A salient feature of eukaryotic proteins is their modular construction. Proteins can gain new functionalities by incorporating new modules. The study of domain composition may suggest hypotheses regarding a protein function(s), and thus it has become desirable to define the function of individual protein domain modules. This is particularly important for SAM (sterile alpha motif) domains, which are among the most common protein modules found in eukaryotic cells. In contrast to many other well-characterized protein-protein interaction modules, SAM domains have considerably more diverse interaction modes. Under the hypothesis that learning module functions can provide biological insights, we propose to assign functions to thousands of uncharacterized SAM domains by a combination of bioinformatic and proteomic techniques.
Aim 1 : Expand our predictions of polymeric and non-polymeric SAM domains. Our first publication partly addressed Aim 1. We computationally identified those SAMs likely to be polymeric by calculating their interaction energy when threaded onto known polymeric structures. We found 694 likely polymers, including SAM domains from the proteins Lethal Malignant Brain Tumor, Bicaudal-C, Liprin-beta, Adenylate Cyclase, and Atherin. About half of all known SAM domains could not be evaluated, however, because of low homology to known SAM structures. As part of our continuing studies, we will expand the number of predictions by including a large number of recently determined structures.
Aim 2 : Experimentally validate and characterize predicted SAM polymers. We will first test whether the predicted polymers actually form polymers by examining the purified SAM domains by EM. We then will further investigate their function by creating mutants that block polymerization to see whether their known functions are interrupted.
Aim 3 : Assign functions to non-polymerizing SAM domains. A common alternative function is hetero- oligomerization with other SAM domains. Thus, for those SAMs that don't polymerize we will investigate whether they bind to one another instead, thereby developing a SAM domain interactome. This will be accomplished by means of yeast 2-hybrid screens. Another class of SAM domains bind to nucleic acids. We will identify these computationally, by looking for nucleic acid binding site features. The methodology outlined here to investigate global SAM domain functions is likely to be widely applicable to other domain types found in a variety of other proteins. Further, the discovery new SAM functions may help identify new drug targets for different diseases.