MLK3 (mixed lineage kinase 3) is a MAP3K [MAPK (mitogen-activated protein kinase) kinase kinase] that activates multiple MAPK pathways, including the JNK (c-Jun N-terminal kinase) pathway. Immunoblotting of lysates from cells ectopically expressing active MLK3 revealed an additional immunoreactive band corresponding to a CTF (C-terminal fragment) of MLK3. In the present paper we provide evidence that MLK3 undergoes proteolysis to generate a stable CTF in response to different stimuli, including PMA and TNFα (tumour necrosis factor α). The cleavage site was deduced by Edman sequencing as between Gln251 and Pro252, which is within the kinase domain of MLK3. Based on our homology model of the kinase domain of MLK3, the region containing the cleavage site is predicted to reside on a flexible solvent-accessible loop. Site-directed mutagenesis studies revealed that Leu250 and Gln251 are required for recognition by the ‘MLK3 protease’, reminiscent of the substrate specificity of the coronavirus 3C and 3CL proteases. Whereas numerous mammalian protease inhibitors have no effect on MLK3 proteolysis, blockade of the proteasome through epoxomicin or MG132 abolishes PMA-induced production of the CTF of MLK3. This CTF is able to heterodimerize with full-length MLK3, and interact with the active form of the small GTPase Cdc42, resulting in diminished activation loop phosphorylation of MLK3 and reduced signalling to JNK. Thus this novel proteolytic processing of MLK3 may negatively control MLK3 signalling to JNK.
- c-Jun N-terminal kinase (JNK)
- mitogen-activated protein kinase (MAPK)
- mixed lineage kinase 3 (MLK3)
- tumour necrosis factor α (TNFα)
The activities and signalling pathways of many protein kinases are controlled through post-translational modifications. Reversible phosphorylation is the most common post-translational alteration impacting protein kinase signalling. For example, phosphorylation within the so-called activation loop promotes a catalytically competent conformation and is responsible for the activation of the majority of protein kinases.
Proteolysis is another post-translational modification that can alter protein kinase activity. Activation of several protein kinases by proteolytic cleavage of their inhibitory subunits has been reported. For instance, removal of the N-terminal inhibitory region of GSK3 (glycogen synthase kinase 3) by the calcium-activated protease calpain increases its kinase activity ; deregulation of GSK3 activity has been implicated in the pathogenesis of neurodegenerative diseases and diabetes [2–5]. Likewise, proteolytic cleavage of PKCδ (protein kinase Cδ) by active caspase 3 releases the N-terminal regulatory domain of PKCδ from its catalytic region, thus activating PKCδ [6,7]. The PKC relative, PKN (protein kinase novel), is also activated in the same way .
On the other hand, proteolysis can also destroy protein kinases, resulting in disruption of their signalling pathways. The common mechanism for degradation of protein kinases is through poly-ubiquitylation, followed by proteasome-mediated degradation. In the case of the MAP3K MEKK1 [MAPK/ERK (extracellular-signal-regulated kinase) kinase kinase 1], its PHD (plant homeodomain) functions as an E3 ligase to facilitate the poly-ubiquitylation of its downstream MAPK (mitogen-activated protein kinase), ERK1/2, thereby attenuating this signalling pathway . Notably, catalytically active MEKK1 itself is regulated by ubiquitylation, mediated through its own PHD, leading to proteasome-mediated degradation of MEKK1 . A number of pathogens utilize proteases to disrupt host signalling pathways. In the case of Bacillus anthracis, a key virulence factor is anthrax lethal factor, which is a zinc metalloprotease that blocks mammalian MAPK pathways through proteolytic cleavage of the N-terminal region of MAPK kinases including MKK1, 2, 3, 4, 6 and 7 [11,12].
MLK3 (mixed lineage kinase 3) is a serine/threonine kinase that regulates multiple MAPK pathways. MLK3 functions as a MAP3K  to primarily activate the JNK (c-Jun N-terminal kinase) signalling pathway [14–16]. In certain experimental settings, MLK3 has also been implicated in the activation of the p38 [15,17,18] and ERK pathways [19,20].
In addition to its kinase domain, MLK3 contains several protein-interaction domains. An N-terminal SH3 (Src homology 3) domain is followed by a catalytic domain, leucine zipper regions and a CRIB (Cdc42/Rac-interacting binding) motif. Work from our laboratory has shown that the activity of MLK3 is regulated by these domains [21–24]. For instance, binding of the activated forms of the small GTPases, Cdc42 and Rac, promotes activation loop (auto)phosphorylation  and catalytic activity [21,25] of MLK3, which phosphorylates MKK4/7 to activate JNK [25–27]. In the present study, we provide evidence for a novel mechanism through which MLK3 activity and signalling can be regulated, namely proteolysis. We demonstrate that a proteolytic event occurs within the catalytic domain of MLK3, leading to the production of a stable CTF (C-terminal fragment). Further investigation revealed that extracellular stimuli can induce this proteolytic event. Based on site-directed mutagenesis studies, the ‘MLK3 protease’ displays remarkable sequence specificity. Inhibitor studies suggest that the proteasome is required for generation of the CTF of MLK3. Finally, we show that the CTF of MLK3 inhibits MLK3-mediated JNK activation, suggesting that this may be a novel mechanism through which MLK3 signalling is negatively regulated.
Protease inhibitors lactacystin, MG132, epoxomicin, MDL28170, leupeptin, L-685458, Z-VAD-FMK (benzyloxycarbonyl-Val-Ala-DL-Asp-fluoromethylketone) and E64d were purchased from Calbiochem. Pepstatin A, chloroquine and PMA were purchased from Sigma. Human TNFα (tumour necrosis factor α) was from Genentech.
Construction of mammalian expression vectors and site-directed mutagenesis
Construction of cytomegalovirus-based expression vectors carrying the cDNAs for wild-type MLK3 (pRK5-mlk3), and the kinase-defective mutant of MLK3 (pRK5-mlk3 K144A) have been described previously . The expression plasmid encoding the N-terminal FLAG-tagged constitutively active variant of Cdc42 (pRK5-NFLAG-Cdc42Val12) was provided by Avi Ashkenazi (Genentech). For the construction of the HA (haemagglutinin)-tagged CTF, the cDNA corresponding to MLK3 amino acids 252–847 was amplified by PCR with pRK5-NFlag.mlk3 as the template and then inserted into the pCGN-HA vector. The wild-type MLK3 expression vector (pRK-mlk3) was used as a template for site-directed mutagenesis (to create MLK3 L250A, L250V, Q251A, Q251E, Q251L, Q251N, P252A, P252L and P252D) by the QuikChange site-directed mutagenesis method (Stratagene) with the following primers and their reverse complements: L250A, 5′-CAACATTTTGCTGGCGCAGCCCATTGAGAG-3′; L250V, 5′-CAACATTTTGCTGGTGCAGCCCATTGAGAG-3′; Q251A, 5′-CATTTTGCTGCTGGCGCCCATTGAGAG-3′; Q251E, 5′-CATTTTGCTGCTGGAGCCCATTGAGAG-3′; Q251L, 5′-CATTTTGCTGCTGCTGCCCATTGAGAG-3′; Q251N, 5′-CATTTTGCTGCTGAACCCCATTGAGAG-3′; P252A, 5′-GCTGCTGCAGGCCATTGAGAGTGAC-3′; P252L, 5′-GCTGCTGCAGCTCATTGAGAGTGAC-3′; and P252D, 5′-GCTGCTGCAGGACATTGAGAGTGAC-3′.
The presence of the desired mutation was confirmed by automated DNA sequencing in the Genomics Technology Support Facility at Michigan State University.
Cell culture, transfections and lysis
HEK (human embryonic kidney)-293 cells were cultured in Ham's F12/low glucose Dulbecco's modified Eagle's media (1:1) (Invitrogen) supplemented with 8% fetal bovine serum (Invitrogen), 2 mM L-glutamine and 100 units/ml penicillin/streptomycin (Invitrogen). Immortalized human T-lymphocyte Jurkat TAg cells were cultured in RPMI 1640 (Invitrogen) supplemented with 8% fetal bovine serum, 2 mM L-glutamine, 50 μM 2-mercaptoethanol and 100 units/ml penicillin/streptomycin. HEK-293 cells were transfected with plasmid DNA using the calcium phosphate method or Lipofectamine™ 2000 (Invitrogen) following the manufacturer's instructions. Jurkat TAg cells were transfected by electroporation (250 V, 950 μF) using a Bio-Rad Gene Pulser electroporator. Cells were harvested, washed with ice-cold PBS and lysed for 5 min on ice by the addition of 1 ml of lysis buffer (50 mM Hepes, pH 7.5, 150 mM NaCl, 1.5 mM MgCl2, 2 mM EGTA, 1% Triton X-100, 10% glycerol, 10 mM NaF, 1 mM Na4PPi, 100 μM β-glycerophosphate, 1 mM Na3VO4, 2 mM PMSF and 0.15 units/ml aprotinin). To disrupt cells under denaturing conditions, 0.2% SDS and/or 8 M urea were added to the lysis buffer.
Anti-MLK3 or -FLAG (Sigma–Aldrich) antibody (0.4 μg/μl slurry) was prebound to Protein A–agarose beads for 0.5 h at room temperature (21 °C). Clarified lysate (300 μl) was incubated with 20 μl of antibody-bound Protein A–agarose for 90 min at 4 °C. Immunoprecipitates were washed with HNTG buffer (20 mM Hepes, pH 7.5, 150 mM NaCl, 0.1% Triton X-100 and 10% glycerol).
SDS/PAGE and Western blot analysis
Lysates and immunoprecipitates of proteins were resolved by SDS/PAGE. Proteins were transferred on to nitrocellulose membranes and immunoblotted using anti-MLK3 antibody (7.7 μg/ml) or the antibodies against the protein of interest, followed by the appropriate horseradish peroxidase-conjugated secondary antibody (Bio-Rad). Western blots were developed by chemiluminescence.
N-terminal sequence analysis
HEK-293 cells (in 16×150 mm tissue culture plates) transiently expressing MLK3 were disrupted in lysis buffer and immunoprecipitated with 77 μg of MLK3 antibody (0.4 μg/μl slurry) for 1.5 h as described above. Immunoprecipitates were washed six times with HNTG buffer and resolved by SDS/PAGE. Proteins were transferred on to a PVDF membrane and stained with Coomassie Blue. The membrane slice corresponding to the CTF of MLK3 was excised from the membrane and washed several times with water. The N-terminus of the fragment was sequenced by automated repetitive cycles of Edman degradation in the Genomics Technology Support Facility at the Michigan State University.
Homology modelling of the kinase domain of MLK3
The Clustal W program was utilized to generate multiple (local) amino acid sequence alignments of the kinase domain of MLK3 (amino acids 111–382) using structures from the PDB database. Based on their high sequence homology (Z-score), four structures of the kinase domain of murine Abl (PDB numbers 1FPU, 1IEP, 1M52 and 1OPJ) and a structure of the kinase domain of human B-Raf (PDB number 1UWH) were used as templates to generate a model of the kinase domain of MLK3 from the aligned templates using the program MODELLER 9v.1  with the model default options. The quality of the model was assessed using Ramachandran plots in conjunction with PROCHECK  and the software package MOE (Molecular Operating Environment) from Chemical Computing Group. The side-chain residues were relaxed and evaluated using the AMBER force field implemented in MOE followed by backbone torsion evaluation, once again, using Ramachandran plots.
MLK3 undergoes activity-dependent proteolysis in HEK-293 cells
When wild-type MLK3 was transiently expressed in HEK-293 cells, immunoblotting of cellular lysates revealed a band corresponding to full-length MLK3 as well as a band of approx. 65–70 kDa (Figure 1A). Since the anti-MLK3 antibody was raised against the C-terminal eight amino acids of MLK3 , the observed 65–70 kDa immunoreactive band could correspond to a stable CTF of MLK3.
To exclude the possibility that the CTF was generated in vitro after cellular lysis, denaturing agents, including urea and/or SDS, in addition to protease inhibitors, were included in the lysis buffer. As shown in Figure 1(A), the CTF of MLK3 was observed even when cells were lysed in buffer containing 8 M urea, 0.2% SDS and a cocktail of protease inhibitors, suggesting that proteolysis of MLK3 occurs within cells. An antibody raised against the N-terminal SH3 domain of MLK3 detects full-length MLK3, but fails to detect a band corresponding to a predicted N-terminal fragment (results not shown), perhaps indicating that it is unstable.
When transiently expressed in HEK-293 cells, MLK3 displays high basal activity [13,21–24], which is increased upon co-expression with activated forms of the small GTPase Cdc42 [21–23]. Expression of wild-type MLK3 leads to production of the CTF, which is retained upon co-expression with the constitutively active variant of Cdc42, Cdc42V12 (Figure 1B, lanes 1 and 2). However, when the catalytically inactive version, MLK3 K144A, is expressed alone or with Cdc42V12, no CTF is detected, consistent with the idea that MLK3 activity is required for generation of the CTF in HEK-293 cells.
Induction of MLK3 proteolysis by exogenous stimuli
Upon overexpression in HEK-293 cells, MLK3 has high basal activity and induces activation of the JNK, p38 and ERK MAPK pathways ([22,26,30], and results not shown). Our finding that MLK3 catalytic activity is required for its proteolysis in HEK-293 cells presents two formal possibilities: either an active conformation of MLK3 is the substrate for the ‘MLK3 protease’ or an MLK3-induced signalling pathway promotes proteolysis of MLK3. To more readily distinguish between these possibilities, we opted to use Jurkat TAg cells in which transiently expressed MLK3 has low basal activity and undetectable activation of MAPK pathways. To induce activation of the MAPK signalling pathways, we used the phorbol ester PMA. When wild-type MLK3 is transiently expressed in Jurkat TAg cells, no CTF is observed. Upon treatment with PMA, the CTF accumulates in a time-dependent manner (Figure 2A), indicating that MLK3 proteolysis is an inducible event. Notably, multiple CTF bands of MLK3 appeared in response to 4 h of stimulation of with PMA (Figure 2A). These hyper-shifted CTF bands resulted from PMA-induced phosphorylation, since treatment of the cellular lysates with lambda phosphatase rendered the bands undetectable (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/427/bj4270435add.htm). The kinase-defective mutant MLK3 K144A, which fails to autophosphorylate within its kinase domain, also undergoes proteolysis in response to PMA treatment (Figure 2B). These results taken together indicate that it is not the active conformation of the MLK3 kinase domain that dictates proteolysis but, rather, proteolysis of MLK3 is promoted through activation of signalling pathways that are shared by MLK3 and PMA (Figure 2B). This argument is further strengthened by our observation that PMA also induces the proteolysis of the catalytically inactive mutant MLK3 K144A in HEK-293 cells (results not shown). Finally, treatment of Jurkat TAg cells with TNFα also generates the CTF of MLK3 (Figure 2C), suggesting physiologically relevant stimuli can promote MLK3 proteolysis.
ERK activation is required for MLK3 proteolysis
PMA activates the ERK, the JNK and the p38 pathways , as does MLK3 . It is possible that activation of one or more of the MAPK pathways is required for the proteolytic event of MLK3. To test this hypothesis, the ability of specific inhibitors of ERK, JNK or p38 pathways to block PMA-induced MLK3 proteolysis was assessed in Jurkat TAg cells. Jurkat TAg cells transiently expressing MLK3 were pretreated with the MEK inhibitor U0126 , the JNK inhibitor SP 600125 [30,33], the p38 inhibitor SB 203580  or a combination of all three for 0.5 h prior to treatment with PMA for 3 h. As shown in Figure 3(A), the MEK inhibitor U0126 alone largely blocks the generation of the CTF of MLK3 upon PMA stimulation. Neither the JNK inhibitor nor the p38 inhibitor alone substantially blocked the PMA-induced MLK3 proteolysis. Generation of the CTF of MLK3 is potently suppressed by using all three inhibitors. Consistently, TNFα-induced CTF generation of MLK3 is also blocked by pretreatment with the MEK inhibitor U0126 in Jurkat TAg cells (Figure 3B). These results suggest that ERK activation is required for the MLK3 proteolytic event. These results are supported by our findings in HEK-293 cells, where expression of MLK3 alone is able to activate ERK and promotes production of the CTF, whereas Cdc42V12 expressed alone cannot activate ERK (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/427/bj4270435add.htm) and therefore cannot promote proteolysis of the kinase-defective MLK3 K144A to generate the CTF. Thus this finding is consistent with our results from HEK-293 cells transiently expressing MLK3 (Figure 1B).
Molecular determinants for MLK3 proteolysis
We sought to determine the cleavage site within MLK3 that generates the CTF. Following large-scale transient expression of MLK3 in HEK-293 cells, the full-length and CTF of MLK3 were immunopurified using the C-terminal-specific anti-MLK3 antibody, resolved by SDS/PAGE, transferred on to PVDF membrane, and visualized by Coomassie Blue staining. The band corresponding to the CTF was excised and subjected to Edman sequencing. The first amino acid of the CTF of MLK3 was deduced as Pro252, which lies within the catalytic domain of MLK3. To better understand the features required for MLK3 proteolysis, we constructed a homology model of the kinase domain of MLK3 (Figure 4A). Similar to other protein kinases, the model of the kinase domain of MLK3 comprises a small N-terminal lobe composed primarily of β-pleated sheets and a larger C-terminal lobe composed largely of α-helices. The active site cleft in which Mg2+, ATP and the protein substrate bind is found between the two lobes. This region is predicted to form two flexible loops, one of which contains the activation loop phosphorylation sites and the other of which contains the deduced site of proteolysis, Gln251–Pro252. Notably, the proteolytic site is predicted to be in an unstructured region which could be accessible to the ‘MLK3 protease’ (Figure 4A).
Most proteases display sequence specificity toward their substrates, particularly at the so-called P1 and P2 sites, where the amino acid residues flanking the cleavage sites of the substrate are designated as P3-P2-P1-P1′-P2′-P3′ and the cleavage site is between P1 and P1′. A series of MLK3 site-directed mutants containing amino acid substitutions at the putative P1 and P1′ positions, Gln251 and Pro252 respectively, were constructed and transiently expressed in HEK-293 cells. As shown in Figure 4(B), multiple amino acid substitutions of the proline residue at 252, including alanine, leucine and aspartate have no impact on the generation of the CTF of MLK3. However, substitution of Gln251 with alanine, glutamate, leucine or asparagine at the P1 position (Figure 4B) abolishes generation of the CTF. Furthermore, when Leu250 at the putative P2 position was substituted with either alanine or valine, formation of the CTF of MLK3 is also abrogated. These results indicate that both Leu250 and Gln251, but not Pro252, are critical for MLK3 proteolysis, consistent with the notion that amino acids at P1 and P2, but not at P1′ positions, are key residues for protease recognition and subsequent cleavage.
To determine whether the sequence requirements for MLK3 proteolysis determined in HEK-293 cells are retained in Jurkat TAg cells, several of the proteolysis-resistant MLK3 engineered variants were expressed in Jurkat TAg cells. Whereas wild-type MLK3 generated a CTF, the MLK3 mutants Q251A, L250A and L250V failed to undergo proteolysis to produce the CTF in response to PMA treatment (Figure 4C), providing further evidence suppporting Leu250 and Gln251 as sequence determinants for the ‘MLK3 protease’.
Proteasome inhibitors block MLK3 proteolysis
In order to identify the protease(s) responsible for MLK3 proteolysis, we tested the ability of various protease inhibitors to block PMA-induced proteolysis of MLK3 expressed in Jurkat TAg cells. Treatment of Jurkat TAg cells with pepstatin A, a general aspartyl protease inhibitor; leupeptin or E64d, general cysteine protease inhibitors; Z-VAD-FMK, a Pan-caspase inhibitor; MDL 28170, a calpain inhibitor; and chloroquine, a lysosomal protease inhibitor, had no effect on PMA-induced MLK3 proteolysis (results not shown).
Pretreatment of Jurkat TAg cells with MG132, a general proteasome inhibitor , completely blocked PMA-induced MLK3 proteolysis (Figure 5A). However, pretreatment with another proteasome inhibitor, lactacystin, only partially suppressed PMA-induced MLK3 proteolysis, perhaps because lactacystin is relatively unstable. Since MG132 has also been reported to inhibit γ-secretase [36,37], it is conceivable that the MLK3 proteolysis is mediated through γ-secretase. To exclude this possibility, Jurkat TAg cells transiently expressing MLK3 were pretreated with L685,458, a specific inhibitor of γ-secretase, followed by PMA stimulation. As shown in Figure 5(A), L685,458 fails to prevent PMA-induced proteolysis of MLK3.
To further corroborate the requirement for the proteasome in mediating MLK3 proteolysis, the stable and highly selective proteasome inhibitor epoxomicin  was used to pretreat MLK3-expressing Jurkat TAg cells for 1 h followed by PMA stimulation for 2 h. As shown in Figure 5(B), epoxomicin inhibits PMA-induced MLK3 proteolysis in a dose-dependent manner. Taken together, these results suggest that MLK3 proteolysis is mediated through the proteasome or through a proteasome-regulated protease.
The CTF of MLK3 negatively regulates MLK3-JNK signalling
Interestingly, we have observed that the expression of the cleavage site mutant MLK3 Q251E enhances JNK activation as compared with wild-type MLK3 (see Supplementary Figure S3 at http://www.BiochemJ.org/bj/427/bj4270435add.htm). This observation has led us to hypothesize that the CTF may negatively regulate MLK3/JNK signalling. The CTF of MLK3 includes a truncated kinase domain, the leucine zipper, the CRIB motif and the C-terminal proline-rich region. Activation loop trans(auto) phosphorylation of MLK3 requires dimerization [22,39]. Since activation of MLK3 is believed to result from transphosphorylation within the context of a homodimer, heterodimerization of the CTF with full-length MLK3 would be predicted to down-regulate MLK3 signalling. To determine whether the CTF forms heterodimers with full-length MLK3, coimmunoprecipitation experiments using differentially-tagged CTF and MLK3 were performed. Total cellular lysates of HEK-293 cells transiently expressing the HA-tagged CTF of MLK3 with or without full-length FLAG–MLK3 were immunoprecipitated using the anti-FLAG antibody and the presence of HA–CTF in the FLAG–MLK3 immunoprecipitates was assessed by immunoblotting. As shown in Figure 6(A), HA-tagged CTF complexes with MLK3.
In addition, the CTF of MLK3 contains an intact CRIB motif, and it is conceivable that the CTF of MLK3 may bind and sequester the active form of the GTPase Cdc42 to regulate MLK3 downstream signalling. To test this idea, cellular lysates from HEK-293 cells expressing Cdc42V12 and CTF of MLK3 were immunoprecipitated with an MLK3 antibody, and the presence of bound Cdc42V12 was assessed by immunoblotting (Figure 6B). As predicted, the CTF of MLK3 is capable of binding to active Cdc42V12.
Activated Cdc42 promotes dimerization  and activation loop phosphorylation of MLK3 , leading to activation of the JNK pathway. To test whether the CTF might inhibit this signalling pathway by hetero-dimerizing with full-length MLK3, and/or by sequestering active Cdc42, we examined the ability of the CTF of MLK3 to block Cdc42-induced activation of MLK3 and its signalling to JNK. As shown in Figure 6(C), co-expression of activated Cdc42 with MLK3 leads to activation of both MLK3 and JNK, as judged by immunoblotting with activation loop phospho-specific antibodies. Notably, expression of the HA-tagged CTF of MLK3 significantly decreases Cdc42V12-induced phosphorylation of MLK3 and JNK (Figure 6C), suggesting that the CTF of MLK3 attenuates Cdc42-induced activation of MLK3 and JNK. These results suggest that the CTF of MLK3 is able to negatively regulate MLK3 activation and signalling.
In addition to phosphorylation, proteolysis is another post-translational alteration with the potential to impact protein kinase activity and signal transduction. In the present study, we have identified and characterized a novel proteolytic event within the catalytic domain of MLK3 that results in the production of a stable CTF that can disrupt MLK3 activity and signalling.
Our evidence indicates that generation of the CTF of MLK3 occurs within cells, as lysis of cells in the presence of protease inhibitors and harsh denaturants failed to block formation of the CTF of MLK3. Although we have not been able to detect whether endogenous MLK3 undergoes proteolysis and generates CTF, our results have shown that production of the CTF is induced by extracellular factors, including PMA and TNFα, bolstering the idea that the CTF of MLK3 may play physiologically important roles.
Based on our observations in both HEK-293 cells and Jurkat TAg cells, we hypothesize that proteolysis of MLK3 is promoted through activation of signalling pathways that are shared by MLK3 and PMA. Our results from the present study have shown that ERK activation may be required for this proteolytic event, as an ERK inhibitor diminished MLK3 proteolysis. However, p38 and JNK inhibitors had modest effects on MLK3 proteolysis, suggesting that these pathways may partially contribute to the generation of the CTF of MLK3 as well. Since PMA is a potent activator of all three MAPK pathways, cross-talk among the different pathways may be important.
For most proteases, substrate specificity is dictated by amino acids occupying the P1 and P2 positions, whereas a range of amino acids is generally tolerated at the P1′ position. Our site-directed mutagenesis studies revealed that proteolysis occurs only when a glutamine residue occupies P1 and a leucine residue occupies the P2 position, whereas multiple diverse amino acid substitutions at the P1′ position had no effect on MLK3 proteolysis, indicating that MLK3 proteolysis is quite specific. A search of the MEROPS peptidase database (http://merops.sanger.ac.uk)  failed to identify any mammalian proteases with these substrate requirements. However, some substrates of mammalian cysteine proteases of the calpain and cathepsin families do contain glutamine residues at the P1 site. Yet pretreatment of cells with the general cysteine protease inhibitor leupeptin, the calpain inhibitor MDL 28170, or the cathepsin inhibitor E64d (results not shown) failed to block MLK3 proteolysis, indicating that these proteases are probably not responsible for the generation of the CTF of MLK3.
Interestingly, the proteasome inhibitors MG132 and epoxomicin, were able to block the MLK3 proteolytic event. It is conceivable that the proteasome is directly responsible for generation of the CTF of MLK3, although the CTF has a predicted molecular mass of 66 kDa, and the proteasome typically degrades its ubiquitylated substrate proteins into small peptides of approx. 8 amino acids. However, there are a handful of examples in which the proteasome is responsible for endoproteolytic processing of proteins to yield stable bioactive fragments . For instance, the p105 precursor subunit of the transcription factor NF-κB (nuclear factor κB) undergoes proteasome-mediated endoproteolytic processing in response to signals involved in cell survival and inflammation to generate a p50 fragment that forms an active heterodimer with RelA (p65) [43,44]. Likewise, the endoplasmic reticulum-bound yeast transcription factor, Spt23/Mga2, which is important in maintaining cellular membrane fluidity, is cleaved by the proteasome to produce a cytosolic p90 form that can translocate into the nucleus to drive gene expression . In the case of the developmentally regulated Drosophila transcription factor Ci (cubitis interruptus), alterations in Hedgehog signalling promote proteasome-mediated processing that transforms the transcriptional activator, Ci155, into a transcriptional repressor, Ci75 [46,47]. This proteolytic processing mechanism is conserved in Gli3 (glioblastoma 3), the mammalian homologue of Ci . In each of these cases, the proteasome is believed to initiate cleavage in an unstructured region of the protein and proteolytic processing continues until a region of high structural complexity is encountered [42,49]. In our homology model of the kinase domain of MLK3 (Figure 4A), the region that includes the proteolytic cleavage site is predicted to form a long, flexible loop that appears to be solvent accessible. It is unclear whether this loop would be able to insert into the proteasome channel as would be required for endoproteolytic processing .
The proteasome's general lack of sequence specificity towards its substrates is discordant with our finding that MLK3 proteolysis is highly dependent upon the cleavage site sequence. One explanation, consistent with our findings, is that the proteasome does not directly process MLK3 but, rather, that the proteasome may regulate a sequence-specific MLK3 protease. Intriguingly, the 3C and 3CL proteases of the positive-stranded RNA viruses, picornavirus and coronavirus, display substrate sequence requirements remarkably similar to that of the MLK3 cleavage site . Picornaviruses and coronaviruses are responsible for human and animal diseases, including SARS (severe acute respiratory syndrome). The 3C and 3C-like proteases are primarily responsible for proteolytic processing of viral polyproteins, though host substrates may also exist [51,52]. These proteases have a strong preference for leucine residues at the P2 position and glutamine at the P1 position (MEROPS, ). For example, 27 out of 33 experimentally determined cleavage sites of substrates of the murine coronavirus picornain 3C-like peptidase have a glutamine residue at the P1 position and a leucine at the P2 position. It is possible that a coronavirus protease-related protein exists within human cells. However, a BLAST search (http://blast.ncbi.nlm.nih.gov/Blast.cgi) of translated open reading frames of the human genome failed to reveal human sequences with significant sequence similarity to the SARS coronavirus 3C-like or the murine hepatitis coronavirus picornain 3C-like peptidases, two coronavirus proteases that show the same substrate sequence requirements as the putative ‘MLK3 protease’. When available, it would be interesting to test whether inhibitors of the SARS coronavirus protease are able to block MLK3 proteolysis. If so, these inhibitors could be useful in our future studies directed towards identification of the MLK3 protease.
Hua Zhang and Geou-Yarh Liou were involved in experimental design and performed the majority of the experimental work; Eva Miller performed some additional experiments. Steve Seibold generated the homology model of the MLK3 kinase domain. Weiqin Chen first demonstrated the PMA induction of the C-terminal fragment. Kathleen Gallo was involved in the design of the project and supervised all of the experiments. Kathleen Gallo, Hua Zhang, and Geou-Yarh Liou were involved in writing the manuscript.
This work was supported by an American Cancer Society (ACS) Research Scholar Grant [grant number RSG-03–084 (to K.A.G.)]. We acknowledge the estate of Lela M. Soulby for partial funding of the ACS grant to K.A.G.
We thank Dr Michele Fluck for critical reading of this manuscript.
Abbreviations: Ci, cubitis interruptus; CTF, C-terminal fragment; CRIB, Cdc42/Rac-interacting binding; ERK, extracellular-signal-regulated kinase; GSK3, glycogen synthase kinase 3; HA, haemagglutinin; HEK, human embryonic kidney; JNK, c-Jun N-terminal kinase; MAP3K, mitogen-activated protein kinase kinase kinase; MAPK, mitogen-activated protein kinase; MEKK1, MAPK/ERK kinase kinase 1; MKK, MAPK kinase; MLK3, mixed lineage kinase 3; MOE, Molecular Operating Environment; PHD, plant homeodomain; PKC, protein kinase C; SARS, (evere acute respiratory syndrome; SH3, Src homology 3; TNFα, tumour necrosis factor α; Z-VAD-FMK, benzyloxycarbonyl-Val-Ala-DL-Asp-fluoromethylketone
- © The Authors Journal compilation © 2010 Biochemical Society