3.4 Orienting ASN, GLN and HIS side-chains 3.4.1 Introduction To define a structure by X-ray crystallography a protein must be modelled into an electron density map that at usual resolutions rarely shows hydrogen atoms and shows little or no difference between carbon, nitrogen and oxygen atoms. There are now a few structures (three in the October 1993 release of the Brookhaven Protein Database (Bernsteine et. al. 1977)) at resolutions as high as 1.0A where the carbon, nitrogen and oxygen atoms can sometimes be differentiated and some of the hydrogens can be observed. However, given the diffracting power of most protein crystals, these will probably remain the exceptions. For the majority of side-chains the atoms can be uniquely identified from the shape of the electron density map, but for asparagine, glutamine and histidine, whose side-chains appear symmetrical in the electron density, some specific atoms can only be identified on the basis of their environment, principally their hydrogen bonds. It is also difficult to differentiate between the three different protonation states of histidine. As the imidazole ring of histidine has a pK (6.5-7.0) close to physiological pH (~7-8) (Matuszak and Matuszak, 1976), both the basic and charged forms occur in vivo. The positively charged form is protonated on both imidazole nitrogens, whilst the basic form is protonated on only one imidazole nitrogen, and occurs as two tautomers which differ in which nitrogen is protonated. NMR studies on His at basic pH and in the polypeptide antibiotic Bacitracin suggested that the basic His is more usually protonated on NE2 rather than the ND1 (Reynolds et al 1973). Because both charged and basic forms of histidine are stable, histidine often participates in catalysis, and is found in the active sites of enzymes (e.g. serine proteases such as chymotrypsin) or as an axial ligand in metalloproteins such as the cytochromes (e.g. Cytochrome b5). In chymotrypsin, for instance, the active site histidine is involved in every step of catalysis and changes protonation state four times in the entire catalytic cycle. The chemistry of Asn and Gln is simpler. Here the problem is only to distinguish between the side-chain nitrogen and oxygen atoms. Distinguishing between the two, if the hydrogens are not visible, can become difficult however because some nearby side-chain atoms or water molecules can act as either donors or acceptors. Since the nitrogen can donate two H-bonds and the oxygen accept two H-bonds, it is sometimes possible to use the information on whether they form one or two hydrogen bonds to differentiate between the alternative conformations. 3.4.2 The Algorithm The study used the list of hydrogen bonds, including those that could only occur if the Asn, Gln and His side-chains were assumed to be in the alternative conformations. This algorithm works on the assumptions that (i) if an atom is accessible to solvent, however slightly, it can form a hydrogen bond to solvent and (ii) hydrogen bonds that are visible in X-ray structures are generally more energetically favourable than those implied by accessibility to solvent. Assumption (ii) is justified because if any atom appears in the electron density its location is well defined, and it is therefore tightly bound. If the H-bonded water molecule is not visible then by implication the binding site is not as well defined and the H-bonds are weaker. It is generally accepted that, of atoms which can donate more than one or accept more than one hydrogen bond, the additional hydrogen bonds are not as energetically favourable as the first hydrogen bond. Since nearly as many Asn and Gln side-chain donors and acceptors form two visible hydrogen bonds as form one, this implies a significant but lesser energetic gain. Therefore, when analysing Asn and Gln side-chains, whether atoms formed one hydrogen bond rather than two, was used as a "tie-breaker" in cases where the two alternative conformations had the same numbers of both buried unsatisfied atoms and of atoms satisfied by implied H-bonds to solvent. Both hydrogen bonding atoms were examined for both conformations of each Asn, Gln and His side-chain, and classed as either satisfied by a visible hydrogen bond ("satisfied"), satisfied by an "implied" hydrogen bond to solvent ("implied"), or unsatisfied by either visible or implied hydrogen bonds ("unsatisfied"). In His residues, the H-bonds formed by (i) ND1 and NE2 and (ii) CD2 and CE1 were examined. We would expect the atoms labelled as nitrogen to be involved in H-bonds rather than the carbons. Occasionally we found that both nitrogens accepted H-bonds, and neither donated. In principle, since it is not possible for both nitrogens to accept H-bonds, only one of the atoms is counted as satisfied in this situation. In the case of Asn (Gln) residues this means examining the OD1(OE1) and ND2(NE2) twice - once including H-bonds donated by the ND2(NE2) and accepted by the OD1(OE1), and once vice-versa. The degree of hydrogen bond satisfaction of either conformation of any Asn, Gln or His side- chain is described by giving a pair of classifications, one for each atom. For instance "unsatisfied and satisfied", or "implied and implied". The degrees of satisfaction can be compared between the PDB and the alternative conformation. The side-chain is classed as follows: Highly Optimal, if there is an "unsatisfied" atom in the alternative conformation but not in the PDB conformation, or if there are two "unsatisfied" atoms in the alternative conformation but only one in the PDB conformation. Slightly Optimal, if the hydrogen bonding potential is more highly satisfied in the PDB orientation than in the alternative, but the hydrogen bonding patterns does not qualify as "Highly Optimal". For instance, if the PDB conformation is "Satisfied and Satisfied" but the alternative conformation is "Implied and Satisfied". Indifferent, if the PDB and the alternative conformation are equally favourable or unfavourable. Slightly Suspect, if the alternative conformation is more favourable than the PDB conformation, but the number of buried unsatisfied atoms is the same for both conformers (i.e. the converse of "slightly optimal"). Highly Suspect, if the number of buried unsatisfied atoms is lower for the alternative conformation (i.e. the converse of "highly optimal"). 3.4.3 Accessibility and Implied Hydrogen Bonds A buried atom is defined as one having a zero solvent accessibility according to an implementation of the Lee and Richards (1971) algorithm calculated by the program ACCESS (Hubbard, 1992, 1994) with a probe size of 1.4A. Any hydrogen bond donor or acceptor with non-zero accessibility to solvent and no visible hydrogen bonds, is regarded as forming an implied hydrogen bond with solvent.