Ex Pharm software (Experimental – Pharmacology) - Series
Computational Modeling of Drug Disposition: Podcast link
Introduction, Modeling Techniques: Drug Absorption, Solubility, Intestinal Permeation, Drug Distribution ,Drug Excretion, Active Transport; P-gp, BCRP, Nucleoside Transporters, hPEPT1, ASBT, OCT, OATP, BBB-Choline Transporter.
Introdution:
Historical Focus: Efficacy and Safety
Historically, drug discovery focused almost exclusively on efficacy and selectivity against the biological target. This singular focus, however, led to a significant challenge in pharmaceutical development: nearly half of drug candidates failed in Phase II and Phase III clinical trials. The primary reason for these failures was undesirable pharmacokinetic properties, which encompass absorption, distribution, metabolism, excretion, and toxicity (ADMET).
The pressure to control the escalating costs of new drug development led to a paradigm shift since the mid-1990s. To mitigate the high attrition rate at these more expensive later stages, in vitro evaluation of ADMET properties was widely adopted in the early phases of drug discovery. High-throughput in vitro ADMET property screening assays were developed and successfully applied. For example, Caco-2 and MDCK cell monolayers are commonly used to simulate membrane permeability, providing an in vitro estimation of in vivo absorption.
These in vitro results have been instrumental in training in silico (computational) models, which can then be applied to predict the ADMET properties of compounds even before they are synthesized. The emergence of numerous computational programs for modeling drug ADMET properties has been fueled by increasing computational power and significant advancements in in silico modeling algorithms.
In the larger context of drug disposition, this shift highlights the critical importance of understanding how drug molecules move and are handled within the body after administration. Drug disposition refers to the movement and fate of drug molecules, encompassing ADME processes. Any alteration in a drug’s bioavailability directly impacts its pharmacological effects and therapeutic activity, making this knowledge essential for optimizing drug formulations, determining dosing regimens, and ensuring drug safety and efficacy. Consequently, computational modeling of drug disposition has become a foundational pillar in pharmaceutical sciences, offering a quantitative and mechanistic framework to predict drug behavior and reduce the reliance on traditional, more resource-intensive in vitro and in vivo experiments.
High-Throughput In Vitro Assays
The sources highlight the development and widespread adoption of high-throughput in vitro assays as a direct response to a significant challenge in historical drug discovery, thereby becoming a foundational aspect of the “Introduction” to modern drug disposition modeling.
To address this costly problem of failure of drug canddates and reduce the attrition rate at these more expensive later stages of development, a paradigm shift occurred around the mid-1990s. This shift involved the widespread adoption of in vitro evaluation of ADMET properties in the early phases of drug discovery. To facilitate this, many high-throughput in vitro ADMET property screening assays were developed and successfully applied.
A key example of such an assay is the use of Caco-2 and MDCK cell monolayers, which are widely employed to simulate membrane permeability. These cellular models provide an in vitro estimation of in vivo absorption, offering early insights into a drug’s potential for oral bioavailability.
Crucially, the results generated from these high-throughput in vitro assays were instrumental in training in silico (computational) models. This advancement allowed for the prediction of ADMET properties of compounds even before they were synthesized, further accelerating the drug discovery process and mitigating risks at earlier stages. The proliferation of these computational programs was fueled by ever-increasing computational power and significant advancements in in silico modeling algorithms.
Therefore, the development and application of high-throughput in vitro assays were a pivotal step in transforming drug discovery, moving it from a purely target-efficacy-driven process to one that integrates early and comprehensive assessment of drug disposition properties, thereby becoming a foundational pillar in pharmaceutical sciences for predicting drug behavior and reducing the reliance on traditional, more resource-intensive in vitro and in vivo experiments in later stages.
Caco-2 & MDCK Cell Monolayers: Membrane Permeability
Caco-2 and MDCK cell monolayers emerged as key examples of these in vitro screening tools:
- They are widely used to simulate membrane permeability. Caco-2 cells, in particular, are described as immobilized human colorectal adenocarcinoma cells.
- These cellular models provide an in vitro estimation of in vivo absorption. This allows researchers to gain early insights into a drug’s potential for oral bioavailability, which is crucial since oral administration is the most preferred drug delivery form due to its convenience and patient compliance, and oral absorption mainly occurs in the human intestine.
- The permeation process through these membranes involves both passive diffusion and active transport. Most current models aim to simulate in vitro membrane permeation using these cell lines, as they have been useful indicators of in vivo drug absorption. PAMPA (parallel artificial membrane permeability assay) is also mentioned as another highly used intestinal epithelial membrane model for predicting permeation characteristics.
Crucially, the results generated from these high-throughput in vitro assays, such as those from Caco-2 and MDCK cells, have been instrumental in training in silico (computational) models. This advancement means that computational models can now be applied to predict the ADMET properties of compounds even before they are synthesized. The emergence of numerous computational programs for modeling drug ADMET properties has been fueled by ever-increasing computational power and significant advancements in in silico modeling algorithms.
Therefore, the development and widespread use of Caco-2 and MDCK cell monolayers represent a fundamental change in drug discovery. They provided the necessary experimental data for training predictive computational tools, establishing computational modeling of drug disposition as a foundational pillar in pharmaceutical sciences. This integration offers a quantitative and mechanistic framework to predict drug behavior, significantly reducing the reliance on traditional, more resource-intensive in vitro and in vivo experiments in later development stages.
In Vitro Results Enable In Silico Model Training
- The Problem Prior to the Shift: Historically, drug discovery focused “almost exclusively on efficacy and selectivity against the biological target”. This singular focus led to a significant issue: “nearly half of drug candidates fail at phase II and phase III clinical trials because of undesirable drug pharmacokinetics properties, including absorption, distribution, metabolism, excretion, and toxicity (ADMET)”. These later stages of development are particularly expensive.
- The Paradigm Shift and In Vitro Adoption: To address this high attrition rate and control escalating costs, a “paradigm shift since the mid-1990s” occurred. This involved the “widespread adoption” of “in vitro evaluation of ADMET properties in the early phases of drug discovery”. To support this, “Many high-throughput in vitro ADMET property screening assays have been developed and applied successfully”. A key example mentioned is the use of “Caco-2 and MDCK cell monolayers [which] are widely used to simulate membrane permeability as an in vitro estimation of in vivo absorption”. These in vitro models indicate in vivo drug absorption.
- The Enabling Role of In Vitro Results for In Silico Models: “These in vitro results have enabled the training of in silico models”. This capability allows researchers to “predict the ADMET properties of compounds even before they are synthesized”. This ability to predict properties computationally at such an early stage helps in “reducing costs, time, and ethical concerns associated with animal testing”.
- Factors Fueling In Silico Advancement: The emergence of numerous computational programs for modeling ADMET properties has been “fueled by the ever-increasing computational power and significant advances of in silico modeling algorithms”. This implies a synergistic relationship where the availability of in vitro data provided the necessary training ground for computational algorithms, which in turn benefited from technological advancements.
Role of Computational Power and Algorithm Advances
“Fueled by the ever-increasing computational power and significant advances of in silico modeling algorithms, numerous computational programs that aim at modeling drug ADMET properties have emerged”. This indicates a synergistic relationship:
- Increasing Computational Power: The sheer processing capability of computers allowed for the complex calculations and data handling required to develop and run sophisticated in silico models. This power enabled the analysis of large datasets and the execution of computationally intensive algorithms.
- Significant Advances of In Silico Modeling Algorithms: Beyond just raw power, breakthroughs in the algorithms themselves were critical. These advancements refer to the development of more effective and accurate methods for:
- Pharmacophore modeling, which investigates structural requirements for drug-target interactions and identifies essential 3D arrangements of chemical features.
- Flexible docking studies, which predict how drugs interact with targets involved in ADMET processes.
- Quantitative Structure-Activity Relationship (QSAR) and Quantitative Structure-Property Relationship (QSPR) studies, which use multivariate analysis to correlate molecular descriptors with ADMET-related properties.
- The development of statistical algorithms like simple multiple linear regression (MLR), multivariate partial least-squares (PLS), artificial neural networks (ANN), and support vector machines (SVM).
- Molecular descriptors, which represent the chemical structure numerically and are correlated with ADMET-related properties.
Absorption, Distribution, and Excretion:
- Drug Absorption:
- Much of the in silico attention is “focused on modeling drug oral absorption” because oral administration is the most preferred drug delivery form due to its convenience and patient compliance.
- Absorption is primarily determined by the interplay between drug solubility and intestinal permeability.
- Solubility is a critical determinant, as a drug must dissolve before it can be absorbed. In silico modeling can predict solubility “even before synthesizing it”, using approaches based on underlying physiological processes or empirical methods like QSPR. Key descriptors like LogP are used to estimate solvent-solute interaction.
- Intestinal permeation describes a drug’s ability to cross the intestinal mucosa. Due to its complexity, most models aim to simulate in vitro membrane permeation in cell lines like Caco-2, MDCK, or PAMPA, which serve as useful indicators of in vivo absorption.
- Other considerations for absorption include the ionization state of a compound, which affects both solubility and permeability and can be determined using its pKa value and environmental pH. The presence of both influx and efflux transporters in intestinal epithelial cells (e.g., hPEPT1, ASBT, P-glycoprotein, BCRP) can significantly influence oral absorption. Drug metabolism in intestinal epithelial cells by cytochrome P450 enzymes should also be considered. Commercial packages such as GastroPlus and iDEA are available for predicting oral absorption and other pharmacokinetic properties.
- Drug Distribution:
- Distribution is a “critical aspect of a drug’s pharmacokinetic profile”.
- It is mainly reflected by three parameters: volume of distribution (VD), plasma-protein binding (PPB), and blood-brain barrier (BBB) permeability.
- VD measures the relative partitioning of a drug between plasma and tissue and is crucial for predicting drug half-life, which impacts dosing frequency. However, computational models solely based on computed descriptors are still under development due to data scarcity and process complexity.
- PPB is important because only the unbound drug contributes to pharmacological efficacy. Models for PPB should account for binding to multiple proteins.
- BBB permeability assesses a drug’s ability to cross into the central nervous system (CNS). Most current BBB permeation prediction models have limited practical use due to inconsistent experimental data. The presence of efflux transporters at the BBB is also a significant factor that, if not considered, can compromise prediction accuracy.
- Drug Excretion:
- Drug excretion or clearance is quantified by plasma clearance.
- The two main components are hepatic and renal clearances.
- No model currently predicts plasma clearance solely from computed drug structures, with efforts focusing on estimating in vivo clearance from in vitro data.
- Like other pharmacokinetic aspects, the hepatic and renal clearance processes are complicated by the presence of active transporters.
- Active Transporters (Integral to all ADE):
- Transporters are considered “an integral part of any ADMET modeling program” due to their ubiquitous presence and the overlap between their substrates and many drugs.
- Despite previous limited understanding, interest has generated a large amount of in vitro data, enabling the development of pharmacophore and QSAR models for many transporters.
- Their incorporation into current modeling programs is essential for “more accurate prediction of drug disposition behavior”. Specific examples include P-glycoprotein (P-gp), Breast Cancer Resistance Protein (BCRP), Nucleoside Transporters, hPEPT1, ASBT, OCT, OATP, and the BBB-Choline Transporter.
In summary, the sources establish that the focus on ADE modeling, particularly before synthesis, is a direct response to the historical failures and escalating costs in drug discovery. This focus is enabled by the combination of in vitro data and significant advancements in computational power and in silico modeling algorithms, positioning computational modeling as a “foundational pillar in pharmaceutical sciences” for predicting drug behavior and reducing experimental burdens.
Frequently asked questions:
Drug disposition refers to the movement and fate of drug molecules within the body after administration, encompassing the processes of absorption, distribution, metabolism, and excretion (ADME). It is a crucial aspect of pharmaceutical sciences because any alteration in a drug’s bioavailability directly impacts its pharmacological effects and therapeutic activity. Understanding drug disposition allows researchers to predict how a drug will interact with the body’s systems, from its initial entry to its eventual elimination. This knowledge is essential for optimizing drug formulations, determining appropriate dosing regimens, and ensuring the safety and efficacy of new drug candidates.
Computational modeling of drug disposition employs various techniques, broadly categorized into quantitative and qualitative approaches. Quantitative approaches include pharmacophore modeling and flexible docking studies, which investigate the structural requirements for drug-target interactions in ADMET processes. Widely used pharmacophore perception tools include DISCO, GASP, and Catalyst/HIPHOP.
Other significant modeling techniques include:
Physiologically Based Pharmacokinetic (PBPK) Models: These are mechanistic, multi-compartmental models that represent organs and tissues, integrating physiological parameters like tissue volumes and blood flow. They are valuable for simulating drug kinetics in diverse populations and predicting tissue-specific concentrations.
Population Pharmacokinetic (PopPK) Models: These models analyze drug concentration data from populations to quantify variability and identify factors influencing drug kinetics, aiding in dose optimization for different patient groups.
Noncompartmental Analysis (NCA): A model-independent approach that estimates pharmacokinetic parameters directly from concentration-time data without assuming a specific compartmental structure.
Molecular Modeling and Simulation: This includes techniques like molecular docking, molecular dynamics (MD) simulations, and homology modeling to predict drug-protein interactions, binding affinities, and conformational changes.
Hybrid and Multiscale Models: These combine different modeling approaches (e.g., PBPK with Machine Learning or QSAR) to capture processes across multiple biological scales, from molecular interactions to whole-body disposition.
These in silico (computational) methods complement or replace traditional in vitro and in vivo experiments, offering advantages in terms of cost, time, and ethical considerations.
Computational models predict drug absorption by considering several critical factors:
Solubility: This is a key determinant of absorption. Models like GastroPlus™ and ADMET Predictor™ calculate regional solubility along the GI tract using the Henderson–Hasselbalch equation, incorporating pKa values and the effects of bile salts. The ionization state of a molecule (determined by its pKa and environmental pH) significantly affects its solubility and permeability. The Developability Classification System (DCS), a refinement of the Biopharmaceutics Classification System (BCS), further categorizes drugs based on their solubility and permeability, distinguishing between dissolution-rate-limiting (Class-IIa) and solubility-rate-limiting (Class-IIb) compounds.
Intestinal Permeation: This refers to a drug’s ability to cross the intestinal epithelium. Models like GastroPlus, Simcyp, and GI-Sim account for regional differences in intestinal permeability, transporter expression, and dynamic processes such as dissolution and precipitation. Molecular Dynamics (MD) simulations provide atomistic insights into passive permeation through lipid bilayers.
Dissolution and GI Transit: Computational models simulate dissolution using models like Noyes-Whitney, incorporating parameters like particle size, solubility, and formulation characteristics. GI transit is modeled as a series of compartments with specific transit times and absorption properties, often using models like CAT and ACAT.
These models allow for in vitro-in vivo extrapolation and the prediction of total pharmacokinetic behavior, including interactions with transporter systems and plasma drug concentrations.
Pharmacophore models are computational tools that identify the essential 3D arrangements of chemical features required for a molecule to interact with a specific biological target, such as a receptor or a transporter protein. These features typically include hydrophobic centers, hydrogen bond donors, hydrogen bond acceptors, and positively or negatively ionizable groups.
In drug disposition studies, pharmacophore models are used for:
Screening Databases: They can be applied to large databases of drug-like molecules to identify compounds likely to interact with specific transporters or enzymes involved in ADMET processes. For example, a pharmacophore model for the human peptide transporter (hPEPT1) identified two hydrophobic features, one hydrogen bond donor, one hydrogen bond acceptor, and one negative ionizable feature as transport requirements.
Understanding Interaction Requirements: They help elucidate the structural characteristics necessary for a drug to bind to or be transported by a particular protein. The ASBT pharmacophore model, for instance, revealed requirements of one hydrogen bond donor, one hydrogen bond acceptor, one negative charge, and three hydrophobic centers for transport.
Drug Design: By identifying key features, pharmacophore models guide the design of new drug molecules with improved ADMET properties. Tools like PharmaGist can analyze a group of similar compounds to derive a common pharmacophore model with precise spatial measurements.
Pharmacophore modeling, along with molecular docking and QSAR/QSPR, is a pivotal computational approach in drug discovery and understanding biotransformation processes.
Drug distribution modeling aims to predict how a drug moves from the systemic circulation into various tissues and organs. The extent of distribution is primarily determined by the structural and physicochemical properties of the drug and is reflected by three key parameters:
Volume of Distribution (Vd): This is a hypothetical volume representing the relative partitioning of a drug between plasma and tissues. It’s a crucial constant that, when combined with drug clearance, can predict a drug’s half-life, which in turn dictates dosing frequency. While computational models for Vd based solely on computed descriptors are still evolving due to the complexity and scarcity of in vivo data, approaches like Lombardo and colleagues’ model for neutral and basic compounds, using in vitro physicochemical parameters, represent significant progress.
Plasma-Protein Binding (PPB): This refers to the extent to which a drug binds to proteins in the plasma. Only the unbound (free) fraction of a drug in plasma is available for distribution into tissues and for pharmacological action. Computational models predict PPB by analyzing molecular structure and electronic configurations, as plasma protein concentration and protein binding nature significantly influence drug distribution.
Blood-Brain Barrier (BBB) Permeability: This measures a drug’s ability to cross the highly selective BBB. Permeation across the BBB can be calculated using the polar surface area (PSA) of the molecule and its Log P value, or by passive permeability factor (PS) with Log D. Computational models also consider the composition and nature of tight junction (TJ) proteins like claudin and occludin, which regulate permeation.
Overall, the physicochemical properties, such as lipophilicity and ionization, alongside blood flow and tissue permeability, are fundamental inputs for these distribution models.
Active transporters are proteins located in cell membranes that actively move drugs and other molecules across biological barriers, such as intestinal epithelial cells, kidney tubules, and the blood-brain barrier. They can either increase (influx transporters) or decrease (efflux transporters) drug absorption and influence distribution and excretion.
Several specific types of active transporters are crucial in drug disposition:
hPEPT1 (Human Peptide Transporter 1): A low-affinity, high-capacity oligopeptide transport system mainly expressed in the intestine and kidney, affecting drug absorption and excretion. It transports a diverse range of substrates, including β-lactam antibiotics and ACE inhibitors.
ASBT (Apical Sodium-Dependent Bile Acid Transporter): A high-efficacy, high-capacity transporter on the apical membrane of intestinal epithelial cells and cholangiocytes. It assists in the absorption of bile acids and their analogs, serving as an intestinal target for improving drug absorption.
OCT (Organic Cation Transporters): Facilitate the uptake of many cationic drugs (e.g., antiarrhythmics, antihistamines) across different barrier membranes in the kidney, liver, and intestine.
P-gp (P-glycoprotein): An efflux transporter that pumps drugs out of cells, often reducing their absorption and increasing their elimination.
Nucleoside Transporters: Responsible for transporting nucleosides (starting materials for DNA/RNA synthesis) and regulating neuronal modulation. They are classified into sodium ion-dependent (CNT1-3) and independent (ENT1-2, SLC29A1-2) systems.
OATP (Organic Anion Transporting Polypeptides): A superfamily responsible for transporting amphipathic endogenous and exogenous organic compounds and intestinal absorption of drugs.
BBB-Choline Transporter: Facilitates choline transport across the blood-brain barrier, essential for acetylcholine biosynthesis.
Computational modeling techniques, including QSAR, pharmacophore modeling, and structure-based docking, are employed to predict interactions with these transporters.
The ionization state of a drug molecule significantly affects its ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) properties. This is primarily because the charge of a molecule influences its solubility and permeability across biological membranes.
Solubility: Ionized forms of a drug generally have higher aqueous solubility compared to their unionized counterparts. For instance, at its isoelectric point, a molecule’s solubility might decrease, but if it forms an ionic derivative or zwitterion at a particular pH, its solubility increases. Since most drugs are weak acids or bases, their ionization state is pH-dependent. Computational programs use the compound’s ionization constant (pKa) and the environmental pH to estimate the charge, thus predicting solubility.
Permeability: Unionized forms of a drug are typically more lipophilic and can more easily cross non-polar biological membranes (like cell membranes) via passive diffusion. Conversely, ionized forms struggle to permeate these lipid bilayers without the aid of active transport systems. Therefore, the balance between a drug’s hydrophilic and lipophilic nature, influenced by its ionization, is crucial for its movement through cellular membranes and portal systems.
Drug Disposition: The pH-partition hypothesis explains how the pH of different bodily environments (e.g., stomach, intestine, blood) influences the proportion of ionized vs. unionized drug, thereby impacting its absorption from the gastrointestinal tract, its distribution into various tissues (including across the blood-brain barrier), and its excretion by the kidneys.
Several commercial and publicly available programs provide pKa estimation based on the input chemical structure, aiding in the prediction of these crucial ADMET properties.
A variety of databases and software tools are essential for computational modeling of drug disposition, providing data and functionalities for predicting ADMET properties.
Databases:
Experimental pKa database: Contains pKa values for a large number of organic acids and bases.
Phys-chem EPISUITE database: Used for calculating physicochemical properties, including octanol-water partition coefficient, boiling point, melting point, and vapor pressure.
ADME database, Bioconcentration NITE: Provide information relevant to ADME properties.
ZINC database: A large collection of commercially available compounds with their physicochemical and pharmacophoric features, often used for virtual screening.
PubChem database: A comprehensive public database of chemical substances and their biological activities, maintained by the National Center Biotechnological Information.
Human Intestinal Transporter Database: Provides curated data and validated QSAR models for various transporters like P-gp, hPEPT1, and ASBT.
Software Tools:
Pharmacophore Perception Tools: DISCO, GASP, Catalyst/HIPHOP for identifying common features among active compounds. PharmaGist for ligand-based pharmacophore detection.
pKa Estimation Tools: SCSpKa, Pallas/pKalc, ACD/pKa, SPARC online calculator, MoKa, CHARMM, H++Poisson-Boltzmann, MCCE, PROPKA.
Log P Prediction Tools: ALOGP, XLOGP, MLOGP, CLOGP.
ADMET Prediction Suites: ADMET Predictor™, SwissADME (a free web-based tool), GastroPlus™, Simcyp®, PK-Sim®. These platforms can simulate drug absorption, dissolution, pharmacokinetics, and predict a wide range of ADMET properties.
Molecular Modeling Tools: Programs for molecular docking, molecular dynamics (MD) simulations, and homology modeling to predict drug-protein interactions (e.g., PHENIX, AFITT).
QSAR/QSPR Software: DRAGON, E-Dragon, PADEL for calculating molecular descriptors; QSARINS for model development, analysis, and validation.
Pharmacokinetic Analysis Software: NONMEM, Monolix, Pmetrics for PopPK modeling; PhoenixWinNonlin, PKSolver for Noncompartmental Analysis (NCA).
Sequence Alignment Tools: FASTA, BLAST for homology modeling when receptor structures are unavailable.
These tools collectively enable in silico assessment of drug molecules, from predicting solubility and permeability to simulating whole-body disposition and interactions with biological targets.
About the Author
