|
Laboratory of Molecular Biophysics
|
Previous: Regulation of Phosphorylase Kinase, Next: CDK Activating Kinase (CAK), Up: Protein Kinases, Return to: Contents.
Protein kinases interact with a number of different protein molecules during signalling processes. These include substrate molecules, upstream protein kinases, protein phosphatases, kinase inhibitors and other regulatory proteins. Interactions may be relatively transient, such as interactions with substrates, or more enduring, for example with regulatory subunits. The site of the protein-protein interaction is not necessarily exclusive to one protein and may overlap with the binding site of another protein that interacts with the kinase. A common interaction site has already been identified for the interaction of CDK2 with the phosphatase KAP (Song et al. 2001) and CDK2 with the small regulatory protein CKS1, as described in our work last year. A number of protein kinase structures have been compared to determine whether there are common features between different kinases that use a similar surface region for their protein-protein interactions.
|
The protein-kinase interaction between CDK2 and cyclin A was used as a basis for comparison with other protein kinases. Cyclins bind mainly to the N-terminal lobe of the CDKs, contacting the C-helix (or PSTAIRE helix) and the activation segment, causing a conformational change that activates the kinase. A similar region in other kinases was examined for interactions with other enzyme subunits or regulatory proteins. Three kinases (casein kinase 2 (CK2), cyclic-AMP dependent protein kinase (cAPK) and Erk2) were identified that have interactions with the kinase core that overlap the C-helix and the activation segment. A sequence alignment for these three proteins with CDK2 is shown in Figure 3. Residues in the kinase core that are less than 4Å away from the residues in the extraneous sequence are marked in light grey. The N- and C-terminal extensions that mimic the cyclin interaction are marked in dark grey.
There are distinct regions where all of the kinase core domains interact with
the extraneous sequences. These include the C-helix (where contacts to this
region are more extensive in CDK2/cyclin A complexes and in Erk2), the
ß6
strand, the loop region between the 
3 helix and the
ß9 strand just prior to
the catalytic loop, and a number of interactions with the activation segment
(although the latter are less extensive in cAPK). All of the kinases have
some contacts to the C-terminal domain of the kinase core but these contacts
are not numerous and do not overlap in all the kinases. A comparison of the
residues that are found in these interfaces does not reveal any strong sequence
identity except in the case of Ser120 (CDK2 numbering). This suggests that
while the protein- protein interaction site may be on the same face of the
kinase, the nature of the interactions is not conserved.
|
In order to assess the level of conservation between interaction surfaces of different kinases in terms of their global geometrical location, we simplified the geometric relationship between these surfaces by projecting three-dimensional surface patches into two dimensions. First, the coordinates of all four protein kinases were aligned based on conserved helices in the C-terminal lobe of the kinase core by using the LSQ commands in O. The coordinates of all atoms for each kinase were then transformed into a spherical coordinate system with its origin in the center of mass of the aligned proteins. Subsequently for each kinase the coordinates of all atoms involved in interaction with regulatory proteins were selected and the angular components of their spherical coordinates recorded (Figure 4). This is equivalent to projecting all atoms of the interaction surface onto a sphere around the aligned kinases, leading to a 2D "footprint" of the interaction surface for each kinase. It is then possible to assess the degree of overlap between these footprints as a measure of similarity in terms of geometric location between the interaction surfaces of the different kinases.
|
If the coordinates of the atoms from the kinase core that contact the extraneous sequence are all projected onto the surface of a sphere, then the radius value for each coordinate is the same and the footprint of the interaction can be displayed by plotting the angles, phi against theta. A comparison of the phi/theta plots for the 4 kinases is shown in Figure 5. The different plots show that the phi/theta angles are similar and therefore that the interaction sites for the different kinases overlap in three dimensions and are located on a similar face of the kinase. Contacts to cAPK are less numerous than the corresponding footprints of the other three kinases. This is due to a difference in surface area of the different interaction site:
| Kinase | Dimer Partner | Interface area |
| Erk2 | C-terminus | 1809 Å2 |
| CDK2 | Cyclin A | 1671 Å2 |
| CK2 | N-terminus | 1396 Å2 |
| cAPK | N-terminus | 756 Å2 |
A visual comparison of the hydrophobic regions of the different interface regions showed that while the surface on the kinase core had hydrophobic patches that correspond to similar hydrophobic patches on its binding partner, there was no distinct hydrophobic patch that was common to all the kinases. This is not unexpected due to the lack of sequence similarity of the of the interface residues.
|
It has been noted that the N-terminal extension of CK2 has a similar arrangement of helices as cyclin A at its binding site with CDK2. Comparing these sites with Erk2 and cAPK indicates that the N-terminus of cAPK has a similar helix as in CK2 although it is a slightly displaced. The Erk2 C-terminal segment chain also passes across the kinase core in a similar manner but is less helical in character (Figure 6).
In summary, we report methods that have allowed us to show that the kinase core regions from four kinases each have a similar binding site for regulatory subunits or extraneous C- and N-terminal extensions. These interactions occur mainly on the same face of the kinase and pack against the C-helix and the activation segment as well as other parts of the N-terminal kinase lobe. There is some evidence to suggest that this packing interaction is important in all of these different kinases. In CDK2 the binding of cyclin A causes conformational changes in the N-terminal lobe aligning the C-helix correctly for catalysis to occur and causing the activation segment to become more solvent accessible and therefore able to be phosphorylated. In CK2 the N-terminal extension is thought to have a stabilisation role, maintaining the catalytic site in its correct conformation. Activation of Erk2 leads to homodimerisation, a process that uses a dimerisation region on the C-terminal extension and is required for translocation of Erk2 into the nucleus. In cAPK removal of the N-terminal extension does not inactivate the constitutively active enzyme but does reduce the thermostabilty of the enzyme and alters the binding constants for its association with regulatory subunits. Therefore it seems that in all these kinases the packing of these extension regions against the kinase core leads to stability of the active site and in some cases activation of the kinase. It is possible that this same site is utilised by other protein kinases to perform a similar function.
Previous: Regulation of Phosphorylase Kinase, Next: CDK Activating Kinase (CAK), Up: Protein Kinases, Return to: Contents.