Microcalorimetry
in Drug Discovery and Development

Ultrasensitive isothermal titration calorimetry (ITC) and differential scanning calorimetry (DSC) have broad application areas in drug discovery and development processes from the validation of hits and secondary screening through to lead optimisation, thermodynamic based QSAR, early ADME studies and formulation development.
The universal ‘kdometer’ (ITC) is a bioassay system for measuring the activity of all biomolecules free and in solution. Millionths of a degree change in temperature occur when biomolecules interact. These can be detected by ultrasensitive calorimeters and used to precisely determine the kd of an interaction. The main advantages over other techniques are that there is almost no assay development because heat generation is common to practically all biomolecular processes and no labelling, immobilisation or other potentially damaging chemistry is required. It can also be used with DMSO or most other organic solvent systems.
Aliquots of a ligand from an automated titrating device are injected into a solution of target molecules contained within the calorimeter at constant temperature. When molecules bind there is either an uptake (endothermic) or release (exothermic) of heat which is then detected by the calorimeter and recorded as a heat pulse. The experiment is set up such that these heat pulses diminish in size as more ligand is added due to saturation of the binding sites on the target molecule. The affinity and the stoichiometry as well as information about the underlying energetics can be determined from a single experiment.

Macromolecular stability monitor
DSC is a universal bioassay system for measuring the stability of biological macromolecules. In a DSC experiment a biomolecule is typically heated at 0.1 to 5°C/min and the heat (DH) and the melting temperature (TM) of denaturation is monitored. It shares many properties with ITC including its ability to detect very small changes in temperature and the related advantages of no ‘tagging’ requirements and little or no assay development. TM shifts are a readily accessible probe for the stability of biomolecules and have wide applications in the drug discovery and development arenas. Small molecule interactions with protein or nucleic acid targets induce TM shifts, a phenomenon that is readily exploited to screen for drug candidates. In addition the stability of protein biopharmaceuticals can be determined and used to find stable protein constructs and excipients to increase shelf lives. These advantages coupled to automation have made technique an invaluable tool throughout the pharmaceutical industry.

Target characterisation and validation
The ability of ITC to detect binding directly has made it the method of choice for identifying potential targets and the characterisation of components of signalling pathways. The simplicity of the system i.e. that there is no requirement for reporter groups, tags or immobilisation means the elimination of false positives such as those due to interactions with non native components such as GST fusion proteins or linker groups introduced solely for assay design. This has many related advantages including the ability to include any additional activator molecules or allosteric modulators in the assay to ensure that the correct ternary complex is employed in the structure based drug design process further down the discovery pipeline.
Structure is central to the rational design of drugs but can be often be difficult to solve due poor protein stability, size or heterogeneity issues. Domain hunting techniques are employed to find active subunits which are easier to handle and stable long enough for the biochemical and biophysical characterisation required for successful drug design and FDA approval. ITC is ideal for stringent, multiprobe comparison of native and engineered or refolded protein material. Also, because it measures the activity and the stoichiometry independently, yet simultaneously in a single experiment, problems associated with batch to batch variability which can lead to costly misinterpretation of activity assays are eliminated.

Hit validation and secondary screening
The 100s of hits generated from real or virtual HTS can all be validated using DSC with no further assay development. DSC screening uses shifts in the melting temperature of a protein in the presence of a drug candidate to determine the affinity. This has the advantages of simplicity, automation and the ability to identify the biggest sources of false positives, namely weak non specific interactions and those which bind to unfolded protein. Many hits that appear to bind a target actually do so by unfolding and inactivating it making it unsuitable as a drug candidate. This mode of binding will go unnoticed by many techniques and can be responsible for costly failures in the discovery process.
ITC generates highly quantitative binding data for accurate ranking profiles and is most suited when 20s-30s of candidates require characterisation. The universal nature of heat as a probe means that all hits can be characterised without further consideration of the addition of chromophores or flourophores and the size of the ligand or target is irrelevant. Even the binding of a proton can be detected! This is a powerful tool in lead optimisation in the hands of the structural biochemists and medicinal chemists. Recently ITC generated energetic profiles have been used to develop drugs with a greater potency and an ability to respond to target variations that commonly infer drug resistance in viral populations.

Lead optimisation
ITC uniquely combines information about the affinity of an interaction and the changes in heat associated with it. As a result, drug-target interactions can be assigned their own ‘thermodynamic signatures’ which are described in terms of the enthalpy (DH), entropy (TDS) and free energy of binding (DG). All of these can be derived by the integrated software and are represented schematically in figure 1.
The profiles depicted in the figure show the energetics of two binding events with the same affinity but with very different underlying energetics. Figure 1A is characteristic of an interaction involving the formation of hydrogen bonds and a concomitant conformational change, these are common to receptor ligand interactions which use structural changes to transduce a signal and activate a pathway. Figure 1B describes the energetics expected for a drug binding which relies on strong hydrophobic interactions for its efficacy. This thermodynamic discrimination cannot be gleaned from structure alone. Structure gives information about proximity and interatomic distance but not the contribution that each contact makes to the overall affinity. A hydrogen donor on a receptor and hydrogen acceptor on a drug may be close in a crystal structure but contribute very little to the overall affinity. This is often due to an unfavourable ordering of the molecules which manifests itself as an entropy which opposes binding, a phenomena that ITC is uniquely situated to detect and quantify. This is the underlying problem behind design strategies which rely on structure alone and is probably the single most important reason for the failure of many rational drug design programmes.
A group led by Ernesto Freire at Johns Hopkins University in Baltimore , USA has demonstrated with a number of antiviral and metabolic drugs that, even without structure, ITC based QSAR can lead to successful lead optimisation. Several marketed HIV drugs have been thermodynamically characterised and have been shown to bind their target entropically. This suggests that hydrophobic interactions play a dominant role in the binding and is not surprising considering that the addition of non polar groups in lead optimisation protocols is common in design strategies. However, second and third generation drugs with higher affinity have been characterised by an increasingly favourable enthalpy. They have shown a strong correlation of improved efficacy with favourable enthalpy and have successfully used this phenomena to guide the optimisation process of allophenylnorstatine based anti malarial compounds.
Hydrogen bond formation is associated with a favourable, negative enthalpy. These interactions have a strong directionality and therefore engender a greater specificity than hydrophobic interactions. This specificity is difficult to engineer but because this property can be readily identified by ITC then high quality hits with good inherent activity are passed down the discovery pipeline. The serendipitous discovery of enthalpically favourable leads coupled to effective optimisation by the addition of hydrophobic groups results in specific and high affinity drugs which in turn can increase drug efficacy and reduce unwanted side effects. The Freire group have shown that hits with more favourable enthalpies of interaction are more likely to result in a marketable drug than a related small molecule with a higher affinity. The message that successful lead optimisation needs to be guided by ITC derived thermodynamic signatures rather than affinity is beginning to have a big impact on those at the discovery rock face.
Related ITC based QSAR are employed by Astra Zeneca in their drug discovery process. They call it discontinuous SAR and incorporate it in their structure based design programmes. It also involves probing the enthalpy and entropy in the optimisation process. Commercial timescales do not normally allow the structure determination of all leads. ITC can be used to decide which compounds will yield the most valuable structures. Biomolecules can be flexible and bind a range of related small molecules with similar affinity. The source of this adaptability is known as enthalpy/entropy compensation (EEC) and one of the major hurdles that drug discoverers must clear. This UK group used ITC to search for triazine derivatives which bound DNA gyrase with thermodynamic signatures different from the ‘pack’. These outliers had an increased likelihood of breaking the EEC and generated valuable structures which resulted in the discovery of a new class of drugs.

Early ADME and toxicity
Competition between target molecules and natural occurring proteins for a drug candidate is a frequent source of reduced drug potency and increased side effects. Binding measurements can be performed in the presence of these common non-specific competitors such as human serum albumin (HSA) or closely chemically related proteins with potential specific interactions so as to ascertain the extent of any such problems. Indeed the presence of HSA in the optimisation buffers can greatly reduce the number of candidate failures in the late discovery and development processes. A reduction in the apparent affinity of a drug candidate for a target when assayed in the presence of blood proteins indicates that they have a significant affinity for the drug themselves which may greatly reduce its pharmacological activity. These side reactions should also be greatly reduced if enthalpic considerations pertaining to specificity drive the optimisation process rather than affinity alone. The potential is also there to study the interactions with cytochrome P450s which are responsible for drug metabolism and excretion. Drugs with poor affinities for proteins involved in detoxification process will have a longer efficacy which in turn results in lower dosages.

Drug development
The growth in the use of proteins as therapeutic agents themselves has caused a search for quick and easy methods for measuring the shelf life of proteins. Liquid formulations are the preferred medium for protein therapeutics because of the high risk of protein inactivation during lyophilisation and because it removes the requirement of the patient or carer to reconstitute the protein powder in a prepacked buffer solution.
It is very difficult to predict the best formulation for a given biopharmaceutical and as such screening methods to optimise solution conditions and to identify excipient type and concentrations have proved to be the most popular. Automated DSC is ideally suited for this process and this versatile technique can be used in formulation development as well as in the discovery process. DSC can be used to measure the ‘melting temperature’ (TM) of a protein under a range of conditions which in turn correlates directly with the shelf life of the solution. 50 conditions a day can be studied meaning that many hundreds of conditions can be tested and optimised before lengthy and costly real time stability studies are initiated.

Summary
The ability of ultrasensitive calorimeters to detect extremely small changes in temperature coupled to the fact and that all reactions generate or sequester heat means that there is the potential to monitor all reactions. The configuration of, easy to use, commercial instruments are ideally suited to the needs of the scientists involved in the drug discovery and development processes.
The ability of ITC to assay the affinity of any two molecules free in solution makes it an ideal tool in target validation programmes, guarantees the validity of hits and quantifies affinity with precision. The enthalpy and entropy data allow for a more thorough culling of HTS hits than affinity measurements alone and gives additional information about the mechanism of binding. This reduces costs enormously by reducing the number of potential failures. The DSC measures stabilities which has applications in both discovery screening programmes and in formulation development. In the discovery process it is ideally positioned between primary screening and ITC optimisation. It is easy to implement and removes false positives that frequently survive binding assays due to non specific interactions. Detection of these non specific binders is the art of protein biopharmaceutical formulations studies. DSC is used to screen for those stabilising agents which increase the shelf lives of liquid formulations which is clearly fundamental to the market and therapeutic success of a drug.