Protein sequence activity relationships

Protein engineering based on sequence-activity relationships is attractive because of the small numbers of variants that need to be tested in order to obtain improvement. In principle this frees the investigator from the tyranny of high throughput screening. It also allows one to design and create the most informative set of variants, rather than relying upon a stochastically constructed library (i-6). 

Figure L Protein engineering process using sequence activity relationships.

For each of the proteinase K activities tested, we used partial least squares regression (PLSR) to model the relationship between amino acid variation and the variations in proteinase activities (the sequence-activity relationship). The application of these methods to nucleic acids, peptides and proteins has been described previously (7, II, 12, 28-30)... 

We have described a method for engineering proteins based on design, synthesis and testing of small numbers of iiulividual variants followed by mathematical modeling to determine a sequence-activity relationship. We have also shown that sequence-activity models can be used predictively to design improved variants. 

Historically the Shaker (Sh) K channel was the first K channel which was cloned and characterized [6-10]. Subsequently many more channel cDNAs and genes have been isolated and studied. Yet Sh channels remained in the forefront of channel research. The study of Sh channel mutants has provided the most thorough insight into structure-function relationships of K channels to date. I will first discuss in this chapter the primary sequences of voltage-gated channels. I will only use a few selected examples for discussion. As of this time, so many related K channel protein sequences have been published that it is not feasible to discuss all of them. Subsequently, I will describe in detail the present knowledge about functional K" " channel domains which are implicated in activation, inactivation and selectivity of the channel. 

The concept of the similarity of molecules has important ramifications for physical, chemical, and biological systems. Grunwald (7) has recently pointed out the constraints of molecular similarity on linear free energy relations and observed that Their accuracy depends upon the quality of the molecular similarity. The use of quantitative structure-activity relationships (2-6) is based on the assumption that similar molecules have similar properties. Herein we present a general and rigorous definition of molecular structural similarity. Previous research in this field has usually been concerned with sequence comparisons of macromolecules, primarily proteins and nucleic acids (7-9). In addition, there have appeared a number of ad hoc definitions of molecular similarity (10-15), many of which are subsumed in the present work. Difficulties associated with attempting to obtain precise numerical indices for qualitative molecular structural concepts have already been extensively discussed in the literature and will not be reviewed here. 

The idea that a small molecule, active with a particular target protein, is very likely to be active also with a sequence-related protein is by no means new. Over several decades medicinal chemists have acquired valuable experience as to how to systematically explore chemistry space around a given lead structure, how to establish structure-activity relationships, and how to use such knowledge for refining the selectivity profile of the drug candidate. Often, selectivity with a related protein can be achieved by relatively small modifications to the original small molecule s structure. 

---