Comparative Studies of a New Subfamily of Human Ste20-Like Kinases: Homodimerization, Subcellular Localization, and Selective Activation of Mkk3and P38

Comparative studies of a new subfamily of human Ste20-like kinases: homodimerization, subcellular localization, and selective activation of MKK3and p38

Jason T Yustein1, Liang Xia2, J Michelle Kahlenburg3, Dan Robinson2, Dennis Templeton4 and Hsing-Jien Kung*,2

1Department of Molecular Biology and Microbiology, Case Western Reserve University, Cleveland, OH 44106-4960, USA; 2University of California at Davis Cancer Center, UCDMC, Res. III, 4645 2nd Avenue, Sacramento, CA 95817, USA; 3Department of Pathology, Case Western Reserve University, Cleveland, OH 44106, USA; 4Department of Pathology, University of Virginia, Charlottesville, VA 22908, USA

The Sterile-20 or Ste20 family of serine/threonine kinases mammalian Ste20 (MST) (Creasy and Chernoff, is a group of signaling molecules whose physiological roles 1995a, b) kinase, to several dozen individual family within mammalian cells are just starting to be elucidated. members. By sequence comparison to the elucidated Here, in this report we present the characterization of yeast pathways, these mammalian homologues have three human Ste20-like kinases with greater than 90% been classified as MAP kinase kinase kinase kinase similarity within their catalytic domains that define a (MAP4Ks), lying upstream of the MEKK family of novel subfamily of Ste20s. Members of this kinase family kinases, and projected downstream of cytokine recep- include rat thousand and one (TAO1) and chicken KFC tors and small G proteins (Kyriakis and Avruch, 1996). (kinase from chicken). For the lack of a consensus The Ste20 family of kinases can be subdivided into two nomenclature in the literature, in this report, we shall separate branches: the p21-activated kinases (PAKs) call this family hKFC (for their homology to chicken and the non-PAKs or GCK family (Dan et al., 2001). KFC) and the three members hKFC-A, hKFC-B, and These two branches differ both structurally as well as hKFC-C, respectively. These kinases have many simila- functionally. The PAK family kinases have carboxyl- rities including an aminoterminal kinase domain, a serine- terminal kinase domains, which are regulated by the rich region, and a coiled-coil configuration within the small G proteins Rac1 and Cdc42Hs. Binding of the C-terminus. All three kinases are able to activate the p38 small guanine triphosphatases to specific consensus MAP kinase pathway through the specific activation of sequences (Manser et al., 1994), known as the CRIB the upstream MKK3kinase. We also offer evidence, both (Cdc42/Rac1 interactive binding) domain, within the theoretical and biochemical, showing that these kinases aminoterminus of the PAK proteins causes a conforma- can undergo self-association. Despite these similarities, tional change and direct activation of the catalytic these kinases differ in tissue distribution, apparent domain. In general, PAK kinases are implicated in the subcellular localization, and feature structural differences activation of the stress-activated protein kinase/jun N- largely within the carboxyl-terminal sequence. terminal kinase (SAPK/JNK) pathway, as well as Oncogene (2003) 22, 6129–6141. doi:10.1038/sj.onc.1206605 regulating cell morphology and cell motility through interaction and regulation of the cytoskeleton (Brown Keywords: Ste20 kinase; TAO; p38 MAPK et al., 1996; Frost et al., 1996; Eby et al., 1998; Chung and Firtel, 1999; Holly and Blumer, 1999). The GCK family differs from PAKs in that its Introduction catalytic domain resides in the aminoterminus part of the protein, with the C-terminus believed to be a The Ste20 family of serine/threonine kinases was first regulatory domain. However, this latter domain does identified and characterized in Saccharomyces cerevisiae not contain any CRIB consensus sequence or other as part of the mating and pheromone response signaling known p21-binding regions. The regulatory mechanisms mechanism (Kyriakis and Avruch, 1996). Over the past for this branch of kinases appear to be far more variable several years, mammalian counterparts have been than those regulating the PAKs. Recent investigations identified, and this family has grown from a handful have shown that members of this family are involved in of kinases with the initial discoveries of germinal center the activation of the stress-activated pathways, which kinase (GCK) (Katz et al., 1994; Pombo et al., 1995) and include the SAPK/JNK and mammalian high-osmolarity glycerol kinase (p38/mHOG) (Pombo et al., 1995; *Correspondence: H-J Kung; E-mail: [email protected] Hu et al., 1996; Tibbles et al., 1996; Su et al., 1997). Received 19 October 2002; revised 7 March 2003; accepted 21 March Furthermore, many of these same kinases are involved 2003 in numerous variable upstream pathways including the hKFCs, a new subfamily of Ste20s JT Yustein et al 6130 transforming growth factor-b (TGF-b) pathway (Zhou and 57–64% homology in the regulatory domain. All et al., 1999), binding to and interacting with the Crk and three kinases specifically activate the MKK3 dual kinase the CrkL family of adaptor molecules (Oehrl et al., and p38/mHOG pathway, and not any of the other 1998; Ling et al., 1999), and activation of integrins known MKKs or MAP kinases. This is the first through Eph receptors (Becker et al., 2000). Other identified group of human kinases that can be specifi- potential physiological roles recently reported include cally and directly placed upstream of the p38 MAPK. involvement in lymphocyte aggregation through the Finally, we show that these hKFC Ste20-like kinases regulation of leukocyte-function-associated antigen have the capability to self-associate, which is another (LFA)-1-mediated clustering (Endo et al., 2000), cere- unique feature for cytoplasmic proteins. While all three bral development in mice (Dan et al., 2000), and in hKFC kinases have cytoplasmic subcellular localization, T-cell receptor signaling (Ling et al., 2001). These hKFC-C appears to have a slightly different localization physiological interactions and functions exemplify the as it does not have the generalized, dispersed appear- diversity of signaling pathways that these kinases may ance, but is more distinctly perinuclear. This difference be involved in. along with variations in the tissue expression patterns The mitogen-activated protein kinases (MAPKs) are for the kinases are future directions of investigations for comprised of three major, highly homologous proline- potential physiological differences among the three directed serine/threonine protein kinases. The first hKFCs. identified was extracellular-regulated kinases (ERKs), which are activated by growth-promoting reagents such as hormones and growth factors (Boulton et al., 1991; Results Charest et al., 1993; Peraldi et al., 1993; Assefa et al., 1997). The other two MAPK members, SAPK/JNK and The structure of hKFCs p38/mHOG, are involved in stress-induced signaling mechanisms. The most potent activators of these kinases We have isolated three human Ste20-like clones with are most notably cytokines such as interleukin-1b (IL- significant sequence homology to the chicken Ste20-like 1b), tumor necrosis factor-a (TNF-a), and ultraviolet kinase KFC (cKFC, GenBankt Accession #AF263314) (UV) irradiation (Geng et al., 1996; Kyriakis and (Yustein et al., 2000), a member of the GCK-like class of Avruch, 1996; Zanke et al., 1996a, b; Paul et al., 1997). Ste20s. These three kinases, belong to and define a new These MAPKs are activated by phosphorylation on subfamily of human Ste20s, we have called hKFC-A, -B, specific tyrosine and threonine residues by upstream and -C. The hKFC-A kinase (GenBankt Accession dual kinases referred to as MAPK kinases (or MKKs). #AF263311) is the direct human homologue of the Each MAPK is activated by relatively specific MKKs, chicken kinase. This 898 amino-acid protein, like the with minimal crosstalk occurring. For example, MEK1 avian counterpart, has three prominent structural and 2 have preferential specificity towards the ERKs, motifs. They include: (1) an aminoterminal kinase while MKK4 and 7 bind and phosphorylate SAPK/JNK domain; (2) a serine-rich region; and (3) a coiled-coil and MKK3 and 6 preferentially activate p38/mHOG conformation occupying the remainder of the C- (Zanke et al., 1996; Enslen et al., 2000). terminal regulatory portion of the protein. This kinase Previously, we identified a Ste20-like kinase, KFC is identical to JIK, described recently by Tassi et al. (kinase from chicken) (Yustein et al., 2000), from (1999). The second Ste20-like kinase, hKFC-B (Gen- chicken embryo fibroblasts, which is related to rat Bankt Accession #AF263312), is a 1001 amino-acid thousand and one (TAO1) and TAO2 kinases (Hutch- protein with structural features similar to the hKFC-A ison et al., 1998; Chen et al., 1999). Together, they form clone – serine-rich region and coiled-coil conformation, a new subfamily of GCK family kinases. We have but also contains an additional carboxyl-terminal serine/ extended this analysis and obtained cDNA clones of threonine-rich region. Finally, the hKFC-C kinase human homologues of these kinases. They are called (GenBankt Accession #AF263313), a 1049 amino-acid hKFC-A, -B, and -C. The hKFC-A kinase is the protein contains the three structural motifs as seen in orthologue of the chicken KFC and is identical to a hKFC-A and -B, as well as a polyacidic span of 15 Ste-20-like kinase, JIK, independently isolated by Tassi consecutive glutamic acids dividing the serine-rich et al (Tassi et al., 1999). The sequence of hKFC-B, an region, and a proline-rich region downstream from the orthologue of rat TAO1, has never been reported termination of the coiled-coil region. This region has before. The hKFC-C’s sequence is also new, with only several potential src-homology 3 (SH3)-binding sites. the first 744 amino acids being identical to that of Schematic representations, as well as the homologues human prostate-specific kinase (PSK) (Moore et al., and different nomenclatures used to describe this family 2000), and the remaining 306 amino acids completely of kinases are shown in Figure 1a. These three kinases divergent. Likely, these are products of alternative all have significant homology within their kinase domain splicing. to one another. Figure 1b graphically compares the In this report, we describe comparative studies of percent similarities within the kinase and regulatory these three kinases. All three hKFCs are characterized domains for these three human KFCs and yeast by an N-terminal kinase domain and a C-terminal Ste20. coiled-coil regulatory domain. They share among In addition, as we reported, the chicken KFC kinase themselves 80–88% homology in the kinase domain also has significant homology to the rat Ste20-like

Oncogene hKFCs, a new subfamily of Ste20s JT Yustein et al 6131 kinase, TAO1 (GenBankt Accession #T17365) (Hutch- kinase, PSK (GenBankt Accession #AF061943) ison et al., 1998). A comparison of the human sequences (Moore et al., 2000), is also a splice variant of hKFC- to both rat TAO1, and TAO2 (GenBankt Accession C and rTAO2. Analysis of the sequences revealed #AAD39480) (Chen et al., 1999) shows that the hKFC- that the kinase domains (spanning approximately 270 B clone is most likely the direct human homologue of amino acids) are 80% identical and greater than 90% rTAO1 and hKFC-C is more closely related, but not as similar. However, upon exiting the kinase domain there homologous to rTAO2. The sequence of hKFC-B is is a significant decrease in sequence similarities. As 98.7% identical to that of the rat homologue, while shown in the phylogenic relationship of KFCs with hKFC-C is greater than 98% identical to rTAO2 within other STE-20-like kinases (Figure 1c), the KFC family is the first 745 amino acids, but varies significantly in the closest to SULU kinase in Caenorhabditis elegans. remainder of the C-terminal sequence (only 14.8% Finally, amino-acid sequence alignment for the three identity). These kinases appear to be splice variants of human KFCs, along with yeast Ste20, is shown in one another. It also appears that a recently identified Figure 1d.

Figure 1 Schematic representation, family nomenclature, percent similarities, phylogenetic tree, and amino-acid alignment for the three human KFC Ste20-like kinases. (a) Schematic representations of the open reading frames for all three hKFCs show the key structural motifs observed. All three kinases show aminoterminal catalytic domains followed by a serine-rich region and a long carboxyl-terminal coiled-coil domain. Table outlining family homologues and their nomenclature. (b) Quantitative comparison of percent similarities within the kinase and regulatory domains for the hKFCs and yeast Ste20. (c) Phylogenetic tree showing the homology of the hKFCs to other members of the Ste20 family. (d) Sequence alignment of the three full-length human KFCs, along with the yeast Ste20 kinase domain (KD)

Oncogene 6132 Oncogene KC,anwsbaiyo Ste20s of subfamily new a hKFCs, TYustein JT tal et

Figure 1 Continued hKFCs, a new subfamily of Ste20s JT Yustein et al 6133 Tissue distribution of hKFCs a significant expression in the prostate, testes, and pancreas. Overall, there is ubiquitous expression, as RNAs encoding the hKFC kinases were examined by moderate to low levels of this message could be detected Northern blot analysis of mRNA samples from multiple in all tissues analysed. In contrast, neither hKFC-B nor human tissues (Figure 2). b-Actin transcript was hKFC-C (Figure 2b, c) had high expression levels in checked as a loading control (Figure 2d). The hKFC- PBLs, thymus, or spleen. The hKFC-B clone, with a A kinase (Figure 2a) has an apparent transcript size of transcript length of about 10 kb, had its highest levels of 4.5 kb. It is most prominent in hematopoietic cells and expression in the testes, with significant levels also related tissues. These included peripheral blood leuko- present in the skeletal muscle and placenta. It is cytes (PBLs), thymus, and spleen. In addition, there was interesting to note that a few tissues had absolutely no apparent hKFC-B message. They included the liver, ovary, and kidney. Finally, hKFC-C, with a 4.5 Kbp transcript length, appears to have a ubiquitous expression pattern like hKFC-A, but in contrast, it has its highest levels in the testes and brain. It was also interesting to note a higher molecular weight variant specifically expressed in the heart and skeletal muscle for hKFC-C.

Expression and catalytic activities of hKFCs We constructed plasmid expression vectors encoding aminoterminal hemagglutinin (HA)-tagged forms of each of the three hKFC kinases. The hKFC-A kinase has an apparent molecular weight of 118 kDa, hKFC-B is 130 kDa, and hKFC-C is approximately 140 kDa. To determine their respective catalytic activity, whole cell lysates were immunoprecipitated with monoclonal anti- HA antibody and subjected to an in vitro kinase assay as described in experimental procedures. Using myelin basic protein (MBP) as the substrate, the hKFC-B and -C kinases had signiﬁcantly greater kinase activity than the hKFC-A, reﬂected both by autophosphorylation and phosphorylation of the exogenous MBP substrate (Figure 3). Approximately equal levels of expressed kinases were used in each lane (immunoblot shown under in vitro kinase assay in Figure 3). We also tested different exogenous substrates, including histone 2B and histone 1, and found activity similar to that seen with MBP (data not shown). The low activity of hKFC-A is similar to the low activity we found previously with the chicken KFC (Yustein et al., 2000).

Signaling pathways for hKFCs As mentioned in the introduction, the Ste20 family has been implicated in the activation of the SAPK/JNK and/or p38 pathways. We also tested the hKFC kinase activities toward SAPK/JNK or p38. Through the use of phosphospeciﬁc antibodies, we found that all three human KFC kinases were able to induce the activating phosphorylation of the p38 MAP kinase (Figure 4a, top) and not able to activate SAPK/JNK (Figure 4a, middle) Figure 2 Expression pattern for the three human KFC Ste20-like kinases. Human Multiple Tissue Northern (MTNt) blots (Clon- or to induce activating phosphorylation of ERK tech) were probed separately using the C-terminal (regulatory (Figure 4a, bottom). The level of activation was domain) cDNA fragments of the three human KFC kinases labeled comparable to that of the positive control, being with [a-32P]dCTP, as described under ‘Experimental procedures’. overexpression of the direct upstream regulator Tissue distribution patterns of the hKFC-A (a), hKFC-B (b), and hKFC-C (c) transcripts are depicted. The blots were also reprobed MKK6. Expression levels of p38, SAPK/JNK, and for b-actin, which revealed relatively equal mRNA loaded for each ERK were equivalent in all lanes, as detected by tissue sample (d) antikinase antibodies (panels beneath each part).

Oncogene hKFCs, a new subfamily of Ste20s JT Yustein et al 6134 To ensure that the heightened phosphorylation of MKK3 reflects the activation of this kinase, the activity of glutathione S-transferase (GST)-MKK3, cotransfected with hKFC-A, was assayed for its ability to phosphorylate MBP as well as undergo autophosphorylation. The GST-tagged MKK3 was precipitated with GST beads, and incubated with MBP in an in vitro kinase reaction. The results show that KFC-A activates MKK3 based on MKK3 autophosphorylation (Figure 5, upper panel) and increased in vitro phosphorylation of MBP (Figure 5, lower panel). These results confirm the activation observed using anti-phosphopeptide immu- noblotting presented in the previous figure.

Subcellular localization of the hKFCs Indirect immunofluorescence was performed using the HA-tagged forms of each hKFC that were transiently transfected into COS-1 cells. As shown in Figure 6, all of the kinases show cytoplasmic staining; however, the exact pattern for each one has potentially significant differences. The hKFC-A (Figure 6a) kinase shows a general cytoplasmic immunofluorescence with a notice- able increase in peripheral cell membrane staining. This could be indicative of a highly upstream signaling molecule being recruited to the cell surface at the source of a signal initiation. Figure 6b shows the generalized cytoplasmic staining seen for the hKFC-B. This kinase shows a greater perinuclear staining and some strong punctuates cytoplasmic patterns of localization. Finally, the hKFC-C clone (Figure 6c) shows a very distinct perinuclear staining that does not extend very far out into the cytoplasm like the other human Ste20-like kinases. This type of staining may be representative of endoplasmic reticulum (ER) localization. However, treatment with Brefeldin A, which disrupts the Golgi complex and causes ER swelling, had no apparent effect Figure 3 Comparison of hKFC kinase activities. HA-tagged on the localization of this kinase. forms of each human KFC were transiently transfected into COS-1 cells. Aliquots from whole cell lysates were immunoprecipitated with the anti-HA antibodies and subjected to an in vitro kinase Regulation of hKFC activities assay as described under ‘Experimental procedures’ using MBP as Following the premise that these kinases are involved in the substate (top panel). Immunoblot, showing relatively equal hKFC protein precipitated, of the in vitro kinase assay is displayed the regulation of p38, we tested numerous potential (bottom panel) activators that have been known to activate this stress- related kinase. Using both transiently transfected COS-1 cells, as well as by examining endogenous hKFC kinases To identify MAPK kinases (MKKs) that transmit in DU-145 and Jurkat cells using antipeptide antibodies signals from hKFCs, and to confirm the specificity of that recognize each isoform specifically, we applied these kinases towards the p38 pathway, we examined the cytokines such as TNF-a and IL-1, stressors such as activation of MKK forms again using phosphospecific osmolar shock, UV irradiation and anisomycin, as well antibodies detecting activating phosphorylation events. as DNA-damaging reagents (i.e., cis-platinum), all of We found that all three hKFC kinases were able to which caused no changes in endogenous or over- activate the MKK3 kinase (Figure 4b, top), but not the expressed hKFCs kinase activities (autophosphorylation similar MKK6 (Figure 4b), which is also capable of or substrate phosphorlylation) (data not shown). We activating the p38 pathway. Confirming the inability of also tested several growth factors, which have been hKFC kinases to activate SAPK/JNK or ERK path- shown to activate either JNK/SAPK (Bost et al., 1999) ways, we found that they were unable to induce or p38 (Assefa et al., 1997) (Rousseau et al., 1997; Xing activating phosphorylation of MKK4 (SEK1), MKK7, et al., 1998) in certain situations, such as epidermal or MEK1/2 (Figure 4b, panels as indicated). These data growth factor (EGF), platelet-derived growth factor support the finding of p38 specificity for these three (PDGF), and insulin, with no detectable changes in related kinases. kinase activities observed. The elusive nature of

Oncogene hKFCs, a new subfamily of Ste20s JT Yustein et al 6135

Figure 4 The hKFC family kinases stimulate MKK3 and p38 phosphorylation. A, COS-1 cells were transiently cotransfected with GST-fused ERK1, SAPK, or p38 along with the hKFCs. Westerns blots were run using the respective phosphospeciﬁc antibodies to each MAP kinase. Samples were also run to check for expression of each isoform of KFC (data not shown). (b) MKK3, MKK6, MKK4, GST-MKK7, or MEK1 along with the hKFCs were transiently cotransfected into COS-1 cells. For all of the MKKs, but MKK7, phosphospeciﬁc antibodies were used to blot for activation, while an in vitro kinase assay was performed for MKK7. MKK7 was pulled down with GSHbeads and tested for activity using bacterial-expressed GST-SAPK KR as the substrate. Positive controls for these experiments included the addition of anisomycin (10 mg/ml) in the MKK7 experiment, cotransfection of MKK6 in the p38 assay, EGF (100 ng/ml) addition to ERK1 and MEK1, and MEKK1 for all other experiments in (a) and (b)

stimulators as a whole has been a common occurrence for many members of the Ste20-like family.

Self-association of hKFCs In our previous report on the original KFC (Yustein et al., 2000), we presented evidence that each hKFC has a protein region containing a large amount of a-helical secondary structure within the carboxyl-terminus that has extremely high probability of forming a coiled-coil conformation. Using the same coiled-coil prediction program (Lupas et al., 1991) used in our report, we have found that all three of these kinases also have the same probability of forming such a conformation within their C-terminal domains (data not shown). Taking this theoretical analysis one step further, we have used another protein sequence analysis program, Multicoil. Figure 5 hKFC-A activates MKK3. COS-1 cells were transfected Through the use of the Multicoil analysis (http// with GST-MKK3 alone, or with increasing amounts of hKFC-A. nightingale.lcs.mit.edu/cgi-bin/multicoil, Wolf et al., Cells were serum starved and lysed 48 h after transfection. GST- MKK3 was isolated via a GSHpull-down and then subjected to an 1997), which assigns an association probability value, in vitro kinase assay alone (top panel), or using MBP as a substrate dimer and trimer, to each protein sequence containing a (bottom panel) coiled-coil region, we found that all three of these

Oncogene hKFCs, a new subfamily of Ste20s JT Yustein et al 6136

Figure 6 Potential differences in subcellular localization for the human KFCs. HA-tagged constructs expressing each hKFC isoform were transiently transfected into COS-1 cells that had been previously plated on sterile coverslips lying within six-well plates. Cells were ﬁxed and stained, using the anti-HA antibody, as described under ‘Experimental procedures’. Coverslips were then mounted onto slides and viewed using a ﬂuorescent microscope (Leica). (a) HA-hKFC-A, (b) HA-hKFC-B, (c) HA-hKFC-C

kinases have regions of predicted dimerization, or oligo- We show that all three of these human KFC Ste20- merization, within their C-terminal regulatory domain. like kinases are involved in the activation of only As shown in Figure 7a, the hKFCs have regions at the MKK3, not MKK6, and the downstream p38 kinase. It beginning and end of the coiled-coil conformation with is uncommon for an activator of the p38 pathway to high predicted values of association. Confirming the stimulate only one of the upstream MEKs and not both theoretical analysis, we found that differently epitope- known regulators. This division of signaling opens up tagged forms of both hKFC-A and hKFC-C proteins the probability of defined activators leading to specific can self-associate (Figure 7b). Attempts have been made differences in p38 signaling modules. This notion, being to identify association of the hKFC kinases with one that the MKK3 and MKK6 kinases participate in another, but these have been unsuccessful. This could be nonredundant physiological functions, has recently been explained by the fact that the precise regions of supported through the use of knockout studies of these predicted association for each hKFC are different, and two genes in mice (Lu et al., 1999; Wysk et al., 1999). that the regions for interaction between the different One purpose of conducting a comparative study of isoforms are not compatible like that for self-associa- these kinases, as presented in this report, is to ensure tion. However, this should not rule out the possibility that the cell context, the experimental conditions, and that this region can associate with other coiled-coil the vectors used are all identical. Given the plasticity of containing proteins that could be vital to kinase the signal pathways, it would not be surprising that the regulation. In addition, the exact regulatory role of this high specificity toward MKK3/p38 may be modulated self-association for the kinases has not been determined, under other conditions. Indeed in previous reports of but we do know that in the chicken KFC, deletion of the this subfamily of kinases, varied results have been C-terminal tail results in higher kinase activity. Con- obtained, likely because of the different cell types, sistent with the above findings are the recent interesting different spliced isoforms, and different induction report that PAK1 homodimerization leads to autoinhi- conditions used. In complete agreement with our bition of kinase activity (Parrini et al., 2002). finding, rTAO1, the rat homologue of hKFC-B, was found to activate specifically in vivo MKK3 and p38, but not MKK6, MKK4, MKK7, and JNK (Hutchison et al., 1998). The N-terminal kinase domain (1–320) of Discussion baculovirus-expressed rTAO2 was found to activate primarily MKK3 and to a lesser extent, MKK6 and Here, we have presented the further characterization of MKK4 (Chen et al., 1999). By contrast, overexpression a new subfamily of human Ste20-like kinases, with high of PSK, which shares the N-terminal 1–744 amino acids homology to the previously reported chicken KFC. In of hKFC-C, was found to activate JNK (Moore et al., this report, we show that these human Ste20-like kinases 2000). Our own studies of chicken KFC A showed that hKFC-A, -B, and -C are involved in a specific pathway it has a significant potential to activate JNK, only if the leading to the activation of the p38 MAP kinase by its C-terminal regulatory domain is deleted (Yustein et al., upstream regulator MKK3. This specificity for regula- 2000). This discrepancy could potentially be explained tion of the p38 MAPK by a class of Ste20-like signaling by the differences in full-length C-terminal structure, molecules has not been previously described. which could provide for differences in substrate

Oncogene hKFCs, a new subfamily of Ste20s JT Yustein et al 6137 a MultiCoil score b total prob 1 hKFC-A dimer prob trimer prob 0.8

0.6

0.4 Probability 0.2

0 100 200 300 400 500 600 700 800 900 Amino Acid MultiCoil score

hKFC-B total prob 1 dimer prob trimer prob 0.8

0.6

0.4 Probability 0.2

0 200 400 600 800 1000 1200 Amino Acid MultiCoil score

1 hKFC-C total prob dimer prob trimer prob 0.8

0.6

0.4 Probability 0.2

0 200 400 600 800 1000 1200 Amino Acid Figure 7 Human KFCs can undergo self-association. (a) The graphical results of the theoretical Multicoil protein analysis program for each hKFC. The program plots the sequence probability of forming either a dimer or trimer. (b) Using HA and T7 epitope-tagged constructs (described in experimental procedures), transient cotransfections were performed in COS-1 cells. The first set of blots (top panels) shows self-association for hKFC-A with immunoprecipitations using anti-T7 monoclonal antibodies and Western blotting with the anti-HA antibody, as well as the reciprocal experiment (IP: anti-HA, WB: anti-T7). The next set of blots (middle panels) shows hKFC-C self-association using the same set of antibodies as used for hKFC-A. The final set of Western blots (bottom panels) were carried out on aliquots taken from each cotransfection whole cell lysates, showing that both HA- and T7-tagged forms of hKFC-A, and hKFC-C, were simultaneously expressed in the COS-1 cells specificity. The MKK3/p38 pathway is preferred by the reinforces the notion that the kinase activity and full-length hKFCs; whereas the SAPK/JNK pathway specificity can be modulated by different conditions may be a ‘default’ pathway for these kinases upon and underscores the importance of comparing all these deregulation of their activity, either by truncation or by kinases in an identical cell context. other means. That the C-terminal structure of this While we have determined that the downstream family of kinases plays a regulatory role on kinase signaling and one mode of regulation seem to be activity has already been demonstrated by our previous common to all three human KFCs, we have also work (Yustein et al., 2000) and by others (Hutchison observed some apparent differences. The most striking et al., 1998; Chen et al., 1999). Future work will be difference arises from comparison of hKFC-A expres- required to investigate the mechanism by which the sion profile to the other two human KFCs. This C-terminal structure regulates the substrate specificity. particular kinase seems to be more ubiquitously Finally, JIK was first shown to inhibit SAPK/JNK expressed, consistent with the JIK expression pattern activity in EGF-treated cells (Tassi et al., 1999), but (Tassi et al., 1999). In comparison, the other two kinases more recently shown to activate SAPK/JNK in ER- have a more limited and different expression profile. induced stress response (Yoneda et al., 2001). This Their expression is highest in the testes (for both), in the

Oncogene hKFCs, a new subfamily of Ste20s JT Yustein et al 6138 brain for hKFC-C, and in the placenta, skeletal muscle, kinases that are placed between the human KFCs and and colon for hKFC-B. The human KFC-B expression the MAP2K (MKK3). They include the mixed-lineage pattern we identified varies considerably from that kinases (MLKs) (Tibbles et al., 1996; Fanger et al., reported for rTAO1, which was exclusively expressed 1997), MAP three kinase 1 (MTK1) (Takekawa et al., in the rat testes and brain only (Hutchison et al., 1998). 1997), apoptosis signal-regulating kinase 1 (ASK1) The expression profile for hKFC-C follows those (Ichijo et al., 1997) and TGF-b-activated kinase 1 reported for rTAO2 and PSK (Moore et al., 2000) (TAK1) (Moriguchi et al., 1996) all of which have been much more closely with highest levels in both the testes shown to activate p38 through MKK3 (and MKK6). and brain. We find it very noteworthy that these highly Determination of this kinase is a future goal in order to related kinases appear to have such significant variation truly elucidate the full KFC-downstream signaling in expression profiles, especially the hKFC-A kinase and pathway. its levels in hematopoietic tissues in contrast to that for As previously mentioned, hKFC-C (and rTAO2) are the other two kinases. This variation in tissue expression splice variants of the recently reported PSK kinase. patterns suggests potential physiological differences for However, while those authors determined that PSK these three family members depending upon the tissue- overexpression leads to the activation of the SAPK/JNK specific stimulation. We are presently investigating kinase, we have determined a relatively specific mechan- potential tissue-specific functions for these kinases ism leading to the activation of the p38 MAP kinase. through the use of various cell lines. This discrepancy could potentially be explained by the Although the overall motifs are strikingly similar for differences in full-length C-terminal structure, which all three kinases: they all contain the aminoterminal could provide for differences in substrate specificity. As kinase domain, followed by a small serine-rich region, for the ability of the truncated, ‘activated’ chicken KFC which is then followed by a long coiled-coil domain kinase to stimulate the SAPK/JNK kinase may be within the C-terminal region; there are some apparent because of the loss of kinase regulation by either loss of structural differences, which could lead to physiological association with a p38-associated or structural protein, differences within the cell. For instance, the hKFC-B or loss of the ability to self-associate, which may be kinase has another serine/threonine-rich region at the important for regulating KFC kinase activity. There- very C-terminal end, whereas the hKFC-C clone has an fore, the SAPK/JNK pathway may be a ‘default’ acid-rich region dividing the serine-rich region. In pathway for these kinases upon deregulation of their addition, hKFC-C has a proline-rich region at the very activity. Future work investigating the regulatory roles C-terminus with multiple potential SH3-binding re- for the structural motifs, including the coiled-coil gions. This region alone could allow numerous other domain and serine-rich region, of the human KFCs is specific protein–protein interactions that are not possi- ongoing. ble for the other two kinases. Using techniques such as Future endeavors include determining the true phy- yeast two-hybrid could assist in the determination of siological regulators for these kinases both individually these novel protein interactions and provide information and as a family. As mentioned, we are presently using regarding cellular function and regulation. Finally, there many different cell lines, in order to determine whether also appear to be unique tyrosine residues, outside of the specific tissue-related signals are responsible for the catalytic region, for hKFC-C that can be appropriate regulation of kinase activities. We are excited about this phosphorylation sites for tyrosine kinases, and therefore new human subfamily of Ste20s and their role in lend another mode of regulation for this particular specifically regulating the p38 MAP kinase pathway, kinase. Initial experiments have shown that hKFC-C can as well as further elucidating their biological functions be tyrosine phosphorylated upon pervanadate stimu- within the cell. lation (anti-phosphotyrosine blot, data not shown). All of this evidence lends credence to the fact that while all three of these human KFCs are upstream Experimental procedures regulators of the p38 MAPK, the exact mechanism of activation for these three may be somewhat different. Materials While we have presented a significant amount of data on this new subfamily of p38-specific regulators, there are Phosphospecific rabbit polyclonal antibodies against the also numerous other questions that need to be activated form of each protein kinase were obtained addressed, and are presently topics of investigation. from New England Biolabs. These were specifically: They include determining the downstream effector MEK1/2 (Ser217/Ser221), MKK4 (Thr223), MKK3/6 kinase(s) directly transducing the signals of hKFCs. (Ser189/Ser207), ERK (Thr202/Tyr204), SAPK The placement of the hKFC family of kinases within (Thr183/Tyr185), and p38 (Thr180/Tyr182). Rabbit the activation pathway leading to p38 is not at this point antisera were raised against each of the three individual clear. While a truncated form of TAO1 has been kinases using synthesized peptides designed from reported to phosphorylate directly MKK3 (Hutchison unique internal sequences to each hKFC by Genemed et al., 1998), we have not observed this to date using the Synthesis, Inc. The peptides used were full-length kinase forms containing regulatory regions. CGHGDPRPEPRPTQSVQSQ for hKFC-A, Since Ste20s are genetically considered to be MAP4Ks, CRTRASDPQSPPQVSRHKS for hKFC-B, and it remains possible that the hKFCs could activate other CTSTTSSARRAYCRNRDHfor hKFC-C. Monoclonal

Oncogene hKFCs, a new subfamily of Ste20s JT Yustein et al 6139 anti-T7 and anti-HA (12CA5) antibodies were pur- maintained in RPMI 1640 supplemented with 10% FBS chased from Novagen and Upstate Biotechnology, and 1% penicillin/streptomycin. respectively. Horseradish peroxidase (HRP)-conjugated goat anti-rabbit and rabbit anti-mouse IgG secondary antibodies were obtained from Jackson ImmunoRe- Transfections, cell lysate preparation, and Western search Laboratories, Inc. All oligonucleotides for PCR blotting reactions were obtained from Integrated DNA Tech- COS-1 cells were plated onto six-well tissue culture nologies (IDT). plates 18 h prior to transfection. COS-1 cells, at about 50–60% confluency, were transfected with 2 mg of DNA s Isolation of cDNA clones encoding hKFCs (total) using FuGene6 transfection reagent (Roche) according to the manufacturer’s protocol. After 48 h, Total RNA was isolated from DU-145 prostate cells were rinsed twice with ice-cold PBS and then lysed carcinoma cell line using TriZol RNA reagent (Gibco, in 500 ml of lysis buffer (20 mm Tris pH8.0, 137 mm Life Technologies) according to the manufacturer’s NaCl, 10% glycerol, 1% NP-40 and 10 mm NaF), protocol. The cDNA was produced using 2 mgof supplemented with protease and phosphatase inhibitors isolated RNA with AMV reverse transcriptase (Roche) (1 mm PMSF, 2.5 mg/ml aprotinin, 2.5 mg/ml leupeptin, 1 0 0 and oligo dT20 primer for 1 h at 42 C. 5 and 3 rapid 1mm Na3VO4, and 10 mm b-glycerolphosphate). The amplification of cDNA ends (RACE) was performed lysates were clarified by centrifugation in a TOMY using primers designed from expressed sequence tags refrigerated centrifuge. Protein concentrations were (EST) analysed from GenBankt, which had sequence determined by using Bio-Rad Bradford reagent accord- homology to the initial chicken KFC reported (Yustein ing to the manufacturer’s protocol. et al., 2000). Full-length clones were amplified, using a Equivalent protein aliquots were mixed with an equal high-fidelity Taq polymerase (Roche), and cloned into volume of 2 Â SDS–PAGE sample buffer (125 mm Tris the pCR2.1-TOPO cloning vector (Invitrogen). The pH6.8, 4% SDS, 2 m m EDTA, 20% glycerol, 1% b- primers used for full-length amplification were: forward mercaptoethanol, and 0.6% bromophenol blue), boiled 50 ATG CGT AAA GGG GTG CTG AAG GAC C 30, for 10 min and subjected to SDS–polyacrylamide gel reverse 50 TTA ATC TCA TCT GTA GTC CTC CTT electrophoresis (PAGE). Predetermined molecular AG 30 for hKFC-A; forward 50ATG CCA TCA ACT weight markers were also loaded to determine the AAC AGA GCA GGC AGT CTT AAG GAC C 30, protein size. The proteins were transferred to PVDF reverse 50CTA GCC AAT TCT TTT CAC CAC ATG C membranes (Biotechnology Systems) using a semidry 30 for hKFC-B; forward 50 AGG AAG AGG TGG apparatus (Bio-Rad). Membranes were then blocked in CAA GGG GAA AG 30, and reverse 50 GAG CTG 5% nonfat milk/TBST (20 mm Tris pH7.6, 137 mm GAG CCT ACA TCT TGT CC 30 for hKFC-C. All NaCl, 1% Tween-20) for 1 h at room temperature. clones were then transferred into pcDNA3.1 ( þ ) After blocking, membranes were incubated with pri- (Invitrogen) that had been modified to express 50 T7 mary antibody for either 1 h at room temperature or or HA epitope tag preceding the start codon of the overnight at 41C, then washed three times with TBST clone. All clones were sequenced using WM Keck for 10 min, incubated with HRP-conjugated secondary Biotechnology Resource/HHMI Biopolymer Lab antibody for another hour, and finally washed three DNA Sequencing Group, Yale University. more times with TBST before being developed using Renaissance Western Chemiluminescence Plus reagent (NEN) and exposed to film. Northern blotting analysis C-terminal DNA segments of 1.5–1.7 kbp from each hKFC kinase and a human b-actin probe (Clontech) Immunoprecipitations and in vitro kinase assays 32 were labeled with radioactive [a- P]dCTP by random Kinases were precipitated for 2–3 h at 41C with constant priming (Roche). The labeled probes were purified rotation using either specific antibody/protein-G- through Sephadex G-50 quick-spin columns (Pharma- sepharose beads or reduced glutathione beads (to cia). Each labeled probe was hybridized to the human precipitate GST fusion proteins). The beads were multiple tissue Northern blots (Clontech) and washed washed three times with lysis buffer supplemented with according to the manufacturer’s protocol. The blots both protease and phosphatase inhibitors (described were then exposed to Kodak X-ray film using the above) and one time with a 50 mm Tris pH7.5 plus 1 m m intensifier at À701C for 24–48 h. DTT solution. Finally, the beads were incubated with 20 ml of kinase buffer (50 mm Tris pH7.5, 10 m m MgCl2, m m Cell cultures 1m DTT, 20 m ATP) supplemented with 1 mgof substrate along with 10 mCi [g-32P]ATP per reaction. COS-1 monkey kidney and DU-145 prostate carcinoma After 20 min, the reactions were terminated by the cell lines were maintained in Dulbecco’s Modified addition of 2 Â SDS sample buffer. The samples were Eagle’s Media (DMEM) high glucose plus 10% fetal then loaded onto a 15% SDS–PAGE gel and electro- bovine serum (FBS) supplemented with 1% penicillin/ phoresed until the dye front was approximately one inch streptomycin. Jurkat T-cell lymphoma cell line was from running off the gel. The gel was transferred to

Oncogene hKFCs, a new subfamily of Ste20s JT Yustein et al 6140 PVDF and then exposed to Packard InstantImager signal-regulated kinase; FITC, fluorescein isothiocyanate; electronic autoradiograhy, as well as to film at À701C. GCK, germinal center kinase; GST, glutathione S-transferase; HA, hemagglutinin; HRP, horseradish peroxidase; IL- 1b, interleukin-1b; KFC, kinase from chicken; LFA, Immunofluorescence and subcellular localization leukocyte-function-associated antigen; MAPK, mitogen- COS-1 cells were plated onto sterile coverslips placed in activated protein kinase; MBP, myelin basic protein; MEK/ six-well plates and incubated overnight at 371C. Cells MKK, MAPK kinase; MEKK, MEK-kinase; MLK, mixed- were then transfected with HA-tagged forms of each lineage kinase; MST1, mammalian Ste20; MTK1, MAP three kinase 1; p38/mHOG, mammalian high-osmolarity glycerol kinase. After 48 h, cells were fixed with a 50/50 kinase; PAGE, polyacrylamide gel electrophoresis; PAK, methanol/acetone solution for 10 min at 41C, air-dried, p21-activated kinase; PBL, peripheral blood leukocytes; washed once with PBS, and incubated with 3% BSA/ PDGF, platelet-derived growth factor; PSK, prostate-specific TBST blocking solution for 1 h at room temperature. kinase; RACE, rapid amplification of cDNA ends; SAPK/ Afterwards, the coverslips were incubated with mono- JNK, stress-activated protein kinase/jun N-terminal kinase; clonal anti-HA antibody (1 : 250 dilution) for 1 h at SH3, src-homology 3; Ste20, sterile-20; TAK1, TGF-b- room temperature, washed three times with PBS, and activated kinase 1; TAO1, thousand and one protein kinase; then incubated with fluorescein isothiocyanate (FITC)- TGF-b, transforming growth factor-b; TNF-a, tumor conjugated goat anti-mouse secondary antibody for 1 h necrosis factor-a; UV, ultraviolet. at room temperature. Finally, the coverslips were washed two times with PBS prior to examining the cells Acknowledgements via a Leica fluorescent microscope. We thank Dr Ichijo (Tokyo Medical and Dental University) for providing the HA-tagged ASK1 clone. This work is supported by NIHgrants (CA39207, CA46613 and CA57179) and a DOD prostate cancer grant (DAMD17-02-1-0020) to Abbreviations HJK JTY is supported through the NIH-supported Medical ASK1, apoptosis signal-regulating kinase 1; CRIB, Cdc42/ Scientist Training Program. We also acknowledge the support Rac1 interactive-binding domain; EGF, epidermal growth of Cancer Center Core Grant to UC Davis and the factor; ER, endoplasmic reticulum; ERK, extracellular Environmental Health Science Center Grant to UC Davis.

References

Assefa Z, Garmyn M, Bouillon R, Merlevede W, Vandenheede Fanger GR, Gerwins P, Widmann C, Jarpe MB and Johnson JR and Agonstinis P. (1997). J. Invest. Dermatol., 108, 886– GL. (1997). Curr. Opin. Genet. Dev., 7, 67–74. 891. Frost JA, Xu S, Hutchison MR, Marcus S and Cobb MH. Becker E, Huynh-Do U, Holland S, Pawson T, Daniel TO and (1996). Mol. Cell. Biol., 16, 3707–3713. Skolnick EY. (2000). Mol. Cell. Biol., 20, 1537–1545. Geng Y, Valbracht J and Lotz M. (1996). J. Clin. Invest., 98, Bost F, McKay R, Bost M, Potapova O, Dean NM and 2425–2430. Mercola D. (1999). Mol. Cell. Biol., 19, 1938–1949. Holly SP and Blumer KJ. (1999). J. Cell Biol., 147, 845–856. Boulton TG, Nye SH, Robbins DJ, Ip NY, Radziejewska E, Hu MC, Qiu WR, Wang X, Meyer CF and Tan TH. (1996). Morgenbesser SD, DePinho RA, Panayotatos N, Cobb MH Genes Dev., 10, 2251–2264. and Yancopoulos GD. (1991). Cell, 65, 663–675. Hutchison M, Berman KS and Cobb MH. (1998). J. Biol. Brown JL, Stowers L, Baer M, Trejo J, Coughlin S and Chant Chem., 273, 28625–28632. J. (1996). Curr. Biol., 6, 598–605. Ichijo H, Nishida E, Irie K, ten Dijke P, Saitoh M, Moriguchi Charest DL, Mordret G, Harder KW, Jirik F and Pelech SL. T, Takagi M, Matsumoto K, Miyazono K and Gotoh Y. (1993). Mol. Cell. Biol., 13, 4679–4690. (1997). Science, 275, 90–94. Chen Z, Hutchison M and Cobb MH. (1999). J. Biol. Chem., Katz P, Whalen G and Kerhl JH. (1994). J. Biol. Chem., 269, 274, 28803–28807. 16802–16809. Chung CY and Firtel RA. (1999). J. Cell Biol., 147, 559–576. Kyriakis JM and Avruch J. (1996). Bioessays, 18, 567–577. Creasy CL and Chernoff J. (1995a). Gene, 167, 303–306. Ling P, Meyer CF, Redmond LP, Shui JW, Davis B, Rich RR, Creasy CL and Chernoff J. (1995b). J. Biol. Chem., 270, Hu MC, Wange RL and Tan TH. (2001). J. Biol. Chem., 21695–21700. 276, 18908–18914. Dan I, Watanabe NM, Kobayashi T, Yamashita-Suzuki K, Ling P, Yao Z, Meyer CF, Wang XS, Oehrl W, Feller SM and Fukagaya Y, Kajikawa E, Kimura WK, Nakashima TM, Tan TH. (1999). Mol. Cell. Biol., 19, 1359–1368. Matsumoto K, Ninomiya-Tsuji J and Kusumi A. (2000). Lu HT, Yang DD, Wysk M, Gatti E, Mellman I, Davis RJ FEBS Lett., 469, 19–23. and Flavell RA. (1999). EMBO J., 18, 1845–1857. Dan I, Watanabe NM and Kusumi A. (2001). Trends Cell Lupas A, Van Dyke M and Stock J. (1991). Science, 252, Biol., 11, 220–230. 1162–1164. Eby JJ, Holly SP, van Drogen F, Grishin AV, Peter M, Drubin Manser E, Leung T, Salihuddin H, Zhao ZS and Lim L. DG and Blumer KJ. (1998). Curr. Biol., 8, 967–970. (1994). Nature, 367, 40–46. Endo J, Toyama-Sorimachi N, Taya C, Kuramochi-Miyagawa Moore TM, Garg R, Johnson C, Coptcoat MJ, Ridley AJ and S, Nagata K, Kuida K, Takashi T, Yonekawa H, Yoshizawa Morris JD. (2000). J. Biol. Chem., 275, 4311–4322. Y, Miyasaka N and Karasuyama H. (2000). FEBS Lett., Moriguchi T, Kuroyanagi N, Yamaguchi K, Gotoh Y, Irie K, 468, 234–238. Kano T, Shirkabe K, Muro Y, Shibuya H, Matsumoto K, Enslen H, Brancho DM and Davis RJ. (2000). EMBO J., 19, Nishida E and Hagiwara M. (1996). J. Biol. Chem., 271, 1301–1311. 13675–13679.

Oncogene hKFCs, a new subfamily of Ste20s JT Yustein et al 6141 Oehrl W, kardinal C, Ruf S, Adermann K, Groffen J, Feng Tibbles LA, Ing YL, Kiefer F, Chan J, Iscove N, Woodgett JR GS, Blenis J, Tan THand Feller SM. (1998). Oncogene, 17, and Lassam NJ. (1996). EMBO J., 15, 7026–7035. 1893–1901. Wolf E, Kim PS and Berger B. (1997). Protein Sci., 6, 1179– Parrini MC, Lei M, Harrison SC and Mayer BJ. (2002). Mol. 1189. Cell, 9, 73–83. Wysk M, Yang DD, Lu HT, Flavell RA and Davis RJ. (1999). Paul A, Wilson S, Belham CM, Robinson CJ, Scott PH, Proc. Natl. Acad. Sci. USA, 96, 3763–3768. Gould GW and Plevin R. (1997). Cell Signal, 9, Xing J, Kornhauser JM, Xia Z, Thiele EA and Greenberg ME. 403–410. (1998). Mol. Cell. Biol., 18, 1946–1955. Peraldi P, Scimeca JC, Filloux C and Van Obberghen E. Yoneda T, Imaizumi K, Oono K, Yui D, Gomi F, Katayama (1993). Endocrinology, 132, 2578–2585. T and Tohyama M. (2001). J. Biol. Chem., 276, 13935– Pombo CM, Kehrl JH, Sanchez I, Katz P, Avruch J, Zon LI, 13940. Woodgett JR, Force T and Kyriakis JM. (1995). Nature, Yustein JT, Li D, Robinson D and Kung HJ. (2000). 377, 750–754. Oncogene, 19, 710–718. Rousseau S, Houle F, Landry J and Huot J. (1997). Oncogene, Zanke BW, Boudreau K, Rubie E, Winnett E, Tibbles LA, 15, 2169–2177. Zon L, Kyriakis J, Liu FF and Woodgett JR. (1996a). Curr. Su YC, Han J, Xu S, Cobb M and Skolnik EY. (1997). EMBO Biol., 6, 606–613. J., 16, 1270–1290. Zanke BW, Rubie EA, Winnett E, Chan J, Randall S, Parsons Takekawa M, Posas F and Saito H. (1997). EMBO J., 16, M, Boudreau K, McInnis M, Yan M, Templeton DJ and 4973–4982. Woodgett JR. (1996b). J. Biol. Chem., 271, 29876–29881. Tassi E, Biesova Z, Di Force PP, Gutkind JS and Wong WT. Zhou G, Lee SC, Yao Z and Tan TH. (1999). J. Biol. Chem., (1999). J. Biol. Chem., 274, 33287–33295. 274, 13133–13138.

Oncogene