Selection of the Kinase Panel

The selection of the kinase domains to be used in the kinase panel to test the screen is an important part of the testing of the screen and to the application of the results to identify similarities in the refolding of protein kinases. If the protein kinases selected were too closely related then this could lead to false positives when considering the similarities observed in the refolding. To avoid closely related kinases being selected, the expanded dendrogram poster of Manning et al. (2002) was used (expanded dendrogram available online at http://kinase.com).The kinases were chosen from different areas on the dendrogram to avoid close sequence relationships. Four kinases were chosen, to form a total kinase panel of five kinases

The first consideration of the kinase panel was to include a protein kinase which is known to refold from inclusion bodies. This would provide extra support

for the results obtained with p38α showing that the refolding screen is able to

identify refolded protein. For this purpose the kinase domain of Phosphorylase kinase was selected. Owen et al. (1995) had refolded this protein from inclusion bodies in order to produce soluble protein which was crystallised and the structure solved from those crystals. Phosphorylase kinase (PhK) is a multi domain protein,

and the functional protein is composed of several subunits. The γ-subunit contains

the catalytic kinase domain, with the other subunits being involved in substrate recognition and control of the catalytic activity. The isolated kinase domain is constitutively active. Phosphorylase kinase is a member of the CAMK family and is

not closely related to p38α. The expression construct for this protein was created by

gene synthesis, and cloned into an expression vector as described in section 3.3.2.

Although tyrosine kinase and tyrosine kinase like (TK and TKL) kinase domains had been excluded from the kinase panel, the isolated kinase domain of TTK was included as it is a dual specificity kinase, and it was thought that changes in the active site required to accommodate the tyrosine side chain might result in differences in the folding of the kinase domain. The TTK kinase domain has been isolated from the context of a much larger multi-domain protein which regulates the activity of the kinase domain as well as performing other functions. The isolated TTK kinase domain is constitutively active, like the kinase domain of PhK. TTK does not fit into one of the families of protein kinases in the analysis of Manning et al. (2002), but is found on the arm of the dendrogram which terminates in the

CMGC group, which contains p38α (Figure 1.4). An expression construct for TTK

(514-804) was supplied by AstraZeneca (Section 3.2.1). This expression construct consisted of a codon optimised expressed sequence contained in a pT7#3.3 vector.

The isolated kinase domain AKT2 or PKBβ was included as a representative

of the AGC group which comprises the protein kinase A, G and C families. These kinases have a interesting sequence feature. As described, previous work on the

equilibrium folding of p38α had identified an absolutely required, conserved core

tryptophan, W207. In the protein kinase B family and other AGC family members there is a double tryptophan motif in the core of these kinases. This raised the possibility of different folding pathway driven by the double tryptophan motif in the

core of the protein. The AGC group also contains many important kinases, so the inclusion of a member of this group was important for representing the kinase domain. The expression construct for AKT2 was also supplied by AstraZeneca (Section 3.2.1).

The final kinase selected to form this kinase panel was included as a kinase regarded to be a challenging kinase to produce in a soluble form. KIS (kinase interacting with stathmin) is a two domain protein which has yet to have its structure solved and limited information on thein vivofunction of the kinase is available. The two domains comprise a kinase domain at the N-termini of the protein and a second C-terminal domain whose function is currently unknown, but based on sequence details is suspected to be an RNA binding domain. The production of the kinase domain in KIS is difficult and has yet to be achieved in significant amounts. Soluble protein can be generated as a GST fusion; however upon removal of the GST domain, the protein aggregates indicating that the protein was mis-folded, or partially folded (AstraZeneca unpublished data). If the protein were able to be produced in a soluble, folded form from a refolding screen this would be especially advantageous as it could lead to the possibility of a crystal structure of the protein. The Manning et al.,(2002) analysis indicates that KIS is closely related to TTK. Despite this feature, and the possibility of similarities between the kinases being due to their close relationship, the very different behaviours of the two proteins, and the lack of structural information for KIS, it was included as an interesting target for refolding. The KIS kinase domain expression construct, representing residues 1-313 of the full length protein, was supplied by Astrazeneca in a pT7#3.3 vector (Section 3.2.1).

The proteins which comprised the kinase panel share certain sequence features which are common to the kinase fold. However, overall sequence identity is low. A sequence alignment of the five members of the kinase panel identifies little common sequence (Figure 3.1). The sequence features identified as common, such as the DFG motif are mostly related to the catalytic function of the proteins. The conserved tryptophan can however be seen.

Figure 3.1:Sequence alignment of the kinase domains of the members of the kinase panel. Consensus sequence indicates residues at least 80% conserved. Key kinase domain motifs such as the DFG motif and the conserved core tryptophan can be noted. Alignment performed using CLC Bio Sequence Viewer, using proprietary pairwise alignment function, utilising default parameters. Background colour indicates degree of residue conservation between kinases from blue to red; unique residues to completely conserved residues.

Although the members of the kinase panel show low sequence homology the crystal structures of the kinase domains which have been solved for all of the members of the panel apart from KIS, show a common kinase fold (Figure 3.2).

Figure 3.2:Crystal Structures of the members of the kinase panel. (A) – crystal structure of AKT2. (B) – crystal

structure of p38α. (C) – Crystal structure of PhK. (D) – Crystal Structure of TTK. (E) – Alignment of the members of the kinase panel to p38α. p38α shown in red, AKT2 in green, PhK in blue and TTK in magenta. All images

produced using ray tracing module of Pymol (Delano, 2008). Alignment performed in the same software.

A B

C D

The alignment of the kinase domains indicates that the members of the kinase

panel have very similar folds. p38α shows several additional α-helices in both the N-

terminal lobe and the C-terminal lobe. There is more variability in the arrangement

of the β-sheets of the N-terminal lobe than in the α-helices of the C-terminal lobe.

From the sequence alignment and the structural alignment of the proteins selected for the kinase panel it is clear that the selected proteins are sequence diverse but structurally similar. The proteins of the kinase panel have a diverse set of physical properties. A selected set of these properties are summarised in table 3.1

Table 3.1:Selected physical properties of the members of the kinase panel

constructs. pI calculated according to Bjellqvistet al.,(1993).

Kinase # of Residues in Kinase Domain # of Residues in Construct Calculated Mass of Construct Calculated pI of Construct N- Terminal 6His tag AKT2 340 354 41.3 kDa 5.65 Y KIS 313 339 37.5 kDa 6.65 Y p38α 360 360 41.3 kDa 5.48 N PhK 299 326 37.5 kDa 6.06 Y TTK 290 315 37.7 kDa 8.42 Y