Chemogenomic screens were performed in both budding and fission yeasts, allowing for a cross-species comparison of drug–gene interaction networks.Drug–module interactions were more conserved than individual drug–gene interactions.Combination of data from both species can improve drug–module predictions and helps identify a compound's mode of action.

To test whether the structure of a protein is determined in a manner akin to the assembly of a jigsaw puzzle, up to 10 adjacent residues within the core of T4 lysozyme were replaced by methionine. Such variants are active and fold cooperatively with progressively reduced stability. The structure of a seven-methionine variant has been shown, crystallographically, to be similar to wild type and to maintain a well ordered core. The interaction between the core residues is, therefore, not strictly comparable with the precise spatial complementarity of the pieces of a jigsaw puzzle. Rather, a certain amount of give and take in forming the core structure is permitted. A simplified hydrophobic core sequence, imposed without genetic selection or computer-based design, is sufficient to retain native properties in a globular protein.The cores of globular proteins consist of buried, primarily hydrophobic, amino acids. Tight packing of the amino acid side chains (1) has led to the idea that the size and shape of the nonpolar amino acids within the core may constrain or define the overall protein fold (2, 3). This "jigsaw puzzle" model of protein folding was originally introduced by Crick (4) as a "knobs into holes" description of a-helix packing and has been elaborated by Chothia et al. (5), and by Alber and co-workers (6). Here the jigsaw puzzle model refers to shape complementarity (3), not to the pathway of folding (7). The model is supported by the observation that changes in the sizes and shapes of residues within the cores of proteins are usually destabilizing (8-10). Also in support of the model, the structures of a-helical coiled coils appear to be determined by the shapes of the buried side chains (6). In contrast with this view, it has been shown that alternative core sequences that lead to viable proteins could be selected by random mutagenesis for both A-repressor (11) and T4 lysozyme (12), among others (13,14). It is possible, however, that a limited number of combinations of amino acids are viable and that they are the ones identified by the mutagenic selection. Here we explore an approach in which there is no selection other than the sites of substitution. MATERIALS AND METHODSWe chose methionine as a generic core-replacement residue for a combination of reasons. First, a methionine side chain occupies roughly the same volume as the frequently observed core residues leucine, isoleucine, and phenylalanine. It is, however, more flexible and can more readily adapt to occupy whatever space might be available. In this sense methionine contrasts with the rigid, predetermined shape of a piece of a jigsaw puzzle. Methionine also occurs relatively infrequently in known proteins (15). Thus multiple methionine substitutions would be expected to substantially change the composition of the core. Finally, we wondered if the intro-

In order to determine the thermodynamic cost of introducing a polar group within the core of a protein, a series of nine Ala-->Ser and 3 Val-->Thr substitutions was constructed in T4 lysozyme. The sites were all within alpha-helices but ranged from fully solvent-exposed to totally buried. The range of destabilization incurred by the Ala-->Ser substitutions was found to be very similar to that for the Val-->Thr replacements. For the solvent-exposed and partly exposed sites the destabilization was modest (approximately less than 0.5 kcal/mol). For the completely buried sites the destabilization was larger, but variable (approximately 1-3 kcal/mol). Crystal structure determinations showed that the Ala-->Ser mutant structures were, in general, very similar to their wild-type counterparts, even though the replacements introduce a hydroxyl group. This is in part because the introduced serines are all within alpha-helices and at congested sites can avoid steric clashes with surrounding atoms by making a hydrogen bond to a backbone carbonyl oxygen in the preceding turn of the helix. The three substituted threonine side chains essentially superimpose on their valine counterparts but display somewhat larger conformational adjustments. The results illustrate how a protein structure will adapt in different ways to avoid the presence of an unsatisfied hydrogen bond donor or acceptor. In the most extreme case, Val 149-->Thr, which is also the most destabilizing variant (delta delta G = 2.8 kcal/mol), a water molecule is incorporated in the mutant structure in order to provide a hydrogen-bonding partner. The results are consistent with the view that many hydrogen bonds within proteins contribute only marginally to stability but that noncharged polar groups that lack a hydrogen-bonding partner are very destabilizing (delta delta G approximately greater than 3 kcal/mol). Supportive of other studies, the alpha-helix propensity of alanine is seen to be higher than that of serine (delta delta G = 0.46 +/- 0.04 kcal/mol), while threonine and valine are similar in alpha-helix propensity.

No abstract