A Dimeric Kinase Assembly Underlying Autophosphorylation in the p21 Activated Kinases
Michelle Pirruccello, Holger Sondermann, Jeffrey G. Pelton, Patricia Pellicena, André Hoelz, Jonathan Chernoff, David E. Wemmer and John Kuriyan
Summary: The p21-activated kinases (PAKs) are serine/threonine kinases that are involved in a wide variety of cellular functions including cytoskeletal motility, apoptosis, and cell cycle regulation. PAKs are inactivated by blockage of the active site of the kinase domain by an N-terminal regulatory domain. GTP-bound forms of Cdc42 and Rac bind to the regulatory domain and displace it, thereby allowing phosphorylation of the kinase domain and maximal activation. A key step in the activation process is the phosphorylation of the activation loop of one PAK kinase domain by another, but little is known about the underlying recognition events that make this phosphorylation specific. We show that the phosphorylated kinase domain of PAK2 dimerizes in solution and that this association is prevented by addition of a substrate peptide. We have identified a crystallographic dimer in a previously determined crystal structure of activated PAK1 in which two kinase domains are arranged face to face and interact through a surface on the large lobe of the kinase domain that is exposed upon release of the auto-inhibitory domain. The crystallographic dimer is suggestive of an engagement that mediates trans-autophosphorylation. Mutations at the predicted dimerization interface block dimerization and reduce the rate of autophosphorylation, supporting the role of this interface in PAK activation.
Figure 1. PAK domain structure and production of phosphorylated PAK2 kinase domain. (a) The domain organization of PAK1 and PAK2 with domain boundaries indicated for each isoform. The CRIB (Cdc42/Rac interactive binding) domain is comprised of residues 70−95 (PAK1 numbering). (b) The structure of the active conformation of PAK1 (PDB code 1YHV). All molecular graphics images were generated with PYMOL [http://www.pymol.org]. (c) The inactivated kinase domain was phosphorylated for 30 min at room temperature by the addition of catalytically active full length PAK2. The phosphorylated and unphosphorylated PAK2Inact samples were run on a denaturing SDS-polyacrylamide gel and probed with a PAK2-specific phosphothreonine antibody (PT402), or a PAK2-specific antibody (PAK2).
Figure 2. NMR spectroscopy of the phosphorylated PAK2Inact kinase domain. 2D 1H-15N TROSY correlation spectrum recorded at 300 K on an 800 MHz spectrometer. (a) The spectrum for a kinase sample at 500 μM protein concentration does not display many resolved peaks. The expanded region of the spectra (117−124 ppm 15N and 7.5−8.5 ppm 1H) is highlighted in yellow. There are also sets of peaks with different intensities (red and green arrows). (b) and (c) Effects of peptide substrates on the NMR spectra of the phosphorylated kinase domain. Same expanded section as above. A tenfold molar excess of LIMKtide or RAFtide caused the peaks to sharpen and increase in number. This did not occur with a stoichiometric amount of LIMKtide or the addition of PYE, which is not a PAK2 substrate. (d) 10 mM MgCl2 and 5 mM ATP were added to kinase at 500 μM protein concentration with no apparent change in the line widths and peak intensities. The pink arrow indicates a peak that disappears upon nucleotide addition.
Figure 3. The phosphorylated PAK2Inact kinase domain dimerizes in solution. (a) Samples containing 2 mg/ml of purified phosphorylated or unphosphorylated PAK2Inact kinase domain were analyzed by multi-angle light scattering coupled to gel filtration. The calculated molecular mass (dotted lines) is plotted as a function of elution volume, with the phosphorylated kinase domain displaying a larger apparent molecular mass than the unphosphorylated sample. (b) The phosphorylated PAK2Inact sample was analyzed by analytical equilibrium ultracentrifugation. Sedimentation was measured at 280 nm after the system had reached equilibrium and analyzed as described in Materials and Methods. The calculated curve for a monomer-dimer equilibrium model is shown in blue with a dissociation constant of 0.91 μM. The residuals to the fit are shown above the absorbance curve indicating the reliability of the fit.
Figure 4. Analysis of the potential dimerization interface. (a) The crystal symmetry of the active conformation of PAK1 (PDB code 1YHV) was used to identify a potential association interface of the active kinase domain, shown with PAK2 numbering (these residues are absolutely conserved in the two isoforms). (b) The interface is expanded to show the residues described in the text, in the orientation shown above the expanded sections. The kinase domain from the inactive PAK1 structure (PDB code 1F3M) (red) was aligned onto the active kinase (blue) to survey possible interfacial contacts with the dimer partner (grey). The observed steric clash provides a possible explanation for the monomeric behavior of the unphosphorylated kinase domain.
Figure 5. Alignment of PAK family sequences. The conservation of residues that interact with the autoinhibitory domain in the inactive conformation (red), and those that make contact with both the autoinhibitory domain in the inactive state and the dimer partner in the active state (blue) are indicated.
Figure 6. NMR spectroscopy of the phosphorylated PAK2Inact/L449Q kinase domain. 2D 1H-15N TROSY correlation spectra of 500 μM kinase were recorded at 300 K on an 800 MHz spectrometer. (a) The mutant, which is deficient in dimerization, has a markedly improved 2D spectrum. (b) Addition of 500 μM LIMKtide shows even greater improvement, with the appearance of many new peaks, which was not seen with PAK2Inact sample. (c) The PAK2Inact/L449Q mutant and PAK2Inact spectra are well overlaid at high peptide concentrations, indicating that the mutant is folded correctly. Note the enhanced peptide sensitivity of the PAK2Inact/L449Q construct, with many more peaks seen in a spectrum obtained with fourfold lower stoichiometric ratio of peptide than used for the PAK2Inact spectrum (purple).
Figure 7. Basis for peptide enhancement of the NMR spectra. The heptapeptide substrate of phosphorylase kinase (teal, PDB code 2PHK) modeled onto the kinase domain of PAK2 shows the overlap between the substrate and the G helix of the dimer partner (blue).
Figure 7. The dimerization surface is utilized in the trans-autophosphorylation reaction. (a) A model for the trans-autophosphorylation event, based on the orientation of a substrate peptide in the active site of the catalytic kinase. The activation loop of the substrate kinase can easily fit in the correct orientation into the active site of the partner kinase. (b) 32P incorporation into the PAK2Inact samples by full length active PAK2 kinase. Each data point is the average of at least three independent experiments. Comparison of initial reaction rates for the dimerization mutants shows a reduction in the rate of phosphoryl transfer upon mutation of the interface. The inset shows the data for the PAK2Inact sample. (c) The catalytic efficiency of the reaction is significantly reduced for the PAK2Inact/L449Q mutant.