How similar are protein folding and protein binding nuclei? Examination of vibrational motions of energy hot spots and conserved residues

Abstract

The underlying physico-chemical principles of the interactions between domains in protein folding are similar to those between protein molecules in binding. Here we show that conserved residues and experimental hot spots at intermolecular binding interfaces overlap residues that vibrate with high frequencies. Similarly, conserved residues and hot spots are found in protein cores and are also observed to vibrate with high frequencies. In both cases, these residues contribute significantly to the stability. Hence, these observations validate the proposition that binding and folding are similar processes. In both packing plays a critical role, rationalizing the residue conservation and the experimental alanine scanning hot spots. We further show that high-frequency vibrating residues distinguish between protein binding sites and the remainder of the protein surface.

FIGURE 1

(a) The distribution of the mean-square vibrations in the fastest four modes of dynamics of the monomer A structure of a serine protease, from the complex serine protease-inhibitor (1tfx), which is the representative structure of an interface cluster. The number of the HFV residues is 26 (0.12 × 223; 223 is the residue number), with the parameters 0.005 and 4 for the lower threshold in the height of the peaks and the number of fast modes incorporated, respectively. The blue dots display the interacting and the nearby conserved residues (6). (b) The ribbon diagram of 1tfx with the high-frequency fluctuating residues grouped into one cluster are depicted in green on monomer A drawn together with the inhibitor C. (c) The vibrations in the fastest 10 modes of a ribonuclease inhibitor A from its complex with angiogenin (PDB, 1a4y). The binding hot spot residues from alanine scanning data, ASEdb (Thorn and Bogan, 2001), are marked with red dots. Blue circles display those residues reported in the literature as involved in hydrogen bond interactions and forming contacts at the interface (Tables II and III in Papageorgiou et al., 1997). The correspondence between the peaks and the red dots and blue circles is remarkable.

FIGURE 2

The vibrations in the fastest five modes of barnase, 1brsD, from its complex structure with barstar, 1brs. The folding core residues of the wild-type barnase from hydrogen exchange experiments (Perrett et al., 1995) are marked with black dots.

FIGURE 3

The number of conserved residues overlapping and not overlapping the HFV residues for 90 cases. A distance in space and in sequence is allowed for the comparison, up to three residues along the sequence and 7 Å in space. The outliers: 1g1kA, B (structural protein; with another binding region); 1irxA, B (ligase; a lower threshold value <0.005 of the high-frequency peaks is needed to be able to identify the HFV residues); 1j46A (oxyreductase; a lower threshold required for the height of the peaks); 1pmaA, B (protease; multiple interfaces); 1dubA, B (lyase; multiple interfaces); 1fntC (hydrolase activator; multiple interfaces); 1dz4A (oxyreductase; two clusters at other regions on the surface); 1fpuA (transferase; one large folding core and a cluster somewhere else on the surface). The outliers are depicted by ×. The number of cases below the line is 14.

FIGURE 4

(a) The average distribution of the distances of the closest 15 α-carbon or side-chains centers of the HFV residues to a conserved residue by GNM and by random sampling. The distances are considered at intervals of 1 Å. The number of sites, 15, may represent ∼7.5 residues, which is close to the average size of a HFV residue cluster. The distribution of the HFV residues centered at the first and the second coordination shell of a conserved residue in native and random packing, respectively. (b) The position of the peak of the average distribution of the distances of the closest centers to a conserved residue versus the number of the closest centers considered. The closest centers are the α-carbons or the side chains of the high-frequency vibrating residues by GNM and by random sampling. The peaks of the distributions of the latter centers level off at a distance ∼6.5 and ∼10.5 Å, respectively, as the number of the closest centers is increased.

FIGURE 5

The number of HFV residues contacting with the interface residues versus contacting with the rest of the surface residues for 100 cases. It is normalized by the number of interface residues and the number of residues in the rest of the surface, respectively. The width of the shell and the distance criteria for the interaction are taken as 7 Å; n is used to represent the number of the respective cases. The outliers: 1is7L (hydrolase/protein binding; no conservation neither in interacting nor in nearby residues); 1irxA, B (ligase; a lower threshold value <0.005 of high-frequency peaks is needed to be able to identify HFV residues); 1j46A (oxyreductase; lower threshold required for the fast-mode peaks); 1pmaA, B (protease; multiple interfaces); 1fntC (hydrolase activator; many interfaces); 1dz4A (oxyreductase; two clusters at other binding regions); 1fpuA, B (transferase; a large folding core and a cluster of residues somewhere else on the surface). The outliers are depicted by ×. The number of cases off the diagonal on the space of the noninterface surface residues is 25.

FIGURE 6

(a) The vibrations of the residues in the fastest six modes for the native monomer structure glutathione S-transferase, 1b8x (red), and the monomer A from its complex structure, 1c72 (dashed black). The interacting and nearby conserved residues and the residues of EPNP site are marked with blue dots and blue circles, respectively. The surface residues are labeled by blue squares. (b) The ribbon diagrams of the isolated monomer and the complex structures on which the HFV residues from panel a are marked. Red, blue, green, and yellow represent four clusters of the HFV residues on 1b8x (A) and red, blue, and green represent the three clusters of the HFV residues on 1c72A (B). The number of the HFV residues identified corresponds to ∼13% of the protein size. The root mean-square deviation between the two structures is 8 Å (on 212 residues; considering C^α-atoms).

Figure