High-resolution crystal structure of cAMP-dependent protein kinase from Cricetulus griseus

The high-resolution crystal structure of a cAMP-dependent protein kinase is reported.

Abstract

Protein kinases (PKs) are dynamic regulators of numerous cellular processes. Their phosphorylation activity is determined by the conserved kinase core structure, which is maintained by the interaction and dynamics with associated domains or interacting proteins. The prototype enzyme for investigations to understand the activity and regulation of PKs is the catalytic subunit of cAMP-dependent protein kinase (PKAc). Major effects of functional regulation and ligand binding are driven by only minor structural modulations in protein–protein interactions. In order to resolve such minor structural differences, very high resolution structures are required. Here, the high-resolution X-ray structure of PKAc from Cricetulus griseus is reported.

1. Introduction  

Protein kinases (PKs) catalyze chemical modifications of their interaction partners (peptides and proteins) mostly by transferring the γ-phosphates of nucleotide triphosphates (NTPs) to the free hydroxyl groups of the side chains of the amino acids serine, threonine and tyrosine (Hanks & Hunter, 1995 ▸). Hence, a change in the functional activity of the substrate is induced, resulting in changes in either enzymatic activity, cellular localization or interaction properties. The activity of PKs must be highly regulated, since their activity has profound effects on cell cycles. Dysregulation of kinases commonly results in severe diseases, including cancer, neurodegeneration and metabolic disorder (Noble et al., 2004 ▸). PK activity can be modulated by autophosphorylation, interaction of activator or repressor proteins or the binding of small-molecule inhibitors to the catalytic protein kinase domain.

The catalytic subunit of the PK superfamily is highly conserved. Numerous homologous PK crystal structures have been reported in the Protein Data Bank (PDB; Berman et al., 2000 ▸), but a number of these structures lack essential details of the kinase domain owing to high flexibility. Therefore, the crystallization of the catalytic domain still needs to be improved. Furthermore, key insight into the catalytic mechanism of kinases and their interaction with regulators, inhibitors and substrates has come from the availability of several kinase structures. New structures therefore continue to improve our understanding of the interactions between PKs and substrates at different steps of the reaction cycle (Adams, 2001 ▸).

The catalytic subunit of cAMP-dependent protein kinase (PKAc) has served as a model kinase as it can be expressed in Escherichia coli owing to its autophosphorylation activity. Furthermore, PKAc is a surrogate kinase to study PKB/Akt kinase and also Rho-kinase, which are validated cancer targets and cannot be expressed in bacteria for structural studies. Structurally, PKAc is described by a single conserved catalytic core domain comprising a small amino-terminal N-lobe mainly characterized by β-sheets and a large carboxy-terminal C-lobe assembled from α-helices (Knighton et al., 1991 ▸). The NTP-binding pocket is located in a cleft between the lobes. Nucleotide and substrate binding induce considerable changes in conformation and dynamics. These changes lead to movements of both lobes, resulting in an opening and closure of the active pocket, which is essential for the catalytic cycle (Taylor et al., 2012 ▸). To date, 173 crystal structures of different PKAc homologues have been reported in the PDB with resolutions from 1.23 to 3.7 Å. Only two full-length ATP-bound crystal structures with a resolution below 1.5 Å have been described [PDB entry 1rdq (Yang et al., 2004 ▸) and PDB entry 4dfx (Bastidas et al., 2012 ▸)], providing more detailed information on the water network surrounding PKAc, which influences ligand binding and activity. Here, we report the full-length ligand-free crystal structure of PKAc from Chinese hamster (Cricetulus griseus; UniProt ID P25321) at a resolution of 1.14 Å.

2. Materials and methods  

2.1. Protein production and crystallization  

Full-length PKAc (residues 1–351) from C. griseus (UniProt ID P25321) was previously cloned into the NdeI/BamHI sites of a pET-TEV vector (a modified pET-16b vector; Novagen) with an N-terminal His₇ tag followed by a tobacco etch virus (TEV) protease cleavage site (Langer et al., 2004 ▸; see Table 1 ▸). PKAc was recombinantly expressed in E. coli BL21 (DE3) cells (Life Technologies). The cells were grown at 310 K to an OD₆₀₀ of 0.6, stored on ice for 20 min, induced with 1 mM IPTG and grown for 18–20 h at 291 K. Cultures were centrifuged and microfluidized in a buffer containing 50 mM Tris pH 8.0, 5 mM β-mercaptoethanol, 5 mM imidazole, 500 mM NaCl and ultracentrifuged. PKAc was purified from the supernatant using immobilized metal-ion affinity chromatography (IMAC; HisTrap HP, GE Healthcare). The His₇ tag was removed from the protein by TEV protease digestion while dialyzing the protein against lysis buffer. After an additional IMAC run, the flowthrough protein fractions were dialyzed against a buffer containing 50 mM HEPES pH 7.4, 50 mM NaCl, 5 mM dithiothreitol (DTT). To separate the different phosphorylation states of recombinantly expressed PKAc, the protein was applied onto a strong cation-exchange column (Mono S, GE Healthcare) and eluted with a gently inclined linear salt gradient to 150 mM NaCl. Protein fractions eluted at the highest salt concentration were pooled and concentrated to 13–15 mg ml⁻¹ in a buffer containing 25 mM MES/bis-tris buffer pH 6.4–6.6, 1 mM DTT, 0.1 mM EDTA, 75 mM LiCl. After the addition of 1.5 mM non-ionic detergent MEGA 8 (G-Biosciences) and 0.8 mM cAMP-dependent protein kinase inhibitor peptide PKI-tide (IAA­GRTGRRQAIHDILVAA; Sigma–Aldrich), crystals were grown by the vapour-diffusion method at 277 K in 3 µl hanging drops against freshly prepared reservoir solution (400 µl) containing 16–23%(v/v) methanol (see Table 2 ▸).

Table 1. Protein-production information.

Source organism	C. griseus
Expression vector	Modified pET-16b
Expression host	E. coli
Complete amino-acid sequence of the construct produced
cAMP-dependent protein kinase catalytic subunit †	MGHHHHHHHASENLYFQGHMGNAAAAKKGSEQESVKEFLAKAKEEFLKKWESPSQNTAQLDHFDRIKTLGTGSFGRVMLVKHKETGNHYAMKILDKQKVVKLKQIEHTLNEKRILQAVNFPFLVKLEFSFKDNSNLYMVMEYVPGGEMFSHLRRIGRFSEPHARFYAAQIVLTFEYLHSLDLIYRDLKPENLLIDQQGYIQVTDFGFAKRVKGRTWTLCGTPEYLAPEIILSKGYNKAVDWWALGVLIYEMAAGYPPFFADQPIQIYEKIVSGKVRFPSHFSSDLKDLLRNLLQVDLTKRFGNLKNGVNDIKNHKWFATTDWIAIYQRKVEAPFIPKFKGPGDTSNFDDYEEEEIRVSINEKCGKEFTEF
cAMP-dependent protein kinase inhibitor PKI-tide	IAAGRTGRRQAIHDILVAA

^†The cleaved squence is underlined.

Table 2. Crystallization.

Method	Vapour diffusion
Plate type	Cryschem plate
Temperature (K)	277
Protein concentration (mgml1)	1315
Buffer composition of protein solution	25mM MES/bis-tris buffer pH 6.46.6, 1mM DTT, 0.1mM EDTA, 75mM LiCl, 1.5mM MEGA 8, 0.8mM PKI-tide
Composition of reservoir solution	2025% methanol
Volume of drop (l)	3
Volume of reservoir (l)	400

2.2. Data collection, structure solution and refinement  

Rod-shaped crystals (Fig. 1 ▸ a) were harvested using mounted cryoloops (Hampton Research), transferred into 30%(v/v) MPD and flash-cooled in pucks immersed in liquid nitrogen for storage until X-ray diffraction data collection. Diffraction data were collected on BL14.1 operated by the Joint Berlin MX-Laboratory at the BESSY II electron-storage ring (Berlin-Adlershof, Germany; Mueller et al., 2012 ▸). The best data set was processed using XDSAPP (Krug et al., 2012 ▸). Data-collection statistics are shown in Table 3 ▸. The structure was determined by molecular replacement with Phaser (McCoy et al., 2007 ▸) using the crystal structure of bovine PKAc (PDB entry 1yds; Engh et al., 1996 ▸) as a search model. A final translation-function Z-score (TFZ) of 8.8 indicated successful structure solution. Model building was performed using Coot (Emsley & Cowtan, 2004 ▸) and the structure was refined using PHENIX (Afonine et al., 2012 ▸). Figures containing molecular graphics were prepared using PyMOL (Schrödinger). Structure-solution and refinement statistics can be found in in Table 4 ▸.

Figure 1.

(a) PKAc protein crystals. (b) The high-quality 2F _obs − F _calc electron-density map reveals the existence of phosphorylated residues Ser11 (i), Thr198 (ii) and Ser339 (iii) and alternative conformations, for example His143 (iv). (c) Interaction network of PKAc and PKI-tide (generated with PDBsum). Hydrogen bonds are marked in blue and nonbonded contacts are shown as orange dashed lines. (d) Stereo cartoon representation (wall-eyed) of the PKAc–PKI-tide complex. PKI-tide (red) binding introduces a closed conformation of the N-lobe (green) and C-lobe (blue). The N-terminal linker helix (light green) and the C-terminal loop region (dark blue) are fully visible in the electron density.

Table 3. Data collection and processing.

Values in parentheses are for the outer shell.

Diffraction source	BL14.1, BESSY
Wavelength ()	0.918409
Temperature (K)	100
Detector	Pilatus 6M
Crystal-to-detector distance (mm)	180
Rotation range per image ()	0.1
Total rotation range ()	180
Exposure time per image (s)	0.1
Space group	P212121
a, b, c ()	58.28, 72.95, 109.35
, , ()	90, 90, 90
Mosaicity ()	0.064
ISa†	25.51
Resolution range ()	45.531.139 (1.211.139)
Total No. of reflections	1101841
No. of unique reflections	169712
Completeness (%)	99.7 (98.7)
Multiplicity	6.49 (6.00)
I/(I)	17.59 (2.3)
R meas ‡	0.054 (0.76)
Overall B factor from Wilson plot (2)	12.5

^†ISa = I/(I)^asymptotic represents the highest possible signal-to-noise ratio of a data set (Diederichs, 2010 ▸). Data sets with ISa values of 25 or greater are considered to be very good and should allow straightforward structure determination.

^‡ R _meas = (Diederichs Karplus, 1997 ▸; Weiss Hilgenfeld, 1997 ▸).

Table 4. Structure solution and refinement.

Values in parentheses are for the outer shell.

PDB code	4wih
Resolution range ()	39.8741.139 (1.15201.1391)
Completeness (%)	99.7
No. of reflections, working set	161224 (5043)
No. of reflections, test set	8485 (265)
Final R cryst	0.177 (0.2510)
Final R free	0.191 (0.2633)
No. of non-H atoms
Protein	2978
Ligand	114
Solvent	585
R.m.s. deviations
Bonds ()	0.006
Angles ()	1.013
Average B factors (2)
Protein	25.0
Water	36.9
Ramachandran plot
Most favoured (%)	92.8
Allowed (%)	7.2

3. Results and discussion  

3.1. Protein expression and crystallization  

Previous studies reporting the NMR backbone assignment of PKAc (Langer et al., 2004 ▸) required extensive long-term protein stability investigations on different PKAc homologues. These studies revealed that the cAMP-dependent protein kinase from C. griseus can be expressed in prokaryotic hosts with high yields (>10 mg per litre of culture). Furthermore, the protein is phosphorylated and catalytically active. The most important advantage of choosing this particular homologue for in vitro experiments is its good stability at high protein concentrations at 298 K. These features allow a systematic transfer from NMR studies to different biophysical methods requiring good protein stability at high concentration (including X-ray crystallography and ITC measurements). Initial crystallization attempts indicated that the homogeneity of the phosphorylation state of the sample is a crucial requirement to obtain highly diffracting crystals. Only by strong cation-exchange chromatography (Mono S) was it possible to separate the different phosphorylation states properly. Generally, the UV₂₈₀ chromatogram of a typical Mono S run is characterized by two well separated peaks of approximately equal heights at different elution volumes. Homogeneously threefold-phosphorylated PKAc eluted at a higher NaCl concentration as a second UV₂₈₀ peak and yielded very well diffracting crystals, while attempts to produce good diffracting crystals from higher phosphorylated protein samples of the first peak did not succeed.

3.2. Overall structure of C. griseus PKAc  

The ternary structure of the PKAc–PKI-tide complex in the absence of nucleotides or small molecules bound to the active site was solved at 1.14 Å resolution with a crystallographic R factor of 17.7% and an R _free of 19.1% (Table 4 ▸). Although the quality of the electron-density map was not completely homogeneous, all expected residues could be placed without ambiguity. Three phosphorylation sites important for membrane association (Ser11), kinase activation (Thr198) and facilitating protein–protein interaction (Ser339) could be clearly mapped (Fig. 1 ▸ b, i–iii; Taylor et al., 2012 ▸) Several side chains with alternative conformations were found [Glu26, Met72, Gln97, Met121, His143 (Fig. 1 ▸ b, iv), Ser160, Thr184, Lys218 and Asn287]. Additionally, four C-terminal residues of the PKI-tide (LVAA) were missing, but the core binding motif (RTGRRQAI) is fully visible. Overall, 17 strong hydrogen-bond interactions mainly mediated by Arg8 and Arg9 stabilize PKI-tide binding. Additionally, 148 nonpolar contacts increase the affinity substantially (Fig. 1 ▸ c). In the PKI-tide bound state the N- and C-lobes are in close proximity. The catalytic core domain adopts a closed conformation, as reported for all PKAc crystal structures with bound inhibitor peptides (Fig. 1 ▸ d). The ATP-binding site is not accessible to substrates. The globular heterodimer consisting of 351 PKAc and 15 PKI-tide residues is surrounded by an extensive water network consisting of 585 water molecules.

3.3. Structural comparison of PKAc homologues  

Sequence-based alignments of PKAc from different vertebrates demonstrate their high homology (Fig. 2 ▸ a). PKAc from C. griseus (KAPCA_CRIGR) shows the following sequence identities to its homologues: mouse (Mus musculus; KAPCA_MOUSE), 98.6%; human (Homo sapiens; KAPCA_HUMAN), 98.3%; cattle (Bos taurus; KAPCA_BOVIN), 97.7%. Less surprisingly, the overall structure of PKAc reported here (PDB entry 4wih) was very similar to that of the nucleotide-free protein ternary complexes from mouse (PDB entry 4o22; Gerlits et al., 2014 ▸) and human (PDB entry 3nx8; Behnen et al., 2012 ▸). Both structure backbones align very well with hamster PKAc, with a root-mean-square deviation of 0.5 Å (Fig. 2 ▸ b). Nevertheless, several structural differences can be observed. Besides the fully exposed N-terminal helix carrying a phosphorylation site (Ser11), hamster PKAc (Fig. 2 ▸ b, blue) shows a slightly more closed conformation than human PKAc (Fig. 2 ▸ b, green), while mouse PKAc (Fig. 2 ▸ b, red) represents the most open conformation. The C^α—C^α distance between Leu83 of α-helix 2 and Pro244 of α-helix 8 (hamster, 15.5 Å; human, 16.8 Å; mouse, 17.9 Å) indicate movement of the N- and C-lobes. Further differences are located in the loop connecting α1 and β1. The course of the protein backbone of hamster PKAc does not align well with the others. Twofold to threefold increased B factors and a weak electron density suggest increased mobility of this region. The glycine-rich loop of human PKAc is tilted into the active site as a result of the binding of a phenol molecule in the ATP pocket. In the ligand-free structures from mouse and hamster the glycine-rich loops have a straight orientation (Fig. 2 ▸ b, right). Hence, protein crystallization of hamster PKAc was performed with a truncated protein kinase inhibitor peptide in which the N-terminal PKI-tide α-helix was missing. Residues 1–4 interact differently with PKAc, leading to a different orientation of the PKI-tide N-terminus. This did not affect the kinase inhibition. With few exceptions, all residues of the further PKI-tide core binding motif align very well. Interestingly, the side chain of Arg8 interacts via hydrogen-bond contacts with the glycine-rich loop of human and mouse PKAc but not hamster PKAc. Here, Arg8 contacts the C-terminal tail loop.

Figure 2.

(a) Amino-acid sequence alignment between PKAc from C. griseus (KAPCA_CRIGR), M. musculus (KAPCA_MOUSE), H. sapiens (KAPCA_HUMAN) and B. taurus (KAPCA_BOVIN) with secondary-structure annotation from PDB entry 4wih reprocessed with ESPript (Robert & Gouet, 2014 ▸). Identical residues are boxed in red and differences are marked in white. (b) Comparison of PKAc–PKI-tide complexes from C. griseus (PDB entry 4wih; PKAc, blue; PKI, cyan), M. musculus (PDB entry 4o22; PKAc, red; PKI, orange) and H. sapiens (PDB entry 3nx8; PKAc, green; PKI, yellow). The positioning of the phosphate-anchoring ‘glycine-rich loops’ is magnified in the circle. Structural alignments were performed with PyMOL (Schrödinger).

4. Conclusion  

The reported crystal structure of PKAc from C. griseus represents the current limit for high resolution of all available cAMP-dependent protein kinase crystal structures. In the context of further structural studies on small-molecule inhibitor binding, we provide a prototype protein kinase model offering the highest possible resolution and the most detailed information about the water network. PKAc represents a tool kinase for future studies investigating the selectivity and affinity of kinase inhibitors by various biophysical techniques (crystallography, NMR, ITC and SPR). For a wide analysis of structure, solvation patterns, mobility and energetic aspects of ligand binding, the reported crystal structure of PKAc from C. griseus represents a very good starting point.

Supplementary Material

PDB reference: cAMP-dependent protein kinase, 4wih