Energetics of aliphatic deletions in protein cores

Abstract

Although core residues can sometimes be replaced by shorter ones without introducing significant changes in protein structure, the energetic consequences are typically large and destabilizing. Many efforts have been devoted to understand and predict changes in stability from analysis of the environment of mutated residues, but the relationships proposed for individual proteins have often failed to describe additional data. We report here 17 apoflavodoxin large-to-small mutations that cause overall protein destabilizations of 0.6-3.9 kcal.mol(-1). By comparing two-state urea and three-state thermal unfolding data, the overall destabilizations observed are partitioned into effects on the N-to-I and on the I-to-U equilibria. In all cases, the equilibrium intermediate exerts a "buffering" effect that reduces the impact of the overall destabilization on the N-to-I equilibrium. The performance of several structure-energetics relationships, proposed to explain the energetics of hydrophobic shortening mutations, has been evaluated by using an apoflavodoxin data set consisting of 14 mutations involving branching-conservative aliphatic side-chain shortenings and a larger data set, including similar mutations implemented in seven model proteins. Our analysis shows that the stability changes observed for any of the different types of mutations (LA, IA, IV, and VA) in either data set are best explained by a combination of differential hydrophobicity and of the calculated volume of the modeled cavity (as previously observed for LA and IA mutations in lysozyme T4). In contrast, sequence conservation within the flavodoxin family, which is a good predictor for charge-reversal stabilizing mutations, does not perform so well for aliphatic shortening ones.

Figure 1.

Cα trace of Anabaena apoflavodoxin (PDB code 1FTG) showing, in green, the mutated side-chains.

Figure 2.

Circular dichroism spectra of wild type and some of the mutants analyzed (see Campos et al. 2004a for the rest of the mutants) at 25.0° ± 0.1°C. (A) Near-UV, in 50 mM MOPS (pH 7.0). (B) Far-UV, in 5 mM MOPS (pH 7.0), with 15 mM NaCl. Signals of the W66F mutant (and derived) are shown normalized, for the sake of comparison.

Figure 3.

Urea denaturation curves of wild-type and mutant apoflavodoxins at 25.0° ± 0.1°C (pH 7.0), 50 mM MOPS. The fluorescence signal of the W66F mutant is shown normalized, for the sake of comparison.

Figure 4.

Comparison of ΔΔG_NU values obtained for wild-type apoflavodoxin and mutants thereof by adding the ΔΔG_NI and ΔΔG_IU terms calculated from global analysis of three-state thermal unfolding curves (Table 3) with those directly measured from two-state urea denaturation (Table 1). The filled circle represents data from mutant I156V, excluded for the correlation due to possible structural relaxations (see text). The dotted line is a unity fit.

Figure 5.

Partitioning of the global destabilization (N-to-U equilibrium) associated to the mutations implemented into the N-to-I equilibrium, relevant destabilization (gray bar), and the I-to-U equilibrium, residual destabilization (white hatched bar).

Figure 6.

Apoflavodoxin secondary structure cartoon showing new (continuous thick lines) and previously reported (Campos et al. 2004a) (discontinuous thin lines) ϕ-values reporting on the integrity of native side-chain interactions in the equilibrium thermal unfolding intermediate at 317.3 K. The native-like region of the intermediate is formed by the packing of helices 1–5, and strands 1–4 (where the ϕ-values are >0.6) and possibly strand 5a. In contrast, the long loop splitting strands 5a and 5b (which contains a small three-stranded b-sheet comprising strands 6, 7, and 8), and two additional loops, are markedly weakened (all ϕ-values <0.4) (Campos et al. 2004a). The ϕ-values are colored from dark green (native interactions) to red (lost interactions). ID numbers of helices (circles) and strands (rectangles) are marked in light gray, and the positions analyzed in this work are indicated in black.

Figure 7.

Plots of ΔΔG (wild-type apoflavodoxin free energy of unfolding minus that of mutants) vs. different parameters that describe the environment of the mutated residues. (A) Number of Cα groups within a sphere of 10 Å radius centered in the Cα of the mutated residue, (B) differences in side-chain solvent-accessible area buried between wild type and mutant, (C) number of methyl or methylene side-chain groups surrounding (within a sphere of 6 Å radius) the methyl or methylene group deleted upon mutation, (D) change of contacts (calculated as contact parameter n_H) relative to the wild-type protein, (E) thermal factors of the mutated residues, as extracted from the PDB file 1FTG, and (F) cavity volume created by the amino acid substitution.

Figure 8.

Plots of ΔΔG of unfolding vs. different parameters that describe the surrounding environment of mutated residues in seven different proteins: apoflavodoxin, ROP (Vlassi et al. 1999), barnase (Serrano et al. 1992), T4 lysozyme (Eriksson et al. 1992a; Xu et al. 1998), staphylococcal nuclease (Shortle et al. 1990), CI2 (Otzen and Fersht 1995), and human lysozyme (Takano et al. 1995, 1997). (A) Number of Cα groups within a sphere of 10 Å radius centered in the Cα of the mutated residue; (B) differences in side-chain solvent-accessible area buried between wild type and mutant; (C) number of methyl or methylene side-chain groups surrounding (within a sphere of 6 Å radius) the methyl or methylene group deleted upon mutation; (D) change of contacts (calculated as contact parameter n_H) relative to the wild-type protein; (E) thermal factors of the mutated residues as extracted from the PDB file of the wild-type protein; (F) cavity volume created by the amino acid substitution.

Figure 9.

Correlation between transfer energies and interceptors of linear relationships between cavity volume and experimental ΔΔG for different types of mutations (Leu to Ala, Val to Ala, Ile to Val, and Ile to Ala) implemented in any of the seven proteins of the data set (see Materials and Methods). The dashed line shows a unity fit.

Figure 10.

The experimentally determined ΔΔG values shown in Figure 8F have been brought to a common reference state (hydrophobicity-independent) by subtraction of the intercepts of mutation-type separated plots of ΔΔG_NU vs. volume (see Discussion).