signals (Slack- important point of functional convergence between Davis et al., 2003). Thus, WNK2 targets the phosphory- signalling pathways and has been found to enhance lation status of MEK1 at Ser-298 and by counteracting the interaction between MEK1 and Raf-1 (Frost et al., this modification modulates the efficiency with which 1997), to prime MEK1 for its activation by Raf-1 (Coles MEK1 can activate ERK1/2 in response to growth and Shaw, 2002) or to promote a molecular scaffold factors. We demonstrated that inhibition of WNK2 containing both MEK1 and ERK1/2 (Eblen et al., released this constraint at MEK1 Ser-298, initiating a 2002). Recent data also suggest that the phosphoryla- more sustained ERK activation. Since this mechanism is tion of MEK1 at Ser-298 can result in Raf-independent dependent on the WNK2 catalytic activity, we assumed activation of MEK1 through its autophosphorylation at that WNK2 may phosphorylate MEK1 directly and Ser-218 and Ser-222 (Park et al., 2007). MEK1 can be interfere with Ser-298 phosphorylation. However, we directly phosphorylated on Ser-298 by protein kinase found no evidence for co-immunoprecipitation between PAK (Frost et al., 1997; Coles and Shaw, 2002). This WNK2 and MEK1, and no evidence for phosphory- phosphorylation is induced during cellular adhesion to lation of recombinant MEK1 by immunoprecipitated

Oncogene WNK2 negatively modulates MEK1/ERK1/2 activation S Moniz et al 6078 WNK2 in vitro (not shown). Therefore, we speculated Materials and methods that WNK2 acts through another kinase or phospha- tase, which regulate the phosphorylation of MEK1 Cell culture, transfections and RNA interference Ser-298. Alternatively, WNK2 may compete with the HeLa (cervix carcinoma) and human embryonic kidney cells pathways that regulate PAK activity (Frost et al., 1997). HEK293 were maintained in Dulbecco’s minimal essential medium (DMEM) and HT29 colon carcinoma cells in RPMI, The EGF-induced Ras-Raf-MEK-ERK/MAP kinase both supplemented with 10% fetal calf serum (Gibco Invitro- pathway is frequently activated in human tumors due to gen Corporation, Barcelona, Spain) and regularly checked for various oncogenic mechanisms, including amplification the absence of mycoplasma infection. For ectopic expression of EGFR genes and activating point mutations in of plasmid cDNAs, HEK293 cells were transfected at 80–90% EGFR, Ras or B-Raf (Gschwind et al., 2004; Wellbrock confluence using Metafectene (Biontex, Martinsried/Planegg, et al., 2004; Rodriguez-Viciana et al., 2005). It is Germany) and HeLa cells at 60% confluence using Lipofec- interesting to note that a partial clone of WNK2 has tAMINE 2000 (Invitrogen, Barcelona, Spain), both according previously been isolated as the colon cancer antigen to the manufacturer’s instructions. Transfection efficiencies SDCC43 (Scanlan et al., 1998) and colon tumors also were determined with a GFP expression vector and found to show frequent hyper-stimulation of the EGFR signal- be 90% for HEK293 and 50% for HeLa cells. Total amounts of transfected plasmid DNA were kept constant at 4 mg per ling pathway. Therefore, this oncogenic pathway is 60 mm dish or 2 mg per 35 mm dish and adjusted with empty considered as an important therapeutic target for vector if required. Cells were analysed after 22 h for anticancer drugs (Sebolt-Leopold and Herrera, 2004). biochemical assays or after 16 h for immunofluorescence. The design of clinically successful signalling inhibitors of Small interfering RNA oligos (siRNAs) were obtained from this pathway requires detailed knowledge about the MWG-Biotech AG (Ebersberg, Germany) and transfected mechanisms involved in fine-tuning of the ERK1/2 using LipofectAMINE 2000 (Invitrogen), either into HT29 activation. In this respect, our data identify WNK2 as a cells at 20% confluence or into HeLa cells at 30% confluence. novel fine-tuning signalling factor, which downregulates Transfection efficiencies were estimated using an FITC- the intensity of EGFR signalling at the MEK1/ERK coupled siRNA (Qiagen, Hilden, Germany) and were found level. Conversely, the loss of WNK2 expression may to be around 70% in HT29 or >90% in HeLa cells. Expression of endogenous WNK2 and WNK1 was inhibited upregulate ERK activation levels so that growth is by the transfection of HeLa cells with 200 pmol of siRNA stimulated and senescence circumvented. These func- per 35 mm dish or 400 pmol per 60 mm dish, whereas HT29 tional properties indicate that WNK2 may function as a cells required twice the amount. The targeted sequences were tumor suppressor gene. To validate this concept, it will siWNK2-a (50-GCU CGA GGA UGC UGA CAU ATT), be interesting to determine whether loss-of-function siWNK2-b (50-GGA CGC ACC CGA UGA AAU UTT), mutations or altered WNK2 expression occurs in patient siWNK1-a (50-GCA GGA GUG UCU AGU UAU ATT) and tumors from tissues where WNK2 is expressed (Verı´ s- siWNK1-b (50-AAU UAU GGC UAC GUA UUG ATT). simo and Jordan, 2001). Such molecular defects can be As controls, pre-designed luciferase GL2 siRNA (50-CGU ACG CGG AAU ACU UCG ATT) or a GFP-specific siRNA expected to commit premalignant epithelial cells to 0 neoplasia and/or collaborate with other oncogenic (5 -GGC UAC GUC CAG GAG CGC ACC TT) were used (MWG-Biotech AG). Cells transfected with siRNAs were defects during the cancer cell progression and survival. analysed after 24 h and the achieved reduction in target gene On the basis of our analysis of WNK2, we have expression was determined in each experiment by removing a further shown that two members of the WNK subfamily 30 ml aliquot from the cell lysate (see below), which was mixed act in parallel MAP kinase cascades downstream of the with extraction buffer RLT and RNA extracted as described EGFR. WNK1 is an activator of the ERK5 cascade, below. whereas WNK2 is an inhibitor of the ERK1/2 pathway. Both WNK1 and WNK2 target cell cycle progression cDNA cloning and expression plasmids and may thus integrate various signals controlling Details on the identification and subcloning strategy of WNK2 cancer cell proliferation. Indeed, WNK kinases may are available under S1 in the Supplementary Information file. affect the activity of other WNK family members The nucleotide sequence for the human WNK2 cDNA has because in vitro recombinant WNK1 was shown to been deposited in the GenBank/EBI databank under the phosphorylate the catalytic domains of WNK2 and accession number AJ242724. The WNK2 cDNA was sub- WNK4 (Lenertz et al., 2005). It will be challenging to cloned as an EcoRI/EcoRV fragment into the pcDNA3-Myc elucidate the individual and coordinated activity of or pEGFP-C2 (Clontech, Mountain View, CA, USA) expres- WNK kinases in signalling pathways and whether these sion vectors. The kinase-dead WNK2-K207M mutant was obtained using the Quickchange mutagenesis kit (Stratagene, are integrated with the known role of WNK kinases in La Jolla, CA, USA). All constructs were verified by automated regulating ion transport channels. DNA sequencing. pRK5-Myc-Ras-V12 was a gift from Dr Jonathan Morris (Kings College, London, UK) and pEGFP- B-Raf V600E a gift from Dr Carla Oliveira (IPATIMUP, Porto, Portugal). Note added Transcript expression analysis and semiquantitative PCR While this paper was under revision, a genome-wide Total RNA was extracted from cell lines or lysates (see below) scan identified a consistent epigenetic silencing of the with the RNAeasy kit (Qiagen) and 1 mg reverse transcribed WNK2 gene in glioma, indicating a role for WNK2 as a using random primers (Invitrogen) and Ready-to-Go You- cell growth suppressor (Hong et al., 2007). Prime Beads (GE Healthcare, Buckinghamshire, UK). Primers

Oncogene WNK2 negatively modulates MEK1/ERK1/2 activation S Moniz et al 6079 for specific amplification of WNK1 and WNK2 were as with Raf1 RBD Agarose beads from Cell Biolabs Inc. (San described (Verı´ ssimo and Jordan, 2001) and for RNA Diego, CA, USA) according to the manufacturer’s instruc- polymerase II (pol II) were pol II-F (50-GAG CGG GAA tions. Washed beads were directly solubilized in 2 Â SDS TTT GAG CGG ATG C) and pol II-R (50-GAA GGC GTG sample buffer. Human tissue protein medleys were purchased GGT TGA TGT GGA AGA). Amplification reactions were from Clontech and 50 mg was analysed. All samples were performed using AmpliTaq polymerase (Perkin-Elmer, Well- separated in a 10% SDS–PAGE Protean III mini-gels (Bio- esley, MA, USA) in 35 cycles of 941C, 45 s; 601C, 45 s; 721C, Rad, Hercules, CA, USA). Proteins were transferred onto a 45 s with an initial denaturation of 5 min at 941C and a final polyvinylidene fluoride membrane (Bio-Rad) using a Mini extension of 10 min at 721C in the case of WNK2, or in 28 Trans-Blot cell (Bio-Rad) at 100 V for 60 min and Coomassie- cycles of 941C, 30 s; 641C, 30 s; 721C, 30 s for WNK1 and pol stained to check for equal transfer. Membranes were blocked II. These semiquantitative amplification conditions were in Tris-buffered saline, 0.1% Triton X-100, 5% milk powder, experimentally determined to correspond to the linear probed using the indicated antibodies, then incubated with a amplification phases. Products were separated on 1.5% secondary peroxidase-conjugated antibody (Bio-Rad) and Agarose gels and band intensity was quantified using Image specific binding detected in a chemiluminescence reaction. J software followed by Pol II normalization. No amplification For densitometric estimation of protein quantities, the was obtained when RNA was mock reverse transcribed without luminescence film exposures from at least three independent adding reverse transcriptase. A multiple tissue northern blot experiments were digitalized and analysed using the ImageJ (Clontech) was hybridized using the standard procedures to a software (NIH). radioactively labelled WNK2 probe of 1042 bp corresponding to nucleotides 681–1722 according to the GenBank accession In vitro kinase assays number AJ242724. For in vitro protein kinase assays the beads containing immunoprecipitated kinases were resuspended in 20 ml kinase Antibodies reaction buffer (30 mM Tris-HCl, pH 7.5, 10% glycerol, 1 mM A rabbit polyclonal antibody SPT81 was raised and affinity- DTT, 1 mM Na3VO4,75mM MgCl2 and 500 mM ATP). Assays purified against the synthetic WNK2 peptide 1264DTDAD for WNK2 autophosphorylation activity were performed in RGSDPGTSP1277 by Eurogentec, Belgium (www.eurogentec. the presence of 5 mCi g-[32P] ATP and in the presence or com) according to the company’s standard procedures and absence of 0.75 mg/ml of MBP (Upstate Biotechnologies). used at a 1:100 dilution. Anti-WNK1 serum 308 was as Assays for B-Raf or Raf-1 activity were non-radioactive and described (Verı´ ssimo and Jordan, 2001), anti-Myc (clone the endogenous immunoprecipitated Raf kinases were incu- 9E10), anti-a-tubulin (clone B5-1-2), anti-ERK1/2 and anti- bated at 301C for 30 min with 200 ng recombinant kinase- phospho ERK1/2 (clone MAPK-YT) were from Sigma- dead MEK1 protein as substrate (Biosource International, Aldrich Quimica (Madrid, Spain); anti-GFP (ab290) from Camarillo, CA, USA). All results were confirmed in at least AbCam (Cambridge, UK); anti-phospho-MEK 217/221 from three independent experiments. Cell Signalling (Danvers, MA, USA); anti-phospho-MEK S298 from Upstate Biotechnologies (Charlottesville, VA, Immunofluorescence microscopy USA); anti-B-Raf (clone F7), anti-Raf-1 (clone E 10 for HeLa cells were grown on coverslips, transfected with GFP- immunoprecipitation and clone C20 for western blot) from WNK2orMyc-WNK2andfixedafter16hwith3.7% Santa Cruz Biotechnologies, (Santa Cruz, CA, USA); anti- paraformaldehyde in PBS followed by permeabilization with cyclin D1 (clone DCS-6, BD Biosciences Pharmingen, 0.1% Triton X-100 in PBS. Internalization of the EGF receptor Erembodegem, Belgium) and anti-BrdU-FITC from Roche in siRNA-treated cells was determined following the incubation Applied Science (Indianapolis, IN, USA). of HeLa cells on ice with 0.4 mg tetramethylrhodamine-labelled EGF (Molecular Probes, Eugene, OR, USA) for 30 min. Cells Cell lysis, immunoprecipitation and western blot procedures were then washed twice with cold PBS, transferred to growth 1 In all experiments involving the transfection of siRNAs, a non- medium at 37 C and fixed at time 0 or after incubations for 2 or denaturing cell lysis buffer was used to allow simultaneously 5 min. Coverslips were mounted on microscope slides with for immunoprecipitation and analysis of RNA and protein. Vectashield (Vector Laboratories, Burlingame, CA, USA); Cells were lysed on ice in either 100 ml (35 mm dishes), 250 ml images were recorded on a Zeiss LSM 410 confocal microscope (60 mm dishes) or 750 ml (100 mm dishes) lysis buffer (20 mM and processed with Adobe Photoshop software. Tris-HCl, pH 7.5, 1% NP-40, 130 mM NaCl, 10% glycerol, 10 mM MgCl2) containing 10 mM NaF, 0.1 mM Na3VO4 and G1/S-progression assay 1mM dithiothreitol (DTT) and a protease inhibitor cocktail Progression into S-phase was monitored by the incorporation was composed of 1 mM phenylmethylsulfonyl fluoride, 1 mM of bromodeoxyuridine (BrdU) as described previously (Matos 1,10-phenanthroline, 1 mM EGTA, 10 mM E64 and 10 mg/ml of and Jordan, 2005). Briefly, HeLa cells were transfected with each aprotinin, leupeptin and pepstatin A (all from Sigma- siRNAs and after 24 h incubated with 60 mM 5-bromo-2- Aldrich). For RNA extraction, a 30 ml aliquot was removed deoxyuridine (BrdU) (Sigma-Aldrich) for 30 min, then fixed and processed as described above. For protein analysis the with formaldehyde, permeabilized and the DNA denatured by remaining 70 ml (35 mm dishes) or a 40 ml aliquot (60 mm or treatment with 4 N HCl. BrdU incorporation was visualized by 100 mm dishes) was added to 17.5 or 10 mlof5Â Laemmli immunofluorescence with mouse anti-BrdU (Roche Applied sample buffer, boiled for 10 min and centrifuged at 2500 g for Science) followed by goat anti-mouse-TexasRed (Jackson 30 s. In the case of immunoprecipitation experiments, followed Immunoresearch Laboratories, West Grove, PA, USA), and or not by kinase assays, the remaining lysate was incubated for nuclei were counterstained with 4,6-diamidino-2-phenylindole 2 h at 41C with the specified antibodies (2.4 mg/ml anti-B-Raf (DAPI). Fluorescence images of seven randomly chosen fields or anti-Raf-1, 48 mg/ml anti-Myc or 26 mg/ml anti-WNK2), from different coverslips were digitally recorded on a Zeiss then further incubated for 1 h with protein G-Agarose beads Axiovert 100 fluorescence microscope equipped with a cooled (Roche Applied Science) and finally washed three times in cold Hamamatsu CCD camera (RGB3-chip). The numbers of lysis buffer. For Ras activation assays the lysate was incubated BrdU-positive as well as that of total nuclei were counted

Oncogene WNK2 negatively modulates MEK1/ERK1/2 activation S Moniz et al 6080 with a minimum of 1000 cells scored for each transfection and automated ABI sequencer and Jonathan Morris (London) and derived from three independent experiments. Carla Oliveira (Porto) for providing plasmids used in this study. This work was supported by INSERM and the Acknowledgements Portuguese Fundac¸a˜ o para a Cieˆ ncia e Tecnologia (grant POCI/56294/04, Programa de Financiamento Plurianual do We thank Estelle Le Gall for her contribution in the CIGMH and fellowship BD 11180/02 to SM) as well as a Differential Display analysis, So´ nia Pedro for running the FEBS short-term visitor fellowship to SM.

References

Alessi DR, Saito Y, Campbell DG, Cohen P, Sithanandam G, Rapp U Lenertz LY, Lee BH, Min X, Xu BE, Wedin K, Earnest S et al. (2005). et al. (1994). Identification of the sites in MAP kinase kinase-1 Properties of WNK1 and implications for other family members. phosphorylated by p74raf-1. EMBO J 13: 1610–1619. J Biol Chem 280: 26653–26658. Balmanno K, Cook SJ. (1999). Sustained MAP kinase activation is Li W, Han M, Guan KL. (2000). The leucine-rich repeat protein SUR- required for the expression of cyclin D1, p21Cip1 and a subset of 8 enhances MAP kinase activation and forms a complex with Ras AP-1 proteins in CCL39 cells. Oncogene 18: 3085–3097. and Raf. Genes Dev 14: 895–900. Coles LC, Shaw PE. (2002). PAK1 primes MEK1 for phosphorylation Manning G, Whyte DB, Martinez R, Hunter T, Sudarsanam S. (2002). by Raf-1 kinase during cross-cascade activation of the ERK The protein kinase complement of the human genome. Science 298: pathway. Oncogene 21: 2236–2244. 1912–1934. Davies H, Bignell GR, Cox C, Stephens P, Edkins S, Clegg S et al. Marshall CJ. (1995). Specificity of receptor tyrosine kinase signalling: (2002). Mutations in the BRAF gene in human cancer. Nature 417: transient versus sustained extracellular signal-regulated kinase 949–954. activation. Cell 80: 179–185. Dobrowolski S, Harter M, Stacey DW. (1994). Cellular ras activity is Matheny SA, Chen C, Kortum RL, Razidlo GL, Lewis RE, White required for passage through multiple points of the G0/G1 phase in MA. (2004). Ras regulates assembly of mitogenic signalling BALB/c 3T3 cells. Mol Cell Biol 14: 5441–5449. complexes through the effector protein IMP. Nature 427: 256–260. Ebisuya M, Kondoh K, Nishida E. (2005). The duration, magnitude Matos P, Jordan P. (2005). Expression of Rac1b stimulates NF-kB- and compartmentalization of ERK MAP kinase activity: mechan- mediated cell survival and G1/S-progression. Exp Cell Res 305: 292–299. isms for providing signaling specificity. J Cell Sci 118: 2997–3002. Min X, Lee BH, Cobb MH, Goldsmith EJ. (2004). Crystal structure of Eblen ST, Slack JK, Weber MJ, Catling AD. (2002). Rac-PAK the kinase domain of WNK1, a kinase that causes a hereditary form signaling stimulates extracellular signal-regulated kinase (ERK) of hypertension. Structure 12: 1303–1311. activation by regulating formation of MEK1-ERK complexes. Mol Moriguchi T, Urushiyama S, Hisamoto N, Iemura S, Uchida S, Cell Biol 22: 6023–6033. Natsume T et al. (2005). WNK1 regulates phosphorylation of Frost JA, Steen H, Shapiro P, Lewis T, Ahn N, Shaw PE et al. (1997). cation-chloride-coupled cotransporters via the STE20-related Cross-cascade activation of ERKs and ternary complex factors by kinases, SPAK and OSR1. J Biol Chem 280: 42685–42693. Rho family proteins. EMBO J 16: 6426–6438. Morrison DK, Davis RJ. (2003). Regulation of MAP kinase signalling Gagnon KB, England R, Delpire E. (2006). Characterization of SPAK modules by scaffold proteins in mammals. Annu Rev Cell Dev Biol and OSR1, regulatory kinases of the Na-K-2Cl cotransporter. Mol 19: 91–118. Cell Biol 26: 689–698. Murphy LO, MacKeigan JP, Blenis J. (2004). A network of immediate Gamba G. (2005). Role of WNK kinases in regulating tubular salt and early gene products propagates subtle differences in mitogen-activated potassium transport and in the development of hypertension. Am J protein kinase signal amplitude and duration. MolCellBiol24: 144–153. Physiol Renal Physiol 288: F245–F252. Park ER, Eblen ST, Catling AD. (2007). MEK1 activation by PAK: a Gschwind A, Fischer OM, Ullrich A. (2004). The discovery of novel mechanism. Cell Signal 19: 1488–1496. receptor tyrosine kinases: targets for cancer therapy. Nat Rev Rodriguez-Viciana P, Tetsu O, Oda K, Okada J, Rauen K, Cancer 4: 361–370. McCormick F. (2005). Cancer targets in the Ras pathway. Cold Hanks SK, Hunter T. (1995). The eucaryotic protein kinase super- Spring Harb Symp Quant Biol 70: 461–467. family: kinase (catalytic) domain structure and classification. Scanlan MJ, Chen YT, Williamson B, Gure AO, Stockert E, Gordan FASEB J 9: 576–596. JD et al. (1998). Characterization of human colon cancer antigens Hong C, Moorefield KS, Jun P, Aldape KD, Kharbanda S, Phillips HS recognized by autologous antibodies. Int J Cancer 76: 652–658. et al. (2007). Epigenome scans and cancer genome sequencing Sebolt-Leopold JS, Herrera R. (2004). Targeting the mitogen-activated converge on WNK2, a kinase-independent suppressor of cell protein kinase cascade to treat cancer. Nat Rev Cancer 4: 937–947. growth. Proc Natl Acad Sci USA 104: 10974–10979. Sewing A, Wiseman B, Lloyd AC, Land H. (1997). High-intensity Raf Ito M, Shichijo S, Tsuda N, Ochi M, Harashima N, Saito N et al. signal causes cell cycle arrest mediated by p21Cip1. Mol Cell Biol 17: (2001). Molecular basis of T cell-mediated recognition of pancreatic 5588–5597. cancer cells. Cancer Res 61: 2038–2046. Slack-Davis JK, Eblen ST, Zecevic M, Boerner SA, Tarcsafalvi A, Jiang ZY, Zhou QL, Holik J, Patel S, Leszyk J, Coleman K et al. Diaz HB et al. (2003). PAK1 phosphorylation of MEK1 regulates (2005). Identification of WNK1 as a substrate of Akt/protein kinase fibronectin-stimulated MAPK activation. J Cell Biol 162: 281–291. B and a negative regulator of insulin-stimulated mitogenesis in 3T3- Verı´ ssimo F, Jordan P. (2001). WNK kinases, a novel protein kinase L1 cells. J Biol Chem 280: 21622–21628. subfamily in multi-cellular organisms. Oncogene 20: 5562–5569. Kahle KT, Wilson FH, Lalioti M, Toka H, Qin H, Lifton RP. (2004). Verı´ ssimo F, Silva E, Morris JD, Pepperkok R, Jordan P. (2006). WNK kinases: molecular regulators of integrated epithelial ion Protein kinase WNK3 increases cell survival in a caspase 3- transport. Curr Opin Nephrol Hypertens 13: 557–562. dependent pathway. Oncogene 25: 4172–4182. Kerkhoff E, Rapp UR. (1998). High-intensity Raf signals convert Vitari AC, Deak M, Collins BJ, Morrice N, Prescott AR, Phelan A mitotic cell cycling into cellular growth. Cancer Res 58: 1636–1640. et al. (2004). WNK1, the kinase mutated in an inherited high-blood- Kostich M, English J, Madison V, Gheyas F, Wang L, Qiu P et al. pressure syndrome, is a novel PKB ()/Akt substrate. (2002). Human members of the eukaryotic protein kinase family. Biochem J 378: 257–268. Genome Biol 3: RESEARCH0043. Vitari AC, Deak M, Morrice NA, Alessi DR. (2005). The WNK1 and Lee BH, Min X, Heise CJ, Xu BE, Chen S, Shu H et al. (2004). WNK1 WNK4 protein kinases that are mutated in Gordon’s hypertension phosphorylates synaptotagmin 2 and modulates its membrane syndrome phosphorylate and activate SPAK and OSR1 protein binding. Mol Cell 15: 741–751. kinases. Biochem J 391: 17–24.

Oncogene WNK2 negatively modulates MEK1/ERK1/2 activation S Moniz et al 6081 Wellbrock C, Karasarides M, Marais R. (2004). The RAF proteins kinase lacking the catalytic lysine in subdomain II. J Biol Chem 275: take centre stage. Nat Rev Mol Cell Biol 5: 875–885. 16795–16801. Wilson FH, Disse-Nicodeme S, Choate KA, Ishikawa K, Nelson- Xu BE, Stippec S, Chu PY, Lazrak A, Li XJ, Lee BH et al. (2005). Williams C, Desitter I et al. (2001). Human hypertension caused by WNK1 activates SGK1 to regulate the epithelial sodium channel. mutations in WNK kinases. Science 293: 1107–1112. Proc Natl Acad Sci USA 102: 10315–10320. Woods D, Parry D, Cherwinski H, Bosch E, Lees E, McMahon M. Xu BE, Stippec S, Lenertz L, Lee B-H, Zhang W, Lee YK et al. (2004). (1997). Raf-induced proliferation or cell cycle arrest is determined WNK1 activates ERK5 by an MEKK2/3-dependent mechanism. by the level of Raf activity with arrest mediated by p21Cip1. Mol J Biol Chem 279: 7826–7831. Cell Biol 17: 5598–5611. Zheng CF, Guan KL. (1994). Activation of MEK family kinases Xu BE, English JM, Wilsbacher JL, Stippec S, Goldsmith EJ, Cobb requires phosphorylation of two conserved Ser/Thr residues. EMBO MH. (2000). WNK1, a novel mammalian serine/threonine protein J 13: 1123–1131.

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

Oncogene