Chemical rescue of a mutant protein-tyrosine kinase.

Protein-tyrosine kinases contain a catalytic loop Arg residue located either two or four positions downstream of a highly conserved Asp residue. In this study, the role of this Arg (Arg-318) in the protein-tyrosine kinase C-terminal Src kinase (Csk) was investigated. The observed k(cat) for phosphorylation of the random copolymer poly(Glu,Tyr) substrate by Csk R318A is approximately 3000-fold smaller compared with that of wild type Csk, whereas the K(m) values for ATP and poly(Glu,Tyr) are only mildly affected. The k(cat) value for poly(Glu,Tyr) phosphorylation by the Csk double mutant A316R,R318A is 100-fold greater than the k(cat) value for the single R318A mutant, suggesting that an Arg positioned at the alternative location fulfills a similar function as in wild type. Csk R318A kinase activity can also be partially recovered by several exogenous small molecules including guanidinium and imidazole. These molecules contain key features whose roles in catalysis can be rationalized from a known x-ray structure of the insulin receptor tyrosine kinase. Imidazole is the best of these activators, enhancing phosphorylation rates by Csk R318A up to 100-fold for poly(Glu,Tyr) and significantly stimulating Csk R318A phosphorylation of the physiologic substrate Src. This chemical rescue of mutant protein kinase activity might find applications in cell signal transduction experiments.

Protein-serine/threonine and -tyrosine kinases are important catalysts in cell signal transduction (1)(2). As a consequence, there has been considerable interest in developing novel tools to enhance our current understanding of protein kinase function (3)(4)(5)(6)(7)(8)(9)(10)(11). One particularly interesting method to study the function of protein kinases has been developed by McMahon and co-workers (3). In this approach, the kinase domain is genetically fused to the estrogen receptor ligand binding domain. It has been found that the catalytic activity of the fused kinase is stimulated if estrogen ligands are added to the cell culture media. This powerful technology has been used to characterize the targets and functions of raf kinase (3,4). However, the potentially unwanted biological impact of the additional estrogen receptor domain suggests that the development of alternative, complementary approaches is desirable.
There are a number of reported examples where mutant, catalytically defective enzymes have been reactivated by introducing small molecules that have the requisite properties of the altered residues (12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23). By making slight structural changes in the small molecule activators, it is possible to learn more about the specific molecular contributions of the altered amino acid residue toward catalysis. Our goal was to apply this chemical rescue technique to a mutant, catalytically impaired protein kinase for eventual use in signaling studies.
The catalytic loop of protein-tyrosine and -serine/threonine kinases is highly conserved among each of these enzyme families. In serine/threonine kinases, a lysine residue is found at the n ϩ 2 position, where n is a critical catalytic loop aspartate residue. In protein-tyrosine kinases, an arginine residue is instead present in the catalytic loop at either the n ϩ 2 position or n ϩ 4 position (Fig. 1). X-ray crystallography of a peptide substrate and ATP analogue complexed to the insulin receptor protein-tyrosine kinase catalytic domain has shown that this catalytic loop arginine residue is within hydrogen bonding distance of the substrate tyrosine phenol, as well as the highly conserved aspartate (Fig. 2) (24). To date, the functional role and contributions of this arginine to protein-tyrosine kinase catalysis have not been investigated in detail but we hypothesized that it might be a good candidate for external replacement given its variable site location in protein-tyrosine kinases.
In this study, we examined the role of the protein-tyrosine kinase Csk 1 catalytic loop arginine in substrate phosphorylation. Csk is a protein-tyrosine kinase with importance in inhibiting Src family members, because it catalyzes the site-specific phosphorylation of Src protein tails (25)(26). Csk has proven to be a particularly useful system for exploring mechanistic and substrate selectivity issues in protein-tyrosine kinase function (27)(28)(29)(30)(31)(32), because it does not undergo autophosphorylation to any significant extent.

EXPERIMENTAL PROCEDURES
Materials-Reagents were purchased from commercially available sources unless otherwise indicated. Appropriate primers for the QuikChange® mutagenesis were purchased from Integrated Data Technologies and were purified by either preparative PAGE or standard desalting techniques prior to use. Catalytically inactive chicken Src K295M protein was expressed and purified from Escherichia coli. 2 Preparation of GST-Csk and Its Mutant-The Csk gene was cloned into the pGEX-3Xb vector (Amersham Pharmacia Biotech) via N-terminal NdeI and C-terminal HindIII restriction sites. All GST-Csk mutants were prepared using the Quik-Change® procedure by Stratagene. For all mutants, DNA sequencing confirmed the presence of only the desired mutation. The proteins were overexpressed and purified analogously to previously described methods (27) with the exception that the DH5␣ strain of E. coli cells were grown at 37°C to A 595 ϭ 0.6 prior to induction with 0.4 mM IPTG at 16°C for 22 h. GST-Csk and its mutants were estimated to be ϳ70 -80% pure by 10% SDS-PAGE stained with Coomassie Blue (Fig. 3). The protein yields ranged from between 2-5 mg/liter of E. coli cell culture as calculated from Bradford analysis against BSA standard. Kinase Activity Assays-All kinetic data reported were based on duplicate experiments. Results generally agreed within 20%. The limiting substrate in each case did not exceed 10% for the conditions used.
Poly (Glu,Tyr) As Substrate-Wild type and mutant GST-Csk kinase assays were performed analogously to previous methods monitoring phosphoryl transfer from [␥-32 P]ATP to 4:1 poly(Glu,Tyr) (28). Optimal MnCl 2 concentration was determined separately for each mutant and found to be ϳ2 mM for wild type Csk and ϳ12 mM for each mutant in the absence of activator. Substrate concentrations were at least 5-fold greater than enzyme concentrations for steady-state kinetic measurements. Reaction velocities were determined, and data was fit to the Michaelis-Menten equation as described previously (29).
Src K295M As Substrate-Conditions were essentially iden-tical to those for the poly(Glu,Tyr) phosphorylation assay except that phosphorylated Src was visualized using 10% SDS-PAGE with Coomassie stain, and the protein band was cut out and immersed in scintillation fluid to determine cpms. Background (non-Csk mediated) phosphorylation of the Src K295M substrate was subtracted to determine the final reaction velocities.

RESULTS AND DISCUSSION
As in the insulin receptor tyrosine kinase, the catalytic loop arginine of Csk is present at the n ϩ 4 position as referenced to the catalytic loop aspartate. To dissect the role of this arginine (Arg-318) in Csk action, this residue was initially replaced by site-directed mutagenesis with Ala, Gln, Ile, and Lys residues. To simplify the preparation of the mutant proteins, these enzymes were expressed and purified as GST fusion enzymes in E. coli. Previously, it was shown that the kinetic parameters obtained from bacterially expressed Csk are in reasonable agreement with those from rat-derived enzyme (30). Contrary to an earlier report (33), the GST moiety did not significantly affect the K m of ATP or k cat value using the artificial substrate poly(Glu,Tyr), although it did result in a modest (4-to 5-fold) drop in K m for poly(Glu,Tyr) compared with that of the GSTfree material ( Table I).
Each of the above mutants exhibited greatly decreased catalytic efficiency compared with wild type GST-Csk. For GST-Csk R318A and R318Q, the k cat values were reduced 2000-to 3000-fold (Table I). The effects on K m for substrate poly (Glu,Tyr) were relatively minor, with 8-to 10-fold elevations compared with wild type GST-Csk. Interestingly, the optimal manganese concentration for the mutant GST-Csk-catalyzed reactions (12 mM) was higher than that for the wild type reactions (2 mM). A reasonable explanation for this change in Mn dependence is a change in the rate-determining step for the wild type versus the mutant Csk-catalyzed reactions. The higher Mn concentration may impede ADP release, which is likely the rate-limiting step for the wild type enzyme-catalyzed reaction (29), yet augment the rate of the chemical step, which is likely rate-determining for the mutant Csk-catalyzed reactions. Similar behavior was seen in kinetic analyses with a previous Csk mutant (D314E) and in studies in which Mg was used in place of Mn (30).
The GST-Csk R318K mutant showed only about a 180-fold drop in k cat compared with that of wild type enzyme GST-Csk (Table I). The greater activity of R318K compared with R318A and R318Q suggests that the positive charge of the residue side chain could be important in facilitating catalysis. An alternative possibility for the enhanced activity of R318K compared with R318A and R318Q is the greater hydrophobicity of the lysine side chain. Arguing against this hypothesis however was the discovery that the hydrophobic GST-Csk mutant R318I was actually less catalytically active than R318A or R318Q mutants (data not shown).
Given the great importance of the Arg residue in catalysis and its conserved location in the n ϩ 2 or n ϩ 4 positions of Chemical Rescue of a Mutant Protein-tyrosine Kinase 38128 tyrosine kinases, we decided to investigate the impact of the GST-Csk double mutant A316R,R318A. We were intrigued to find that this second mutation (A316R) led to substantial recovery of the activity lost with the single Csk mutant (R318A). The overall k cat for the kinase activity of the double mutant was about 100-fold greater than that of the single mutant (Table I). This rate acceleration confirms that the Arg residue could be displaced N-terminally two positions along the Csk protein backbone and still subserve a salutary effect on kinase action. This result also provides a functional explanation for the presence of a key Arg at two different positions in protein-tyrosine kinases. This catalytic flexibility further bolstered our hypothesis that rescue of R318A by intermolecular addition of small guanidinium-like molecules might be successful.
Analysis of the kinase activity of R318A as a function of guanidinium concentration showed that substantial stimulation (20-fold) of the activity could be observed with maximal activation at around 50 mM guanidinium for 1 M enzyme. As in other examples of chemical rescue, a high concentration is required for saturation of the active site with a small molecule, indicating that the affinity of the activator for the enzyme is relatively weak (12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23). The K m values for ATP (K m ϭ 40 Ϯ 4 M) and poly(Glu,Tyr) (K m ϭ 115 Ϯ 10 g/ml) were only modestly affected by the addition of 50 mM guanidinium to R318A. As a control, it was shown that neither wild type GST-Csk nor GST-Csk R318Q was significantly activated (Ͻ2-fold) by 50 mM guanidinium. These data suggest that the guanidinium molecule substitutes for the role played by the missing arginine side chain of the R318A mutant. As discussed above, chemical rescue of mutant enzymes has been observed in a number of systems, but to our knowledge this result represents the first such example for a mutant protein kinase.
Based on the guanidinium rescue, a number of analogs were studied to learn more about the detailed structural requirements for kinase activation (Table II and Fig. 4). Alkyl-substituted guanidiniums were somewhat less active in kinase rescue than the unsubstituted version. Replacing one of the nitrogens of guanidinium with an oxygen atom resulted in an analog (urea) that showed no ability to activate the kinase mutant. Interestingly, replacement of one of the guanidinium NH 2 groups with a methyl group led to an analog (acetamidinium) that showed kinase rescue similar to guanidinium. The truncated analog ethylamine, however, showed no kinase rescue activity. From the findings with these analogs, it appeared that two proton-donating nitrogens with at least one carrying a positive charge were necessary and sufficient for kinase rescue.
In an attempt to optimize kinase rescue, a series of 5-and 6-membered ring compounds were tested, and of these, imidazole was found to be the best activator of all the small molecules examined (Table II and Fig. 4). ϳ100-fold activation was observed with imidazole at 100 mM concentration, yielding a k cat value that is only ϳ30-fold below wild type and similar to that of the Csk A316R,R318A double mutant (see Fig. 5A and Table I). As in the case of guanidinium, imidazole did not significantly activate wild type GST-Csk (Ͻ2-fold activation). Other analogs showed a range of rescue activities, but the general rule was upheld that two hydrogen bond donating nitrogens with one carrying a positive charge were necessary for activation. The above results are consistent with the x-ray structure of the insulin receptor protein-tyrosine kinase where two nitrogens of Arg-1136 are within hydrogen bonding distance of (ϳ3 Å from) the substrate tyrosine-phenol hydroxyl and the highly conserved, catalytically important aspartate (24,30) (Fig. 2).
The enhanced rescue activity of imidazole versus guanidinium was unexpected and to our knowledge is the first example in enzymology of a chemical rescue of an arginine mutation with imidazole. In the case of the protein-tyrosine kinase rescue, the enhanced action of imidazole over guanidinium is rationalized as likely resulting from 1) the planar, aromatic 5-membered ring with its extra ethylene group having favorable intermolecular interactions with the enzyme and/or substrates or 2) the lower pK a value of the imidazolium ion versus the guanidinium ion increasing hydrogen bond strength with the phenol oxygen or Asp carboxylate. Further experimentation will be needed to reveal which structural features are important for rescue.
Interestingly, GST-Csk R318H is a very poor kinase compared with wild type GST-Csk (observed velocity similar to unactivated GST-Csk R318A). This suggests that the geometry of the covalently attached imidazole side chain in R318H is not optimal for forming key interactions with the enzymes or substrates and is consistent with the absence of catalytic loop histidines in place of arginines in protein-tyrosine kinases in nature.
Because imidazole and its analogs should be cell permeable, and significant activation of poly(Glu,Tyr) phosphorylation at concentrations as low as 10 mM imidazole is observed, the use of this activator as a rescue agent for mutant protein-tyrosine kinase activity may hold promise for cell biology applications. As a precursor to examining activation within cells, we felt it was important to examine the ability of imidazole to rescue the phosphorylation of a physiologic substrate by GST-Csk R318A in vitro. To function as an in vitro substrate of GST-Csk and  Chemical Rescue of a Mutant Protein-tyrosine Kinase 38129 mutants, the Src used in this experiment contained a K295M mutation that reduced its kinase activity (34). A comparison of velocities of the phosphorylation of Src K295M showed a drop in Src K295M phosphorylation (ϳ500-fold) catalyzed by GST-Csk R318A versus wild type GST-Csk. This diminished rate of Src K295M phosphorylation by GST-Csk R318A was stimulated ϳ20-fold by 100 mM imidazole (Fig. 5B), whereas imidazole showed no effect on the phosphorylation rate catalyzed by wild type GST-Csk. Therefore the arginine rescue approach also extends to a physiologic protein substrate for the proteintyrosine kinase Csk. It is our hope that the imidazole rescue results shown here may ultimately be exploited in providing direct, kinetically controlled information about kinase function in cell signaling pathways.