Biophysical Techniques Used In Fragment Based Drug Design Biology Essay
It has already been noted that the main biophysical techniques currently employed the fragment-based drug design approach are NMR, X-ray crystallography, SPR, MS, fluorescence spectroscopy as well as calorimetry. The first work done in which X-ray crystallography was used in the identification of fragment hits was reported by Ringe et al in 1996 when they used a method known as multiple solvent crystal structures (MSCS). The MSCS approach involves elucidating the 3D structure of the target protein in a variety of solvents, with the solvent molecules acting as probes of complementary binding sites on the target molecule, thereby allowing the identification of favorable fragment-binding positions, the orientation of bound fragments and characterization of fragment interactions within these sites (16,19 ).
However, it was Philip J. Hajduk et al of Abbott Laboratories who first published the pioneering work of using a biophysical technique, namely NMR to detect structural-activity relationships (SAR) in fragment and target binding experiments. In this approach, small molecules with nM affinities for the FK506 binding protein (FKBP12); a 12kDa proline isomerase part of the calcium release channel complex, were developed by the optimization of two fragment compounds in the mM affinity binding range (20,21).
Other techniques have also been experimented with in fragment screening although 3D structure elucidation still utilizes NMR and X-ray crystallography due to the high level of detail that can be obtained with these two techniques. As a result, other techniques that are used in FBDD are merely complementary to NMR and X-ray crystallography but what is required of any biophysical technique used is that it must be highly sensitive enough to detect the low affinity hit compounds in FBDD (6).
Each of the biophysical techniques and its use in FBDD is described in more detail below.
3.1 Nuclear Magnetic Resonance (NMR)
Nuclear magnetic resonance spectroscopy measures the unique chemical shifts of nuclei in proteins as resonance peaks and can be one dimensional (1D) or multi-dimensional; i.e. 2D, 3D or more, owing to the large size of most proteins which cause an overlap in the resonance peaks. NMR experiments can further be classified as those that measure through-bond frequencies or those that measure the through-space frequencies. Through-bond analyses provide information that is used to assign the unique chemical shifts to a specific nucleus, while through-space analyses provide information essential for assigning distance restraints in structure calculation, and also in assignment with unlabelled proteins.
Isotopically labeled proteins are typically analyzed using two dimensional (2D) Heteronuclear Single Quantum Correlation (HSQC) spectrum which theoretically provides resonance peaks of hydrogen (H) bound to any heteronucleus; i.e. any nucleus other than 1H. 15N HSQC shows peaks for all amine groups, and side chain N nuclei, essentially showing the backbone structure of the protein, and how many residues are present. For this reason it is known as the fingerprint of a protein.
Unlabelled proteins are analyzed using 2D homonuclear NMR experiments through COSY, TOCSY and NOESY. COSY and TOCSY provide information on through-bond resonance frequencies, and together these 2D experiments are used to build spin systems of all the protons in a protein sequence. NOESY experiments provide information on through-space resonance frequencies, which are used to connect the different spin systems in a sequential order. NOESY shows cross-peaks for nuclei less than 5A apart. Homonuclear NMR is restricted to small proteins due to the problem of overlaps. Dynamics in a protein structure can also be studied using NMR, by studying relaxation times which in turn provide information on correlation times and chemical exchange rates.
The success of NMR in FBDD has been due its wide availability of measurable parameters that can be used to detect the association between ligand and target protein. These can be chemical shift perturbations in 1H, 15N or 13C isotopically labeled proteins, chemical shift anisotropy using 19F, longitudinal and transverse relaxation times (T1 and T2), cross-relaxation in the fragment-protein complex; measured using transferred NOE (tNOE-NMR) or saturation transfer difference (STD-NMR), and finally relaxation between the fragment and protein-bound water molecules; measured as Water-LOGSY. Additionally, small ligands are sensitive to interactions with larger molecules and changes in the line widths, relaxation times and NOE values are used to characterize and quantitate the binding mechanism of a ligand to the target protein (23,25,26).
There are two ways that NMR can be used in FBDD, by either observing changes in the target protein structure or changes in the ligand.
Protein-observed NMR is by 2D analysis of 1H/15N resonance assignments of amide groups, and provides principle information on the interactions between ligand and labeled protein. It requires expensive equipment, the protein target is usually of molecular size less than 30kDa and has high protein requirements. Ligand-observed NMR can distinguish between active and non-active site binding ligands using 1H or 19F spectra analysis, requiring expensive equipment, but is well suited for use in screening mixtures. Target protein molecular sizes are less than 20kDa, with moderate protein requirements (6).
Observing changes in the ligand is favored because it requires less time on instruments, does not need isotopically labeled proteins and can work for any size of target molecule. However, one disadvantage of the latter method is that it is required that the ligands have fast off-rates which can reduce on the potency of the molecule and screening candidate compounds (26).
In the SAR by NMR approach developed by Hajduk et al, changes observed in the NMR spectra of the protein were used to confirm ligand binding, in the process also providing useful information of where the binding site is located on the protein surface and the ligands binding mode. 2D spectroscopy was used to detect two fragment leads that were able to bind to proximal sites on the protein target and then 3D structural information on the bound ligands was used to join the fragments to produce a high affinity ligand, i.e. using the linked-fragment approach (26). SAR by NMR senses local chemical shift perturbations (CSPs) when a ligand binds the target protein. Due to the low affinity binding of fragment molecules, high concentrations of both protein and fragments were used in this experiment (20,22,24). In fluorine detected NMR experiments do not require high concentrations of either protein target or ligand compounds. In these 19F experiments, the chemical shift resonances of the ligand are observed in absence and presence of target protein thereby allowing the detection of very weakly binding affinity molecules (15).
Following the initial success of SAR by NMR, the SHAPES approach was developed by Fejzo et al (23) using NMR to detect binding between ligand and protein, using a limited but diverse library of 132 small compounds that could be applied to larger molecular weight target molecules. The SHAPES approach differs from SAR by NMR in its use of scaffold compounds derived from known therapeutic agents, and it does not require isotopic labeling. The objective of the team was to design a fragment library that optimized a large number of factors in terms of cost, synthetic accessibility, solubility, separation of NMR peaks and diversity. These scaffold compounds in the SHAPES fragment library were composed of rings, linkers and side chains broken down from ‘drug-like’ therapeutic compounds. The final library was composed of small molecules ranging in molecular weight 68-341Da, with 6-22 heavy atoms and a calculated logP of –2.2 to 5.5. The candidate fragments were also required to be able to yield a simple, well-resolved 1H NMR spectrum and at possess at least two protons within 5A of each other for NMR-based fragment screening.
Screening of fragment libraries can be either by single compounds, or mixtures of compounds which can result in challenges such as competitive binding which may in turn result in potential hits being missed as well as spectral overlap in 1D NMR. Appropriate selection of mixtures can help to reduce on these challenges in such as in SHAPES by NMR (23). Additionally, the need to use high concentrations of weakly binding small molecules in FBDD, can affect sensitivity in NMR-based studies. This is overcome by using cryogenically cooled probes, miniaturized receiver coils and samples, or enhancing NMR relaxation effects using paramagnetic ions (26).
In an NMR-based technique known as Rapid Analysis and Multiplexing of Experimentally Discriminated Uniquely Labeled Proteins using NMR (RAMPED-UP NMR), multiplexing of both the ligand and target protein was accomplished by labeling them with one amino acid type. The resulting spectra were reportedly simple, easy to interpret and allowed several proteins to be labeled at once and in the same tube (27). This unique advantage can this be employed to screen related proteins against the same fragment (4). Another unique approach in NMR-based FBDD is target immobilized NMR screening (TINS) which reduces on the amount of protein required for screening. In TINS, binding is detected by the comparing between the 1D NMR spectra of compound mixtures in the presence of a target immobilized on a solid support and that of a control sample. TINS has the advantage of being utilized on membrane samples which is not possible with any other biophysical technique (4,28).
3.2 X- Ray Crystallography
X-ray crystallography has remained the method of choice for determining the 3D structures of proteins. In this biophysical technique, protein crystals of the structure to be elucidated are bombarded with X-rays, and then the diffraction pattern obtained is used to determine the amplitudes of the X-ray waves after they are diffracted by the crystals. X-rays are excellent for 3D structure determination because their wavelengths are in the same range as bond lengths in proteins 1.5A. X-rays are generated using tube generators or synchrotrons, and are focused on a mounted protein crystal. The diffraction pattern is collected on an image plate and the intensities of the diffraction spots calculated. It is the intensities of the diffraction spots that provide information on the amplitudes of the X-ray electromagnetic waves but their phases are determined using other methods such as anomalous scattering, molecular replacement or isomorphous replacement.
In the early days of X-ray crystallography, homologues of protein targets were used as comparative models of protein targets to try and develop inhibitors (29) until the potential of applying X-ray crystallography in designing drugs was realized. HIV protease inhibitors were developed using this method after determining its 3D structure as was the neuraminidase inhibitor Relenza (30). Other examples of drugs that have been developed from X-ray crystallography include anti-cancer kinase inhibitors such as Gleevac, Tarceva and the anti-coagulant Exanta (31).
X-ray crystallography in FBDD has recently developed due to the high-throughput methods that are being applied in the crystallization of proteins and solving multiple protein-ligand complexes. These new methods allow up to thousands of small molecules to be screened and thus identify fragments that can bind to target protein molecules, as well as to define the binding site. For the first time ever, automation and miniaturization has made 3D structure elucidation by X-ray crystallography much faster, reliable and cheaper (22,32).
According to Congreve et al, fragment screening by X-ray crystallography is made possible because protein crystals have extensive solvent-filled channels accounting for up to 50% of the crystal volume, so small fragment molecules can easily diffuse into the crystals to interact with their respective binding sites as if they were in solution themselves, unless the binding site is occluded by the crystal packing. Libraries of up to 1000 small molecule fragments are used in X-ray crystallography-based screening methods either as single compounds or mixtures of cocktails. These fragments are then soaked or co-crystallized with the target protein molecule and the structures of the resulting crystal complexes solved. The number of fragments in a cocktail mixture usually varies between 2-8 depending on the protein target, the likely affinity of the fragment compounds as well as the anticipated hit rate. Difference Fourier Transform (FT) techniques are used to visualize the bound small molecules by collecting data on the soaked crystals. Automation methods are used to rapidly interpret and analyze data, and in mixtures of fragments with target protein, the different fragment molecules can be fitted to the difference electron density and ranked. The 3D structure of the ligand-protein target obtained is what provides the necessary structural information to develop hits to leads by optimization in FBDD (26,32-34).
Prior to the development of automation for X-ray crystallography in FBDD, the important step of fitting the ligand structure into the electron density structure was manually performed, a method which had challenges in being slow and biased by the crystallographer. Astex Therapeutics; who discovered the cancer drug CDK inhibitor AT7519 which is currently in clinical trials using X-ray crystallography-based techniques (33), developed a platform known as Auto Solve, based on a protein –ligand docking program known as GOLD which addresses several challenges including ligand fitting by using a genetic-algorithm-based search engine to sample the conformations and orientations of the ligands, and scoring the fit depending on the agreement with the electron density. The AutoSolve platform integrates the ligand fitting step with data processing, molecular replacement, water placement and structures refinement (26,34).
3.3 Mass Spectrometry (MS)
MS is a technique which uses the differential deflection of proteins in their ionic form in a magnetic field to determine their molecular mass based on the mass/charge ratios of proteins.
Biomolecules may be ionised by one of several methods currently in use today. The method of choice is electrospray ionisation but the challenge with it is that it is limited to small sample components, so larger macromolecules require other more efficient ways to be ionised. Some ionisation modes used in mass spectrometry today include;
Fast Atom Bombardment
Plasma Desorption Ionisation
Matrix-Assisted Laser Desorption Ionisation (MALDI)
Ion Spray Ionisation
The use of MS in fragment-based drug design focuses on the identification of fragments by measuring the amount of unbound ligands, i.e. in non-covalent interaction in the absence and presence of the protein target, or by analyzing the non-covalent target-ligand complex in gas phase. The data obtained from ESI-MS for example, is complementary to techniques such as fluorescence and NMR in FBDD, and, it is quickly and easily obtained by directly infusing both the ligand fragment and target protein. ESI-MS is measures the molecular mass of a protein-ligand complex, and is thus able to identify two closely related ligands that bind a common site on the target protein (35,36).
Alix Pharmaceuticals (www.alix-pharma.com) report successful implementation of ESI-MS-based fragment screening using their innovative Fragment Analysis by MASs Spectrometry (FAMASS) technology. The fragment library is screened by in-silico docking followed validation of virtual hits by non-denaturing MS. The validated hits are then co-crystallized with target protein and their binding mode determined by crystallography. In this platform, ESI-MS is thus used as a complementary technique to X-ray crystallography to characterize the binding mode of the ligand to protein target site, provide useful information on the binding stoichiometry, and specificity of the binding to the active site.
Other recent advances in the application of MS in FBDD include, but are not limited to Fourier Transform (FT)-ion cyclotron resonance (ICR) mass spectrometry and a combination of Frontal Affinity Chromatography with MS (FAC-MS).
FT-ICR MS is able to provide ultra-high mass resolution and accurate and non-destructive detection with high sensitivity by trapping ions in a FT-ICR cell, detecting their cyclotron frequency and finally determining the desired mass/charge ratio of the ions. Together with ESI, this method has been successfully implemented in target identification and validation, as well as in lead identification and optimization (37).
FAC-MS is based on the continuous infusion of ligands onto immobilized target protein on a solid support column and the eluting ligands are then detected by MS. The technique differs from the traditional ‘capture and release’ affinity chromatography and instead based on on-going equilibrium between the ligands in mobile phase flowing through the immobilized target protein in the stationary phase in the column. FAC-MS can be used to measure the relative binding strengths of ligands, thereby allowing rapid ranking and ligand identification. It can screen cocktails of ligands and target, and provides data from which the dissociation constants of the resulting complexes can be determined (38). FAC-MS can be automated to increase throughput and because of its sensitivity, requires reduced amounts of target protein when compared to other techniques (35).
Calorimetry is a technique that measures one of the most important physical quantities in biological systems; heat, and the instrument used to achieve this is known as a calorimeter. This is experimentally achieved by maintaining a biochemical reaction at a single temperature and then measuring the amount of heat added or removed to maintain this single temperature. The measurement of thermodynamic changes in biochemical reactions involving proteins-protein and protein-DNA interactions is known as biocalorimetry.
Biochemical reactions can be either endothermic; absorbing heat in order for reactants to collide, react and form products, or exothermic; producing heat as a by-product of the reaction between the reactants. The main thermodynamic parameters used in biocalorimetry are:
• Activation energy of reactions- the energy (Ea) required by reactants to undergo a chemical reaction and form products.
• Enthalpy- this is the energy produced or absorbed by a reaction. In endothermic reactions, enthalpy of the product H > that of reactants, while in exothermic reactions enthalpy of the product H < that of reactants.
• Entropy- is the measure of randomness in a molecule or process (ΔS). This occurs naturally in ordered structures such as biomolecules which tend to adopt more random forms. Absorption of heat results in an increase of disorder and thus an increase in entropy as well.
• Free energy- is the energy available to do work, it is also known as Gibbs free energy (G). Spontaneous reactions result in a decrease in G.
Binding constants are related to the free energy of interactions, which are dependent on the association between enthalpy and entropy in a biochemical reaction, and are influenced by the nature of the solvent (39).
This can be represented as below, in a modification of the Gibb’s equation:
-RT lnKD = ΔG = ΔH -TΔS
An emerging method in calorimetry is isothermal titration calorimetry which is one of two calometric methods used to measure enthalpy in chemical reactions and processes, the other being differential scanning calorimetry (DSC). ITC is a highly accurate method used to directly measure the change in heat in processes such as ligand binding and complex formation while DSC is used for studying heat-initiated phase changes in biopolymers and gives enthalpy and other thermodynamic measurements (40).
In protein binding experiments, ITC essentially measures the heat change resulting from the addition of a ligand to a protein:
P + L <=> PL
and allows the estimation of both the binding constant (Ki) and dissociation constant (KD) which is related to ΔG by the following equation:
ΔG = -RT lnKD
ΔH for a binding reaction decreases as the system reaches equilibrium, thus it is a useful and sensitive probe for measuring the extent of the binding reaction. The binding energy of the ligand per atom or ligand efficiency (ΔG) can be calculated by converting the dissociation constant into free energy of binding at 300K and then dividing by the number of heavy atoms in the molecule (40,41). Ligand efficiency can be used to assess the quality of hits during fragment screening and also monitor the quality of leads during the optimization stage (42).
A survey of the Structure/Calorimetry of Reported Protein Interactions Online (SCORPIO) database by Tjelvar et al (40) revealed that the majority of interactions are actually enthalpically driven and that synthetically designed inhibitors have greater affinity due to more favorable entropy changes on binding. ITC experiments make determination of enthalpic and entropic measurements easier, compared to other techniques by measuring the incremental heats of reaction as one component from the non-linear fitting of the titration curve obtained. Variations of Gibb’s free energy equations (as shown below) are then used to accurately determine binding energy of protein-ligand interactions (40).
Binding energy per atom (ligand efficiency):
Δg =ΔG/N non-hydrogen atoms
From these measurements, it is determined that the logarithmic relationship between free energy of binding and dissociation constant potency implies that for every change in G of -1.4kcal/mol, there is a 10-fold increase in potency of the ligand molecule. Thermodynamic measurements by ITC are therefore able to help strike a balance between the ligand efficiency of a potential lead and its final molecular weight (42).
ITC has successfully been used in studies on HIV protease inhibitors, anti-malarial drugs and DNA gyrase inhibitors by generating thermodynamic inhibitors of their interactions with their respective protein targets and the development of ITC instrumentation has increased throughput using this technique (47)
3.5 Surface Plasmon Resonance (SPR)
SPR is a label-free technique that measures tha interactions between molecules by measuring the changes in refractive index between two phases separated by a sensor which is usually a thin gold film. This technique has been employed in macromolecule analysis because the refractive index measured is directly linked to the change in mass at the sensor surface which has limited sensitivity. In an experiment to increase sensitivity by researchers at Graffinity Pharmaceuticals (www.graffinity.com), fragment molecules of molecular weight less than 350Da were immobilized on the gold film surface/sensor while the heavier target protein is in solution (40).
SPR has the advantage of low protein requirements, and the traditional approach of immobilizing the target protein onto the chip surface allows studies of thousands of molecules to be screened sequentially using the same surface. Parameters measured in SPR include a change in the SPR angle when binding occurs between ligand and protein, which is measured as resonance units (RU) where 1RU is equal to a shift of 10^-4 degrees (15). In 2008, Hämäläinen et al (43) used SPR for screening a directed fragment library composed of known thrombin binders and a general five hundred compound fragment, and providing affinity ranking of fragment libraries on four different protein targets. They highlighted this technique as an alternative to NMR and X-ray crystallography for fragment screening due to its low resource demands; only 12ug of protein was needed to analyze a three thousand molecules fragment library, and also as a ‘clean-up’ technique to get rid of promiscuous fragments.
The limitations of SPR arise from the need to work at concentrations equal to the dissociation constant of the fragment (KD) as well as the decrease in dynamic range as the size of the target protein increases.
However, SPR is successful in being a unique technique with the ability to provide information on association rate and dissociation constants of the ligand-protein interaction, as well as dissociation constant of the fragment (15).
3.6 Fluorescence Spectroscopy
Light energy absorbed by molecules raising chromophores through vibrational levels in the ground state to a higher excited state. As these excited chromophores return to the ground state, the energy will be lost through a series of non-radiative transitions, or by radiative transitions in which the energy is emitted as light. This is known as fluorescence and fluorescent molecules have a characteristic emission spectrum which can be measured using a spectrofluorimeter.
Fluorescence measurements can be used in binding studies to detect conformational changes in proteins as a result of ligand binding by observing changes in the emission spectrum of a target protein. If a fluor has a unique position in the target protein, changes in the emission spectrum at a fixed wavelength can be used to determine the KD of the protein for the ligand, thereby measuring the binding affinity between the two molecules:
P + L PL
The fluorescence parameters measured are fluorescence quantum yield and observed fluorescence lifetime, which are determined by summation of the radiative and non-radiative rate constants. The fluorescence quantum yield is a measure of the quanta emitted or absorbed, while the observed fluorescence lifetime relates to the time a fluor spends in the excited state (44,45).
In FBDD, fluorescence anisotropy (FA) and fluorescence lifetime (FL) are used to measure the changes in fluorescent-labeled molecules. In FA, the emission of polarized excitation light is measured in parallel and perpendicular orientations, and the relationship between fluorescence anisotropy and molecular properties can be predicted because anisotropy is a function of molecular properties, specifically Brownian molecular rotation. FL represents an intrinsic property of the fluor and can detect minute changes in its direct environment and its measurements are performed directly by pulsed or phase-modulation techniques and, similarly to FA, possess quality control measures by discriminating the real lifetime signal associated with the biological interaction from background signals (44).
Both FA and FL are highly sensitive techniques in FBSS and are thus favoured for use in fragment screening in competition mode with a fluor-labelled reporter. FA-based studies firstly require the chemical derivation of a known high affinity binding ligand and then covalently linking it to a fluorescent dye with an appropriate excited state lifetime. Exciting a fluor with polarized light results in a polarized emission spectrum which can be depolarized by rotational diffusion.
Anisotropy (A) is calculated by:
A=(I1 – I2)/(I1+2I2)
Where I1 and I2 are the emission intensity parallel and perpendicular, respectively, to the plane of the excited light (15). Since A is interchangeable with polarity, its value depends on the fluor rotational correlation time and lifetime of the excited state therefore it can be calculated from these values and a limiting anisotropy (about 0.25 for fluorescein) which depends on the wavelength and dye used. Ultimately, these measurements are used to calculare the observed A in a fluor-free and fluor-bound state with the target receptor (44).
FA is similar to SPR in working within a limited dynamic range and fragment concentrations comparable to the KD of the fragment. FA studies should be performed in high fraction conditions of bound fluor-tagged molecule and comparable to the dissociation constant. With FA reporting a number of false positives or false negatives as a result of optical interferences, FL are added to FA studies to reduce this and identify autofluorescent molecules. FA and FL are very similar, including the need to test fragments within their KD range, but they differ in their target protein size sensitivity, and that in FL the change in fluorescence lifetime cannot be predicted following binding (15).
Overall, both these techniques are highly sensitive and with their ability to be automated, as well as compact size of assays, can lead to rapid fragment screening of up to hundreds of thousands of compounds in a matter of days (15).
Fluorescence correlation spectroscopy (FCS) is another method that is applied in FBDD. Its measurements are based on detecting the temporal fluctuations in the fluorescence signal from the diffusion of individual fluors in and out if a small tightly focused confocal element by autocorrelation. FCS provides data on the concentration of fluors, diffusion times of the components, and average light intensity per molecule (44).
Based on the principle of detecting conformational changes in fragment-protein interactions, Mikuni et al (46) successfully used FCS in a competitive binding assay to screen fragments that could weakly bind to the c-Cbl TKB domain and phosphopeptides derived from the tyrosine-protein kinases zeta-chain (TCR) associated protein kinase 70kDa (ZAP-70), adapter protein with a PH and SH2 domain (APS) and epidermal growth factor receptor (EGFR) from which six hits with an affinity for the c-Cbl site were identified. Detection was in the increased diffusion time of a labeled probe upon binding to target protein and the FCS data obtained were fitted in models of fluorescently labeled unbound and bound ligand probes. The heterogeneity of the probes resulting from the addition of compound was used to select hit fragments and analyzed using the fluorescent particle number and chi square statistical values to provide evidence of a good enough fit inot the models. The results of this experiment were confirmed using SPR to directly measure the binding amounts of the fragments from which five that bound selectively were identified. FCS therefore also has an advantage in the study of time-dependent binding of fragments to target proteins which can be applied in FBDD.
If you are the original writer of this essay and no longer wish to have the essay published on the UK Essays website then please click on the link below to request removal: