Structural basis for the aminoacid composition of proteins from halophilic archea

Abstract

Proteins from halophilic organisms, which live in extreme saline conditions, have evolved to remain folded at very high ionic strengths. The surfaces of halophilic proteins show a biased amino acid composition with a high prevalence of aspartic and glutamic acids, a low frequency of lysine, and a high occurrence of amino acids with a low hydrophobic character. Using extensive mutational studies on the protein surfaces, we show that it is possible to decrease the salt dependence of a typical halophilic protein to the level of a mesophilic form and engineer a protein from a mesophilic organism into an obligate halophilic form. NMR studies demonstrate complete preservation of the three-dimensional structure of extreme mutants and confirm that salt dependency is conferred exclusively by surface residues. In spite of the statistically established fact that most halophilic proteins are strongly acidic, analysis of a very large number of mutants showed that the effect of salt on protein stability is largely independent of the total protein charge. Conversely, we quantitatively demonstrate that halophilicity is directly related to a decrease in the accessible surface area.

Figure 1. The effect of charge and side chain length in protein haloadaptation.

The effect of the chain size (panels A to F) and charge (panels G to L) in haloadaptation of ProtL, Hv 1ALigN, and Ec 1ALigN was investigated by measuring the free energy at 3.2 M salt (KCl or NaCl) versus the number of substituted residues in the multiple mutations (n in XYxnWZ). In addition, the variation in protein stability induced by salt (m_KCl or m_NaCl) is also reported. For each panel, proteins and mutation classes are specified in the enclosed legend. Δe⁻ is defined as the residual theoretical charge upon mutation (mutant minus wild type). Error bars result from propagation of the experimental uncertainties in the T_m values, by Montecarlo analysis. Dashed lines represent the or the m_salt values for wild type proteins. A negative m_salt value means that the cosolute destabilizes the protein.

Figure 2. The low prevalence of lysines in the amino acid composition of halophilic proteins.

The effect of lysines on protein haloadaptation was investigated by measuring the change in stability induced by salt (, m_KCl, and m_NaCl) versus the number of substituted residues in the lysine involving mutations (n in XYxnWZ). For each panel, proteins and mutation classes are specified in the enclosed legend. Error bars result from propagation of the experimental uncertainties in the T_m values, by Montecarlo analysis. Dashed and dotted lines represent the or the m_NaCl and m_KCl for wild type ProtL, Hv 1A LigN, and Ec 1ALigN. A negative m_salt value means that the cosolute destabilizes the protein.

Figure 3. Structural characterization of the KxnQE mutants.

Using NMR spectroscopy, the high-resolution structures of ProtL Kx5Q and ProtL Kx6E were obtained. The 10 lowest energy-refined conformers are shown in (A). The mutated residues are highlighted in yellow or red for K to Q and K to E mutations, respectively. The alignment between the mutant structures and wild type ProtL (1hz6) , shown in (B), reveals that the changes introduced upon mutation are minimal. The spatial distribution of the mutated side chains is generally very similar to the wild type. (C) shows close-up views for a representative selection of wild type and mutant side chains that have been aligned. Lysine side chains are colored in magenta whereas glutamine and glutamate side chains are colored in pink and blue, respectively.

Figure 4. A metric for ProtL haloadaptation.

(A) The mesophilic ProtL can be converted from a folded protein that does not become stabilized by salt (ΔG^U−F>0, m_NaCl = 0, wild type) to an obligate halophile (ΔG^U−F<0, m_NaCl>0, Kx7E). This transition is a continuous process as demonstrated by the empirical correlation found between the m_NaCl and ΔASA^WT−Mut for the ProtL mutants: K for Q (black squares) and K for E (black circles). This data plot uses the left ordinate axis for the m_NaCl units. (B) An equivalent correlation can also be found for the D for E and E for D mutants Hv 1ALigN (black circles) and Ec 1ALigN (black squares). Error bars result from propagation of the experimental uncertainties in the T_m values, by Montecarlo analysis. The solid black line corresponds to the best linear fit of the data. In (B), black line corresponds to the linear fit including Hv and Ec 1ALigN whereas the fit of Hv 1ALigN is depicted with a blue line. For each protein, the stability upon unfolding at 298 K is shown in gray circles (units on the right ordinate axis). Error bars reflect the mean value from all the experimental measurements. Dashed gray lines highlight the ΔG^U−F values for wild type ProtL (ΔG^U−F (WT)) and when ΔG^U−F = 0, as indicated.

Figure 5. Salt-dependent folding of ProtL Kx7E.

In the absence of salt the ¹H,¹⁵N-HSQC spectrum for ProtL Kx7E shows a peak pattern typical of an unfolded protein, with a small peak dispersion in the proton dimension (A). The similarity in the transversal relaxation times (T ₂(¹⁵ N)) for the main chain (NH^bb, red) and the side chain (, blue) nitrogen nuclei (inset of panel A) indicates fast tumbling of the polypeptide chain, a feature of unfolded proteins. Upon addition of salt (2 M NaCl), the peak pattern in the spectrum expands (B) and the T ₂(¹⁵ N) of the main chain and the side chains can be discriminated (inset of panel B), as expected for a folded protein of this size. The same trend is manifested in the peak intensities of the HSQC spectra and the traces for the considered peaks are also shown in the bottom right-hand corners of the spectra. Solid lines in the figure insets correspond to the best exponential fitting to the experimental data (filled circles).