Evidence against concerted evolution in a tandem array in the clumping factor gene of Staphylococcus aureus.

Naturally occurring tandemly repeat sequences are widely distributed among eukaryotes. Their evolution is characterized by genetic instability, which may arise from slippage, unequal crossover, and gene conversion (Charlesworth et al. 1994); this instability is seen in noncoding DNA (Jeffreys et al. 1994), RNA-coding genes (reviewed in Li and Graur 199 1, pp. 162169), and protein-coding genes (Sharp and Li 1987). The consequence of these events is that elements within an array are more similar to each other than would otherwise be expected: the homogenization of these repeats is described as concerted evolution (Li and Graur 199 1, pp. 162169). Tandem arrays are rarer in prokaryotes. The recently characterized (McDevitt et al. 1994) clumping factor protein-coding sequence (cl&t) in Staphylococcus aureus (strain Newman; GenBank Z 18852) contains one of the most extensive prokaryote simple tandem arrays known. A two-amino acid Serine-Aspartate motif spans 308 residues, with most departures from the exact motif at the two ends of the array. The DNA encoding this motif is an array of 18-nucleotide repeats, whose elements follow the consensus GAY TCN GAY TCN GAY AGY, where N is any base and Y is T or C. Thirty-one of the repeats conform exactly to the consensus and are located in the center of the array. This array is large and homogeneous enough to test whether concerted evolution is occurring at high rates. Concerted evolution of a tandem array resulting from the physical proximity of similar sequences can have some or all of the following consequences: (1) the sequence identity between repeat members within a species is higher than that between members from two different species (Dover 1986); (2) there is a visible homogeneity of the arrayed repeats compared with monomers elsewhere in the genome; (3) the repeats that are physically closer within the array are more similar; (4) the distribution of distances between array elements is multimodal, as sequences that have recently undergone homogenization will be very similar, while those that have not done so will be quite different; more complex events such as multimeric homogenizations will also distort the distribution of distances from a unimodal