A common problem in biological research is being faced with a protein that is difficult to subdivide into functional domains. The problem is intensified if the essential region of the protein is large and/or sequence homology to known polypeptides is lacking. Targeted mutagenesis strategies may be cumbersome or prohibitive under these conditions. A few years ago, we confronted all of the above difficulties while attempting to separate functions within the essential ~300 amino acid N-terminus of Gcr1p. We reasoned that random mutagenesis (we used error-prone PCR), followed by selection for fully functional variants, would identify the equivalent of conserved and non-conserved subsections of the N-terminal region. We recovered 24 functional alleles of the GCR1 gene, based on the ability of each to complement the severe glucose non-responsive phenotype resulting from the removal of Gcr1. The alignment among these variants is similar to one found for a group of homologous proteins from different organisms. Since in our approach they are selected in the same organism (ie., in an isogenic background), we call the method, "unigenic evolution" (Deminoff et al. 1995).

The 24 variants contained a total of 315 nucleotide substitutions, 200 of which were missense mutations. As expected, certain subregions were either over- or under-represented for missense substitutions. However, the distribution could have been due to hot or cold spots, respectively, in the template for mutagenesis. We therefore normalized to the background of silent mutations, which contribute to the total density of nucleotide substitutions but have no effect on the protein. We fine-tuned the normalization to account for the fact that different codons produce a different spectrum of missense and silent mutations, given all possible single-nucleotide changes; ordinarily, stop codons would not occur in a functional allele. We then plotted the expected vs. the observed ratio of missense to total mutations on average for a moving 20-codon window. The resulting bar graph (see Figure below) displays regions that are either hypo- or hyper- mutable, depending on normalized frequency of missense mutations. We used the chi-square test to assess the statistical significance of these data. We also found good agreement between unigenic evolution analysis and independent experiments that used deletions and site-directed point mutations to probe Gcr1p function.

Together, the four hypomutable regions we identified (A, B, C, and D) occupy less than half of the region we set out to analyze. Further work confirmed that each contains individual amino acids that are required for Gcr1p function. Analysis of individual hypomutable regions has yielded intriguing results. (1) A LexAp fusion to region A, the smallest of the four hypomutable domains, is able to coimmunoprecipitate with Gcr2p. (2) Although many inactivating point mutations in region C do not destabilize the protein, its deletion does, which suggests that some or many of the hydrophobic residues that predominate in region C could contribute to the formation of important internal regions of Gcr1p. (3) Region B can be subdivided into two smaller hypomutable domains, and has not yet been correlated with known Gcr1p functions (ie., activation and Rap1p contact). (4) Further inspection of the primary sequence of region D revealed an excellent match to leucine zipper motifs, which we showed was essential for Gcr1p homodimer formation. We have since gone on to characterize the role this dimerization domain in Gcr1p function (Deminoff et al. 2001).

Others have adapted the unigenic evolution technique to their studies (Friedman et al. 2003; Zeng et al. 2003; San Filippo and Lambowitz 2002; Guo et al. 2000; Friedman and Cech 1999). We feel that the method should be widely applicable, particularly now, as genomic studies provide previously uncharacterized genes as candidates for involvement in various processes. The requirements of the method are modest: (1) that the gene of interest leads to an observable phenotype when mutated, (2) an organism that can be transformed with a library of randomly mutagenized alleles, and (3) that the latter can be recovered and sequenced. Thus, although it is ideally suited to genetic studies in Saccharomyces or other easily manipulated microorganisms, unigenic evolution in higher eukaryotes can be envisioned.

References:

Friedman KL, Heit JJ, Long DM, and TR Cech, 2003. N-terminal domain of yeast telomerase reverse transcriptase: Recruitment of Est3p to the telomerase complex. Molecular Biology of the Cell 14:1-13.

Zeng X, Zhang D, Dorsey M, and J Ma, 2003. Hypomutable regions of yeast TFIIB in a unigenic evolution test represent structural domains. Gene 309:29-56.

San Filippo J and AM Lambowitz, 2002. Characterization of the C-terminal DNA-binding/DNA endonuclease region of a group II intron-encoded protein. Journal of Molecular Biology 324:933-951.

Deminoff SJ, and GM Santangelo, 2001. Rap1p requires Gcr1p and Gcr2p homodimers to activate ribosomal protein and glycolytic genes, respectively. Genetics 158:133-43.

Guo H, Karberg M, Long M, Jones JP 3rd, Sullenger B, and AM Lambowitz, 2000. Group II introns designed to insert into therapeutically relevant DNA target sites in human cells. Science 289:452-457.

Friedman KL, and TR Cech, 1999. Essential functions of amino-terminal domains in the yeast telomerase catalytic subunit revealed by selection for viable mutants. Genes Dev. 13:2863-2874.

Deminoff SJ, Tornow, J., and GM Santangelo, 1995. Unigenic evolution: a novel genetic method localizes a putative leucine zipper that mediates dimerization of the Saccharomyces cerevisiae regulator Gcr1p. Genetics 141:1263-74.