Beta turns are among the most common non-repetitive structural motifs in proteins, serving as critical architectural elements that allow polypeptide chains to reverse direction and achieve compact, globular folds. On top of that, understanding the intrinsic tendency of specific amino acids to populate these tight turns is fundamental to protein structure prediction, folding studies, and rational protein design. One of the most enduring and widely referenced frameworks for quantifying this tendency is the Levitt beta turn propensity values for amino acids, derived from the pioneering statistical analysis conducted by Michael Levitt in the late 1970s Simple, but easy to overlook..
This article provides a comprehensive exploration of these propensity values, their methodological origins, the structural rationale behind the observed preferences, their application in modern bioinformatics, and the nuances that distinguish them from subsequent propensity scales Practical, not theoretical..
The Historical Context: Levitt’s 1978 Analysis
Before the explosion of structural data in the Protein Data Bank (PDB), Michael Levitt performed a seminal statistical survey of known protein structures to decipher the conformational preferences of amino acids in reverse turns. Published in 1978, his work analyzed a dataset of roughly 30 to 40 high-resolution protein structures—a tiny fraction of today’s database, but statistically significant for the era Most people skip this — try not to..
Levitt’s approach was rigorous for its time. On top of that, he defined a beta turn (often called a $\beta$-turn or hairpin bend) as a four-residue segment ($i, i+1, i+2, i+3$) where the distance between the $C_\alpha$ atoms of residues $i$ and $i+3$ is less than 7 Å, and the central two residues ($i+1$ and $i+2$) are not helical. He then calculated the frequency of each of the 20 standard amino acids occurring at each of the four positions within the turn.
The result was a position-specific propensity scale. Which means unlike simple amino acid composition scales, Levitt recognized that the structural constraints at position $i+1$ (the first corner residue) differ vastly from those at position $i+2$ (the second corner residue) or the flanking positions $i$ and $i+3$. This positional specificity is the hallmark of the Levitt scale and remains its most valuable feature for structural biologists.
Quick note before moving on.
The Levitt Propensity Scale: Values and Interpretation
The propensity value ($P$) for an amino acid at a specific position is typically calculated as the ratio of the observed frequency of that residue at that position in turns to its expected frequency based on its overall abundance in the dataset. A value greater than 1.Also, 0 indicates a preference for that position; a value less than 1. 0 indicates avoidance.
While exact numerical values vary slightly depending on the specific dataset reproduction (original 1978 vs. later re-calculations on larger datasets), the qualitative hierarchy established by Levitt has proven remarkably strong. The following summarizes the characteristic preferences for the two critical central positions ($i+1$ and $i+2$), where the backbone dihedral angles ($\phi, \psi$) are most constrained Easy to understand, harder to ignore..
Position $i+1$: The First Corner
This position requires a sharp change in chain direction. The backbone dihedral angles typically cluster around $\phi \approx -60^\circ, \psi \approx -30^\circ$ (Type I/II turns) or $\phi \approx 60^\circ, \psi \approx 30^\circ$ (Type I'/II' turns).
- High Propensity ($P > 1.5$): Asparagine (Asn), Aspartic Acid (Asp), Cysteine (Cys), Serine (Ser).
- Rationale: These residues possess short, polar side chains (or thiol in Cys) that can form stabilizing hydrogen bonds with the main chain (side chain-to-main chain H-bonds) without steric clash. Asn and Asp are the quintessential "turn starters."
- Moderate Propensity: Glycine (Gly), Proline (Pro), Threonine (Thr).
- Glycine: Extreme conformational flexibility allows it to adopt the positive $\phi$ angles required for Type II turns, though it is statistically more dominant at $i+2$.
- Proline: Its rigid cyclic structure locks $\phi \approx -60^\circ$, perfectly matching the Type I/II $i+1$ conformation. On the flip side, its lack of an amide hydrogen prevents it from donating an H-bond, slightly moderating its score compared to Asn/Asp in some analyses.
- Low Propensity ($P < 0.5$): Valine (Val), Isoleucine (Ile), Leucine (Leu), Phenylalanine (Phe), Tyrosine (Tyr), Tryptophan (Trp), Methionine (Met).
- Rationale: Branched $\beta$-carbons (Val, Ile, Thr) or bulky aromatic/long aliphatic side chains create severe steric hindrance with the main chain atoms of the $i+2$ and $i+3$ residues when the backbone folds tightly.
Position $i+2$: The Second Corner
This position often adopts conformations with positive $\phi$ angles (Type I: $\phi \approx -90^\circ, \psi \approx 0^\circ$; Type II: $\phi \approx 80^\circ, \psi \approx 0^\circ$). Positive $\phi$ is sterically forbidden for all residues except Glycine due to clash between the $C_\beta$ atom and the carbonyl oxygen of the preceding residue.
- Dominant Propensity ($P \gg 2.0$): Glycine (Gly).
- Rationale: Lacking a $C_\beta$ atom, Glycine is the only residue that can comfortably adopt the positive $\phi$ angle required for Type II and Type II' turns. It is the single most overrepresented residue in beta turns overall, specifically at $i+2$.
- High Propensity: Asparagine (Asn), Aspartic Acid (Asp), Serine (Ser).
- These residues can adopt the necessary conformations (often Type I or VIII) and provide side-chain H-bonding stabilization.
- Moderate/Low Propensity: Proline (Pro) is occasionally found here (Type II turns require $i+1$ Pro, $i+2$ Gly usually, but Pro can fit Type I $i+2$). Most hydrophobic and bulky residues are strongly disfavored.
Positions $i$ and $i+3$: The Flanking Residues
Levitt’s analysis showed these positions are less conformationally restricted. Their propensity values hover closer to 1.0 (random expectation). Even so, Proline shows a notable preference for position $i$ (preceding the turn) because its fixed $\phi$ angle pre-organizes the chain for the turn initiation. Large hydrophobic residues are often found at $i$ and $i+3$ in "hydrophobic staple" turns, where side chains pack against each other to stabilize the fold.
Structural Basis for the Propensities
The Levitt values are not arbitrary statistics; they are direct readouts of steric exclusion and hydrogen bonding potential governed by Ramachandran plot constraints.
1. The Ramachandran Constraint
The $\phi/\psi$ angles required for the central residues of a beta turn ($i+1, i+2$) often fall in regions of the Ramachandran plot that are sterically inaccessible to residues with $C_\beta$ atoms Small thing, real impact. Simple as that..
- Type II Turns: Require residue $i+2$ to have $\phi \approx +80^\circ$. Only Glycine can do this easily.
- Type I Turns: Require $i+1$ at $\phi \approx -60^\circ, \psi \approx -
