Where did HO come from? It seems very likely that HO and VDE share a recent common ancestor in the hemiascomycetes because their sequences and phylogenetic distributions are similar (Gimble and Thorner 1992; Keeling and Roger 1995; Liu 2000). One scenario is that a mobile intein (presumably from a bacterium) first invaded VMA1 to form VDE, and has subsequently been horizontally transmitted with frequent gains and losses in various hemiasco-mycete species to give its current phylogenetic distribution (Koufopanou et al. 2002). The phylogenetic range over which VDE can spread is limited by the requirement for a suitable target site in the VMA1 sequence of the host species, but this target site can co-evolve with the protein sequence of VDE (Posey et al. 2004). At some point in the cycle of horizontal transfers, gains and losses, the VDE of one species duplicated to give rise to a homing endonuclease gene that was the ancestor of HO. An alternative scenario is that the first mobile intein to invade the hemiascomycetes was located in some other unknown gene, and both VDE and HO genes were derived from it by later duplication.
We speculate that HO may have been formed when a mobile intein invaded a gene coding for a protein with a zinc finger domain. The integration site was either very close to the 5' end of the gene, or else the original 5' end of the zinc finger protein gene was lost later. Loss of protein-splicing activity led to the formation of a chimeric protein with an endonuclease domain and two DNA-binding regions: one in the zinc finger and one in the N-terminal protein-splicing domain. Somehow this fusion protein began to bind DNA specifically at a site in MATa 1, rather than in its own gene, thus creating a DSB that was repaired using the silent cassettes of mating-type information. This step may not be quite as far-fetched as it seems, because VDE recognition sites have been shown to drift (Posey et al. 2004), and there is a small amount of DNA sequence similarity between the recognition sites of HO and S. cerevi-siae VDE (Gimble and Wang 1996; Bakhrat et al. 2004).
The origin of HO may also be connected to the origin of an unlinked gene, SIR1. A crucial feature of HO activity is that it cleaves the recognition site present in the active MAT locus, but not the sites in the HM cassettes which have identical DNA sequences. Cleavage at the silent cassettes is repressed by the presence of heterochromatin, whose formation is directed by the Sir proteins. Whereas Sir2, Sir3 and Sir4 proteins are also involved in transcriptional silencing at rDNA and telomeres, Sirl is uniquely involved in silencing at the HM loci, by recruiting the NAD-dependent his-tone deacetylase Sir2 (Chien et al. 1993). SIR1 genes have almost exactly the same phylogenetic distribution as HO genes: present in Saccharomyces sensu stricto, Saccharomyces castellii and Zygosaccharomyces rouxii, but absent in
K. lactis, K. waltii, S. kluyveri, A. gossypii and more distantly related species. The only discrepancy is that C. glabrata has HO but not SIR1.
The co-option of HO by yeasts led to a dramatic increase in the rate of mat-ing-type switching, from once per million cells to almost one switch per cell generation, resulting in a change from a life cycle where the major growth phase was haploid (like K. lactis today) to one where most cells are diploid because budding of a haploid cell is followed almost inevitably by mother-daughter mating. This change was made possible by the fortuitous invasion of a parasitic genetic element but had a profound effect on the biology of a large clade of yeast species.
Acknowledgements. We thank Bernard Dujon, Cecile Fairhead and Gilles Fischer for stimulating conversations. Research in K.H.W.'s laboratory is supported by Science Foundation Ireland. Research in J.E.H.'s laboratory is supported by the US National Institutes of Health.
Was this article helpful?