Describe Up To Now How People Combine Both Approaches Of Theoretical And Experimental Techniques To Determine 3D Structures Of Complexes.
Ali, I – 100150403
Introduction
Glycosylation is a post-translational modification in which a sugar group is added to a protein. It's utilised by cells to increase the diversity, stability and intracellular traffic of proteins1. Experimental and theoretical approaches are used to determine the 3D structure of protein-ligand complexes. The techniques are often used together to maximise the validity of results2.
Currently, determination of the structure of protein-ligand complexes are based on theoretical approaches which validate the results from experimental approaches, such as X-Ray Crystallisation and NMR Spectroscopy3,4. The issue with using traditional NMR techniques is that they require Nuclear Overhauser Effect (NOE) which require a lot of time and resources to measure5 Because of this, theoretical approaches are being investigated as they could improve the efficiency of elucidating 3D structures of protein-ligand complex6.
The objective of this literature review is to look into the combination of experimental and theoretical approaches in creating of 3D images of protein-ligand complexes.
Experimental Approaches to Determine 3D Structures
STD NMR is based on NOE and is a 1H7 NMR technique in which protein protons are energised in full saturation and then ligands are introduced to be saturated by the magnetised protein with a greater binding ligand seeing more of the energy
efficiently transferred, this is reflected in the intensity of signals7 The values are compared against an off-resonance experiment in which no irradiation occurs2 and this is how STD NMR produces its results. The results produced are shown as intensity peaks and relate to the efficiency of energy transfer to protons in which percentage saturation is yielded, in which greater outcome percentage indicates the approximation of the ligand proton in respect to the protein. Equation 1 is used to calculate this8.
Equation 1.
STD (tsat) = STDmax (1 − e −ksat·tsat).
Experimental approaches are used to show interactions between proteins and ligand binding. This can be used to discover new drugs or to develop drugs which have greater affinity to their receptors and so increased efficacy9. This is because ageing populations and increasing antimicrobial resistance means there's a greater demand for new medicines. Production of new medicines is becoming more expensive and so there's a need not to pursuing false positives or positive negatives which can be avoided using multiple techniques to validate results.
Figure 1. Transfer of Energy in STD NMR3
The NOE is an underpinning concept of NMR spectroscopy10 and is defined as the relative changes in intensity of one resonance spin as it comes near the spin of another in which an increase or a decrease of intensity can occur and this allows for 3D structures of protein-ligand complex to be determined10. There is a comparison between the off-resonance, which sees the same conditions as the on-resonance but the frequency is steered away from any resonance in the structure. In the on-resonance experiment very similar spectrums are created but the difference between the experiments in the resonance intensities and these are compared against each other10 . Equation 2 is used to calculate this10.
Equation 2
ISTD= Io – ISat.
The difference describes the affinity of ligand binding to protein11. The technique yields epitope maps to give structural images of a protein-ligand complex. Such use is seen in the binding affinity of human albumin to varying proteins such as 6-CH3-Trp and 7-CH3-Trp10. STD NMR showed intense STD peak intensity for 6-CH3-Trp, indicating great affinity of ligand to protein to form a complex whilst 7-CH3-Trp showed no STD peak intensity showing how the ligand didn't bind with the protein3.
Figure 2. Off and On Resonance10
The limitation of STD NMR is that whilst it produces 3D mapping structures of the protein-ligand complex it doesn't give information regarding the amino acids around the ligand18. This is achieved by DEEP STD NMR. DEEP STD NMR utilises the difference in energy saturation of protein protons when they come into contact with amino acids when running the same experiment but using D2O and H2O as solvents12 and results produced run through calculations to find the differential epitome map of the complex. This is shown in equation 312.
Equation 3
∆STDi = (STDexp1,i)/(STDexp2,i) – 1/n∑ni (STDexp1,i)/(STDexp2,i).
The technique is relatively new and as such there's a lack of viable articles. In a study12 3NPG and CTB(at the end of the paper have a acronym section in which we write the full name of these) were bound together but the complex couldn't yield a differential epitome and this is because the change of solvent from D2O to H2O didn't change the exchange of protons slow enough and therefore gave a low ∆STD value12. DEEP STD NMR is a useful technique to establish the surrounding amino acids of the complex. The properties of DEEP STD NMR can be improved if an STD NMR already exists for a complex in which orientation of 3D structure can also be determined12.
Another NMR technique is NMR2 which yields information regarding the binding site of the protein-ligand complex at high resolution13. NMR2 is a new technique and requires already known structures of the protein-ligand complex to derive its structural determinations and the use of experimentally derived NOE values of the complex at hand. The experimental approach gives fast determinations because it does not use a force field aspect, rather its simplified approach utilises protein labelling using 13C and 15N13. The technique is in its early stages and therefore only 8 experiments have utilised the technique as of this article13.
Whilst the use of NMR techniques to determine protein-ligand complexes are becoming more prevalent14, the techniques have a reoccurring disadvantage to them, namely the low resolving power of NMR means data from other experimental approaches are required like X-ray crystallography and the increasing cost of the techniques14.
Theoretical Approaches to Determine 3D Structures
Experimental approaches aren't the only techniques used to determine the relative 3D structures of proteins and their ligand complexes. Theoretical approaches are used because they don't rely on the slow
collection of NOE distances used in experimental NMR techniques and so are used as a faster means of determining structures5. The technique sees a protein suspended in space and a ligand's relation to space altered to form a complex and the fit scored to determine how good the fit between protein and ligand is5. But because of the large possibilities of a fit the calculations are strenuous and heavy to be dealt with and therefore a single solution is rarely generated5. Theoretical approaches can be molecular modelling based or molecular dynamics based. However, the results yielded from such calculations are of high benefit for drug discovery and development and as such there are varying theoretical techniques developed to give molecular docking calculations2,5,15.
One such theoretical approach is Glide. Glide combines the use of HTS and combinatorial chemistry to look at a molecules ability to form protein-ligand complexes to yield a biological activity2. Glide utilises the molecules' X-ray crystallography to determine the 3D structure of the complex which is used to calculate a score for a ligand against its protein based on favourable bonding interactions and desolvation properties of the complex formed, but devalued based on steric clashes as this would signify that the atoms of the ligand at a region don't conform to the required structure in space to be bound as well16,17. It's been found that the utilisation of molecular dynamics in determining 3D structures of complexes can give very accurate determinations and can help to discriminate well binding ligands to those which aren't17. The calculations from Glide are then used in QSAR models to boost improved predictions17,18.
The shortcomings of Glide, and
other theoretical approaches, is the required data handling capabilities to deal with thousands possible conformations and comparing this to vast libraries of experimentally determined 3D structures of complexes. However, improvements in computer RAMs have meant that the issue of initiating calculations are getting less tiresome and instead more calculations can take part which can give better predictions and higher resolution 3D structures with some calculations running on 8-core desktop computers13,14,17.
Figure 3. Distribution of pKa and Glide Scores17
HADDOCK also uses combinational chemistry, biochemical and biophysical bond interactions based on already know and experimentally determined data to produce AIRs19 and in effect is based in Monte Carlo schemes. The demerit of such a technique in drug discovery is that target specific data in most cases aren't available to such scientists and so other means of experimental techniques must be used instead to determine 3D structures of complexes20. HADDOCK also experiences ambiguous results if there aren't enough data to determine the surfaces of the protein and ligand and as such interfaces of both must be known to reduce this, but the technique also uses free-moving side chains to allow for possible better conformational symmetry between the protein and ligand, which is starkly different from earlier approaches which were rigid in nature5. The technique, based on its merit of being flexible with the special configurations of side groups, is a worthwhile theoretical approach to be used in looking at the sugary parts of proteins because of steric differences occur naturally in an organic system17,21.
Another theoretical approach is AutoDock which utilises, as previous techniques of the theoretical approach have, experimental data of already known docking areas of proteins from protein databanks without the presence of ligand, water or other buffers as these will have an impact on the binding22. The area of binding will require polar hydrogen atoms and every atom seeing Gasteiger charges to create a map of the binding area. The data produced will be bond distance which will be interpreted as affinity of ligand to protein in the complex to indicate how well the bonding interactions are in the complex22,23. This is then compared against experimentally derived results of 3D structure and the similarities looked at. Aforementioned points about the AutoDock technique means that it's used in fragment growth during drug development and because AutoDock user friendly interface the technique is used quite often in pharmaceuticals22.
Combined Use of Experimental and Theoretical Approaches
Aforementioned points regarding both theoretical and experimental approaches are valid but the concurrent use of them both together gives a greater depth of structural determination than they do individually, with this being outlined in many articles dealing with 3D structural determination of protein-ligand complexes3,12,24,25. Theoretical techniques are based on experimentally generated data, whether from X-ray crystallography or NMR, including experimentally yielded NOE values and thermodynamic parameters5,26. Both approaches are used together with the experimental experiments undertaken used to find possible molecules required for the initial discovery of drugs10,26 with the effectiveness of using NMR being seen in by the amount of drugs discovered like this being in clinical trials10,26. Computational approaches are used in the development stage to increase the potency of the drug27 to improve the overall biological activity of the drug.
Glide and STD NMR are used in conjunction with each other quite often in the pursuit of determining 3D structures of complexes2,5,25,28. Results from STD NMR are inputted into the Glide program which takes into consideration parameters like pendent groups for possible bonding interactions and the conformation of glyosidic bonds to give a score5,17 and the better scoring molecule is then furthered onto possible molecular dynamic simulations to look at possible ways to improve the efficacy and affinity of drug5. The reason why Glide is more often used rather than HADDOCK is due to their docking types. Glide is a flexible ligand docking system whilst HADDOCK a rigid ligand docking system and because of this their relative accuracies reflect this with Glide seeing
90% accuracy and HADDOCK around 70%29. Because of the large computational power18 required by theoretical approaches and the varying additional computer programs required by HADOOCK3,12,24,25 means that differences in accuracy are enough to mean Glide is used more often than HADDOCK.
However, the legitimacy of using either theoretical techniques are still upheld because of the large number of 3D structures analysed in such way, the number of biomolecules in clinical trials and the amount of these molecules commercially available17,26. The output of both qualitative and quantitative data from theoretical and experimental approaches allows for high resolution 3D structural determinations to be produced, meaning greater information about molecule interactions can be found and documented which can be used at a later stage to produce complexes with better binding affinities.
Figure 4. An oligosaccharide bound to a protein30.