The work explores a pervasive feature of the genomic architecture of flowering plants: polyploidy. The basic model for the eukaryotic genome is diploidy. Diploid cells – like the ones found in most animals, including us - contain two copies of the genome: one inherited maternally, the other paternally. Polyploid cells harbor more than two genomic copies. Polyploidy is quite common in flowering plants, or angiosperms – the most diverse plant phylum, with at least 250,000 species. Polyploidy in plants often accompanies hybridization – mating between individuals from separate species – giving rise to offspring termed allopolyploid.
Allopolyploidization in plants has been associated with hybrid vigor, and may have provided adaptive value in various taxa during their evolutionary histories. “There is a time when these guys have specific advantages,” says Luca Comai, a biologist at the University of California, Davis – who was not associated with the present research, but who assessed it in a minireview appearing May 1 in the Journal of Biology – “and these are probably times of intense stress – or times when there is a need for colonizing an environment which is not as easily colonized by the quote-unquote regular plants.”
An over-arching question in the study of allopolyploids is: how do two divergent genomes reconcile their patterns of gene expression in the allopolyploid offspring? “You have this divergence – this lineage splitting into two lineages that for thousands and then millions of years are evolving in isolation,” begins Jonathan Wendel, who chairs the Department of Ecology, Evolution, and Organismal Biology at Iowa State and co-authored the current report. “You like to think of their 40,000 genes all working in this beautiful 40,000-part harmony together in development and reproduction. And yet, they acquire different lifestyles, and different morphologies - maybe they’re on different continents - and then you put them back together: this wonderful biological reunion. They get slammed back together in the same nucleus and only one of two parental cytoplasms, because of hybridization, and in the case of polyploidy, there’s genome doubling. And there’s this huge black box. What happens? 40,000 now-duplicate genes? and why is this evolutionarily successful? Why is it so common? It’s a fantastic mystery.”
To address these questions, investigators Ryan Rapp and Jonathan Wendel of Iowa State, and Joshua Udall of Brigham Young, evaluated patterns of gene expression in diploid angiosperms and their allopolyploid progeny. They conducted two crosses utilizing closely related cotton species from the genus Gossypium. In both crosses the maternal genome was supplied by G. arboreum. In one cross, the paternal genome was from G. bickii; in the other it was from G. thurberi. The researchers induced genome duplication in the hybrids by treatment with colchicine – a chemical compound which inhibits chromosome segregation. To evaluate genome-wide gene expression, the researchers extracted RNA from each diploid parent and both allopolyploid progeny, converted it to cDNA, and performed hybridization experiments using custom micro-arrays for more than 40,000 genes. From the results they also calculated mid-parent values – that is, the average of the parental values for each gene.
Their initial observation was a surprisingly high level of expression divergence between parents. The three Gossypium species diverged from a common ancestor within the last five to ten million years and their gene sequences only differ by about 3%. Yet more than 40% of genes are differentially expressed between G. arboreum and G. bickii, and more than 50% between G. arboreum and G. thurberi. Of the differentially expressed genes, approximately equal numbers were up-regulated in each parent, relative to the other.
When they compared expression levels in the allopolyploids with those in the parents, the researchers found – not surprisingly – that the vast majority of genes were expressed at values not significantly different from the calculated mid-parent values (99% of genes in the G. arboreum x G. bickii cross, and 94% of genes in the G. arboreum x G. thurberi cross). This observation was consistent with an earlier study, by another group, that reported mid-parent expression levels for 95% of genes in the allopolyploid progeny of two Arabidopsis species. On the surface, these observations in cotton might seem to imply a sort of genomic averaging out of expression levels between parents. However, a closer look revealed that something very different was going on inside the hybrid cotton nuclei.
The researchers discovered thousands of genes for which the expression level in the allopolyploid lay between the expression level of one parent and the mid-parent value and – while not significantly different from either of these values – was significantly different than the expression value for the other parent. Such genes can be described as mirroring the expression level of one parent, while diverging in expression from the other. Next, according to Wendel, lead author Ryan Rapp – a PhD student in Wendel’s lab – had a very good idea. He partitioned the data into the 12 possible categories of differential expression (for example: a<b=c, a=b<c, a=b>c, and so on). It was only after this partitioning that a striking pattern emerged from the shadows.
The results were clearly and dramatically biased toward just two of the twelve categories, in each cross. In the G. arboreum / G. thurberi cross, the favored categories represented 1.)the allopolyploid gene up-regulated relative to maternal expression and approximately equal to paternal expression, and 2.) the allopolyploid gene down-regulated relative to maternal expression and approximately equal to paternal expression. In the G. arboreum / G. bickii cross, the pattern was reversed, with a preponderance of genes either up-regulated or down-regulated to mimic the expression observed in the maternal diploid. “Unless you look at the data that way,” Wendel says of Rapp’s insight, “you can’t see expression dominance. It’s just not evident – because you don’t have the lens that allows you to see it.”
Furthermore, in each case, the expression dominance was truly bi-directional, with roughly equal proportions of allopolyploid genes up-regulated, and down-regulated, to the expression level of the dominant parent. These results differ from the earlier Arabidopsis study, in which differential expression in the allpolyploid appeared to be unidirectional, reflecting global repression, biased toward greater repression of genes which were up-regulated in one particular parent – A. thaliana.
The authors speculate that the dramatic expression dominance observed in cotton allopolyploids may be the result of epigenetic mechanisms – mechanisms which have a heritable effect on phenotype, not determined directly from DNA sequence. Further research will be required to elucidate the mechanism by which these plants orchestrate this bi-directional, genomic expression dominance.
“I think there’s still a tremendous amount we don’t understand about the process of genome merger and doubling,” Wendel muses. “Even though it’s seemingly so unlikely, there must be, occasionally, extremely fit outcomes that are visible to selection and are preferred – because the problems it has to surmount to reconcile regulation of 40,000 genes and have a positive evolutionary outcome, those problems would seem insurmountable and yet it’s common.”
Wendel is enthusiastic about what he calls a new integrative era in science. “The promise for understanding ecology and evolution is much higher because of genome science, and bioinformatics, and systems biology, and computational science. That’s a very promising future direction for us all.”
RESEARCH ARTICLE: Ryan A Rapp, Joshua A Udall, Jonathan F Wendel, “Genome expression dominance in allopolyploids,” BMC Biology 7:18, 1 May 2009, (doi:10.1186/1741-7007-7-18).
MINIREVIEW: Daniela Pignatta and Luca Comai, “Parental squabbles and genome expression: lessons from the polyploids,” Journal of Biology 8:43, 1 May 2009, (doi:10.1186/jbiol140).
