MIT OpenCourseWare http://ocw.mit.edu 6.047 / 6.878 Computational Biology: Genomes, Networks, EvolutionFall 2008 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.Comparative Genomics II: From Genomes to Evolution Lecture: Manolis Kellis. November 4, 2008 1 Introduction This is the second lecture in a series of two lectures on the work being done in Prof. Kellis’s lab on comparative genomics. Comparing genomes of related species at different evolutionary distances can teach us both about the genetic code and about evolution. In this lecture we discussed two examples of what we can learn about evolution using comparative genomics. The topics of this lecture are discussed in greater detail in the papers [1] and [2]. A brief summary is given in these notes with references to figures in lecture slides. 2 Whole genome duplication Evolution requires infrequent random “errors”, i.e. mutations, in cell division to create variation in different species. Genomic duplication is a particular type of such error which can be useful to explain innovations in evolution[3]. Most of the time genomic duplication will lead the daughter cell to be less fit (“sick”) and to get selected out. However, if the daughter species survives and adapts, one copy of an original gene can perform its task while the other gene can evolve to gain new function increasing gene content and fitness. In class, we discussed one type of genomic duplication, namely whole genome duplication (WGD), in the study of which comparative genomics has brought in novel information. 2.1 Before comparative geno mi cs In his 1970 book “Evolution by Gene Duplication” where he postulated the role of genomic duplication in evolution, Ohno also s uggested the occurance of whole genome duplications, in particular that the vertebrate genome is the result of one or more whole genome duplications[3]. Such large scale duplications would explain, for example, how there are 4 Hox genes in humans compared to 1 Hox gene in flies. 1WGD has been suggested in various other cases but conclusive evidence was not found until 2004. In particular, the possibilty of WGD in the yeast S. cere-visiae has been debated since 1997. When S. cerevisiae genome was sequenced, large duplication blocks were observed. Wolfe and colleagues suggested that these duplications were due to a WGD[4]. Others have argued that the ob-served paralogous gene rate of 8% was too small to suggest a WGD and could be explained with independant local duplications[5]. 2.2 Comparative evidence for WGD in yeasts Assume we label a number of neighboring genes in a segment of DNA in order 1-16. Then a WGD occurs and there are two copies of the segment in the genome. As this dual-genome species evolves, some of the redundant genes are lost and some gain new functions such that one of the chromosomes has orthologous genes to gene 1, 3, 4, 6, 9, 10, 12, 13, 14, 16 while the other chromosome has 2, 3, 5, 7, 8, 11, 13, 15. Comparing the two chromosomes we would see paralogues 3 and 13 and we would not be able to tell if these paralogues are from WGD or individual local duplications. However, if we compare genomes of different species before and after WGD, we would see that a region in the species before WGD will correspond to two regions in the species after duplication. (See slide 5.) Here before and after duplication refer to spec ies that descend from a branch without and with WGD respectively. This is exactly what was observed in Prof. Kellis’s lab. Com-parative analysis between S. cerevisiae and K. Waltii gave a less clean signal than the comparison of different Saccharomyses species. (See slide 6.) Look-ing closely at individual matching regions S. cerevisiae and K. Waltii, the dual match signal that would be expected from a WGD was noticed. (See slide 7.) During the analysis of the data, sequencing of the K. Waltii was completed. It was seen that 16 chromosomes in S. cerevisiae corresponds to 8 chromosomes in K. Waltii. (See slide 9.) Looking at slide 9, one can see that the correspon-dence between the chromosomes in the m odern species is not perfect. This is due to many chromosome crossing that have happened since the two species branched. (Chromosome crossing is not expected to have any major selective effects so can happen relatively often.) Another evidence for WGD in the data comes from the pos itions of chro-mosome centromeres relative to genes. Assume there is a centromere between genes 6 and 7 in the above example. In the two corresponding S. cerevisiae chromosomes, we would expect centromeres to be between labels 6 and 9 and 5 and 7 respectively. This kind of ce ntromere position prediction was seen to hold true in all the chromosomes in the comparison of S. cerevisiae and K. Waltii. (See slide 10.) An interesting observation is that WGD event in the yeasts happened ap-proximately 190M years ago. This was also when fruit bearing plants evolved. This was a time when there was an abundence of sugar available in the envi-ronment and the first generations of inefficient/sick species of yeasts with too many redundant genes would be able to survive. Also the new functions that 2would develop with new derived genes could give them useful features needed in these new conditions. (Such as making beer.) 2.3 Post-duplication evolution K. Waltii has approximately 5000 genes. After genome duplication, the ances-tor of S. cerevisiae had approximately 10000 genes. Soon after, most of the duplicate genes were lost. S. cerevisiae now has 5500 species. Do the new paralogous g enes “share” their old task and evolve at a similar high rate (as proposed by Lynch in 2000)? Or is one better preserved to do the old task while one evolves rapidly to gain new function (as proposed by Ohno in 1970)? Comparing genes in S. cerevisiae and K. Waltii, it was seen that 95% of the gene pairs showed asymmetric accelaration rates, supporting Ohno’s model. This means we can define “ancesteral” and “derived” functions for the two paralogous genes. Indeed, biological experiments have shown that such a dis-tinction exists. Ancesteral genes are more vital (gene removal is more likely to be lethal), are expressed more abundantly and serve general functions whereas derived genes are used