Heat-shock proteins help to protect organisms from external stresses. The idea that they can also buffer against internal — genetic — variations has received support from studies of fruitflies and, now, of plants.
Living organisms are caught between a hammer and an anvil, evolutionarily speaking. On the one hand, they need to buffer the influences of genetic mutations and environmental stresses if they are to develop normally and maintain a coherent and functional form. On the other, stabilizing one's development too much may mean not being able to respond at all to changes in the environment and starting down the primrose path to extinction. On page 618 of this issue, Queitsch et al.1 propose that, in plants, the balance between stability and the potential for change is made possible in part through a protein involved in 'heat-shock responses' in a wide variety of species, from plants to insects.
Heat-shock responses are a fundamental and widespread type of cellular defence against environmental stress. They have been studied for their effects on the fitness of organisms2; for their co-evolution with other cellular functions3; for their role in response to stresses not related to temperature4; and for the level of natural variation in the genes that encode the heat-shock proteins (Hsps)5, which mediate heat-shock responses. Notwithstanding the continuing debate6 about the actual function of these proteins, it is now clear that they are a complex family of gene products that are involved in protecting other proteins. Some Hsps are expressed continuously in the organism, whereas others are triggered by several environmentally harsh conditions (not only increases in temperature).
Given the ubiquity of these proteins and their role in protecting organisms from environmental changes, it makes sense to ask a more subtle question: can they also help to protect against disruptive genetic variations? After all, the systems that allow organisms to develop from a fertilized egg to the adult form have been honed over millions of years of evolution, and it is likely that a mutation in any of the tens of thousands of genes involved would disrupt the entire process, just as a severe environmental stress does. This idea is rooted in the 1940s, in Waddington's classic studies7,8 of 'canalization' — the resistance of developing organisms to change when perturbed genetically or environmentally. More recently it has been suggested that, from the point of view of development, internal disturbances are simply another form of environmental change that needs to be properly 'canalized' to maintain a viable form (phenotype) tailored to specific functions9.
So can the Hsp proteins buffer genetic as well as environmental change? Rutherford and Lindquist10 first tested the idea of a connection between Hsp activity and genetic variation by looking at a popular animal model of developmental genetics — the fruitfly Drosophila melanogaster. The results were stunning. When the authors disrupted Hsp90, by either mutating or inhibiting it, phenotypic variation in nearly every structure of adult D. melanogaster ensued, with the details depending on the genetic background of the insects used (that is, on which other specific genes were present in each individual). This led the authors to conclude that D. melanogaster accumulates hidden genetic variation, which is somehow kept by Hsp90 from affecting the phenotype. If the function of Hsp90 is partly compromised, the buffer breaks and we can see previously 'unavailable' phenotypic variants.
Queitsch, Sangster and Lindquist1 have now expanded this research to another model of developmental genetics, the plant Arabidopsis thaliana11,12. These two species, D. melanogaster and A. thaliana, are of course very different in many ways. They have evolved separately over hundreds of millions of years. As one is a plant and the other an animal, they develop radically differently. And their breeding systems are not at all alike: fruit flies are obligatory 'outcrossers', meaning that they need a partner to produce offspring, whereas A. thaliana is mostly a 'selfer' — it fertilizes its own female gametes. Nonetheless, in A. thaliana, as in D. melanogaster, changes in Hsp90 release previously hidden genetic variation, resulting in the production of novel phenotypes. These include altered leaf shapes, the accumulation of a purple pigment in hypocotyls (embryonic stems), and variations in hypocotyl length.
These latest findings1 are also interesting for their differences from the previous results and for the potential follow-up that they make possible. For example, the range of phenotypes released by compromising Hsp90 function in A. thaliana seems less wide than the corresponding variation seen in D. melanogaster (in as much as it is possible to compare the phenotypes of insects and plants). This might be a result of the differences in developmental plasticity inherent to the two types of organism. Furthermore, by using sophisticated approaches to genetic analysis such as microarrays, it should be possible to further characterize the molecular basis of the phenotypic variation induced by blocking Hsp90 in A. thaliana.
More generally, there is no question about the importance of this sort of study for our understanding of the complex relationships between genes, environments and phenotypes. The mere ability to affect Hsp90 and possibly other similar proteins experimentally, thus producing an array of new phenotypes in a potentially wide range of organisms, provides evolutionary biologists with a powerful new research tool. It is less clear and more controversial whether these findings have long-term evolutionary implications, and the nature of the mechanisms that make the relationship between heat-shock proteins and hidden genetic variation possible is still unknown.
An obvious question is whether Hsp90 was actually selected to act as a 'capacitor' of morphological evolution, as it has been characterized by Rutherford and Lindquist10, or whether the storage of hidden genetic variation is instead a by-product of its normal physiological function. Although I would bet on the latter, only a phylogenetically informed comparative study, tracing the evolution of both the Hsp family itself and its phenotypic effects, will be able to shed light on this question.
Queitsch et al.1 also suggest that, because of the breadth and strength of Hsp90's effects, it might have a large impact on evolutionary processes, and I am more inclined to agree on this point. But statements10 to the effect that the conditional release of hidden genetic variation may have allowed the rapid diversification of forms (radiations) that are occasionally seen in the fossil record might be too premature a connection between developmental genetics and palaeontology. Then again, it is exactly this sort of connection that will be needed to enlarge modern evolutionary theory to include molecular and developmental genetics9.