Hydrophobicity Scales: What They Measure and Where They Disagree

Why There Is No Single Correct Scale

A hydrophobicity scale assigns a numerical value to each amino acid side chain representing how hydrophobic it is. The concept is straightforward; the measurement is not. Hydrophobicity is not a fixed molecular property like mass or bond length. It is a thermodynamic quantity that depends on the reference environment: the hydrophobicity of leucine relative to glycine in water-to-octanol transfer is not the same as its relative hydrophobicity in water-to-membrane transfer, or in water-to-vapor transfer, or in the context of a folding protein interior. Different scales measure these different things, and each is correct for its intended application. The disagreements between scales are not experimental errors; they reflect genuine differences in what is being measured.

The Kyte-Doolittle Scale: Protein Interior Focus

The most widely cited hydrophobicity scale in the literature was developed by Kyte and Doolittle in 1982, originally for identifying transmembrane helices in protein sequences. ^[4] Their scale was empirically derived from a combination of water-to-vapor transfer free energies and interior-to-surface distributions of amino acids in crystal structures, normalized to run from most hydrophilic to most hydrophobic. The scale assigns the highest hydrophobicity to isoleucine and the lowest to arginine. Applied as a sliding window average along a sequence, it identifies hydrophobic stretches likely to be buried in protein interiors or inserted in membranes.

The Kyte-Doolittle scale is appropriate when the question concerns burial in a protein hydrophobic core or spanning a membrane bilayer. It is less appropriate for predicting behavior at water-membrane interfaces or for peptides that do not form compact folded structures. Its wide adoption reflects historical precedent and ease of application rather than demonstrable superiority across all contexts.

The Eisenberg Consensus Scale

Recognizing that individual experimental scales carry measurement-specific artifacts, Eisenberg and colleagues constructed a consensus scale by normalizing and averaging five existing scales. ^[5] The resulting values represent the central tendency of the hydrophobicity concept across different measurement approaches and are less subject to the idiosyncrasies of any single method. The Eisenberg scale also introduced the hydrophobic moment as a quantitative descriptor of amphipathicity, allowing systematic identification of sequences with one hydrophobic face and one hydrophilic face in a helical conformation. This descriptor proved particularly useful for analyzing antimicrobial peptides and membrane-active sequences, where amphipathicity is a key determinant of activity.

The Wimley-White Interfacial Scale

The Wimley-White scale, derived from measurements of peptide partitioning into palmitoyloleoylphosphatidylcholine, POPC, bilayer interfaces, addresses a limitation of earlier scales: they measured transfer into bulk organic phases such as octanol or cyclohexane, which do not capture the unique properties of the membrane-water interface. ^[6] The interface is not simply a boundary between water and hydrophobic core; it is a distinct chemical environment with a gradient of polarity, specific hydrogen bonding geometry, and electrostatic properties that differ from both bulk phases. Peptides that associate with membrane interfaces, including many antimicrobial peptides, cell-penetrating peptides, and amphipathic helices, experience this interfacial environment rather than the membrane interior.

The Wimley-White scale shows significant differences from octanol-based scales for polar and charged residues, which make unfavorable contributions to octanol partitioning but can be accommodated at the bilayer interface. These differences have practical consequences: peptides designed using octanol-based scales for membrane activity may be mispredicted compared to those designed using interfacial parameters.

The Moon-Fleming Biological Scale

A more recent contribution to the hydrophobicity debate is the Moon-Fleming scale, derived from the measured efficiency of translocon-mediated membrane insertion of designed helical segments in a cellular context. ^[7] Rather than measuring partitioning into a model system in vitro, this approach measures biological membrane insertion directly, using the Sec61 translocon machinery of the endoplasmic reticulum. The resulting scale reports on the thermodynamic cost of inserting each side chain into a biological membrane under physiological conditions, which is directly relevant to predicting the transmembrane topology of membrane proteins.

The Moon-Fleming scale differs from earlier scales in several important ways: it assigns more favorable hydrophobicity to certain aromatic residues, particularly tryptophan and tyrosine, which are abundant at the lipid-water interface of biological membranes but score poorly on octanol-based scales. It also provides a more accurate framework for predicting which sequences will be recognized as transmembrane by the cellular insertion machinery, a property that earlier scales were applied to but not specifically designed to predict.

Choosing the Right Scale for the Application

The practical guidance that emerges from this landscape is straightforward, though it requires knowing which question is being asked. For identifying sequences likely to be buried in protein hydrophobic cores, Kyte-Doolittle or the Eisenberg consensus scale are appropriate. For analyzing membrane-active peptides that interact with the lipid-water interface, the Wimley-White interfacial scale is the better tool. For predicting transmembrane segments in polytopic membrane proteins in biological membranes, the Moon-Fleming biological scale provides the most relevant thermodynamic parameters. For calculating the hydrophobic moment of a helical peptide to assess amphipathicity, the Eisenberg formalism remains the standard.

No scale should be applied without consideration of what it measures. The widespread use of Kyte-Doolittle for all hydrophobicity calculations regardless of context is a persistent source of misprediction in the literature. When a peptide's hydrophobicity is reported without specification of which scale was used, the value is incompletely described. This is a field where the units, so to speak, determine the meaning, and choosing them carelessly produces results that are consistent with themselves but not with the biology.