The server attempts to identify a small set of amino acid residues in a query protein that have a high probability of being buried (side-chain accessibilities less than 5%, expressed in terms of residue depth). The server suggests substitutions at these buried positions that are most likely to result in a temperature sensitive (Ts) phenotype.
The Ts phenotype has been shown to correlate with decreased protein stability and reduced levels of the protein, in vivo.
It has also been shown that substitutions of buried hydrophobic residues often result in significant destabilization of the protein, often much larger changes in protein stability than mutations at surface positions. Hence, our approach to predict substutions that result in a Temperature-sensitive mutant is to predict positions of hydrophobic residues in the protein that are likely to be buried.
Several cases of substitution of buried residue positions have been shown to result in a Ts phenotype, for example in the case of T4 lysozyme, and gene V protein.
The Rose Hydrophobicity scale is chosen to quantify the hydrophobicity of amino acid residues, as it most closely correlates with the degree of residue burial. In this study, the hydrophobicity values of the scale were chosen to be equal to the average extent of burial of the residue in the training set, i.e.
7 types of amino acid residues with rescaled hydrophobicity greater than 80, namely Cys, Phe, Ile, Val, Trp, Met, and Leu are defined as hydrophobic residues in this study. As Cys could be involved in disulfide bonds or metal ion coordination, they are not included in the prediction candidates.
| Amino Acid | Hydrophobicity | Amino Acid | Hydrophobicity | Amino Acid | Hydrophobicity | Amino Acid | Hydrophobicity |
|---|---|---|---|---|---|---|---|
| Cys | 100 | Met | 85 | Gly | 51 | Asn | 28 |
| Phe | 92 | Leu | 85 | Thr | 46 | Gln | 26 |
| Ile | 92 | His | 67 | Ser | 36 | Glu | 26 |
| Val | 87 | Tyr | 62 | Arg | 31 | Asp | 26 |
| Trp | 85 | Ala | 56 | Pro | 31 | Lys | 0 |
Burial is quantified using two parameters: 1) average hydrophobicity and 2) Hydrophobic moment.
Average Hydrophobicity of a residue (averaged over a seven residue window) is given by:
where the H(n)s are the rescaled individual residue hydrophobicities listed above.
The hydrophobic moment, Hmom is calculated over a nine residue window as follows:
where δ is the phase angle and is dependent on the periodicity of the secondary structure that the sequence is assumed to adopt:
δ(Α-helix) = 100°.
δ(flat β-sheet) = 180°
δ(curved β-sheet) = 160°
Hydrophobic moment is introduced because both helices and β strands often have one solvent exposed hydrophilic face and one buried hydrophobic face. Buried regions of such sequences therefore cannot be identified using only Hav. In contrast, they can be indentified by average Hav and high Hmom values.
The following prediction rules were generated by large scale analysis of data (Varadarajan et al, 1996).
| Burial Prediction Confidence Level(%) | Prediction criteria |
|---|---|
| >=95 | Residue, as well as both flanking residues, are hydrophobic and Hav >= 75 |
| >=90 |
|
| >=80 | Residue is hydrophobic and any of the following conditions are met:
|
The free energies of unfolding of typical globular proteins are in the range of 5 - 15 kcal/mol at room temperature. A Ts mutation should destabilize the protein by an amount that is an appreciable fraction of the free energy of folding at the nonpermissive temperature.
The exact amount of destabilization produced by a mutation will depend on the effect of the mutation on ΔG, ΔH, and ΔCp. In general, these will not be known for the protein of interest. It is therefore desirable to make both conservative and nonconservative substitutions at predicted buried sites, so that at least one of these will result in a Ts phenotype.
Our approach is therefore to suggest five different substitutions at each predicted buried site that differ in the stereochemistry and polarity of the substituted residue. These substitutions span a wide range of free energy and, we assume that at least one of these substitutions would destabilize the protein to an extent appropriate for a Ts phenotype.
One effective set of stereochemically diverse set of residues was found to be {Ala, Trp, Gln, Asp, Pro}
Server pipelineThe server uses sequence and structure information to make predictions of temperature sensitive mutant positions. When structures are input, the server computes residue depth and uses this in conjunction with the sequence-based burial prediction computations. When only sequences are input, structural information is inferred from a homology model (constructed using stringent modeling criteria). If it is not possible to construct a homology model, the predictions are based solely on the sequence based computations. |
|
