Advanced Strategies to Study Plasma Protein Binding, Improve PK/PD, and Develop Better Enzyme Inhibitors
In early drug discovery, the molecular hits generated from a high throughput screen (HTS) are subject to an initial assessment to identify promising lead compounds. Following this process, the leads identified undergo more extensive optimization, for both efficacy and safety. The object of this optimization phase is to maintain the desirable properties in lead compounds while improving on deficiencies in the lead structure for better target specificity and selectivity. More often than not, only one or two molecules out of hundreds survive the entire screening process and make the transition to drug candidate.
During optimization, to produce a strong candidate profile for preclinical studies, the molecules must be evaluated for pharmacokinetic/pharmacodynamic (PK/PD) and toxicological properties. All this data, plus manufacture and control considerations, will guide the creation of a regulatory submission that will allow the drug candidate to be tested in humans.
However, despite rigorous drug lead analysis, there’s still a chance that certain adverse events related to the metabolism, transport, and clearance of these molecules, and/or drug-drug interactions may appear much later in the downstream development stages. To avoid such issues, the pharma industry is moving towards novel “fail fast, fail early” approaches by developing improved in vitro assays, and predictive tools to empower more effective PK/PD testing.
Measuring Unbound Drugs in Plasma Protein Binding Studies
The fraction of unbound drug (fu) is a critical pharmacokinetic parameter that is routinely used in establishing PK/PD relationships, predicting human dose and drug-drug interaction liability (DDI), and assessing drug safety. According to Zhengyin Yan, Ph.D., senior scientist, department of drug metabolism and pharmacokinetics, Genentech, this is an important aspect of lead optimization because binding of drug molecules to plasma proteins can significantly lower free drug concentrations and impact not only their therapeutic potency and half-life, but also tissue distribution, metabolism, and excretion.
Currently, there are three commonly used methods for measuring drug plasma binding: equilibrium dialysis (ED), ultrafiltration (UF), and ultracentrifugation (UC). Each of these methods has its own advantages and limitations depending on the properties of the investigated drug. “The UF method is straightforward but suffers from high nonspecific binding to the filtration membrane, while UC is a low throughput method that is more suitable for drugs with high molecular weight or instability in plasma,” explains Dr. Yan. “ED is actually regarded as the ‘gold standard’ because its underlining principal is based on equilibrium, and it allows higher throughput.”
However, it’s important to note that the concentration of unbound drug is very much dependent on whether true equilibrium is reached in the ED assay. In this regard, Dr. Yan and his team at Genentech have optimized a strategy to reliably measure fu values of high binders and ensure true equilibrium is attained. “We are working on an orthogonal approach, called the bi-directional equilibrium dialysis (Bi-ED),” says Dr. Yan. “It enables us to simultaneously measure a pair of fu values for each drug based on equilibration in two opposite dialysis directions; from plasma to buffer and from buffer to plasma.”
Dr. Yan went on to explain that, hypothetically, if true equilibrium is obtained in both dialysis directions, two measured fu values for a given drug should converge and the ratio of the two values would become unity (1.0). So, the ratio can be used as a tangible readout for data reliability.
Although the Bi-ED approach represents a simple and effective alternative to determine low quantities of unbound drug with increasing confidence, it still remains a challenge to accurately measure low fu values for those compounds with high nonspecific binding (due to slow equilibrium). “For such extensively bound drugs with high logD7.4, only a range of fu can be reported with confidence because of uncertainty in the true equilibrium,” remarks Dr. Yan.
Importance of a Combined PK/PD Approach
In the lead identification stage of small molecule drug discovery, the goal is to achieve optimum potency and absorption, distribution, metabolism, and excretion (ADME) properties that translate into desirable pharmacokinetics (PK), i.e., systemic clearance, exposures, half-life, and oral bioavailability. These, in turn, are expected to yield the desired pharmacological effect in animal models. During lead identification, the ADME properties of the lead series are evaluated typically using a combination of in silico and in vitro assays where hundreds of compounds are screened in high throughput mode—often referred to as front loading of drug metabolism and pharmacokinetics (DMPK).
Ramesh Jayaraman, CSO, TheraIndx Lifesciences , believes that this process happens relatively quickly, leading to compounds being prioritized for in vivo PK studies. “Techniques such as snap shot PK, sparse sampling, and cassette dosing, with the optimization of bioanalytical methods for minute samples, has significantly increased the speed of in vivo PK screening,” he explains. However, according to Jayaraman, the same cannot be said about pharmacodynamics (PD). “PD characterization does not receive the same attention when compared to PK. It is very important to characterize PK and PD effects equally early on when looking for potential leads.”
When a molecule is ready for efficacy testing, the common questions that arise are related to the dose, frequency, and duration needed to achieve efficacy. All these are answered if PD is also characterized simultaneously with PK. Relationships between time course of concentration (PK) and the time course of effect such as onset, intensity, and duration (PD) are best characterized early on in lead optimization, to confirm the hypothesis that optimum PK has produced the desired pharmacological effect in the relevant animal model of disease.
“Such a combined approach tells the discovery team a lot about the key PK parameters of the compound that contribute to efficacy, which in turn will help the medicinal chemist to design compounds with optimized physico-chemical properties, or in other words, developing structure-activity relationships for PK and PD,” notes Jayaraman. “A thorough understanding of PK/PD of the lead compound and developing a PK/PD model helps prioritize optimized compounds for pharmacology studies based only on PK and in vitro potency. This becomes particularly important for long duration pharmacology studies where savings in resources and cost become substantial.”
TheraIndx provides services in preclinical research including integrated drug discovery services, whereby sponsors are advised on the importance of a combined PK/PD approach to help optimize their compounds based on quantitative pharmacological parameters.
Predicting CYP Inhibitors with an Integrated In-silico Approach
Cytochrome P450 (CYP) enzymes are major drug-metabolizing enzymes; they are responsible for the metabolism of approximately 75% of human drugs. CYP enzymes mediate the oxidation of many endogenous molecules and xenobiotics involved in numerous physiological and pathophysiological processes. Malfunctioning of these enzymes can result in dire metabolic effects.
“Metabolism is a key mechanism for detoxification allowing drugs to be eliminated from the organism,” says Maria A. Miteva, Ph.D., research director, Molécules Thérapeutiques in silico (MTi), Inserm Institute. “But, in some cases, drug oxidation by CYPs can produce toxic metabolites that cause adverse drug reactions. Also, administration of more than one drug can provoke drug-drug interactions via inhibition or induction of CYPs.”
Dr. Miteva goes on to add that several key CYP isoforms exhibit genetic polymorphisms that can affect the enzyme expression and/or function, leading to distinct phenotypes as either poor, intermediate, or extensive metabolizers, and sometimes, ultra-metabolizers. “Loss-of-function variants of CYP can lead to reduced intrinsic clearance and increased drug plasma concentrations provoking toxicity, while gain-of-function variants can lead to increased clearance and lower drug concentrations, resulting in diminished drug efficiency.”
For decades, quantitative structure-activity (property) relationships (QSARs/QSPRs) and statistical approaches have been used to develop computer models to predict CYP inhibition. However, these approaches are limited by an inadequate amount of publicly available high-quality experimental data. Moreover, they tend to neglect the 3D structures of CYPs. The fascinating natural flexibility of CYP structures enables them to accommodate many different ligands into their active sites. But the flip side is, it can strongly disturb the prediction of the ligand binding, which is critical for understanding inhibition.
“For several years, we have been developing a new in silico approach by integrating knowledge of CYP 3D structures, their dynamic behavior in response to the binding of various inhibitors, and machine-learning techniques,” says Dr. Miteva. “This integrated strategy takes into account both the physico-chemical properties of CYP inhibitors and their interactions with the enzyme at the atomic level.”
During the last few years, there has been a growing number of research studies applying various machine-learning methods in drug discovery. Dr. Miteva emphasizes that these methodologies present great potential for the development of powerful, predictive artificial intelligence-enabled drug discovery programs. She adds that including mechanistic knowledge of drug-target/off-target interactions at the atomic level, as well as genetic variations, is critical to reliably simulate complex molecular mechanisms associated with drug metabolism and toxicity.