The Role of Aromatic-Aromatic Interactions in Strand-Strand Stabilization of β-Sheets

Abstract

Aromatic-aromatic interactions have long been believed to play key roles in protein structure, folding, and binding functions. However, we still lack full understanding of the contributions of aromatic-aromatic interactions to protein stability and the timing of their formation during folding. Here, using an aromatic ladder in the β-barrel protein, cellular retinoic acid-binding protein 1 (CRABP1), as a case study, we find that aromatic π stacking plays a greater role in the Phe65-Phe71 cross-strand pair, while in another pair, Phe50-Phe65, hydrophobic interactions are dominant. The Phe65-Phe71 pair spans β-strands 4 and 5 in the β-barrel, which lack interstrand hydrogen bonding, and we speculate that it compensates energetically for the absence of strand-strand backbone interactions. Using perturbation analysis, we find that both aromatic-aromatic pairs form after the transition state for folding of CRABP1, thus playing a role in the final stabilization of the β-sheet rather than in its nucleation as had been earlier proposed. The aromatic interaction between strands 4 and 5 in CRABP1 is highly conserved in the intracellular lipid-binding protein (iLBP) family, and several lines of evidence combine to support a model wherein it acts to maintain barrel structure while allowing the dynamic opening that is necessary for ligand entry. Lastly, we carried out a bioinformatics analysis and found 51 examples of aromatic-aromatic interactions across non-hydrogen-bonded β-strands outside the iLBPs, arguing for the generality of the role played by this structural motif.

Keywords: ASA; COM; CRABP1; FABP; H-bond; HSQC; IFABP; ILBP; MD; MSA; PDB; Protein Data Bank; TS; WT; accessible surface area; cellular retinoic acid-binding protein 1; center of mass; dynamics; fatty acid-binding protein; folding; heteronuclear single quantum coherence; hydrogen bond; iLBP; ileal lipid-binding protein; intestinal fatty acid-binding protein; intracellular lipid-binding protein; molecular dynamics; multiple sequence alignment; transition state; wild type; β-barrel; π stacking.

Fig. 1. Conserved Phe ladder in CRABP1, equilibrium stabilities of CRABP1 Phe variants, and MD-simulated geometries of the corresponding aromatic pairs

(a) Ribbon representation of mouse holo-CRABP1 X-ray crystal structure (PDB ID 1CBR) with strands 3, 4 and 5 labeled and F50, F65 and F71 constituting the cross-strand parallel-stacked aromatic ladder shown in spacefill and highlighted in orange. (b) Representative curves for urea-induced equilibrium denaturation of WT CRABP1 and the F50M, F65M and F71M variants. The solid lines on equilibrium denaturation plots are fits of the data to a two-state model using a fixed equilibrium m-value of −2.0 kcal/mol·M. (c) COM-COM distance (R_COM) and (c) dihedral angle (ϕ) variation within the F50-F65 (black) and the F65-F71 pairs (red) extracted from an 8-ns MD trajectory of WT CRABP1.

Fig. 2. Unfolding and refolding kinetics of CRABP1 Phe variants

(a) Dependence of the logarithm of the unfolding rates on [urea] of WT CRABP1 and the F50M, F65M and F71M variants. The solid lines on ln k_U plots represent linear fits. (b) Dependence of the logarithms of the rates of I₂ (fast phase, in squares) and N (slow phase, in triangles) formation (fast and medium kinetic phases, correspondingly) on [urea] for WT CRABP1 and the F50M, F65M and F71M variants. Colors are as in (a).

Fig. 3. Chemical shift perturbation analysis of CRABP1 Phe variants

(a) Histograms of chemical shift perturbation, Δδ, between WT CRABP1 and the F50M (top panel), F65M (middle panel) and F71M (bottom panel) variants. Bars corresponding to the residues showing significant perturbation are colored in red (Δδ > 0.1 ppm, large changes) and yellow (Δδ <= 0.1 ppm, moderate changes); those with insignificant perturbations are shown in gray. (b) The magnitude of Δδ for the F50M, F65M and F71M variants mapped onto the holo-CRABP1 structure (PDB ID 1CBR). Residues with large and moderate Δδs are colored in red and yellow, unperturbed residues are shown in gray (same as in (a)); unassigned residues are colored in black.

Fig. 4. Conformational flexibility of CRABP1 strands 3, 4 and 5

(a) Interstrand distance variation between strands and 4 and 5 in CRABP1 extracted from an 8-ns MD trajectory of WT CRABP1 (black trace) and the F71M variant (red trace). The interstrand distance was defined as the distance between Cα atoms of T61 (strand 4) and E74 (strand 5), both shown as spheres in (b). (b) Overlay of the structures after 1.5 ns of MD simulation: WT CRABP1 (gray) and the F71M variant (red). The Cα atoms of T61 (strand 4) and E74 (strand 5) are shown as spheres. (c) Overlay of the X-ray structures of the holo-CRABP1 (PDB ID 1CBR, chain A (gray)) and apo-CRABP1 (PDB ID 1CBI, chains A (green) and B (blue)).

Fig. 5. The conserved aromatic ladder in iLBPs

Pie diagram representing the different arrangements of the cross-strand aromatic ladder and their respective occurrences in the representative set of iLBP structures acquired from the PDB. Black sector corresponds to the iLBPs having no aromatic ladder.

Fig. 6. Frequency of cross-strand aromatic pairs

Distribution of the ratio between the number of neighboring strands carrying a cross-strand aromatic pair over the total number of neighboring strands at a given interstrand distance (step size 0.2 Å). Black bars represent interstrand distances with good statistics (both numbers are > 5); gray bars represent the ones with poor statistics (either nominator or denominator is ≤5).