Interaction of the Regulatory and Catalytic Subunits of cAMP ...

THE JOURNAL OF BIOLOGICAL CHEMISTRY © 1997 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 272, No. 26, Issue of June 27, pp. 16343–16350, 1997 Printed in U.S.A.

Interaction of the Regulatory and Catalytic Subunits of cAMP-dependent Protein Kinase ELECTROSTATIC SITES ON THE TYPE Ia REGULATORY SUBUNIT* (Received for publication, December 19, 1996, and in revised form, April 17, 1997)

Robin M. Gibson‡, Ying Ji-Buechler§, and Susan S. Taylor¶ From the Department of Chemistry and Biochemistry, University of California, San Diego, La Jolla, California 92093-0654

Since a basic surface on the catalytic (C) subunit of cAMP-dependent protein kinase is important for binding to the regulatory (R) subunit, acidic residues in R were sought that might contribute to R-C interaction. Using differential labeling by a water-soluble carbodiimide (Buechler, T. A., and Taylor, S. S. (1990) Biochemistry 29, 1937–1943), seven specific carboxylates in RIa were identified that were protected from chemical modification in the holoenzyme; each was then replaced with Ala. Of these, rRI(E15A/E106A/D107A)), rRI(E105A), rRI(D140A), rRI(E143A), and rRI(D258A) all were defective in holoenzyme formation and define negative electrostatic surfaces on RIa. An additional conserved carboxylate, Glu101 in RIa and the equivalent, Glu99 in RIIa were mutated to Ala. Replacement of Glu101 had no effect while rRII(E99A) was very defective. RIa and RIIa thus differ in the molecular details of how they recognize C. Unlike wild-type RI, two additional mutants, rRI(D170A) and rRI(K242A), inhibited C-subunit stoichiometrically in the presence of cAMP and show increases in both on- and off-rates. Asp170, which contributes directly to the hydrogen bonding network in cAMPbinding site A, thus contributes also to holoenzyme stability.

Cyclic AMP-dependent protein kinase (cAPK)1 exists as an inactive tetramer (R2C2) that dissociates in the presence of cAMP to release two active catalytic (C) subunits and a dimeric regulatory (R) subunit saturated with four molecules of cAMP (R2-cAMP4). The two classes of cAPK, type I and type II, differ primarily in their regulatory subunits (1). The two known families of physiological inhibitors of cAPK, the regulatory * This work was supported in part by National Institutes of Health Grant GM34921 (to S. S. T.) and National Institutes of Health Training Grant T32 CA0952223-08 (to R. M. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ Present address: University of Washington, Dept. of Pharmacology, Box 357750, Seattle, WA 98195-7750. § Present address: EGen Corp., 3550 General Atomics Ct., 02-544, San Diego, CA 92121. ¶ To whom correspondence should be addressed: University of California, San Diego, 9500 Gilman Dr. 0654, La Jolla, CA 92093-0654. Tel.: 619-534-3677; Fax: 619-534-8193; E-mail: [email protected]. 1 The abbreviations used are: cAPK, cAMP-dependent protein kinase; R, cAMP-dependent protein kinase regulatory subunit; RI, type Ia regulatory subunit of cAMP-dependent protein kinase; RII, type IIa regulatory subunit of cAMP-dependent protein kinase; C, cAMP-dependent protein kinase catalytic subunit; b-ME, b-mercaptoethanol; PAGE, polyacrylamide gel electrophoresis; EDC, 1-ethyl-3(3-dimethylaminopropyl)-carbodiimidezHCl; PKI, heat stable protein kinase inhibitor; PRS, peripheral recognition site; MES, 4-morpholineethanesulfonic acid; MOPS, 4-morpholinepropanesulfonic acid. This paper is available on line at http://www.jbc.org

subunits (RI and RII) and the heat-stable protein kinase inhibitor (PKI) employ a common mechanism for the inhibition of C. Each inhibitor has an autoinhibitory domain that resembles the substrate consensus sequence, “RRXS/TC,” where X is any amino acid, and the P11 site is a hydrophobic group (C). In RII the P-site is a Ser, while in RI and PKI, this site is an Ala (2– 4). Although binding of this autoinhibitor region to the active site of C is essential for inhibition, this interaction alone is not sufficient for high affinity binding. The R-subunits, as well as PKI, bind to C with Kd values that lie in the subnanomolar range (5–7), while peptides containing only the autoinhibitor site bind with micromolar affinity (8, 9). Furthermore, mutant forms of RI and RII with Ala substitutions at the P-2 and P-3 arginines can still form stable holoenzyme complexes in the absence of cAMP (10, 11), indicating that there are additional regions of interaction between R and C that lie beyond the autoinhibitor site. These sites are referred to here as peripheral sites. As shown in Fig. 1, the R-subunit has a well defined domain structure consisting of an N-terminal dimerization domain, followed by the autoinhibitory region and two tandem cAMPbinding domains, A and B. Proteolytic cleavage of the RIIsubunit at a site just prior to the P-3 Arg yielded a monomeric R-subunit fragment that still bound C with high affinity (12) indicating that the sites of interaction that contribute to high affinity binding lie C-terminal to the autoinhibitor consensus sequence. This is in striking contrast to PKI, whose highaffinity binding determinants lie mostly N-terminal to the consensus site (7, 9). Furthermore, deletion mutations that removed the N terminus (D1–91) or the B-domain (D260 –379) or combined mutations that removed both the N terminus and the B-domain, also did not prevent high affinity binding to C (13, 14). The high affinity binding sites are thus localized between the consensus site and the end of the A-domain (Fig. 1). When the crystal structure of the C-subunit (15) and, more recently, a deletion mutant of the R-subunit (16) were solved, these structures provided major insights into the docking surfaces that might be important for interactions between the C-subunit and its inhibitors. The peripheral PKI site, referred to here as peripheral recognition site 1 (PRS1), was originally mapped with peptide analogs (9), and is defined in the crystal structure of the catalytic subunit bound to the inhibitor peptide, PKI(5–24) (17). An additional surface on the C-subunit, referred to here as peripheral recognition site 2 (PRS2), was predicted based on genetic mapping, to be important for Rsubunit binding (18 –21). In contrast to the acidic surface on C where the consensus site peptide and PKI bind, the PRS2 surface importance for binding to R is very basic. This led us to search for acidic residues in R that might complement C. The deletion analysis described above limited the search for acidic residues to the region between the inhibition site and the beginning of domain B.

16343

Mutation of Charged Residues in RIa of cAPK

16344

FIG. 1. Location of mutation sites on the type I regulatory subunit of the cAMP-dependent protein kinase. Schematic diagram of the regulatory subunit: N-terminal dimerization domain (hatched box), autoinhibitor region (black box), cAMP-binding domains A and B (shaded boxes). Top, acidic sites protected from chemical modification in the holoenzyme complex. Bottom, additional charged residues that were mutated to Ala. Sequence alignment of bovine RI and RII subunits with the yeast regulatory subunit, BCY1 is indicated for the segment that is required for high affinity binding to the C-subunit. Conserved residues, black boxes; autoinhibitor region, hatched bar; cAMP-binding domain A, solid line; acidic sites protected from chemical modification in the presence of C-subunit, ,; additional acidic sites targeted by site-directed mutagenesis, f.

To chemically identify potential sites on the RI-subunit involved in R-C interaction, type I holoenzyme was treated with the water-soluble carbodiimide, 1-ethyl-3(3-dimethyl-aminopropyl)-carbodiimidezHCl (EDC), and [14C]glycine ethyl ester in the absence and presence of cAMP (22). Five carboxyl groups in C were identified, Glu170, Asp328, Asp329, Glu332, and Glu333, that were protected in the holoenzyme complex. The amino acids that were labeled in the free RI-subunit were: Glu77, Glu105, Glu106, Asp107, Asp140, Glu143, Glu255, Asp258, and Glu275. The reactivity of seven of these residues (Glu105, Glu106, Asp107, Asp140, Glu143, Glu255, and Asp258) was significantly reduced in the holoenzyme complex.2 To clarify the potential importance of these residues for R-C interaction, each was replaced with Ala. The mutant R-subunits were then purified and tested for their ability to inhibit C. Several additional sites were tested. The highly conserved glutamic acid residue 150 (Glu101 in RI), when mutated to Ala in BCY1, the yeast regulatory subunit, led to defective recognition of C (TPK1).3 This site was therefore mutated to Ala in RI and RII to establish whether this residue was also important for R-C interaction in the mammalian holoenzymes. Two additional charge-to-Ala mutations, D170A and K242A, were made in the RI subunit based on their location in the crystal structure of (D1–91)RI (16). Asp170, a highly conserved residue, lies on b-strand 3 of the A-domain where it forms a hydrogen bond with Arg209, a key residue in the cAMP-binding site (16, 23). The overall charge of the A-domain is acidic, however, there is a basic patch on the C-helix associated with Lys242. Lys242 was therefore targeted as a potential electrostatic interaction site in the holoenzyme complex. EXPERIMENTAL PROCEDURES

Materials—Reagents were obtained from the following sources: phosphoenolpyruvate, magnesium chloride, reduced nicotinamide adenine dinucleotide (NADH), MES, MOPS, Tris, pyruvate kinase (rabbit muscle), and lactate dehydrogenase (bovine heart), Sigma; restriction endonucleases, T4 ligase, T7 polymerase, U. S. Biochemical Corp. or Life Technologies, Inc.; radioactive nucleotides, Amersham or NEN Life Science Products; and media supplies, Difco. The peptide substrate, LRRASLG, was synthesized at the Peptide Facility at the University of 2 3

J. A. Buechler, unpublished results. C. Gibbs, personal communication.

California, San Diego. Oligonucleotides were synthesized using an Applied Biosystem DNA synthesizer, Model 380B. The following bacterial strains were used: Escherichia coli DH5a, E. coli JM101 (ATCC), E. coli BL21-DE3 (gift from William Studier of Brookhaven National Laboratories, Upton, NY), and E. coli E222. The following vectors were used: phagemid pUC119 (ATCC), PLWS-3 (24), and PRSET-B (Invitrogen). Mutagenesis—Mutations in the R- and C-subunits were introduced as described by Kunkel (25). Glu101, Glu105, Glu106, Asp107, Asp140, Glu143, Glu255, Asp258, Asp170, and Lys242 were each mutated to Ala in the pUC118 vector containing the bovine RIa gene (26). A triple mutant was also made where all three acidic residues, Glu105, Glu106, and Asp107, were changed to Ala (E105A/E106A/D107A). The mutant recombinant regulatory subunits, rRI(E101A), rRI(EED/AAA), rRI(E105A), rRI(E106A), rRI(D107A), rRI(D140A), rRI(E143A), rRI(E255A), rRI(D258A), rRI(D170A), and rRI(K242A), were expressed in E. coli 222 as described previously (27). The residue corresponding to Glu101 (Glu99) was mutated to Ala using a PRSET-B vector containing the RII gene fused to a polyhistidine (His6) tag. Protein Purification—The recombinant murine C-subunit was purified from E. coli as described (24, 28). Isoform II was used for all of the following experiments. Wild-type and mutant RI-subunits were purified using anion exchange chromatography (DE52) as described previously (29). Pooled fractions from the DE52 column were precipitated with ammonium sulfate and further purified by FPLC gel filtration on a Superdex 200 column (Pharmacia). Concentration and purity of the RI-subunits were determined by absorbance at 280 nm using an extinction coefficient of 1.0. All RI-subunits were at least 95% pure based on SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and analytical gel filtration. H6-RII(WT) and H6RII(E99A) were expressed in E. coli BL21-DE3 (4L) by inducing with isopropyl-1-thio-b-D-galactopyranoside for 5 h at 37 °C. Cells were harvested and lysed twice in a French pressure cell in 40 ml of lysis buffer (50 mM potassium phosphate, pH 8.0, 300 mM NaCl, 5 mM b-ME, 10 mM benzamidine, 1 mM leupeptin, 70 mg/ml L-1-tosylamido-2-phenylethyl chloromethyl ketone, 37 mg/ml N a-p-tosyl-L-lysine chloromethyl ketone, 0.5 mM phenylmethylsulfonyl fluoride). Cell lysates were centrifuged for 40 min at 15,000 rpm. Supernatant factions were batch bound to 10 ml of Ni21 resin (Quiagen) for 15 min at 4 °C. Resin was washed twice with 20 ml of lysis buffer, twice with 20 ml of 20 mM imidazole, and twice with 20 ml of 50 mM imidazole. The protein was eluted with 100 mM imidazole (100 ml) and collected in 5-ml fractions. The peak containing R-subunit was pooled and dialyzed against 2 liters of (20 mM Tris, 150 mM NaCl, 5 mM b-ME, pH 7.3) for 5 h and then overnight against 2 liters (20 mM Tris, 20 mM NaCl, 5 mM b-ME, pH 7.3). The dialyzed sample was then loaded onto a Mono Q column in Buffer A (20 mM Tris, 5 mM b-ME, pH 7.3) and eluted with a linear gradient from 0 to 300 mM potassium phosphate, pH 7.2. Pooled fractions were concentrated using a Centriprep-30 concentrator (Ami-


16345

TABLE I R-subunit mutations R-subunit

Mutation

Reason for mutation

RI

E105A/E106A/D107A E105A E106A D107A D140A E143A E255A D258A

Sites protected from chemical modification in type I holoenzyme

RI

E101A

RII

E99A

Conserved residue. Equivalent mutation in yeast R-subunit, BCY1, disrupted R/C interaction (Footnote 3). Not protected from chemical modification in type I holoenzyme

RI

D170A K242A

Conserved acidic residue in the A-domain of R. Hydrogen bond with Arg209 Exposed basic residue on the C-helix of domain A.

con) prior to addition of glycerol (final concentration, 30%) and stored at 220 °C. To obtain cAMP-free R-subunit, the R-subunits were unfolded with 8 M urea, dialyzed, and then purified by gel filtration as described by Buechler et al. (10). Kinase Assay—Kinase activity was measured spectrophotometrically using the heptapeptide substrate, LRRASLG (Kemptide), and a coupled enzyme assay (30). The concentration of C in the assay was between 20 and 60 nM. Holoenzyme samples were assayed first in the absence and then in the presence of 100 mM cAMP to obtain a ratio of activity representing the amount of free C in the mixture. Holoenzyme Formation—cAMP-saturated native RI subunit (3 mM) and C-subunit (2 mM) were combined (0.8 ml) and dialyzed in a multiwell dialysis chamber against buffer B (25 mM potassium phosphate, 100 mM ATP, 500 mM MgCl2, 5% glycerol, 5 mM b-ME, pH 6.5) at 22 °C. Aliquots (30 ml) were removed at the indicated times and assayed for catalytic activity. The H6RII holoenzyme was formed at 4 °C in buffer C (25 mM potassium phosphate, pH 6.5, 5% glycerol, 5 mM b-ME, 0.1 mM EDTA, 0.1 mM EGTA) to minimize breakdown of the H6RII-subunit. Dialysis experiments were typically carried out in duplicate and were highly reproducible. As an alternative method for measuring holoenzyme formation, the activity of 20 nM C-subunit was titrated directly in the spectrophotometric assay by the addition of increasing amounts of R-subunit that had been stripped of cAMP by urea. Tryptophan Fluorescence—R-subunit was dialyzed overnight against 1 liter of buffer D (5 mM MOPS, pH 7.0, 0.5 mM EDTA, 100 mM KCl, 5 mM b-ME) and against an additional 1 liter for 4 h to remove free cAMP. Fluorescence measurements were made using 1 ml of 0.2 mM R-subunit in a 1-cm quartz cuvette at 23 °C and a SLM Aminco SPF-500 spectrofluorimeter interfaced with a PC/AT computer. The excitation wavelength was 290 nm with a bandpass of 10 nm. The excited samples were scanned for tryptophan fluorescence from 300 to 450 nm using an emission bandpass of 5 nm. Spectra were analyzed using Spectra-Calc software. For the cAMP titrations, urea-stripped RI-subunit (50 nM) was incubated in buffer E (20 mM MOPS, 150 mM KCl, 5 mM b-ME, pH 7.0) with increasing concentrations of cAMP (0 –100 mM) as described by Leoń and Taylor.4 The excitation wavelength was 295 nm and the emission was measured at 347 nm. Apparent Activation Constants (Ka) for cAMP—Apparent activation constants, Ka(cAMP), were determined according to Herberg et al. (32). Wild-type and mutant holoenzymes were prepared with 1.2 mM Csubunit and 1.0 mM R-subunit by dialysis overnight at room temperature against buffer F (20 mM potassium phosphate, 100 mM KCl, 5 mM b-ME, 5% glycerol, 100 mM ATP, 1 mM MgCl2, pH 6.5). Holoenzyme (30 nM) was incubated for 5 min at room temperature with cAMP (1–180 mM) in 1 ml of assay mixture prior to the addition of Kemptide substrate. Activity was then measured using the spectrophotometric assay. Analytical Gel Filtration—The Stokes radius was determined using an FPLC Superdex 200 column as described previously (28). The column was equilibrated with (20 mM MOPS, 150 mM KCl, 5 mM b-ME, pH 4

D. A. Leoń and S. S. Taylor, manuscript in preparation.

FIG. 2. Holoenzyme formation of rRI(EED/AAA), rRI(E105A), rRI(E106A), and rRI(D107A) with wild-type C-subunit. Wild-type (●) and mutant RI-subunits: RI(EED/AAA) (f), RI(E105A) (å), RI(E106A) (l), and RI(D107A) (ç) were dialyzed in the presence of wild-type C-subunit as described under “Experimental Procedures.” % Activity is the ratio of catalytic activity remaining in the dialysis mixture in the absence or presence of 100 mM cAMP and represents the amount of free C remaining in the mixture at the times indicated. These dialysis were all carried out simultaneously and were highly reproducible with multiple preparations of the proteins. Duplicates were typically within 5%. 7.0) and run with a flow rate of 0.8 ml/min. Surface Plasmon Resonance—Surface plasmon resonance was used to study the interaction between the rC subunit and the rR subunits of cAPK using a BIAcore instrument (Pharmacia/Biosensor) (33). The C-subunit was coupled to a sensor chip by direct amine coupling to the CM dextran surface (Biosensor Amine Coupling Kit) as described previously (32). Kinetic constants were calculated using the BIAcore pseudo-first order rate equation, dR/dt 5 kassCRmax 2 ~kassC 1 kdis!Rt

(Eq. 1)

where kass is the association rate, kdis is the dissociation rate, C is the concentration of the injected analyte, and R is the response. Plots of dR/dt versus Rt have a slope of ks. When ks is plotted against C the resulting slope is equal to the kass. The dissociation rate, kdis, was calculated by integrating the rate equation when C 5 0, yielding ln(Rt1/ Rtn) 5 kdis(tn 2 t1). Affinity constants were calculated from the equation Kd 5 kdis/kass. All binding interactions were performed at 23 °C in 20 mM MOPS, 150 mM KCl, 100 mM ATP, 1 mM MgCl2, 0.5 mM dithiothreitol, 0.005% P20, pH 7.0. After injections of the rR subunits the rC subunit surface was regenerated by injection of 10 ml of 10 mM cAMP. Association rate calculations were performed using concentrations between 30 and 500 nM for rRI(WT), rRI(D170A), and rRI(K242A). All of the rR-subunits were stripped of cAMP prior to injection. RESULTS

To characterize the importance of various charged residues in the R-subunit for recognition of C and formation of holoenzyme, the following charge to Ala mutations were engineered: E101A, E105A/E106A/D107A, E105A, E106A, D107A, D140A, E143A, E255A, D258A, D170A, K242A in RI and E99A in RII (Table I). Each mutant RI-subunit was expressed in E. coli and purified as described under “Experimental Procedures.” The proteins were typically at least 80% pure following DE52 anion exchange chromatography and eluted as a sharp peak from the Superdex 200 gel filtration column. Typical yields were 40 –50 mg of RI-subunit per 4 liters of cells. All of the purified RIsubunits were greater than 95% pure as judged by SDS-PAGE. The poly-His-tagged RII-subunits, H6RII(WT) and H6RII(E99A), were purified from Ni21 resin by batch elution with imidazole. H6RII(E99A) was greater than 95% pure following FPLC Mono Q chromatography; however, degradation occurred upon long-term storage (.6 months) at 220 °C. Importance of the Acidic Patch Glu105-Glu106-Asp107—An

16346


FIG. 3. Holoenzyme formation of mutant RI-subunits with the catalytic subunit of cAPK. Holoenzyme was formed with 3 mM Rsubunit and 2 mM wild-type C-subunit as described under “Experimental Procedures” and in the legend to Fig. 2. % Inhibition corresponds to the activity remaining in the holoenzyme mixture following dialysis for the time indicated. Panel A, rRI(WT) (solid), rRI(D140A) (striped), rRI(E143A) (shaded), 10 h. Panel B, rRI(WT) (solid), rRI(E255A) (striped), rRI(D258A) (shaded), 10 h. Panel C, rRI(WT) (solid), rRI(E101A) (stippled), 20 h. Panel D, H6RII(WT) (solid), H6RII(E99A) (stippled), 17 h. All dialysis in each panel were carried out simultaneously using a multiwelled dialysis chamber. Similar results were obtained in three separate experiments.

acidic patch just after the autoinhibitor site and conserved in all R-subunits was protected against modification by EDC when the RI subunit was associated with C.2 To test whether specific contacts involving Glu105-Glu106-Asp107 were important for interactions between the R- and C-subunits, each carboxylate was replaced with Ala, generating the following mutants: rRI(E105A), rRI(E106A), and rRI(D107A). In addition, the triple mutant, rRI(E105A/E106A/D107A), was made. The purified RI-subunits were combined with purified recombinant murine C-subunit, rC(WT), in a multiwelled microdialysis chamber and the relative rates of holoenzyme formation of the mutant RI-subunits were compared with that of wild-type RI as described under “Experimental Procedures.” The dialysis curves in Fig. 2 indicated that the triple mutant, rRI(EED/AAA), formed 40% holoenzyme after 10 h of dialysis at room temperature, while, after the same length of time, 90% of the wild-type control had formed holoenzyme. The single mutant, rRI(E105A), exhibited a slight defect in holoenzyme formation while the other two single-site mutations, rRI(E106A) and rRI(D107A), were indistinguishable from the wild-type control. Holoenzyme Formation of rRI(D140A), rRI(E143A), rRI(E255A), and rRI(D258A) with Wild-type C—In addition to the acidic patch Glu105-Glu106-Asp107, four other acidic residues in the RI-subunit were protected from chemical modification in the presence of C.2 These sites included two residues at the beginning of the A-domain (Asp140 and Glu143) and two at the end of the A-domain (Glu255 and Asp258). To test whether any of these sites were involved in a direct interaction with C, each was mutated to Ala, and the purified mutant proteins were paired with wild-type C in the holoenzyme dialysis assay. As shown in Fig. 3 (A and B), three of the mutants, rRI(D140A), rRI(E143A), and rRI(D258A), showed similar defects in their rates of holoenzyme formation. After 10 h of dialysis at room temperature, these mutant RI-subunits inhibited only 60% of the C-subunit activity, while the wild-type control was more than 90% inhibited, indicating that Asp140, Glu143, and Asp258 contribute to holoenzyme formation. Mutation of Glu255 to Ala had no affect on holoenzyme formation.

FIG. 4. Inhibition of 20 nM wild-type C-subunit with unstripped RI-WT and rRI(D170A). 20 nM C-subunit was incubated with varying amounts of either rRI(WT) (●) or rRI(D170A) (å) in 1 ml of assay mixture for 2 min prior to the addition of Kemptide. Similar results were obtained on multiple occasions.

The Role of Glu101 in Holoenzyme Formation—Although Glu101 was not modified by EDC in either the cAMP-bound form of RI or in the type I holoenzyme, it is highly conserved in all regulatory subunits. Furthermore, mutation of this site to Ala in BCY1, the yeast regulatory subunit, abolished the ability of BCY1 to inhibit the yeast C-subunit, TPK1.3 As shown in Fig. 3C, however, mutation of this site to Ala in the RI-subunit had no affect on the rate of holoenzyme formation compared with RI-WT. Since BCY1 is a type II R-subunit, the equivalent mutation was also made in the mammalian RII-subunit, RII(E99A), to determine if this site was important for R-C interaction in the type II holoenzyme. Activity assays of the holoenzyme mixtures following 17 h of dialysis indicated that, unlike the wild-type control, which was 95% inhibited, H6RII(E99A) only showed 20% inhibition (Fig. 3D). Following dialysis, SDS-PAGE of the holoenzyme mixtures verified that the lack of holoenzyme formation observed for H6RII(E99A) was not due to proteolysis of the RII-subunit. The fact that Glu101 is important for holoenzyme formation with RII and BCY1 but not for RI, even though this residue is conserved, indicates that the interaction sites between R and C in the type I and type II holoenzymes that lie beyond the autoinhibitor region differ in their molecular details. The Effect of Mutating Asp170 in the A-Domain of RI—Asp170 is another acidic residue that is conserved among the regulatory subunits of the cAPK family (34). It lies in cAMP-binding domain A where it forms a hydrogen bond with Arg209, a residue that is absolutely conserved among the cAMP-binding domains of the regulatory subunits and the catabolite gene activator protein (16, 23). Although Asp170 lies on the surface of the A-domain, it also reaches through the domain to make contact with Arg209 in the cAMP-binding site. This hydrogen bonding between Asp170 and Arg209 which potentially helps to neutralize the charge on Asp170, would have prevented the identification of Asp-170 in the holoenzyme protection study that identified Glu105-Glu106-Asp107, Asp140, Glu143, Glu255, and Asp258 as potential R-C interaction sites by chemical modification. Asp170 was mutated to Ala to test whether it played a role in holoenzyme formation or cAMP binding. The wild-type RIa-subunit binds cAMP very tightly, so that the purified R-subunit is always fully saturated with cAMP. Thus, under the conditions used for holoenzyme formation where C, R, and cAMP are present at relative concentrations of 1:1:2, wild-type RI does not linearly titrate the activity of the C-subunit. Linear and stoichiometric titration is achieved only


16347

FIG. 5. Tryptophan fluorescence of rRI(D170A), rRI(R209K), and rRI(WT). RI-subunit (0.2 mM) was dialyzed against (5 mM MOPS, pH 7.0, 0.5 mM EDTA, 100 mM KCl, 5 mM b-ME) to remove free cAMP. Fluorescence measurements were made as described under “Experimental Procedures.” The excitation wavelength was 290 nm with a bandpass of 10 nm. The excited samples were scanned for tryptophan fluorescence from 300 to 450 nm using an emission bandpass of 5 nm. Spectra were analyzed using Spectra-Calc software. Inset, the tryptophan fluorescence of urea stripped rRI(WT) (●) and rRI(D170A) (f) was titrated by the addition of increasing amounts of cAMP as described under “Experimental Procedures.” The % quench was plotted against the log[cAMP] to obtain apparent Kd values for cAMP binding. The fluorescence titration of rRI(WT) is carried out routinely and is highly reproducible. The mutant was characterized on two separate occasions.

when cAMP is stripped from R prior to reassociation with C. Otherwise, dialysis is required to remove the cAMP. In contrast to wild-type RI, when rRI(D170A) was added to the C-subunit, inactivation was immediate; rRI(D170A) behaved similarly to R-subunit that had been stripped of cAMP with urea. rRI(D170A) was then tested for its ability to titrate the catalytic activity directly in the spectrophotometric assay. The inhibition assay measured the following reaction, R2~cAMP!4 1 2C7R2C2 1 4cAMP

(Eq. 2)

where loss of activity correlates with the formation of the R2C2 complex. As shown in Fig. 4, when rRI(D170A) was added directly to the assay mixture where the C-subunit concentration was 20 nM, inhibition of catalytic activity was linear and stoichiometric. This was in striking contrast to the inhibition of C with wild-type RI which must be stripped of cAMP before stoichiometric inhibition can be achieved under these conditions. The linear and stoichiometric inhibition of C with rRI(D170A) indicated not only that the Kd for this mutant R and rC(WT) was well below 20 nM, but also, under these conditions, rRI(D170A) had a higher affinity for C than for cAMP, causing the equilibrium of reaction (2) to lie to the right. Several factors could account for this. The affinity for cAMP could be reduced so that the purified rRI(D170A) was not saturated with cAMP. Alternatively, the affinity for C could be increased. To resolve these possibilities, the cAMP binding properties of rRI(D170A) were characterized more fully. cAMP Binding Properties of rRI(D170A) and rRI(D170A)2C2—One explanation for the above observations would be that mutation of Asp-170 interferes with the high affinity binding of cAMP. Other mutations in the R-subunit that specifically disrupt cAMP binding in either the A- or B-domain such as rRI(R209K) or rRI(R333K) also inhibit C linearly and stoichiometrically without having to first remove cAMP (6). When purified, only one cAMP-binding site is occupied in these mutant R-subunits and these proteins behave like stripped rRI(WT). Unlike rRI(WT), rRI(R209K) and rRI(R333K) form stable holoenzyme complexes even when cAMP levels are high. The intrinsic tryptophan fluorescence of the RI-subunit is quenched by cAMP (29, 35). Since the B-domain of RI does not have any tryptophans, the change in tryptophan fluorescence in response to cAMP is a sensitive method for measuring cAMP binding to the A-domain. For example, the intrinsic tryptophan

fluorescence of wild-type RI is quenched compared with that of RI(R209K) because no cAMP is bound to site A in RI(R209K).4 The fluorescence spectrum of rRI(D170A) compared with rRI(WT) and rRI(R209K) is shown in Fig. 5. The spectrum of rRI(D170A) was similar to that of the wild-type RI-subunit and showed significant quenching compared with rRI(R209K). Thus cAMP was bound to the A-domain of rRI(D170A), similar to RI(WT) and in contrast to the rRI(R209K) where the A site is not occupied by cAMP. To confirm this, an apparent Kd for cAMP binding was obtained by titrating the tryptophan fluorescence of the urea stripped proteins with increasing amounts of cAMP. As seen in Fig. 5 (inset), the apparent Kd for cAMP binding to rRI(D170A) (Kd 5 62 nM) was similar to that of rRI(WT) (Kd 5 45 nM) and was comparable to the Kd measured by equilibrium dialysis (36). To further characterize the cAMP binding properties of rRI(D170A), the apparent Ka for cAMP activation of the mutant holoenzyme was measured. The Ka for cAMP activation of rRI(D170A) was 127 nM which correlates well with 88 nM for wild-type holoenzyme. This is also similar to the previously reported Ka for wild-type holoenzyme (6). In contrast, the Ka for cAMP activation of RI(R209K) was 1700 nM (6). Stokes Radius—The Stokes radius of rRI(D170A) was measured to determine whether the mutation had altered the global structure of RI. The native form of rRI(D170A) had a Stokes radius of 46.1 Å, similar to that of native RI (32). Furthermore, the urea-stripped cAMP-free form of rRI(D170A) had a Stokes radius that was 1.3 Å smaller than the native, cAMP-bound form. This is also similar to the wild-type protein which exhibited a 2-Å decrease in Stokes radius upon removal of cAMP (32). In contrast, RI(R209K), which is defective in binding cAMP to the A-domain, has a Stokes radius that is similar to the cAMP-free form of RI(WT) (32). The Effect of Mutating Lys242—To establish whether Lys242 in RI contributed to the high affinity binding between R and C, the mutant protein, rRI(K242A), was tested for its ability to inhibit wild-type C. Surprisingly, like the D170A mutant, inhibition of rC(WT) with rRI(K242A) did not require dialysis. rRI(K242A) also inhibited 20 nM rC(WT) in the spectrophotometric assay in a linear and stoichiometric fashion without being stripped of cAMP. This mutation also did not appear to affect cAMP binding, since the apparent Ka of cAMP activation of the rRI(K242A) mutant holoenzyme was similar to that of wild-type holoenzyme.


16348

TABLE II Quantitation of interaction between regulatory and catalytic subunits using surface plasmon resonance Surface plasmon resonance was used to determine the association and dissociation rate constants of cAMP-free rR-subunits using a BIAcore instrument with immobilized wild-type C-subunit. Only the late phase (t 5 605.5– 694.5 s) was used to determine off-rates RIstripped

Kd nM

WT D170A K242A

0.1 0.2 0.1

kass M

21

s21) 3 105

8.1 6 2.8 15.0 6 6.0 11.2 6 4.0

kdiss (s21) 3 1025

8.0 6 2.7 23.3 6 3.6 9.0 6 0.9

Surface Plasmon Resonance—One explanation for the ability of rRI(D170A) and rRI(K242A) to stoichiometrically inhibit C without first removing cAMP is that the mutations increased the affinity of RI for C. Surface plasmon resonance was used to quantitate the affinity of these mutant R-subunits for C. Using a BIAcore instrument (Pharmacia), cAMP-free rRI(D170A) and rRI(K242A) were applied to a chip containing immobilized wild-type C-subunit. R/C binding was measured by monitoring changes in the refractive index on the surface of the chip, allowing real-time binding kinetics to be measured. The results using rRI(D170A) and rRI(K242A) are compared with rR(WT) in Fig. 6. Both the on- and off-rates are monophasic for all three proteins. As summarized in Table II, although the overall binding affinities of rRI(D170A) and rRI(K242A) were similar to rRI(WT), rRI(D170A) exhibited on- and off-rates that were 2and 3-fold faster than the wild-type control. DISCUSSION

Many lines of evidence indicate that high affinity binding of the R and C subunits of cAPK is a multistep process that requires more than just a linear sequence of R that occupies the active site cleft of the C-subunit. The inhibitor site in R resembles a peptide substrate and presumably binds to the active site of C in a manner similar to the consensus sequence of PKI (15, 17). This inhibitor site extends from the P23 site to the P11 site and will be similar for most substrates and inhibitors of cAPK. This site, however, is not sufficient to confer high affinity binding. In addition to the inhibitor recognition site, an additional large surface on C is apparently masked by the R-subunit. This peripheral site on C, referred to here as PRS2, is more complex and will almost certainly depend on the global conformation of the R-subunit. Neither site alone is sufficient for high affinity binding. From the crystal structure of the catalytic subunit we can define the inhibition recognition site very precisely, and, coupled with genetic and mutational mapping, the surface that comprises the peripheral recognition site (PRS2) can be mapped. In contrast to the acidic inhibitor recognition site on C, where the consensus site Arg’s dock, the PRS2 surface is very basic. For the R-subunit, the crystal structure of a deletion mutant (D1–91)RI has defined at least some of the structural features that will be recognized by the PRS2 surface on C. Deletion mutants localized most of the essential docking requirements to the A-domain while chemical studies identified specific residues that may be important. Because the R-docking surface on C is very basic, acidic residues in the A-domain of R were specifically targeted. The location of most of these residues in (D1–91)RI are shown in Fig. 7A. Each of these residues was replaced with Ala. These mutants will be discussed as three groups. First are those that cluster near the inhibitor site. An additional set of acidic residues were identified by differential labeling with EDC. The third set were identified based on their position on the crystal structure of (D1–91)RI. Glu-101, highly conserved in all R-subunits, lies in the P14

FIG. 6. Measurement of association and dissociation rate constants for regulatory and catalytic subunits using surface plasmon resonance. Wild-type C-subunit was immobilized and the binding of wild-type or mutant R-subunits were monitored as described under “Experimental Procedures.” Panel a, rR(WT); panel b, rR(D170A); panel c, rR(K242A). x axis represents time (in seconds), and y axis represents arbitrary response unit. The concentrations (in nM) used for the various R-subunits were indicated.

position just following the inhibitor site. Although this site is not important for R-C interaction in the type I holoenzyme, mutation of this site in both the mammalian and yeast type II R-subunits seriously disrupted holoenzyme formation. The difference in the role of this conserved glutamate in RI and RII highlights that these two types of R-subunit may differ in significant ways in how they interact with C. The interaction of RII and C is potentially much simpler and does not require tight binding of MgATP as does RI and PKI (37). The structure of the region containing the autoinhibitor region, Arg94-Arg95-Gly96-Ala97-Ile98, and extending though the acidic patch, Glu105-Glu106-Asp107, is disordered in the crystal structure of (D1–91)RI and is therefore not seen in the crystal structure which begins with Arg113. The position of Glu105Glu106-Asp107 immediately following the autoinhibitor region, and the fact that these carboxylates were protected from chemical modification in the holoenzyme suggested that these residues interacted with the C-subunit. Based on sequence analysis of RI, RII, and BCY1 (34), the region corresponding to Glu105-Glu106-Asp107 in RI is only partially conserved (Fig. 1). While RII has six acidic residues (Asp104-Glu105-Glu106-Glu107-


16349

FIG. 7. Location of mutation sites on the x-ray crystal structure of (D1–91)RI. A, ribbon diagram. The N terminus, beginning with 113 and the cAMP-binding domain B are dark gray and the A-domain is white. cAMP is occupies each of the cAMP binding sites. B, hydrogen bond network in the cAMP-binding domain A. Asp170 is linked to the highly ordered hydrogen bond network surrounding cAMP via its interaction with NH2 of Arg209. Arg241 is located on the C-helix of domain A and Asp267 lies on the C terminus of the A-helix of domain B.

Glu108-Asp109) in this region, BCY1 contains only two (Asp150 and Asp151). The mutations described here indicated that an acidic patch, Glu105-Glu106-Asp107, in RI was important for binding C; however, the defect exhibited by the triple mutant, rRI(EED/AAA), did not represent an additive effect of each of the single mutants. In fact, rRI(E105A) was the only single-site mutant that was defective in holoenzyme formation. Since the triple mutant resulted in a stronger defect than the single-site E105A mutation, it is likely that an extended negative surface is important for R-C interaction. In addition to the acidic patch Glu105-Glu106-Asp107, four other acidic residues in RI were identified which were protected from chemical modification by EDC in the presence of C. These were: Asp140, Glu143, Glu255, and Asp258. Of these sites, only Asp140 is conserved in RI, RII, and BCY1 (Fig. 1). As shown in Fig. 6A, Asp140 and Glu143 lie on the N terminus of the A-helix at the beginning of the A-domain. Glu255 lies 5 residues beyond the C-helix of the A-domain on a turn that bends back toward the cAMP-binding site and ends at Trp260. Asp258 is on the same turn and points toward the cAMP-binding site. Although removal of none of these residues abolished holoenzyme formation, replacing Asp140, Glu143, or Asp258 with an Ala did significantly reduce the binding efficiency between RI and C. Only mutation of Glu255 had no effect. As shown in Fig. 7B, the hydrogen bond network in the cAMP-binding pocket of the A-domain is extensive. At one end of this extended network is Asp170. Asp170, on b-strand 3, is on the surface of the A-domain and its side-chain forms a hydrogen bond with Arg209, another conserved residue that is an essential feature of the cAMP-binding pocket. When cAMP is removed from the A-domain, this hydrogen bonding network in the cAMP pocket must change significantly. This could change the position of Asp170, causing it to either interact more tightly to Arg209, or to be exposed on the surface where it could interact

with another basic residue. Thus, Asp170 was mutated to Ala to see if this acidic site in the A-domain influenced R-C interactions. Although removal of Asp170 did not significantly alter the overall affinity for R and C, the D170A mutation resulted in onand off-rates that were both slightly faster than those of wildtype. The faster on-rate in the absence of Asp170 indicates that Asp170 may stabilize the cAMP-bound conformation. This constraint would be its interaction with Arg209 as seen in the crystal structure. Removal of Asp170, however, did not alter the affinity for cAMP. Asp170 may help to stabilize the final holoenzyme conformation as well since the dissociation rate for the mutant protein was also 3-fold faster. The overall charge on the surface of the A-domain is acidic, however, there is a basic patch on the C-helix of the A-domain that is associated with Lys242. The adjacent residue, Arg241, is a critical residue for cooperativity between domain A and B (38). As shown in Fig. 7B, Arg241 forms hydrogen bonds with residues that reside both in the A-domain (Glu200) and in the B-domain (Asp267). Asp267 is the only residue in domain B that contributes directly to the hydrogen bond network associated with cAMP binding in site A. Both Arg241 and Lys242 lie on the C-helix of domain A, and the orientation of this helix relative to the b-barrel will very likely change when cAMP is released. When the structures of domains A and B of (D1–91)RI and the nucleotide-binding domain of CAP are superimposed, the most striking difference is the position of the C-helix (16). In the A-domain, the C-helix is longer, slightly bent and held away from the cAMP pocket. Furthermore, although Trp260, at the end of the loop that follows the C-helix was photoaffinity labeled with the cAMP analog, 8-azido-cAMP (8-N3-cAMP), in full-length RI, in a B-domain deletion mutant, RI(D260 –379), Trp244 was labeled instead (31). Thus the orientation of the C-helix in the A-domain relative to the cAMP-binding pocket is capable of a large conformational change. The B-domain with

16350


its strong hydrophobic interactions with the C-helix exerts structural constraints on the conformation of the A-domain (16). The surface of C that is thought to interact with R is mostly basic with the exception of the negative charge associated with the phosphorylated Thr197. Lys242 was thus targeted as a potential electrostatic interaction site for P-T197 in the holoenzyme complex. Like rRI(D170A), rRI(K242A) inhibited rC(WT) in the spectrophotometric inhibition assay without requiring the removal of cAMP by either urea stripping or dialysis. It is unlikely, therefore, that Lys242 interacts directly with P-T197 on C. Furthermore, as with rRI(D170A), the K242A mutation did not alter either the overall affinity for C or the apparent Ka for cAMP activation. Why removal of this Lys actually facilitates R-C interaction is still unclear. To fully understand the precise molecular details of R-C interaction, a crystal structure of the holoenzyme complex is required. Acknowledgments—We thank Dr. Sarah Cox and Dr. Fritz Herberg for valuable discussion. We also thank Lily Jun-shen Huang for useful discussions and assistance in measuring the surface plasmon resonance of the mutant R-subunits. REFERENCES 1. Hofmann, F., Beavo, J. A., Bechtel, P. J., and Krebs, E. G. (1975) J. Biol. Chem. 250, 7795–7801 2. Scott, J. D., Fischer, E. H., Takio, K., Demaille, J. G., and Krebs, E. G. (1985) Proc. Natl. Acad. Sci. U. S. A. 82, 5732–5736 3. Titani, K., Sasagawa, T., Ericsson, L. H., Kumar, S., Smith, S. B., Krebs, E. G., and Walsh, K. A. (1984) Biochem. 23, 4193– 4199 4. Takio, K., Smith, S. B., Krebs, E. G., Walsh, K. A., and Titani, K. (1984) Biochemistry 23, 4200 – 4206 5. Hofmann, F. (1980) J. Biol. Chem. 255, 1559 –1564 6. Herberg, F. W., Taylor, S. S., and Dostmann, W. R. G. (1995) Biochemistry 35, 2934 –2942 7. Walsh, D. A., Angelos, K. L., Van Patten, S. M., Glass, D. B., and Garetto, L. P. (1990) in CRC Reviews (Kemp, B. E., ed) pp. 43– 84, CRC Press, Inc., Boca Raton, FL ¨ . Z., Ragnarsson, U. and Engstrom, L. (1990) in CRC Reviews 8. Zetterqvist, O (Kemp, B. E., ed) pp. 171–187, CRC Press, Inc., Boca Raton, FL 9. Glass, D. B., Cheng, H.-C., Mende-Mueller, L., Reed, J., and Walsh, D. A. (1989) J. Biol. Chem. 264, 8802– 8810 10. Buechler, Y. J., Herberg, F. W., and Taylor, S. S. (1993) J. Biol. Chem. 268,

16495–16503 11. Wang, Y. H., Scott, J. D., McKnight, G. S., and Krebs, E. G. (1991) Proc. Natl. Acad. Sci. U. S. A. 88, 2446 –2450 12. Weldon, S. L., and Taylor, S. S. (1985) J. Biol. Chem. 260, 4203– 4209 13. Saraswat, L. D., Ringheim, G. E., Bubis, J., and Taylor, S. S. (1988) J. Biol. Chem. 263, 18241–18246 14. Ringheim, G. E., and Taylor, S. S. (1990) J. Biol. Chem. 265, 19472–19478 15. Knighton, D. R., Zheng, J., Ten Eyck, L. F., Ashford, V. A., Xuong, N.-h., Taylor, S. S., and Sowadski, J. M. (1991) Science 253, 407– 414 16. Su, Y., Dostmann, W. R. G., Herberg, F. W., Durick, K., Xuong, N.-h., Ten Eyck, L., Taylor, S. S., and Varughese, K. I. (1995) Science 269, 807– 813 17. Knighton, D. R., Zheng, J., Ten Eyck, L. F., Xuong, N.-h., Taylor, S. S., and Sowadski, J. M. (1991) Science 253, 414 – 420 18. Gibbs, C. S., Knighton, D. R., Sowadski, J. M., Taylor, S. S., and Zoller, M. J. (1992) J. Biol. Chem. 267, 4806 – 4814 19. Levin, L. R., and Zoller, M. J. (1990) Mol. Cell. Biol. 10, 1066 –1075 20. Orellana, S. A., and McKnight, G. S. (1992) Proc. Natl. Acad. Sci. U. S. A. 89, 4726 – 4730 21. Orellana, S. A., Amieux, P. S., Zhao, X., and McKnight, G. S. (1993) J. Biol. Chem. 268, 6843– 6846 22. Buechler, J. A., and Taylor, S. S. (1990) Biochemistry 29, 1937–1943 23. Bubis, J., Neitzel, J. J., Saraswat, L. D., and Taylor, S. S. (1988) J. Biol. Chem. 263, 9668 –9673 24. Slice, L. W., and Taylor, S. S. (1989) J. Biol. Chem. 264, 20940 –20946 25. Kunkel, T. A. (1985) Proc. Natl. Acad. Sci. U. S. A. 82, 488 – 492 26. Saraswat, L. D., Filutowics, M., and Taylor, S. S. (1988) Methods Enzymol. 159, 325–336 27. Saraswat, L. D., Filutowicz, M., and Taylor, S. S. (1986) J. Biol. Chem. 261, 11091–11096 28. Herberg, F. W., Bell, S. M., and Taylor, S. S. (1993) Protein Eng. 6, 771–777 29. Leoń, D. A., Dostmann, W. R. G., and Taylor, S. S. (1991) Biochemistry 30, 3035–3040 30. Cook, P. F., Neville, M. E., Vrana, K. E., Hartl, F. T., and Roskoski, J., R. (1982) Biochemistry 21, 5794 –5799 31. Ringheim, G. E., Saraswat, L. D., Bubis, J., and Taylor, S. S. (1988) J. Biol. Chem. 263, 18247–18252 32. Herberg, F. W., Dostmann, W. R. G., Zorn, M., Davis, S. J., and Taylor, S. S. (1994) Biochemistry 33, 7485–7494 33. Karlsson, R., Michaelsson, A., and Mattsson, L. (1991) J. Immunol. Methods 145, 229 –240 34. Toda, T., Cameron, S., Sass, P., Zoller, M. J., Scott, J. D., McMullen, B., Hurwitz, M., Krebs, E. G., and Wigler, M. (1987) Mol. Cell. Biol. 7, 1371–1377 35. Builder, S. E., Beavo, J. A., and Krebs, E. G. (1980) J. Biol. Chem. 255, 2350 –2354 36. Ogreid, D., Ekanger, R., Suka, R. H., Miller, J. P., and Doskeland, S. O. (1989) Eur. J. Biochem. 181, 19 –31 37. Herberg, F. W., and Taylor, S. S. (1993) Biochemistry 32, 14015–14022 38. Symcox, M. M., Cauthron, R. D., Ogreid, D., and Steinberg, R. A. (1994) J. Biol. Chem. 269, 23025–23031