PubMed1994
We show that quantitative analysis of replicated, full-factorial crystal growth experiments and, by implication, similar studies of a wide variety of other phenomena, can be a powerful tool for analyzing macromolecular systems with complex, interacting dependencies on functionally significant factors. Bacillus stearothermophilus tryptophanyl-tRNA synthetase crystallizes in three different crystal forms depending on the ligands present under otherwise identical conditions. Comparison of crystallographic space groups for complexes with different ligands reveals that the three forms entail at least two very different families of packing arrangements that are correlated with specific changes in the enzyme ligation state. One is associated with the ligand-free enzyme, substrate ligands, and the binding of the activated amino acid; the other results from the presence of high ATP concentrations and/or the synthesis of the unusual acyl-transfer product, tryptophanyl-2'(3') ATP. Together with previous physico-chemical studies of aminoacyl-tRNA synthetases, these observations suggest that the two families are related, respectively, to the biochemical processes of amino acid activation and acyl transfer. Further evidence that the crystal polymorphism results from an underlying protein conformational polymorphism has now been obtained by quantitative analysis of how crystal growth depends on pH and the substrates tryptophan and ATP. The analysis consists first in showing that crystallization conditions for the unliganded protein are very favorable, suggesting that variation in crystal growth induced by pH and substrates under otherwise identical conditions is due to their effects on the protein conformation and not on incidental perturbations of crystal growth, per se. Next, crystal growth experiments are shown to be reproducible enough to support statistical analysis of quantitative scores assigned to the results. Finally, the observed variation in scores can be attributed at high confidence levels chiefly to three effects: that of pH alone, the synergistic effects of pH plus tryptophan, and of tryptophan plus ATP. These statistical inferences are consistent with other biochemical data, and support the conclusions based on crystal packing that representative stages of the enzyme mechanism have been trapped in the different crystal forms. The pH-tryptophan interaction implies that there is a pH-dependent conformational change favoring high affinity substrate binding at high pH. The pH-ATP interaction implies that a subsequent conformational change, not previously considered, occurs between tryptophan activation and acyl transfer.