Crystallisation of eukaryotic ADPGKs

Several constructs with N-terminal truncations of ADPGK from M. musculus and H.

sapiens were generated (see Table 3.1 for an overview). The aim of these truncations

was to remove N-terminal regions or the potential signal sequences, with a propensity for disorder (see section 3.1). These regions would likely make unfavourable entropic contributions to the formation of crystals. C-terminal truncations or internal deletions were ruled out from the beginning, based on the assumption that eukaryotic ADPGK would have a fold similar to the archaeal ADPGKs. Therefore, one of the main catalytic sites, the GXGD motif, would be located very close to the C-terminus. Of the tested constructs it was only possible to obtain crystals for rmADPGKΔ51, a construct with a truncation of the first 51 amino acids based on isoform 3 of mouse ADPGK. Crystallisation trials for the human ADPGK or a full-length construct of either mouse or human ADPGK have not been successful thus far. However, the crystal contacts in the solved structure of the mouse ADPGK could shed some light on the underlying problem. Of the contacts in the rmADPGKΔ51 crystal listed in Table 3.7, two sites are of particular interest. The first site of interest is interface number 3 in Table 3.7 (or Figure 65, on the right side), which could be very specific to the isoform 3 of mouse ADPGK. The interaction is depicted in Figure 78.

Figure 78: Crystal contact specific for of M. musculus ADPGK isoform 3.

The figure shows a crystal contact identified by PISA between two symmetry mates (grey and green molecule), which is potentially very specific for the isoform 3 of mouse ADPGK. The isoform 1 of ADPGK incorporates an additional glutamine residue Q314, which would follow Q313 shown in the grey coloured molecule, which could prevent the formation of this crystal contact. The van der Waals radii of contacting atoms are displayed as yellow and purple surfaces.

Due to the high level of sequence conservation between mouse and human ADPGK, one would expect to find an analogous situation for isoform 1 and 2 of human ADPGK. A histidine residue H312 (grey molecule in Figure 78) makes contact with the side chain of aspartate residue D191 and the backbone of valine residue V189 in a different molecule (green molecule in Figure 78). The potential problem could arise in isoform 1 of mouse ADPGK. Two glutamine residues, Q313 and Q314 follow the histidine. Only one glutamine residue is found in isoform 3 (see Figure 68 for an alignment of this particular region). The additional glutamine could disrupt the hydrophobic packing of the inward facing side of the helix (see Figure 69), introducing a larger bulge in the last helical turn or producing a disordered stretch of sequence. In both cases, this would likely have an unfavourable contribution to crystal formation.

The second crystal contact of interest involves mostly the contact surface number 2 in Table 3.7 (see Figure 65), which affects the N-terminus of ADPGK. The first N- terminal helix of rmADPGKΔ51 is capped by proline residue P53 (yellow molecule in Figure 79), which packs against the side chain of the arginine residue R71 in a symmetry related molecule (green molecule in Figure 79). The side chain of arginine

154 residue R71 (green molecule in Figure 79) packs between two different symmetry mates (yellow and cyan molecule in Figure 79), alanine residue A50 (yellow molecule in Figure 79) at the N-terminus and proline and histidine residue P224 and H225 (cyan molecule in Figure 79) of the small domain. For the crystal contact between P53 and R71 to form, the position of the proline could be mandatory. In a truncated construct starting after the proline, the α-helix could extend further making the formation of this particular crystal form impossible. In a full-length construct, P53 is preceded by prolines in positions 50 and 47. Dihedral angles for position 50 fall in the allowed regions of the Ramachandran plot for proline in the rmADPGKΔ51 structures. Given the tightly stacked interaction of A50 and R73, a longer construct incorporating these residues could have a positive or negative effect on the crystallisation behaviour. The rigidity due to the restriction in dihedral angles of the proline residues could make a favourable (or at least negligible) entropic contribution to the crystal formation. On the other hand, in a full-length construct the N-terminal amino acids preceding these crystal contacts could form a small sub-domain by themselves or pack against the rest of the protein and in either case would likely interfere with the crystal formation. In the case with the N-terminal region being disordered, and the rest of the protein fold not altered, obtaining a crystal in this form of a full-length ADPGK would possibly not have yielded much additional structural information.

Figure 79: Crystal contacts of the truncated N-terminus of rmADPGKΔΔ51.

The figure shows the crystal contacts around the N-terminus of rmADPGKΔ51, where three different

symmetry mates are involved (yellow, green and cyan molecules). The van der Waals radii of contacting atoms are displayed as yellow and purple surfaces. The proline P53 at the beginning of the N-terminal helix (yellow) packs against the sidechain of arginine R71 (green). The side chain of arginine R73 (green) packs between alanine A50 (yellow) and proline P224 and histidine H225 (cyan).

When a protein is crystallised, it is always desirable to obtain diffraction data of the protein in different functional states. In the case for ADPGK, this would include the different stages of substrate binding, lid closing and possibly a catalytic transition state. Unfortunately, given the B-factors (shown in Figure 63) and the distribution of the crystal contacts (shown in Figure 65), there is a possibility that the particular crystal form found for rmADPGKΔ51 is unable to undergo a domain movement upon ligand binding in soaking experiments that introduce the ligands into the crystals. The apparent rigidity of the ’axis’ formed by the lid domain, the linking β-sheet hinge and the more N-terminal side of the large domain (indicated by lower B-factors in these particular regions as shown in Figure 63) indicates that the crystal packing could collapse upon ligand binding, dissolving the crystal. Hence to obtain the structure of a ligand bound ADPGK, eventually a different crystal form needs to be grown.