Plasma Protein Binding

Lipophilicity is the primary driver of non-specific binding to serum albumin. A higher LogP or LogD (distribution coefficient at pH 7.4) correlates strongly with increased plasma protein binding (PPB).

• Mechanism: Hydrophobic molecules partition into the lipophilic binding pockets of albumin (Sudlow sites I and II).
• Example: A neutral, highly lipophilic steroid will typically exhibit >95% PPB, while a polar, ionized molecule like metformin shows negligible binding.
• Key Metric: LogD7.4 is often a more relevant descriptor than LogP for ionizable drugs, as it accounts for the partitioning of both neutral and ionized species at physiological pH.

The ionization state at physiological pH (7.4) dictates affinity for specific binding proteins. This is quantified by the molecule's pKa.

• Acids (pKa < 7.4): Predominantly ionized and bind with high affinity to Sudlow site I on albumin. Warfarin is a classic example.
• Bases (pKa > 7.4): Predominantly ionized and bind primarily to alpha-1-acid glycoprotein (AAG) rather than albumin. Lidocaine and propranolol are classic basic drugs with high AAG affinity.
• Neutrals & Zwitterions: Binding is driven almost entirely by lipophilicity, as electrostatic interactions with protein surfaces are minimized.

Binding is a saturable, equilibrium-driven process dependent on the concentration of both the drug and the available protein.

• Albumin Concentration: Normal serum albumin is ~3.5–5.0 g/dL. In hypoalbuminemia (e.g., liver cirrhosis, nephrotic syndrome), the free fraction of highly bound acidic drugs can increase dramatically, raising toxicity risk.
• AAG Concentration: AAG is an acute-phase reactant. Its concentration can double or triple during trauma, infection, or cancer, leading to a decreased free fraction for basic drugs.
• Displacement Interactions: A co-administered drug with a higher affinity for the same binding site can displace the first drug. A seemingly small displacement from 99% to 98% bound doubles the free, pharmacologically active concentration.

The three-dimensional shape and specific atomic arrangement of a molecule determine its complementarity to a protein's binding pocket.

• Planarity: Highly planar, aromatic molecules often intercalate more effectively into hydrophobic pockets than non-planar, flexible ones.
• Stereoselective Binding: Binding is often chiral. For example, S-warfarin is 2–5 times more potent than R-warfarin, partly due to differences in their binding affinity and subsequent clearance.
• Topological Polar Surface Area (TPSA): A high TPSA (>140 Ų) generally correlates with poor membrane permeability and lower non-specific protein binding, as polar surface area disfavors desolvation into a hydrophobic pocket.

Modern machine learning models predict the fraction unbound (fu) directly from molecular structure, bypassing the need for physical experiments in early screening.

• QSAR Models: Traditional models use 2D descriptors (e.g., ECFP4 fingerprints) and algorithms like Random Forest or XGBoost trained on curated PPB datasets.
• Graph Neural Networks (GNNs): These learn directly from the molecular graph, capturing topological features relevant to binding without explicit descriptor engineering.
• Key Challenge: The applicability domain is critical. A model trained on drug-like small molecules will fail to predict the binding of macrocycles or PROTACs, which lie outside its chemical space.

The foundational principle stating that only the unbound (free) fraction of a drug in plasma is available to exert a pharmacological effect. The bound fraction acts as a reservoir, unable to cross membranes or interact with targets. This hypothesis directly links plasma protein binding to efficacy and clearance. A highly bound drug (e.g., >99%) can experience dramatic changes in free concentration from small displacements, leading to toxicity. Key implications:

• Free drug concentration drives pharmacodynamic response
• Hepatic clearance models (e.g., the Well-Stirred Model) use the free fraction to predict extraction ratios
• Drug-drug interactions often manifest through competitive binding displacement

The two primary binding proteins in human plasma with distinct ligand preferences:

• Human Serum Albumin (HSA) : The most abundant plasma protein (~40 mg/mL), preferentially binds acidic and neutral drugs (e.g., warfarin, ibuprofen) at Sudlow's Sites I and II
• Alpha-1 Acid Glycoprotein (AAG) : An acute-phase reactant (~0.5-1.0 mg/mL) that primarily binds basic and lipophilic drugs (e.g., lidocaine, propranolol). AAG levels can double during inflammation, cancer, or trauma, significantly altering the free fraction of basic drugs
• Binding to these proteins is typically non-linear and saturable, described by the Langmuir isotherm

The quantitative parameter representing the ratio of unbound drug concentration to total plasma concentration. A critical input for physiologically-based pharmacokinetic (PBPK) models and in vitro-to-in vivo extrapolation (IVIVE).

• Measured experimentally via equilibrium dialysis, ultrafiltration, or ultracentrifugation
• In silico prediction models often use QSAR and proteochemometric modeling to estimate fu from molecular structure
• Typical fu values range from <0.01 (highly bound, e.g., diclofenac) to >0.5 (weakly bound, e.g., metformin)
• Species differences in protein homology necessitate experimental verification across preclinical species

The theoretical volume required to contain the total amount of drug in the body at the same concentration as in plasma. Plasma protein binding is a primary determinant of Vd.

• High plasma binding restricts a drug to the vascular compartment, resulting in a low Vd (e.g., warfarin, Vd ~0.1 L/kg)
• Extensive tissue binding pulls drug out of plasma, resulting in a high Vd (e.g., chloroquine, Vd >100 L/kg)
• The relationship is formalized as: Vd = Vp + Vt * (fu / fu,t), where Vp is plasma volume, Vt is tissue volume, and fu,t is the unbound fraction in tissue

A transient drug-drug interaction where a co-administered agent competes for the same binding site, displacing the primary drug and transiently increasing its free concentration. Clinical significance is highest for drugs with:

• High extraction ratio: Displacement increases clearance, often compensating for the increased free fraction
• Low extraction ratio and narrow therapeutic index: Displacement can cause toxicity (e.g., warfarin displaced by sulfonamides)
• The effect is often self-limiting due to redistribution and increased clearance. In silico models predict displacement risk using molecular docking at Sudlow's sites and relative binding affinities

The gold-standard in vitro method for measuring fraction unbound (fu) . A semi-permeable membrane separates a plasma compartment from a buffer compartment, allowing only unbound drug to equilibrate.

• Key parameters: 37°C incubation, physiological pH 7.4, and validation of equilibrium time
• Challenges: Non-specific binding to the apparatus, volume shifts due to osmotic pressure, and poor solubility of lipophilic compounds
• Rapid Equilibrium Dialysis (RED) devices enable higher throughput with smaller volumes
• In silico models are trained on equilibrium dialysis data to predict fu for novel compounds before synthesis