- Email: [email protected]
Molecular interactions and structural complementarity between DNA and model histones
on their mutual orientation and alignment in neighbouring helix turns as a function of superhelix radius and pitch and, using interactive molecular graphics, it becomes possible to investigate the effect of basepair unfolding on neighbouring DNA (or RNA) strands. The highly regular and repetitive structures generated neglect any local structural perturbations that are bound to occur in any real biological environment. They are nevertheless of interest as standard, initial constructs for the application of energy minimization procedures and dynamic studies as soon as more powerful computers become available for this formidable task. Large standard protein models have also been constructed using available protein dictionaries. Long segments of a-helices and P-sheets have been combined with DNA segments into initial files for refinement by interactive programs like FRODO and MIDAS and for the study of structural complementarity and alignment. Photos showing a series of particularly striking structural correlations between a class of basic proteins, the histones, and superhelical B-form DNA will be presented. These structures are the direct consequence of geometric calculations and simulations4. Only minor refinements involving the detailed orientations of the histone sidegroups were necessary, in order to simulate the extraordinarily specific complementarity. References 1 Sussman, J L and Trifonov, E N Proc. Natl. Acad. Sci. Vol 75 (1978) p 103 2 Levitt, M Proc. Natl. Acad. Sci. Vol 75 (1978) p 640 3 Arnott, S and Hukins, D W L J. Mol. Biol. Vol 81
(1973) p 93 4 Ohlenbusch, H H Specul. Sci. Tech. Vo14 (198 1) p 359
13
easily simulated from the known coordinates of the nucleotides, and of 8 core histone molecules, which contain about 55% a-helical regions, peptide models of which are also easily simulated by presently available molecular graphics systems. Simultaneous display of DNA fragments and a-helical histone segments show striking structural correlations. The arguments which led to the histone-histone alignment have been described before. They are important in that they fix the location and regions of the histones, free to interact with the DNA. Three particularly compelling histone-DNA binding regions may serve as examples and will be presented as 3D displays together with the overall disposition of these three regions in conjunction with two parallel fragments of core DNA on either side of the chromatin subunit and the intervening homodimer interactions between the centres of the two H4 molecules. Not counting interactions between H4 and its core histone neighbours, this structure is held together by 8 interhistone salt bridges, at least 16 histone-DNA salt bridges in addition to 16 interhistone hydrogen bonds and at least 24 histone-DNA hydrogen bonds. There are about 10 residues situated favourably for hydrophobic interactions with internal regions of the DNA grooves and 7 additional hydrogen bonds between Ser-1, Asn-24, Thr-30, Sln-93 and Thr-96 and particular base pairs are indicated. Since the structural stability in aqueous solution is the result of the relative binding strength between the water molecules and macromolecular sites, it is difficult to suggest definite values for the binding energy of such hypothetical structures. The structural complementarity, however, is a reality, which has to be reckoned with and which is certainly not fortuitous for molecules which have remained evolutionally stable for billions of years.
14
Molecular interactions and structural complementarity between DNA and model histones Computational and graphical approaches to the design and enzyme inhibitors
H H Ohlenbusch
Institut de Physique Biologique, France
F-67085 Strasbourg,
Model studies of superhelical DNA and a-helical regions of histones suggest numerous, DNA-specific binding sites which may be of interest to other studies of nucleic acid-protein interactions. Interactive molecular graphics allows the detailed investigation of specific conformational states of the residues involved in these intermolecular interactions and permits an appreciation of the structural complementarity responsible for each of the binding sites. Comparison of such models with known experimental data and crystal structures may then encourage new studies aimed at interpreting and perhaps predicting 3D configurations and dynamics of biological molecules in vivo. Although ab initio predictions of protein structures are beyond our present capabilities, some favourable systems may exist, which contain a sufficient number of structural constraints to encourage attempts to study their structural complementarity. One such system seems to be that of the chromatin subunit, being composed of 142-144 basepairs of double-helical B-DNA, whose structure is relatively Volume 4 Number 3 September 1986
B L Bush and T A Halgren
Merck Sharp & Dohme Research Laboratories, NJ 07065, USA
Rahway,
When an atomic structure of an enzyme active site is known from crystallography or has been generated by modelling from related enzymes, it may seem relatively straightforward to evaluate proposed inhibitor designs and binding modes by computing and optimizing empirical energies. We rarely make choices based directly on calculated binding affinities, however, because these are subject to large physical uncertainties. Energy optimizations do allow us to compare inhibitor geometries and to assess enzyme-inhibitor complementarity, using 3D graphics to analyse the calculated values. The ‘mapping’ of active sites, using interaction probes corresponding to hydrophobic, polar, or charged ligand groups, shows vividly how an inhibitor design might be modified. Equally important, mapping and other graphical approaches help us to see what the energy calculations are doing, to estimate the omitted effects, and to improve the physical basis of our calculations.
183
(1973) p 93 4 Ohlenbusch, H H Specul. Sci. Tech. Vo14 (198 1) p 359
13
easily simulated from the known coordinates of the nucleotides, and of 8 core histone molecules, which contain about 55% a-helical regions, peptide models of which are also easily simulated by presently available molecular graphics systems. Simultaneous display of DNA fragments and a-helical histone segments show striking structural correlations. The arguments which led to the histone-histone alignment have been described before. They are important in that they fix the location and regions of the histones, free to interact with the DNA. Three particularly compelling histone-DNA binding regions may serve as examples and will be presented as 3D displays together with the overall disposition of these three regions in conjunction with two parallel fragments of core DNA on either side of the chromatin subunit and the intervening homodimer interactions between the centres of the two H4 molecules. Not counting interactions between H4 and its core histone neighbours, this structure is held together by 8 interhistone salt bridges, at least 16 histone-DNA salt bridges in addition to 16 interhistone hydrogen bonds and at least 24 histone-DNA hydrogen bonds. There are about 10 residues situated favourably for hydrophobic interactions with internal regions of the DNA grooves and 7 additional hydrogen bonds between Ser-1, Asn-24, Thr-30, Sln-93 and Thr-96 and particular base pairs are indicated. Since the structural stability in aqueous solution is the result of the relative binding strength between the water molecules and macromolecular sites, it is difficult to suggest definite values for the binding energy of such hypothetical structures. The structural complementarity, however, is a reality, which has to be reckoned with and which is certainly not fortuitous for molecules which have remained evolutionally stable for billions of years.
14
Molecular interactions and structural complementarity between DNA and model histones Computational and graphical approaches to the design and enzyme inhibitors
H H Ohlenbusch
Institut de Physique Biologique, France
F-67085 Strasbourg,
Model studies of superhelical DNA and a-helical regions of histones suggest numerous, DNA-specific binding sites which may be of interest to other studies of nucleic acid-protein interactions. Interactive molecular graphics allows the detailed investigation of specific conformational states of the residues involved in these intermolecular interactions and permits an appreciation of the structural complementarity responsible for each of the binding sites. Comparison of such models with known experimental data and crystal structures may then encourage new studies aimed at interpreting and perhaps predicting 3D configurations and dynamics of biological molecules in vivo. Although ab initio predictions of protein structures are beyond our present capabilities, some favourable systems may exist, which contain a sufficient number of structural constraints to encourage attempts to study their structural complementarity. One such system seems to be that of the chromatin subunit, being composed of 142-144 basepairs of double-helical B-DNA, whose structure is relatively Volume 4 Number 3 September 1986
B L Bush and T A Halgren
Merck Sharp & Dohme Research Laboratories, NJ 07065, USA
Rahway,
When an atomic structure of an enzyme active site is known from crystallography or has been generated by modelling from related enzymes, it may seem relatively straightforward to evaluate proposed inhibitor designs and binding modes by computing and optimizing empirical energies. We rarely make choices based directly on calculated binding affinities, however, because these are subject to large physical uncertainties. Energy optimizations do allow us to compare inhibitor geometries and to assess enzyme-inhibitor complementarity, using 3D graphics to analyse the calculated values. The ‘mapping’ of active sites, using interaction probes corresponding to hydrophobic, polar, or charged ligand groups, shows vividly how an inhibitor design might be modified. Equally important, mapping and other graphical approaches help us to see what the energy calculations are doing, to estimate the omitted effects, and to improve the physical basis of our calculations.
183