Figure 16: Ribbon representations of ID2 homodimer interactions. ID2: chain A in purple, chain B in brown, loop in green and potassium ion in grey (A) Hydrophobic core residues as spheres. The packing of the residues showed how the monomers interacted to form the homodimer. (B) Hydrogen bonds represented as dashed lines with distances in Å. Intrahelix bonds between Asn40 and Lys61. Interhelix bonds Tyr43 (chain A) with Tyr43 (chain B), Tyr71 (chain A) and Leu49 (chain A) shared bond with Gln76 (chain B). cA = chain A, cB = chain B (C) Conserved hydrogen bonds between ID2 crystal structure (Y43.cA-Y43.cB & Y71.cA-Q76.cB) and ID3 NMR structure (Y48.cA-Y48.cB & Y76.cA-Q81.cB). ID3: chain A in pink, chain B in red, loop in olive. (D) & (E) SDS-PAGE gels showing no protein production where expected (red arrow) in pDest565 vector at 17°C induction. U=before induction, P= insoluble pellet fraction, S=soluble fraction. (D) ID2, Y71 mutants, (E) ID2, Q76 mutants ! ! 55! ID2-K61A showed wild-type levels of soluble protein production (Figure 17A). The equivalent ID3-Q66A mutant did not produce any soluble protein (Figure 17C). To test for the effect of Q at equivalent positions, the residue from ID2 was exchanged with the residue in ID3 and vice versa. They were ID3-Q66K and ID2-K61Q. Expression tests showed that ID2-K61Q produced more soluble protein (Figure 17A) than wild-type ID2 (Figure 6D, red arrow) and ID3-Q66K did not produce any soluble protein, thus suggesting the importance of Q66 specifically to ID3’s stability and solubility. Figure 17: Loop region mutants of ID2 and ID3. SDS-PAGE: marker (kDa, lane M), before induction (lane U), insoluble pellet fraction (lane P), soluble fraction (S). Red boxes denote expected expression region. Gel A and B expression vector was pDest-565 induced at 17°C. Gel C expression vector was pDest-HisMBP induced at 17°C. (A) ID2 (N-HLH82-L) mutants: K61A showed wild-type expression in the soluble fraction, double mutant Q55A_K61A showed the same amount of expression as the K61A mutant and K61Q had the most soluble protein. (B) ID2 (N-HLH82-L) mutants: Q55A soluble protein production was similar to that of ID2-K61Q (gel A) (C) ID3 (HLH) mutants: R60A had little soluble protein, R60Q had more soluble protein than R60A. Q66K and Q66A did not express any protein. (D) ID2 (N-HLH82-L) mutants: Q55R had more insoluble protein fraction than soluble fraction. ! 56! 4.3.3 Comparison of ID3 homology model homodimer interactions The ID3 homology model (Wibley, et al., 1996) showed potential ID2 homodimer interactions at equivalent positions as follows: N38.cA repulsing K61.cB, D41.cA hydrogen bonding with Q71.cB and K45.cA forming a salt bridge with D75.cB (Figure 18A). However, these predictions were not observed in the ID2 structure even though structural and sequence alignments were highly similar (Figure 19). Wibley did show the potential interaction of ID3.Y76 with ID3.Q81 in their model but made no mention of it in the text. This interaction corresponded to the Q76.cA-Y71.cB hydrogen bond in ID2 that was found to be important for homodimer formation. Comparison of these predictions with the ID3 NMR structure supported the findings of the ID2 crystal structure. Similar to ID2, these interactions did not exist in the ID3 NMR structure (Figure 18B) and therefore were highly unlikely to support homodimer formation. ! 57! Figure 18: Predicted interactions based on ID3 homology model (Wibley, et al., 1996) were not found in either the ID2 crystal structure nor ID3 NMR structure. (A) ID2 structure showed equivalent residues from ID3 homology model. Wibley’s postulated interactions were i) K61 repelling N38 (ID3: D43 repels Q66), ii) Q66 h-bond with D41 (ID3: H46 hbond Q71), iii) D70 salt bridge with K45 (ID3: R50 salt bridge with D75). Chain A in purple, chain B in brown, loop in green. Interaction distances (Å) shown as dashed lines. (B) ID2 crystal structure superimposed with ID3 NMR structure showed the proposed ID3 homology model interactions from (A). ID2 structure: chain A in purple, chain B in brown, loop in green. ID3 structure: chain A in pink, chain B in red, loop in olive. Interaction distances (Å) shown as dashed lines. ! Figure 19: Structural alignment of the bHLH domain of ID proteins and their binding partners. Alignments were done manually using Pymol’s align function as a guide. ! ! 58! 4.3.4 Disulfide bond in ID2 homodimer formation In an earlier work, a cysteine residue in helix-1 of ID2 was shown to be critical in homodimer formation by the creation of an intermolecular disulfide bond (Liu, et al., 2000). However, the ID2 crystal structure revealed that C42 on each monomer pointed away from each other and did not form a disulfide bond (Figure 14C). Superposition of ID2 crystal structure and ID3 NMR structure showed almost identical conformation of this cysteine residue, thus confirming that it was unlikely to form a disulfide bond. It is possible that a disulfide bond could be a transient interaction, acting to bring the monomers closer together to form the functional homodimer. 4.4 Loop region The loop region of ID2 was unique in terms of having a monovalent positive ion at the start of the loop that was not previously reported in HLH structures. Contoured at 2.5-sigma, the radius of the 2Fo-Fc electron density map was approximately 1.4Å using Coot’s crosshairs. This most likely corresponded to a potassium ion based on size and the fact that the crystallization solution contained 2.5M Potassium Acetate. 2Fo-Fc electron density map contoured at 1.5σ showed the coordination of backbone oxygens from K47 (helix-1), V50 (helix-1), I53 (loop) and a side-chain oxygen from Q55 (loop) pointing towards this ion at favourable angles and at distances less than 3.2Å (Figure 14D). This ionic interaction was mirrored in both monomers and could explain the rigidity of the loop in previous studies where interchanging residues ID1.L76 (ID2.I53) and ID1.Q78 (ID2.Q55) led to total loss and partial loss of binding, respectively, to MYOD1 (Pesce, et al., 1993). The E47 homodimer reported a hydrogen-bonded network of glutamines at positions 364, 373 and 381 (Ellenberger, et al., 1994) (Figure 20) that acted to stabilize the loop. The authors mentioned that these three glutamines were unique to ! 59! the E-proteins, the group that bound to ID proteins. However, only one bond involving the side-chains of E47-Q373 and E47-Q381 had a distance less than 3.5Å. From the E47 structure, the side-chain oxygens of E47-Q364 (helix1), E47-K371 (loop), E47-Q381 (helix2) were coordinated in a very similar way to ID2’s ioninteracting loop residues. Therefore, it was possible that a similar interaction to a positive ion held the E47 loop in place (Figure 20). The potentially conserved ionic interaction of E47-K371 and ID2-Q55 could therefore play a role in binding specificity of IDs with the E-proteins. Figure 20: E47 homodimer showing the network of glutamines that were predicted to form hydrogen bonds but the distances were too far for most of them. Perhaps E47 also had a positive ion in the loop coordinated by two of the glutamines that held it rigid? (grey sphere). ! ! 60! Based on sequence alignments (Figure 19), Q364 corresponded to P51 in ID2. This residue was conserved in ID1, ID2 and ID3 and its mutation in ID1 led to a complete loss of binding to MYOD1 and E47 (Pesce, et al., 1993). Since prolines were known to disrupt secondary structures, such as alpha helices, and were commonly found in turns, where they provided leeway for the structure to change direction, it was possible that mutation of the proline at this position caused a geometrical change that was unfavourable to dimerization. Site directed mutagenesis experiments were done on some of the loop residues (Table 5) and their surrounds to aid in understanding how differences in the loop residues were responsible for ID homodimer formation and binding specificities. Alanine mutants were made to look at residues necessary for dimer formation. Other mutants exchanged ID2 for ID3 residues and vice versa to look for potential differences in binding affinities between the ID proteins. ID2-Q55A, Q55R and the equivalent ID3-R60A, R60Q mutants all expressed some fraction of soluble protein (Figure 17) which suggested that these residues were not fundamental for stability or dimerization. However, under the same expression conditions, ID2-Q55R and ID3R60A produced less soluble protein than ID2-Q55A or ID3-R60Q respectively. The ID2 structure showed a side-chain interaction between Q55 and the positive ion. A change to arginine could potentially repel this interaction and cause loss of protein solubility. This could explain why a change from the arginine to the glutamine residue in ID3 increased solubility. 4.5 N-terminal Helix-1 region Three residues (Table 11) in helix-1 of ID1 and ID3 were proposed to play a role in differential binding affinity to MYOD1 but not to E47 (Langlands, et al., 1997). In both the ID2 and ID3 structures, residues ID2-Y37, ID2-D41, ID3-D42, ID3-H46 pointed away from the dimer interface (Figure 21). Previously, ID2-K47 was shown to ! 61! interact with the loop ion and could potentially have an effect on dimer formation. ID3-R52 pointed towards the loop in a similar fashion as ID2-K47 and had the backbone oxygen pointed towards the ion so it was possible that a mutation in this residue could affect binding in a similar way to ID2. Table 11: Positions of residues thought to be important for heterodimerization with MYOD1. ID2 ID1 ID3 ID4 position residue position residue position residue position residue 37 Y 67 Y 42 D 66 C 41 D 71 G 46 H 70 D 47 K 77 K 52 R 76 R ID1 and ID3 cloning, expression and purification protocols may be found in Appendix 5. ! 62! Figure 21: Ribbon representation of ID2 and ID3 opposing chains to illustrate residues thought to play an important role in heterodimerization with MYOD1. Residues from ID2 (Y37, D41) and ID3 (D42, H46) pointed away from the dimer interface. ID2-K47 and potentially ID3-R52 had interactions with the loop ion that was necessary for homodimer formation of ID2. ! 63! . and ID3 and its mutation in ID1 led to a complete loss of binding to MYOD1 and E47 (Pesce, et al., 1993). Since prolines were known to disrupt secondary structures, such as alpha helices, and. could explain the rigidity of the loop in previous studies where interchanging residues ID1.L76 (ID2.I53) and ID1.Q 78 (ID2.Q55) led to total loss and partial loss of binding, respectively, to. (loop) and a side-chain oxygen from Q55 (loop) pointing towards this ion at favourable angles and at distances less than 3.2Å (Figure 14D). This ionic interaction was mirrored in both monomers and