A global view of pleiotropy and phenotypically derived gene function in yeast

There is a News and Views associated with this document.

Aimée Marie Dudley^1,a, Daniel Maarten Janse^1,a, Amos Tanay², Ron Shamir² & George McDonald Church¹

^aThese authors contributed equally to this work

Article highlights

Synopsis

A major goal of systems biology is to uncover the underlying structure of biological networks in order to develop comprehensive models of biological processes that are both accurate and predictive. In addition to genes that perform a single function, genetic networks also contain genes that perform many functions. To date, the network architecture of these multi-function or pleiotropic genes has not been well characterized. This study uses a combination of functional genomics and computational methods to reveal a non-trivial organization of highly pleiotropic genes in the important model organism S. cerevisiae.

Pleiotropy is a relatively common, but poorly understood phenomenon in biology defined as the ability of a mutation in a single gene to produce multiple effects in vivo. The prevalence of pleiotropy is perhaps most striking in the growing number of human diseases for which a mutation in a single gene produces a broad and seemingly unrelated set of symptoms (Brunner and van Driel, 2004). In yeast, the availability of resources such as a complete set of isogenic deletion mutants (Giaever et al., 2002) makes it possible to identify a comprehensive set of genes required for growth under a given environmental condition. To facilitate such genome-wide mutant analysis under a large number of conditions, we developed a cost-effective strategy to measure the growth of individual strains on standard agar plates. We used this method to assay the complete set of non-essential yeast gene deletions (4,700 mutants) under 21 environmental conditions in duplicate (>10⁵ data points) and identified 216 highly pleiotropic mutants with growth defects in 3 or more conditions.Figure 6

Figure 6: Using phenotype profiles to identify separable functions in pleiotropic genes

(A) General principle. For a pleiotropic gene (gene3) with growth defects in five conditions (1–3, 6, and 7), it is possible to partition these phenotypes into two sets of functions (blue and purple) based on the results of biclustering. (B) SNF1 example. SNF1 belongs to two biclusters with the phenotypes (HU=hydroxyurea, Gly=glycerol, Cd=cadmium, Cyh=cycloheximide, Caff=caffeine, Rap=rapamycin) outlined in blue and purple. Subsets of the genes present and GO functional categories enriched in each bicluster are also listed.

Full figure and legend (175K)Figures & Tables index

It has long been hypothesized that pleiotropy, while frequently observed, poses significant disadvantages for the evolvability of an organism, including limiting the rate of adaptation and reducing the level of adaptation for some traits in response to selection for others (Otto, 2004). By generating phenotype data for a relatively large number of conditions we were able to make an initial estimate of the overall degree of pleiotropy in yeast. We assessed the level of pleiotropy in our data by counting the number of phenotypes observed for each mutant. The results (Figure 8) show that approximately 70% of the mutants have a relatively low degree of pleiotropy, with phenotypes in only one or two conditions, and the remaining 30% are highly pleiotropic, with growth defects in as many as 14 conditions. To test the statistical significance of this amount of pleiotropy, we compared our data to a randomly generated distribution of phenotypes per mutant (Figure 8). A statistical comparison of these distributions demonstrates that the degree of pleiotropy we observe in an important model organism, Saccharomyces cerevisiae, is significantly greater than can be explained by chance. These results, support the hypothesis that pleiotropy plays an important role in biological systems.

Figure 8: Distribution of pleiotropy in our data and 1000 randomly generated sets.

Error bars represent 1 standard deviation. These distributions are significantly different as assessed by the Kolmogorov–Smirnov test with a P-value of 9 10^-70.

Full figure and legend (83K)Figures & Tables index

Our work also yields a number of widely applicable technical advances that have the potential to meet one of the greatest challenges in understanding pleiotropic genes—determining whether the phenotypes associated with a mutation are the result of the loss of a single function or of multiple functions encoded by the same gene. In human disease research, understanding gene function at this level provides important information relevant for devising effective treatments and analyzing drug side-effects. Unfortunately, classical approaches for dissecting the behavior of pleiotropic genes, such as searching for multiple alleles of a gene that exhibit different phenotypes, are time-consuming and often not feasible in a clinical setting. To address this problem, we developed a method for determining the association between gene functions and mutant phenotypes based on a single mutant allele, such as a complete gene deletion.

The strong link between mutant phenotype and cellular function suggests that high throughput methods for measuring phenotypes may be used to identify such functions de novo. In our study, we identified groups of mutants that shared a statistically significant set of phenotypes. The high degree of overlap between these clusters and known biological process annotations, synthetic lethal interactions, and protein complexes supported the hypothesis that this high throughput, unsupervised method can be used to discover genetically defined functional categories. Application of these functional predictions to the phenotype profiles of the highly pleiotropic mutants allowed us to generate hypotheses about the number of functions carried out by these genes and the conditions under which they are required (Figure 6). While extremely powerful in yeast, these methods hold even greater promise for analysis in organisms that are less genetically tractable, such as mammalian cell lines in which endogenous genes can be silenced using RNAi technology (Schutze, 2004).

MORE ARTICLES LIKE THIS

These links to content published by NPG are automatically generated.