Structural characterization of Ca 2+ /CaM in complex with the phosphorylase kinase PhK5 peptide Atlanta G. Cook*, Louise N. Johnson and James M. McDonnell Laboratory of Molecular Biophysics, Department of Biochemistry, Oxford University, UK Phosphorylase kinase (PhK) is a Ca 2+ -regulated pro- tein kinase that controls the breakdown of glycogen through phosphorylation of glycogen phosphorylase (reviewed in [1]). The enzyme is a large, 1.3-MDa hexa- decameric complex consisting of four copies of four subunits, a, b, c and d. The a and b subunits are regu- latory and are the sites of phosphorylation and meta- bolite binding and are also regulated by the binding of extrinsic calmodulin (CaM). The c subunit is the catalytic subunit and the d subunit is an intrinsic mole- cule of CaM that binds to the enzyme even in the absence of Ca 2+ [2]. The regulation of PhK through Ca 2+ ⁄ CaM enables the coordination of muscle contrac- tion with the production of glucose through the action of Ca 2+ on calmodulin and troponin C [3]. PhK is related to other Ca 2+ ⁄ CaM-dependent protein kinases including myosin light chain kinase (MLCK), CaM kinases I, II and IV (CaMKI, CaMKII and CaMKIV, respectively), CaM kinase kinase (CaMKK), titin kinase, and death associated kinase [4]. Structural studies on CaMKI [5], titin kinase [6] and twitchin kinase [7] have upheld the prediction that many CaM-dependent protein kinases are regulated through an autoinhibitory mechanism [8]. In these structures a C-terminal extension to the protein kinase folds back on the kinase domain and interferes with the substrate binding sites. In the case of the titin and twitchin kinases, the autoinhibitory sequence acts as a pseudo- substrate, occluding ATP binding and preventing protein substrates from binding (reviewed in [9]). PhK shows typical traits associated with such an autoinhibitory mechanism. The sequence of the PhKc subunit encodes a C-terminal extension to the protein kinase domain and treatment of the kinase with Keywords calmodulin; kinase regulation; protein– protein interaction; NMR spectroscopy Correspondence J. M. McDonnell, Laboratory of Molecular Biophysics, Department of Biochemistry, Oxford University, South Parks Road, Oxford OX1 3QU, UK Fax: +44 1865 275182 Tel: +44 1865 275381 E-mail: jim@biop.ox.ac.uk *Present address EMBL, Meyerhofstrasse 1, D-69117 Heidelberg, Germany (Received 7 December 2004, revised 23 January 2005, accepted 1 February 2005) doi:10.1111/j.1742-4658.2005.04591.x Phosphorylase kinase (PhK) is a large hexadecameric enzyme consisting of four copies of four subunits: (abcd) 4 . An intrinsic calmodulin (CaM, the d subunit) binds directly to the c protein kinase chain. The interaction site of CaM on c has been localized to a C-terminal extension of the kinase domain. Two 25-mer peptides derived from this region, PhK5 and PhK13, were identified previously as potential CaM-binding sites. Complex forma- tion between Ca 2+ ⁄ CaM with these two peptides was characterized using analytical gel filtration and NMR methods. NMR chemical shift perturba- tion studies showed that while PhK5 forms a robust complex with Ca 2+ ⁄ CaM, no interactions with PhK13 were observed. 15 N relaxation characteristics of Ca 2+ ⁄ CaM and Ca 2+ ⁄ CaM ⁄ PhK5 complexes were compared with the experimentally determined structures of several Ca 2+ ⁄ CaM ⁄ peptide complexes. Good fits were observed between Ca 2+ ⁄ CaM ⁄ PhK5 and three structures: Ca 2+ ⁄ CaM complexes with pep- tides from endothelial nitric oxide synthase, with smooth muscle myosin light chain kinase and CaM kinase I. We conclude that the PhK5 site is likely to have a direct role in Ca 2+ -regulated control of PhK activity through the formation of a classical ‘compact’ CaM complex. Abbreviations CaM, calmodulin; CaMK, CaM kinase; CaMKK, CaM kinase kinase; eNOS, endothelial nitric oxide synthase; MLCK, myosin light chain kinase; PhK, phosphorylase kinase; TFA, trifluoracetic acid. FEBS Journal 272 (2005) 1511–1522 ª 2005 FEBS 1511 proteases causes cleavage of this extension at residue 296 and 298 and loss of Ca 2+ -dependent activity [10] (Fig. 1). However, PhK also differs significantly from other CaM-dependent protein kinases in that it binds to CaM even in the absence of Ca 2+ . Studies on the disruption of the holoenzyme produced two functional complexes, the acd complex and the cd complex dem- onstrating that CaM interacts with the kinase chain of PhK directly [11]. Furthermore, separation of these two chains is only possible through denaturation [12]. CaM is a ubiquitous Ca 2+ sensor that is found in all eukaryotes and shows little sequence variation in metazoans [13]. CaM undergoes large conformational changes both on binding to Ca 2+ ions and on interact- ing with its targets [14,15]. The protein consists of two domains each encoding two EF-hand motifs separated by a linker region that imparts a high degree of conformational flexibility. In the presence of Ca 2+ ions the protein undergoes conformation changes that alter the surface properties of the two domains allowing CaM to recognize its targets. A number of structures of CaM in complex with peptides derived from CaM target proteins have been solved. In the classical case, these CaM binding pep- tides have been identified as basic motifs of approxi- mately 20 amino acids that are able to bind to CaM as amphipathic helices [16]. Binding causes CaM to wrap around the peptide helix forming a hydrophobic chan- nel and a number of acidic residues on the surface of the CaM domains typically form salt bridges with the basic residues that are found in the peptide (Fig. 1). Despite an overall similarity, the closed Ca 2+ ⁄ CaM ⁄ peptide complexes show a variety of conformations with respect to domain orientation and conformation of the EF hands. The CaM binding domains in the different proteins have little sequence similarity. They display a variable distance between the hydrophobic ‘anchor’ residues that bind into each domain of CaM and are classified on the basis of the distance between these two motifs [17]. Two 25-mer peptide regions from the C-terminal extension of PhKc were previously identified as poten- tial CaM binding sites using a series of overlapping synthetic peptides. These two peptides, PhK5 (342– 367) and PhK13 (302–326), were able to inhibit CaM activation of MLCK [18] and Ca 2+ -dependent phos- phodiesterase [19] and were shown to have K I values in the low nanomolar range. Both peptides were dem- onstrated to have an inhibitory effect on PhK with PhK13 demonstrating competitive inhibition and PhK5 showing noncompetitive inhibition kinetics [20]. PhK13, which lies towards the N terminus of the C-terminal extension (Fig. 2), contains a sequence that Fig. 1. The Ca 2+ ⁄ CaM ⁄ smMLCK structure (PDBid 1cdl), demon- strating the ‘compact’ structure of a Ca 2+ ⁄ CaM ⁄ peptide complex. The N-terminal domain is in blue and the C-terminal domain is in pink. The peptide is shown as a green helix and the termini are indicated with N and C. Two side chains are depicted correspond- ing to the two anchor residues of the smMLCK peptide, Trp800 and Leu813. The individual helices are labelled with roman numer- als starting from the N terminus. The Ca 2+ ions are shown as blue spheres and are labelled 1–4. Figure prepared with PYMOL [42]. The alignment shows peptide sequences from various Ca 2+ ⁄ CaM ⁄ pep- tide structures in single amino acid code. The residues in yellow are anchor residues and basic residues are shown in blue. In the two PhK peptides large hydrophobic and aromatic residues are shown in green that have been predicted as potential anchor resi- dues for these two peptides. Both PhK peptides have previously been assigned a ‘1–12 motif’, a structurally uncharacterized motif. Fig. 2. An overview of the PhKc domain structure. The first 296 residues of the subunit encode a protein kinase domain. The struc- ture of the constitutively active kinase domain has previously been solved, coordinates are taken from PDBid 2phk [43,44]. Proteolytic treatment cleaves between the kinase domain and the 90-residue C-terminal extension of the kinase. Two 25-mer peptides in this region, PhK13 and PhK5, were identified as potential CaM binding domains. Ca 2+ ⁄ CaM in complex with the PhK5 peptide A. G. Cook et al. 1512 FEBS Journal 272 (2005) 1511–1522 ª 2005 FEBS suggested it could act as a pseudosubstrate mimicking the substrate phosphorylase. However, this role was not supported by mutagenesis studies that converted a potential pseudosubstrate cysteine residue to a serine residue [21]. Comparative studies of the two peptides in complex with Ca 2+ ⁄ CaM with CD demonstrated that PhK5 was likely to form an a-helix when bound to CaM but that PhK13 showed no changes in secondary structure contributions on binding [22]. Further studies with small angle X-ray scattering [23] demonstrated that PhK5 induced a compact conformation of CaM, sim- ilar to that observed with MLCK peptides, a result that was further confirmed by fluorescence anisotropy studies [24], but that the interactions of PhK13 were anomalous and in the complex CaM adopted an exten- ded conformation. While these studies indicate that there are marked differences in the Ca 2+ ⁄ CaM ⁄ PhK5 complex vs. the Ca 2+ ⁄ CaM ⁄ PhK13 complexes, no direct structural evidence for these two complexes is available. In this paper we have used NMR methods to characterize the structures. The NMR evidence shows that PhK5 does indeed form a classical collapsed complex with CaM, but no interaction of CaM with PhK13 could be detected. A method for identifying structural similarity between Ca 2+ ⁄ CaM ⁄ peptide complexes is presented. Results The binding of PhK5 and PhK13 to Ca 2+ /CaM High resolution analytical gel filtration was used to identify the complexes of CaM with the two PhK pep- tides. Because the previously reported K I values for the peptides were in the low nanomolar range, the samples were mixed together in a 1 : 1 molar ratio and gel fil- tration was carried out in the presence of Ca 2+ . Ca 2+ ⁄ CaM, in the absence of peptides, elutes as a sin- gle peak after a volume of 16.12 mL. As Ca 2+ ⁄ CaM does not contain any tryptophan residues the absorb- ance at 280 nm is entirely contributed by tyrosine residues and is therefore relatively low. When the Ca 2+ ⁄ CaM ⁄ PhK5 complex was loaded onto the col- umn, the peak moved to 16.31 mL and showed a higher absorbance at 280 nm. PhK5 has one trypto- phan residue and two tyrosine residues that accounts for the increase in absorbance. The formation of a Ca 2+ ⁄ CaM ⁄ PhK5 complex was confirmed by analysis of the peak fractions by tris-tricine SDS ⁄ PAGE that shows a smaller band, of % 3 kDa, that coelutes with CaM. That the peak shows an increase in peak elution volume indicates that complex Ca 2+ ⁄ CaM ⁄ PhK5 has a smaller radius of gyration than Ca 2+ ⁄ CaM and this is consistent with a more compact structure. When a similar analysis was carried out with the Ca 2+ ⁄ CaM ⁄ PhK13 mixture no alteration in the peak intensity or the peak volume was observed. Although PhK13 has no tryptophan residues, it does contain sev- eral tyrosine residues and the absence of an absorb- ance increase suggests that PhK13 does not bind. Furthermore, analysis of peak fractions by SDS ⁄ PAGE indicates that no peptide coelutes with Ca 2+ ⁄ CaM (Fig. 3 inset). The lack of PhK13 binding was surprising, so we sought a more definitive experiment to demonstrate peptide binding. Using NMR spectroscopy, titrations were carried out with unlabelled peptides into 15 N labelled Ca 2+ ⁄ CaM. Chemical shift changes in the CaM backbone amides were monitored using 1 H- 15 N HSQC. PhK5 causes a large number of chemical shift changes in the Ca 2+ ⁄ CaM HSQC spectrum indicative of a large conformational change upon binding of PhK5 (Fig. 4A). These changes are observed early in the titration, at as little as 20% saturation changes are readily apparent. Intermediate peaks between the unbound and bound species are not observed indica- ting that the binding of PhK5 to Ca 2+ ⁄ CaM has slow exchange kinetics and this is consistent with the nano- molar K I values that had previously been reported. Fig. 3. Gel filtration analysis of Ca 2+ ⁄ CaM and its complexes with PhK5 and PhK13. In all runs the same concentration of Ca 2+ ⁄ CaM was used and a 1 : 1 molar ratio of peptide was added. The inset shows a Tris ⁄ tricine SDS ⁄ PAGE gel analysis of the peak fractions from the gel filtration run. In all cases CaM is seen as a band of % 17 kDa. Only when Ca 2+ ⁄ CaM is mixed with PhK5 is a smaller peptide band also observed. A. G. Cook et al. Ca 2+ ⁄ CaM in complex with the PhK5 peptide FEBS Journal 272 (2005) 1511–1522 ª 2005 FEBS 1513 When PhK13 was titrated into Ca 2+ ⁄ CaM no chemical shift changes are observed (Fig. 4B. No binding inter- actions are observed between PhK13 and CaM; at the concentrations we performed this experiment the affin- ity of this interaction would need to be 10 mm for us not to detect it by this method. The 1 H- 15 N HSQC for free Ca 2+ ⁄ CaM are essen- tially identical to previously described spectra [25], and so peaks were assigned by comparison. Of the 92 peaks that could be assigned in this way, only 13 in the PhK5 titration do not show any alteration in chemical shift (summarized in Table 1). These include residues that are found in solvent exposed regions (Leu4, Glu6, Asn42 and Glu45) that are unlikely to change on complex formation as they do not interact with peptide ligands. In addition, four glycine residues, found in the second position of the four distinct EF- hand motifs (Gly23, Gly59, Gly96 and Gly132) also show no changes in chemical shift. Three further resi- dues that are found buried between the two EF-hand motifs (Thr62, Ile100 and Val136) are also found not to have any chemical shift changes. These residues form part of the hydrophobic packing between EF- hand motifs and this explains their unaltered chemical environment. Lastly two further residues, Met72 and Glu82 are also unchanged. These two residues are found on the connecting helices between the two domains of Ca 2+ ⁄ CaM. While Glu82 is either solvent exposed or disordered in Ca 2+ ⁄ CaM peptides com- plexes, Met72 is found to interact with peptide ligands in some structures such as the complex with CaMKIIa peptide [26], but not in others, for example the com- plex with smMLCK [27]. T1 and T2 relaxation times for Ca 2+ /CaM and Ca 2+ /CaM/PhK5 The large number of chemical shift changes that occur on binding of PhK5 to Ca 2+ ⁄ CaM suggest that this binding event is not a localized phenomenon and causes large changes in the structure throughout the molecule. This is consistent with a large conforma- tional change occurring on binding of the peptide and could indicate that the binding of PhK5 to Ca 2+ ⁄ CaM is similar to that observed for Ca 2+ ⁄ CaM binding to isolated peptides from other protein kin- ases. Previous studies using fluorescence anisotropy have indicated that PhK5 binds to CaM as a col- lapsed complex that has similar properties to other CaM ⁄ peptide complexes [24]. To determine whether the binding of PhK5 does cause a conformational change in CaM, the T1 and T2 constants were meas- ured for each residue to allow a better understanding of the hydrodynamic behaviour of Ca 2+ ⁄ CaM and Ca 2+ ⁄ CaM ⁄ PhK5. HSQC spectra were taken after a series of increasing relaxation delays to measure the T1 and T2 relaxation times. For each residue identified in the spectrum, a single exponential fit to the peak intensities over the series of spectra was used to calculate R1 (1 ⁄ T1) and R2 (1 ⁄ T2). The R1 and R2 values were plotted against Fig. 4. (Top) Overlay of two spectra from the titration of Ca 2+ ⁄ CaM with PhK5 peptide. The red spectrum shows the position of peaks prior to the addition of PhK5 peptide. The blue spectrum shows a spectrum taken when a 1 : 1.2 molar ratio of Ca 2+ ⁄ CaM to PhK5 had been reached. The inset shows an expanded view of the indi- cated area of the spectrum. (Bottom) An overlay of spectra from the titration of Ca 2+ ⁄ CaM with PhK13 peptide showing in red the spectrum prior to titration and in blue, the spectrum after a 1 : 0.8 molar ratio of Ca 2+ ⁄ CaM to PhK13 had been reached. The inset shows an expanded view of part of the spectrum equivalent to that shown in (A). No chemical shift changes were observed on titration with PhK13. Table 1. CaM residues unaffected by PhK5 binding. Residues that remain unchanged Role in structure Leu4, Glu6, Asn42, Glu45 Solvent exposed Gly23, Gly59, Gly96, Gly132 Second glycine in EF-motif Thr62, Ile100, Val136 Buried between EF-motifs Glu82 Found in connecting helices Met72 Found in connecting helices, binds ligands in some CaM structures Ca 2+ ⁄ CaM in complex with the PhK5 peptide A. G. Cook et al. 1514 FEBS Journal 272 (2005) 1511–1522 ª 2005 FEBS residue number, and average values for R1 and R2 over each helix were calculated (Fig. 5). R1 and R2 are expected to have a reciprocal relationship for movements on the nanosecond to picosecond time- scale, i.e. the timescale for molecular reorientation. The pattern of the average helical R1 and R2 values clearly indicate that the complex of Ca 2+ ⁄ CaM ⁄ PhK5 forms a more compact structure than Ca 2+ ⁄ CaM, and that PhK5 binding does induce conformational change. The R2 plot for Ca 2+ ⁄ CaM ⁄ PhK5 shows unusual periodic increases in R2 over the length of helix I that are not reflected in the R1 values. The periodicity of the changes follow approximately the periodicity of the helix and the residues with higher R2 values are found on one face of helix I that is found to interact with peptide ligands in CaM ⁄ peptide complexes. This effect could be caused by conformational exchange on the millisecond timescale (R ex ) that affects only one side of the helix. Fig. 5. The R1 and R2 relaxation rates were plotted against residue number. At the base of each plot the structural elements of CaM are indicated by the single line plot with heli- ces labelled in Roman numerals in the first plot. The average relaxation rate of each helix is plotted as a black bar. The first helix of the Ca 2+ ⁄ CaM ⁄ PhK5 R2 is highlighted by a box; the periodicity of the R2 effects suggests R ex phenomena in this helix. Residues 21, 57, 94 and 130 (indicated), which occupy identical positions in each EF-hand motif and are involved in chelating the Ca 2+ ions, each show markedly reduced R2 values. A. G. Cook et al. Ca 2+ ⁄ CaM in complex with the PhK5 peptide FEBS Journal 272 (2005) 1511–1522 ª 2005 FEBS 1515 Comparison of relaxation data with other CaM/peptide complexes In anisotropic proteins, 15 N relaxation rates depend on the orientation of the N–H bond vectors relative to the principal axis frame of the rotational diffusion ten- sor. For residues that are known to be structurally rigid relaxation anisotropy can be used to derive orien- tational constraints for structure calculations [28]. The analysis of orientational information from the relaxa- tion characteristics of amide bond vectors is made more straightforward in the case of a-helical proteins. Because the amide bond vectors of an a-helix all point in the same direction, the average R1 ⁄ R2 ratio for resi- dues in that helix provides information about its orien- tation relative to the overall diffusion tensor of the molecule. Upon binding of peptide ligands, calmodulin undergoes a dramatic alteration in structure marked by changes in the relative orientations of its eight a-helical elements. Different peptide ligands result in subtle differences in the CaM conformations. We did not have an experimental model of Ca 2+ ⁄ CaM ⁄ PhK5 structure, but a number of struc- tures of CaM ⁄ peptide complexes are available that reflect wide conformational diversity of the CaM mole- cule. Therefore we used the program rotdif [29,30] to calculate the diffusion tensor of a PDB-derived CaM structure, back-calculate the R1 and R2 values for the helical elements of the PDB structure and then com- pare these values to the experimentally derived relaxa- tion parameters for Ca 2+ ⁄ CaM ⁄ PhK5 and, as a control, for Ca 2+ ⁄ CaM. For these calculations each helix was treated as a structural element; R1 and R2 values for each helix were averaged based on four con- secutive residues in the middle of the helix. For the Ca 2+ ⁄ CaM ⁄ PhK5 data, helix IV had to be excluded because of insufficient data due to spectral overlap and in the case of helix I, only the lower R2 values were used to exclude the R ex effects. Isotropic, axially symmetric (z ¼ long axis, x ¼ y ¼ short axes) and fully anisotropic models (z > y > x axis) were all considered for fitting the relaxation data. The data fit much better to axially symmetric models than to isotopic models (P<1 · 10 )4 , unless noted). The improvement in fit going from an axially symmet- ric model to the fully anisotropic model was generally not statistically significant (the likelihood that the improvements in fit occurred by chance showed a range of P ¼ 4.58 · 10 )1 to P ¼ 2.12 · 10 )2 ) and therefore the data presented here are for an axially symmetric model only. One case, CaMKK, was noted to be borderline as the improvement in fit using an axially symmetric model over an isotropic model gave a poorer probability score (P ¼ 2.29 · 10 )2 for the anisotropic vs. isotropic models) and this is reflected in the D para ⁄ D perp ratio that was calculated for this com- plex (Table 2). However, this structure was included and treated as axially symmetric for the purposes of this study. Table 2 shows the target functions calculated by comparison of experimental relaxation data with data back-calculated from the PDB files of the input models. As expected, poor fits were generally observed between the Ca 2+ ⁄ CaM data, where CaM is in the extended conformation, and the various peptide complex models, where CaM is in the folded conformation. The Ca 2+ ⁄ CaM ⁄ PhK5 dataset showed betters fits in general and several structures (Ca 2+ ⁄ CaM ⁄ smMLCK, Ca 2+ ⁄ CaM ⁄ endothelial nitric oxide synthase (eNOS) and Ca 2+ ⁄ CaM ⁄ CaMKI) appeared to fit the relaxation data better than others. A structural alignment of these three models shows that they are indeed close structural Table 2. The quality of fits (described by the chi-squared per degrees of freedom) for a series of Ca 2+ ⁄ CaM ⁄ peptide complexes against the relaxation data for Ca 2+ ⁄ CaM ⁄ PhK5 and Ca 2+ ⁄ CaM. While all structures show relatively poor fits with the unbound Ca 2+ ⁄ CaM structure, including the compact form of Ca 2+ ⁄ CaM, much better fits are observed with the peptide bound structures, particularly the complexes with smMLCK peptide, CaMKI peptide and eNOS peptide. The s c value, the molecular correlation time, is calculated from the fit and can be defined as the time taken for the molecule to rotate through 1 radian or to translate through its own length. D para ⁄ D perp is the ratio between the length of the longs axis, z, and the length of the shorter, equal x and y axes. CaM ⁄ peptide complex PDB file v 2 Ca 2+ ⁄ CaM ⁄ PhK5 v 2 Ca 2+ ⁄ CaM s c (ns) D para ⁄ D perp smMLCK 1cdl 0.178 2.46 9.41 1.18 CaMKII 1cdm 0.239 2.67 9.41 1.21 CaMKI 1mxe 0.144 2.53 9.41 1.18 CaMKK 1iq5 0.697 3.21 9.37 1.13 eNOS 1niw 0.170 2.15 9.45 1.23 MARCKS 1iwq 0.329 2.68 9.40 1.20 Ca 2+ ⁄ CaM (compact) 1prw 0.449 3.31 9.41 1.18 Ca 2+ ⁄ CaM in complex with the PhK5 peptide A. G. Cook et al. 1516 FEBS Journal 272 (2005) 1511–1522 ª 2005 FEBS relatives. The remaining structures, which fit the relaxa- tion data less well, show a variety of relative helix ori- entations (Fig. 6). Discussion The binding of PhK5 and PhK13 to Ca 2+ /CaM PhKc interacts with the CaM Ca 2+ sensor through a C-terminal extension to the kinase domain of % 90 res- idues. Previous studies on PhKc identified two regions in this C-terminal extension, PhK5 and PhK13, that were able to inhibit Ca 2+ ⁄ CaM-mediated activation of MLCK. Further studies also indicated that there may be substantial differences in the binding of PhK5 and PhK13 to CaM based on small angle X-ray scattering studies and CD measurements. The present studies on the PhK13 and PhK5 peptide complexes have substan- tiated these significant differences in their interactions with CaM. Analytical gel filtration and NMR chemical shift perturbation studies indicated no detectable inter- action between Ca 2+ ⁄ CaM and PhK13. In contrast, PhK5 formed a robust complex with Ca 2+ ⁄ CaM under conditions of gel filtration and the addition of PhK5 caused a large number of NMR chemical shift changes in Ca 2+ ⁄ CaM. Taken together, these data demonstrate that the region of PhKc that is delineated by the PhK13 pep- tide does not interact directly with Ca 2+ ⁄ CaM under the conditions used in this study. The ability of this peptide to inhibit the MLCK activity was well established and it was demonstrated to bind to CaM in a Ca 2+ -dependent manner in gel mobility assays [18]. However the small angle X-ray and neutron scattering data indicated that the PhK ⁄ Ca 2+ ⁄ CaM interaction is anomalous [23]. The CaM remained extended on binding PhK13. PhK13 failed to protect CaM against proteolysis while PhK5 did protect in a manner similar to other CaM binding peptides [22]. Analysis of the PhKc C-terminal extension sequences from a number of different organisms suggests that the PhK5 region is likely to be the more important in Ca 2+ signalling. The PhK5 region is conserved (15 residues out of 25 identical) while the PhK13 region shows only two residues out of 25 identical and does not exhibit a canonical CaM binding sequence (Fig. 7). The differences observed under dif- ferent experimental conditions for the PhK13 ⁄ CaM interactions require further work and could best be resolved by a crystal structure of full length PhKc subunit with CaM, a structure that has so far been elusive. Does the PhK5 peptide cause conformational changes in Ca 2+ /CaM? Both X-ray scattering and fluorescence anisotropy studies of the Ca 2+ ⁄ CaM ⁄ PhK5 complex have indica- ted that this complex has a similar structure to the complexes of Ca 2+ ⁄ CaM with peptides derived from other CaM-dependent enzymes. Our analytical gel fil- tration studies and NMR titration experiments are also Fig. 6. (A) A comparison the three structures that show highest correlation with Ca 2+ ⁄ CaM ⁄ PhK5, that includes Ca 2+ ⁄ CaM ⁄ eNOS in blue (PDBid 1niw), Ca 2+ ⁄ CaM ⁄ smMLCK in pink (PDBid 1cdl) and Ca 2+ ⁄ CaM ⁄ CaMKI in orange (PDBid 1mxe). The Ca 2+ ions are indicated as cyan spheres and the smMLCK peptide is shown as a white helix. Structures are aligned using the C-terminal domain of CaM (residues 86–146). The orientation presented here is similar to that shown in Fig. 1. (B) A similar alignment showing structures that show poorer correlations with Ca 2+ ⁄ CaM ⁄ PhK5. Ca 2+ ⁄ CaM ⁄ smMLCK is shown in pink once again as a reference. Ca 2+ ⁄ CaM ⁄ CaMKK is in red (PDBid 1iq5), Ca 2+ ⁄ CaM ⁄ CaMKIIa is shown in green (PDBid 1cdm) and the Ca 2+ ⁄ CaM ⁄ MARCKS structure (PDBid 1iwq) is shown in yellow. Figure pre- pared with AESOP (MEM Noble, unpublished work). A. G. Cook et al. Ca 2+ ⁄ CaM in complex with the PhK5 peptide FEBS Journal 272 (2005) 1511–1522 ª 2005 FEBS 1517 consistent with a compact structure of the Ca 2+ ⁄ CaM ⁄ PhK5 complex. The spectrum of Ca 2+ ⁄ CaM ⁄ PhK5 is dramatically different from that of the unbound Ca 2+ ⁄ CaM with only 13 peaks out of 146 that show no alteration in chemical shift, indicating that a large conformational change occurs on binding peptide. Comparison of the R1 values for Ca 2+ ⁄ CaM and Ca 2+ ⁄ CaM ⁄ PhK5 show a significant decrease upon PhK5 binding as well as a reduction in the vari- ation of R1 in secondary structural elements, suggest- ing that the two domains no longer tumble independently and that CaM assumes a more compact structure in the bound form. Analysis of the R1 and R2 relaxation times for each assigned peak in the Ca 2+ ⁄ CaM and Ca 2+ ⁄ CaM ⁄ PhK5 samples supported the notion of a significant conformational change. The R1 and R2 relaxation times for a particular nucleus are depend- ent on its reorientation in solution and can give information about the conformational flexibility of different protein regions. For relatively rigid residues the relaxation times will reflect the overall motion of the molecule as it tumbles in solution. These motions, on the nanosecond to picosecond timescale, will have reciprocal effects on the R1 and R2 relaxa- tion times, although this relationship breaks down when the residues are subject to conformational exchange (R ex ) [29,30]. Thus for residues in secon- dary structure elements, such as a-helices, the R1 and R2 relaxation times can give information on the rotation of the molecule in solution. When a mole- cule tumbles anisotropically, the residues with N–H bond vectors aligned with the long axis of the mole- cule will reorient more slowly compared with N–H bond vectors that are orientated along the shorter axes. This causes differences in the average R1 and R2 properties of a given helix depending on how it is oriented with respect to the long axis of the mole- cule. The differing patterns of average R1 and R2 values along each helix for Ca 2+ ⁄ CaM and the Ca 2+ ⁄ CaM ⁄ PhK5 indicate that the orientations of the helices that make up the structure are different in the two species. As these data apply only to the main chain nitrogen atoms of the CaM structures this shows that PhK5 has indeed induced a conform- ational change in Ca 2+ ⁄ CaM. Fig. 7. Sequence alignment of the C-terminal extension of PhKc. Sequences were obtained for both vertebrate and nonvertebrates by using sequence from 296 to 386 of PhKc from rabbit muscle. Identities are shown by the blue boxes with white text, while blue text and green boxes indicate regions of sequence similarity. Numbering is taken from the rabbit muscle sequence. The regions corresponding to the PhK5 and PhK13 peptides are indicated by pink boxes. This figure was prepared using ESPRIPT [45]. Ca 2+ ⁄ CaM in complex with the PhK5 peptide A. G. Cook et al. 1518 FEBS Journal 272 (2005) 1511–1522 ª 2005 FEBS How similar is the Ca 2+ /CaM/PhK5 complex to other Ca 2+ /CaM/peptide complexes? Classical Ca 2+ ⁄ CaM ⁄ peptide complexes are largely similar in their formation of a closed, compact CaM structure around a helical amphipathic peptide. How- ever, many of these complexes also show distinct con- formations, not only in the mode of ligand binding but also in the relative orientation of the two CaM domains. These differences are facilitated by the flexi- bility of the linker between the two domains as well as a certain degree of conformational flexibility in the two EF hand motifs of each CaM domain. The pres- ence of a relatively large number of Ca 2+ ⁄ CaM ⁄ pep- tide complexes in the PDB opens up the possibility of using these structures to perform structural compari- sons against unknown structures. Indeed, comparative studies have previously been performed using NMR NOE data or residual dipolar couplings [31,32] to identify structural similarities between different CaM ⁄ peptide complexes. The study presented here uses rotation anisotropy of CaM and its effects on the relaxation rates of N–H bond vectors of the CaM backbone. Each of the Ca 2+ ⁄ CaM ⁄ peptide structures was used as an input model in lieu of a Ca 2+ ⁄ CaM ⁄ PhK5 model, for the program rotdif that calculates hydro- dynamic parameters based on a structure and its relaxation data [29,30]. The hydrodynamic data were consistent with a monomeric structure of % 20 kDa. Out of the seven PDB files that were used as input models, three structures showed better fits than the rest. These three structures are the complexes with the eNOS peptide, the smMLCK peptide and the CaMKI peptide. All three of these are structural rel- atives and show rmsds of the Ca atoms of 1.705 A ˚ comparing CaMKI to smMLCK, 1.972 A ˚ comparing CaMKI to eNOS and 2.633 A ˚ comparing smMLCK to eNOS. In addition, all three peptides have been identified as binding with a 1–14 motif of anchor residues while the remaining structures in the study show 1–16 binding (CaMKK) and 1–10 binding (CaMKIIa) [17]. The MARCKS peptide structure shows an unusual pattern of binding in that its anchor residues are separated by only one residue. A 1–14 motif is compatible with the PhK5 sequence, with L345 and V358 serving as anchor residues. With this predicted motif a strong similarity between the PhK5 and CaMKI peptide sequences becomes more apparent (Fig. 1). Of the PDB files tested, the CaM ⁄ CaMKI structure was the best fit to the CaM ⁄ PhK5 relaxation data (Table 2). On the other hand, PhK5 does contain one large hydrophobic residue, Trp357, that is perhaps more likely to serve as anchor reside than Val358. Interestingly, the three peptides from MLCK, CaMKI and eNOS each have a large aromatic residue in the N-terminal part of the peptide that binds in the with the C-terminal domain in CaM. However in PhK5 Trp357 is found at the C-terminal end of the peptide. This suggests that either PhK5 might bind with Trp357 in the N-terminal site in CaM, or perhaps could bind in the opposite orientation to incorporate the Trp into the larger C-terminal domain binding site. In summary, PhKc is known to bind to CaM via its C-terminal extension. The data presented here demon- strate that PhK5 interacts directly with Ca 2+ ⁄ CaM and that PhK13 does not. The PhK5 interaction causes a large conformational change to occur in Ca 2+ ⁄ CaM that produces a classical compact Ca 2+ ⁄ CaM ⁄ peptide complex. Analysis of the NMR relaxation data sug- gests that the Ca 2+ ⁄ CaM ⁄ PhK5 complex is a close structural relative of CaM complexes with eNOS, smMLCK and CaMKI. Experimental procedures Preparation of CaM The CaM cDNA from Xenopus leavis was a kind gift from D. Owen (Oxford University, UK) and was cloned into the pPROTet.E232 vector (Clontech, Oxford, UK). The plasmid was transformed into BL21-PRO cells (Clontech) and expressed at 37 ° C by induction with 100 ngÆmL )1 anhydro- tetracycline for 5 h. The cells were harvested by centrifuga- tion and resuspended in 50 mm Tris ⁄ HCl pH 7.5, 2 mm EDTA, 0.2 mm phenylmethanesulfonyl fluoride and were stored at )20 °C. Purification of CaM was carried out using the method of Hayashi et al. with minor modifications [33]. The cells were thawed and lysed by sonication and the sol- uble fraction was collected by centrifugation at 100 000 g in a Beckman L8-M ultracentrifuge for 1 h at 4 °C. CaCl 2 was added to the supernatant to a final concentration of 5 mm and the sample was then applied to a 50-mL phenyl seph- arose column pre-equilibrated in 50 mm Tris ⁄ HCl pH 7.5, 5mm CaCl 2 , 100 mm NaCl. Two steps were carried out to remove the majority of contaminants from the protein sam- ple, a low Ca 2+ wash with 50 mm Tris ⁄ HCl pH 7.5, 0.1 mm CaCl 2 , 100 mm NaCl and then a high salt wash with 50 mm Tris ⁄ HCl pH 7.5, 0.1 mm CaCl 2 , 0.5 m NaCl. Finally, CaM was eluted from the column using a Ca 2+ -free buffer con- taining 0.2 mm EDTA and 50 mm Tris ⁄ HCl pH 7.5. To produce 15 N labelled protein for NMR experiments, the cells were grown on M9 medium instead of Luria– Bertani broth and supplemented with 1.5 lm thiamine and 15 NH 4 Cl as sole nitrogen source. Purification was carried out using the standard protocol and the pure protein was A. G. Cook et al. Ca 2+ ⁄ CaM in complex with the PhK5 peptide FEBS Journal 272 (2005) 1511–1522 ª 2005 FEBS 1519 concentrated to % 1mm and buffer exchanged at least four times into 20 mm deuterated Tris ⁄ HCl pH 6.5. PhK5 and PhK13 peptides The PhK5 peptide (amino acid sequence LRRLIDAYAFRI YGHWVKKGQQQNR) and the PhK13 peptide (amino acid sequence GKFKIVCLTVLASVRIYYQYRRVKP) were custom synthesized using solid phase fmoc chemistry by G. Bloomberg (Bristol University, UK). The peptides were further purified by reverse phase chromatography using a 250 · 10 mm C5 column (Phenomenex, Maccles- field, UK). The peptides were loaded in 0.1% trifluoracetic acid (TFA) and eluted using a gradient from 0.1% TFA to 0.1% TFA and 50% acetonitrile over five column volumes. The peptides were then lyophilized and reconstituted into double deionized H 2 O and dialysed against water at 4 °Cto remove salt impurities. MS of these purified peptides was performed on a Micromass Platform-II ESI mass spectro- meter (Waters, Elstree, UK) and gave expected molecular mass values (3118 ± 3 and 3004 ± 4 Da, for PhK5 and PhK13, respectively), thus confirming the composition of the peptides that were used in subsequent experiments. Analytical gel filtration Analytical gel filtration was carried out using an SD200 high resolution sepharose column (Amersham-Pharmacia, Uppsala, Sweden). The column was pre-equilibrated with 50 mm Tris ⁄ HCl pH 7.0, 10 mm CaCl 2 , 100 mm KCl. The concentration of CaM and the peptides was determined by absorbance at 280 nm using calculated e-values. Peptides were mixed with 30 nmol CaM in a 1 : 1 molar ratio and then analysed by gel filtration chromatography. The peaks were concentrated using 3 lL of Strataclean protein bind- ing beads (Stratagene, Amsterdam, the Netherlands) and then analysed on a 10–20% gradient Tris ⁄ tricine polyacryl- amide gel (BioRad, Hercules, CA, USA). CaM/peptide titrations Samples for NMR titrations contained 5 mm CaCl 2 ,20mm deuterated Tris ⁄ HCl pH 6.5, 5% (v ⁄ v) D 2 Oina400lL vol- ume. Titrations were carried out using a spectrometer with a magnet (Oxford Instruments) operating with a 1 H frequency of 500 MHz. The sample was maintained at a temperature of 25 °C. Gradient enhanced HSQC spectra were collected with a sweep width of 16 p.p.m. in the 1 H dimension and 40 p.p.m. in the 15 N dimension with the 1 H carrier frequency set to 4.74 p.p.m. and the 15 N carrier frequency set to 120 p.p.m. For each experiment 32 scans were taken with 128 increments in the nitrogen dimension. The Ca 2+ ⁄ CaM sample was at a concentration of 0.45 mm and contained % 0.2 lmol protein. The PhK5 peptide was reconstituted in NMR buffer to a concentration of 16 mm. Successive additions of 0.02–0.04 lmol of peptide were made to the CaM sample, to a final molar ratio of 1 : 1.4 CaM to peptide. The PhK13 peptide was treated in the same way and spectra were taken over a similar range, up to a ratio of 1 : 0.8 CaM to PhK13. Data were processed using FELIX 2.3 (Biosym Inc.) and analysed with xeasy [34]. Analysis of 15 N relaxation data NMR experiments were performed on spectrometers oper- ating at 1 H frequencies of 600 MHz at 25 °C. Backbone 15 N relaxation parameters, comprising the rates of 15 N transverse (R2) and longitudinal (R1) relaxation were measured using previously described experimental protocols [35]. 15 N R1 and R2 relaxation data for Ca 2+ ⁄ CaM and for the Ca 2+ ⁄ CaM ⁄ PhK5 complex were obtained by recording a series of gradient enhanced two-dimensional HSQC spectra with a series of T1 (20, 400, 600, 800, 1000, 1200 and 1400 ms) and T2 (8.6, 60.5, 86.4, 103.7, 129.6, 172.8 and 216 ms) delays. For the Ca 2+ ⁄ CaM sample, a 1.4 mm sample of CaM was made up in 20 mm deuterated Tris ⁄ HCl pH 6.5, 100 mm KCl, 10 mm CaCl 2 and 5% (v ⁄ v) D 2 O. For the Ca 2+ ⁄ CaM ⁄ PhK5 complex 350 lLof 1.4 mm CaM was diluted into 6 mL of the NMR buffer and mixed with 0.5 lmol PhK5 peptide. The complex was concentrated using a centricon filter unit with a 3 kDa cut-off to a final concentration of 0.75 mm. Data were processed with felix 2.3. The peaks were assigned by com- parison with the previous assignments for CaM [25] using the program sparky for assignment and measurement of peak intensities. The longitudinal and transverse relaxation rates (R1 ¼ 1 ⁄ T1 and R2 ¼ 1 ⁄ T2, respectively), were calculated by fit- ting a single exponential to the peak intensities for different time points using matlab 6.5. Four residues from each helix in the structure were used and their R1 and R2 values were averaged to produce orientation vectors for each helix for the rotational anisotropy analysis. Fitting of the hydro- dynamic parameters was carried out using rotdif written by D. Fushman (University of Maryland, USA) [29,30]. For the Ca 2+ ⁄ CaM data all eight helices from the CaM structure were represented, however, in the Ca 2+ ⁄ CaM ⁄ PhK5 data the fourth helix was discarded as spectral overlap resulted in too few data points for meaningful analysis. As no hetero- nuclear NOE measurements were taken for either data set, the NOE values assigned for each residue used in the analy- sis was 0.80 ± 0.04, to reflect typical values for residues in stable secondary structure elements and are consistent with previous measurements made for CaM [25,36]. Six Ca 2+ ⁄ CaM ⁄ peptide structures were selected from the PDB along with the collapsed, peptide-free structure of Ca 2+ ⁄ CaM (PDBid 1prw) [37] as input models for the calculations in rotdif. The structures included complexes Ca 2+ ⁄ CaM in complex with the PhK5 peptide A. G. Cook et al. 1520 FEBS Journal 272 (2005) 1511–1522 ª 2005 FEBS [...]... phosphorylase kinase and their potential role in phosphorylation-dependent activation of the holoenzyme J Biol Chem 267, 810– 818 20 Dasgupta M & Blumenthal DK (1995) Characterization of the regulatory domain of the gamma-subunit of phosphorylase kinase The two noncontiguous 1521 Ca2+ ⁄ CaM in complex with the PhK5 peptide 21 22 23 24 25 26 27 28 29 30 31 32 calmodulin-binding subdomains are also autoinhibitory... structure of a MARCKS peptide containing the calmodulin-binding domain in complex with Ca2+-calmodulin Nat Struct Biol 10, 226–231 42 DeLano WL (2002) The PyMOL Molecular Graphics System Delano Scientific, San Carlos, CA, USA 43 Owen DJ, Noble ME, Garman EF, Papageorgiou AC & Johnson LN (1995) Two structures of the catalytic domain of phosphorylase kinase: an active protein kinase complexed with substrate... Calcium control of muscle phosphorylase kinase through the combined action of calmodulin and troponin Ann NY Acad Sci 356, 151–161 4 Manning G, Whyte DB, Martinez R, Hunter T & Sudarsanam S (2002) The protein kinase complement of the human genome Science 298, 1912–1934 5 Goldberg J, Nairn AC & Kuriyan J (1996) Structural basis for the autoinhibition of calcium ⁄ calmodulindependent protein kinase I Cell... scattering studies show distinct conformations of calmodulin in its complexes with two peptides based on the regulatory domain of the catalytic subunit of phosphorylase kinase Biochemistry 29, 9316– 9324 Yao Y & Squier TC (1996) Variable conformation and dynamics of calmodulin complexed with peptides derived from the autoinhibitory domains of target proteins Biochemistry 35, 6815–6827 Ikura M, Kay LE & Bax... structure of native Ca2+-calmodulin Structure (Camb) 11, 1303–1307 38 Clapperton JA, Martin SR, Smerdon SJ, Gamblin SJ & Bayley PM (2002) Structure of the complex of calmodulin with the target sequence of calmodulin-dependent protein kinase I: studies of the kinase activation mechanism Biochemistry 41, 14669–14679 39 Kurokawa H, Osawa M, Kurihara H, Katayama N, Tokumitsu H, Swindells MB, Kainosho M... An inhibitory segment of the catalytic subunit of phosphorylase kinase does not act as a pseudosubstrate J Biol Chem 276, 34560– 34566 Juminaga D, Albaugh SA & Steiner RF (1994) The interaction of calmodulin with regulatory peptides of phosphorylase kinase J Biol Chem 269, 1660–1667 Trewhella J, Blumenthal DK, Rokop SE & Seeger PA (1990) Small-angle scattering studies show distinct conformations of. .. pleckstin homology (PH) domain in solution: analysis of 15N relaxation with monomer ⁄ dimer equilibrium J Mol Biol 266, 173–194 36 Ikura M, Kay LE, Krinks M & Bax A (1991) Tripleresonance multidimensional NMR study of calmodulin complexed with the binding domain of skeletal muscle myosin light-chain kinase: indication of a conformational change in the central helix Biochemistry 30, 5498–5504 37 Fallon... Calmodulin target database J Struct Funct Genomics 1, 8–14 18 Dasgupta M, Honeycutt T & Blumenthal DK (1989) The gamma-subunit of skeletal muscle phosphorylase kinase contains two noncontiguous domains that act in concert to bind calmodulin J Biol Chem 264, 17156– 17163 19 Newsholme P, Angelos KL & Walsh DA (1992) High and intermediate affinity calmodulin binding domains of the alpha and beta subunits of phosphorylase. .. Swindells MB, Kainosho M & Ikura M (2001) Target-induced conformational adaptation of calmodulin revealed by the crystal structure of a complex with nematode Ca2+⁄ calmodulin-dependent kinase kinase peptide J Mol Biol 312, 59–68 40 Aoyagi M, Arvai AS, Tainer JA & Getzoff ED (2003) Structural basis for endothelial nitric oxide synthase binding to calmodulin EMBO J 22, 766–775 41 Yamauchi E, Nakatsu T,... Wilmanns M & Gautel M (1998) Structural basis for activation of the titin kinase domain during myofibrillogenesis Nature 395, 863–869 7 Kobe B, Heierhorst J, Feil SC, Parker MW, Benian GM, Weiss KR & Kemp BE (1996) Giant protein kinases: domain interactions and structural basis of autoregulation EMBO J 15, 6810–6821 8 Kemp BE & Pearson RB (1991) Intrasteric regulation of protein kinases and phosphatases . [8]. In these structures a C-terminal extension to the protein kinase folds back on the kinase domain and interferes with the substrate binding sites. In the. C-terminal extension to the protein kinase domain and treatment of the kinase with Keywords calmodulin; kinase regulation; protein– protein interaction; NMR spectroscopy Correspondence J.