Metabolic gene clusters—functionally related and physically clustered genes—are a common feature of some eukaryotic genomes. Two hypotheses have been advanced to explain the origin and maintenance of metabolic gene clusters: coordinated gene expression and genetic linkage. Here we test the hypothesis that selection for coordinated gene expression underlies the clustering of GAL genes in the yeast genome. We find that, although clustering coordinates the expression of GAL1 and GAL10, disrupting the GAL cluster does not impair fitness, suggesting that other mechanisms, such as genetic linkage, drive the origin and maintenance metabolic gene clusters.
Citation: Lang GI, Botstein D (2011) A Test of the Coordinated Expression Hypothesis for the Origin and Maintenance of the GAL Cluster in Yeast. PLoS ONE 6(9): e25290. doi:10.1371/journal.pone.0025290
Editor: Laura N. Rusche, Duke University, United States of America
Received: July 4, 2011; Accepted: August 31, 2011; Published: September 22, 2011
Copyright: © 2011 Lang, Botstein. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by the National Institute of General Medical Sciences (NIGMS) Centers of Excellence grant GM071508 (to DB) and the individual National Institutes of Health (NIH) grant GM046406 (to DB). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
In eukaryotic genomes functionally related genes are, to a first approximation, dispersed throughout the genome. There are counter examples, however, of the physical clustering of genes whose products function in the same metabolic pathway –. In the yeast, Saccharomyces cerevisiae, metabolic gene clusters exist for biotin synthesis , allantoin degredation , and galactose assimilation . The GAL cluster consists of three genes (GAL1, GAL10, and GAL7), encoding enzymes that catalyze four sequential steps in galactose assimilation, that are clustered in a 7 kb region of Chromosome II (Figure 1). The GAL cluster evolved independently through gene relocation in two fungal phyla (Ascomycota and Basidiomycota) and has been horizontally transferred within Ascomycota .
Figure 1. The GAL1-GAL10-GAL7 gene cluster in Saccharomyces cerevisiae.
(A) GAL1, GAL10, and GAL7 encode enzymes that catalyze sequential steps in the assimilation of galactose. Gal1 is the galactokinase. Gal10 contains two catalytic domains: a mutarotase that interconverts galactose enantiomers, and an epimerase domain that converts UDP-galactose to UDP-glucose. Gal7 is the galactose-1-p uridyl transferase. An intermediate in galactose assimilation, galactose-1-p, is toxic to cells. (B) GAL1, GAL10, and GAL7 are clustered within a 7 kb region on Chromosome II with GAL1 and GAL10 sharing a divergent promoter.doi:10.1371/journal.pone.0025290.g001
It is not clear what evolutionary forces favored the formation and maintenance of gene clusters. As is the case for hypotheses for the origin and maintenance of bacterial operons, there are two attractive ideas: coordinated expression  and genetic linkage –. Many gene clusters encode for metabolic pathways with toxic intermediates, for example thalianol and thalian-diol in the triterpene biosynthesis pathway in plants , , glyoxylate in the yeast allantoin degradation pathway , and galactose-1-phosphate in the yeast GAL pathway. Coordinated expression of individual enzymes of these pathways could facilitate metabolic channeling and lessen the buildup of toxic intermediates. Alternatively, the physical proximity of functionally related genes could reflect selection for genetic linkage, either to maintain alleles of co-adapted genes or as a result of recurrent horizontal transfer of the gene cluster. Neither of these models has been tested experimentally. Using the GAL cluster in S. cerevisiae, we directly test the coordinated expression hypothesis, which makes two experimental predictions: (1) clustering contributes to coordination gene expression and (2) clustering provides a fitness advantage.
To determine whether the GAL cluster organization improves coordinated expression of the GAL genes, we generated diploid strains in which GFP is fused to GAL1 and mCherry is fused to GAL10 (or GAL7) in either the cis or trans conformation (Figure 2). We monitored the correlation between Gal1-GFP and Gal10-mCherry (or Gal7-mCherry) following induction of the GAL genes in a steady-state glucose-limited chemostat (Figure 2). Consistent with the coordinated expression hypothesis, Gal1-GFP and Gal10-mCherry are more correlated when these genes are in cis. This is not surprising since these two genes share a divergent promoter. For Gal1-GFP and Gal7-mCherry, however, we find no difference in the coordination of gene expression between the two conformations. The correlation between Gal1 and Gal7 in either conformation is similar to Gal1-Gal10 in the trans conformation. This is inconsistent with the hypothesis that the physical association of these genes facilitates coordinated expression.
Figure 2. To determine the effect of gene clustering on the coordinated expression of the GAL genes, we monitored production of Gal1-GFP and Gal10-mCherry (or Gal7-mCherry) in both the cis and trans conformations following a 2.5 g/L galactose pulse into a steady-state glucose-limited (0.8 g/L) chemostat.
GFP and mCherry were quantified by flow cytometry. (A) Population profiles showing the correlation (R2) between Gal1-GFP and Gal10-mCherry (or Gal7-mCherry) following the galactose pulse. (B) Correlation coefficients (R2) between Gal1-GFP and Gal10-mCherry as a function of time following the galactose pulse. Gal1-GFP and Gal10-mCherry are more correlated in the cis conformation (0.82 at 200 min) compared to the trans conformation (0.63 at 200 min). For Gal1-GFP and Gal7-mCherry, however, we find no difference in the coordination of gene expression between the two conformations (0.70 and 0.69 for cis and trans, respectively at 200 min). The correlation between Gal1 and Gal7 in either conformation is similar to Gal1-Gal10 in the trans conformation. (C) The average cell density (± one standard deviation) for all eight populations following the galactose pulse as measured by Coulter counter. Although the Gal proteins were detectable 30 minutes, cell number did not increase until 120 minutes subsequent to the galactose pulse.doi:10.1371/journal.pone.0025290.g002
To determine whether GAL gene clustering provides a selective advantage, we generated strains hemizygous for GAL1, GAL10 and GAL7, in either the cis conformation, or with one of the GAL genes in trans (Figure 3). We measured the fitness of the hemizygous strains, as well as homozygous wild-type and galΔ strains in batch culture under three conditions: glucose (GAL genes fully repressed), galactose (GAL genes fully induced), and alternating glucose/galactose. Disrupting the contiguity of the GAL1-GAL10-GAL7 cluster does not decrease fitness in any of the tested conditions (Figure 4). We have reported previously that the error in measurement in competitive fitness assays is approximately 0.4% . The eight hemizygotes, competed in glucose, where we do not expect a difference in fitness, have a standard deviation of only 0.1%. Estimates of the effective population size for natural yeast populations are ~107 , ; therefore, evolution can conceivably act on selection coefficients several orders of magnitude smaller that we can detect in the laboratory. For this reason we can not rule out that very small, but non-trivial, selective forces play some role in the maintenance of the GAL cluster, although our data suggest that GAL10-trans and GAL7-trans may, in fact, have a slight fitness advantage in alternating glucose/galactose (0.5% and 0.6%, respectively) perhaps by alleviating transcriptional interference between GAL10 and GAL7 , . These results fail to support the hypothesis that selection for coordinated gene expression is responsible for the origin or maintenance of the GAL gene cluster, and suggest (1) that the GAL cluster may be maintained in spite of fitness cost and (2) that coordinated expression of Gal1 and Gal10 is a consequence, rather than a cause, of clustering.
Figure 3. Strategy for disrupting the contiguity of the GAL cluster starting from the prototrophic S288c strains DBY12000 (MATa) and DBY12001 (MATα).
Construction of strains hemizygous for each of the three GAL-cluster genes required three rounds of transformation replacing GAL7, GAL10, and GAL1 with HphMX, KanMX, and NatMX, respectively. Prior to mating, the haploid strains were backcrossed to DBY12000 (or DBY12001) carrying either GFP or dTomato in order to fluorescently label strains for the competition experiment. Note that each of the four possible hemizygous was constructed twice independently, and are indicated by open and closed circles in Figure 4.doi:10.1371/journal.pone.0025290.g003
Figure 4. To determine if GAL gene clustering provides a selective advantage, we generated GFP-labeled strains hemizygous for GAL1, GAL10 and GAL7, in either the cis conformation, or with one of the GAL genes in trans (Figure 3) and measured their fitness by competing each against a dTomato-labeled GAL cis strain in three conditions: glucose (YPD), galactose (YPG), and alternating glucose and galactose.
Open and filled circles represent independently constructed biological replicates of the hemizygous strains (Figure 3). In glucose there is no difference in fitness for any of the hemizygous strains. The homozygous GAL (wild-type) strain has a 0.9% advantage and the homozygous galΔ strain has a 0.8% fitness disadvantage compared to the hemizygous reference strain. Since the GAL genes are repressed in glucose, this fitness difference is due to the presence of the three drug cassettes (KanMX, NatMX, and HphMX) absent from the wild-type strain, present in one copy in the reference strain, and in two copies in the galΔ strain. In galactose, the galΔ strain is quickly outcompeted, however all other strains have a slight advantage over the reference strain, suggesting a slight advantage to GFP over dTomato in galactose (0.7±0.3%). The lack of a fitness difference between wild-type and the hemizygous strains in galactose suggests two things: (1) the cost of the drug markers in mitigated in galactose (perhaps because of slower growth in galactose and/or because the GAL genes are overexpressed under this condition and may incur a cost themselves), and (2) the hemizygotes are able to maintain adequate levels of the Gal proteins despite a two-fold reduction in gene copy number. Under alternating conditions, the wild-type strain has a 4.3% advantage indicating that increased copy number of the GAL genes, which does not affect fitness when growing exclusively in galactose, is beneficial in a changing environment, likely by establishing steady state levels of the Gal proteins more rapidly. In the alternating regime, like in galactose, disrupting the contiguity of the GAL1-GAL10-GAL7 cluster does not impair fitness; GAL10-trans and GAL7-trans may, in fact, have a fitness advantage.doi:10.1371/journal.pone.0025290.g004
Our demonstration that disrupting the contiguity of the GAL cluster does not incur a fitness cost lends support to the hypothesis that genetic linkage is the selective force driving the origin and maintenance of GAL gene clusters. Why would genetic linkage of GAL1, GAL10, and GAL7 be selectively advantageous? In S. kudriavzevii, a closely related species to S. cerevisiae, the GAL cluster, as well as the unlinked GAL2, GAL4, and GAL80 exist as degenerate pseudogenes that are maintained, along with functional alleles of these genes, despite historical gene flow between the Gal+ and Gal- subpopulations . In a population maintaining the GAL genes as a balanced unlinked gene network polymorphism, linkage of GAL1, GAL10, and GAL7 prevents the buildup of the toxic galactose-1-phosphate, which occurs in GAL1-proficient strains lacking either GAL10 or GAL7. The loss of the GAL genes in S. kudriavzevii is far more recent than the evolution of the GAL cluster; however, it is not unique: at least five Ascomycota species have recently lost or pseudogenized the GAL genes , . It is possible that the spread of nonfunctional gal genes in an ancient population drove the evolution of the GAL gene cluster. In a population segregating functional and nonfunctional alleles of unlinked GAL genes, the alleles of these genes will assort randomly in the absence of galactose. Upon exposure to galactose, Dobzhansky-Muller incompatibilities will be revealed between the functional GAL1 allele and the nonfunctional gal10 and gal7 alleles. Clustering eliminates this incompatibility by genetically linking GAL1, GAL10, and GAL7. Similarly, the rate of loss of the GAL genes is greater in species where GAL1, GAL10, and GAL7 are clustered ; this is consistent with the spread of nonfunctional alleles being attenuated in species with unlinked GAL genes.
A second mechanism that could favor genetic linkage is horizontal gene transfer. Phylogenic evidence indicates that fungal gene clusters, including the GAL cluster, can be horizontally transferred –, –. The “selfish-operon” hypothesis posits that clustering enhances the spread of genes without producing a direct fitness benefit , , . Given only one documented horizontal transfer of the GAL cluster , it is unclear if the rate of horizontal gene transfer is sufficient to explain the maintenance of the GAL cluster based solely on this mechanism.
We have shown that clustering coordinates the expression of GAL1 and GAL10. Clustering, however, does not coordinate the expression of GAL1 and GAL7, nor does it provide a fitness advantage during continuous induction or alternating induction and repression of the GAL genes. Our results fail to support the coordinated expression hypothesis and suggest that other mechanisms, such as genetic linkage, drive the origin and maintenance of GAL gene clusters in yeast.
Materials and Methods
All strains in this experiment are derived from the prototrophic S288c strains DBY12000 (MATa) and DBY12001 (MATα). Strains for monitoring the correlation between Gal1-GFP and Gal10-mCherry (or Gal7-mCherry) were constructed as follows: In DBY12000, GFP (with a KanMX marker) was fused to the GAL1 gene and mCherry (with a NatMX marker) was fused to the GAL10 (or GAL7) gene. These strains were then crossed to DBY12001 to generate the Gal1-Gal10-cis and Gal1-Gal7-cis strains, respectively. Additionally, mCherry (with a NatMX marker) was fused to the GAL10 (or GAL7) gene in DBY12001. These strains were then crossed to DBY12000 (with a Gal1-GFP fusion) to generate the Gal1-Gal10-trans and Gal1-Gal7-trans strains, respectively.
Our strategy for disrupting the contiguity of the GAL cluster is shown in Figure 3. Starting with DBY12000 and DBY12001, we constructed all possible combinations of gal1Δ, gal10Δ, and gal7Δ, replaced with NatMX, KanMX, and HphMX, respectively. Prior to mating, the haploid strains were backcrossed to DBY12000 (or DBY12001) carrying either GFP or dTomato integrated at the dubious ORF YLR255c (marked with the NatMX) cassette in order to fluorescently label strains for the competition experiment.
Coordinated gene expression measurements
We monitored production of Gal1-GFP and Gal10-mCherry or Gal7-mCherry (in either the cis or trans conformation) following a 2.5 g/L galactose pulse into a steady-state glucose-limited (0.8 g/L) chemostat . Samples were taken at 10, 20, 30, 40, 50, 60, 80, 100, 120, 160, and 200 minutes following the galactose pulse and expression of Gal1-GFP and Gal10-mCherry (or Gal7-mCherry) was determined by flow cytometry. Correlation coefficients were calculated in Matlab and points on the axes were excluded.
We measured the fitness of the hemizygous strains, as well as homozygous wild-type and galΔ strains in three conditions: glucose (GAL genes fully repressed), galactose (GAL genes fully induced), and alternating glucose/galactose. Fitness assays were performed as described previously  with slight modifications. Briefly, prior to mixing, cells were initially grown to mid log in YPD (for the glucose and alternating regimes) or YPG (for the galactose regime) prior to starting the competition. Cultures were diluted every 12 hours; dilutions from YPD and YPG were approximately 1:500 and 1:100, respectively, although the exact dilutions were adjusted to keep the cells per culture consistent between competitions. At each dilution, cells were counted to determine the number of generations between each sample point, and fitness was calculated as the rate of change of the ln ratio of experimental to reference versus generations .
The complete laboratory notebook describing these experiments is available as Notebook S1.
The complete laboratory notebook detailing the strain constructions and experiments presented in this study.
We thank Chris Todd Hittinger and Antonis Rokas for comments on the manuscript and Christina DeCoste for assistance with flow cytometry.
Conceived and designed the experiments: GIL DB. Performed the experiments: GIL. Analyzed the data: GIL DB. Wrote the paper: GIL DB.
- 1. Field B, Osbourn AE (2008) Metabolic diversification--independent assembly of operon-like gene clusters in different plants. Science 320: 543–547.
- 2. Hall C, Dietrich FS (2007) The reacquisition of biotin prototrophy in Saccharomyces cerevisiae involved horizontal gene transfer, gene duplication and gene clustering. Genetics 177: 2293–2307.
- 3. Pedley KF, Walton JD (2001) Regulation of cyclic peptide biosynthesis in a plant pathogenic fungus by a novel transcription factor. Proc Natl Acad Sci U S A 98: 14174–14179.
- 4. Qi X, Bakht S, Leggett M, Maxwell C, Melton R, et al. (2004) A gene cluster for secondary metabolism in oat: implications for the evolution of metabolic diversity in plants. Proc Natl Acad Sci U S A 101: 8233–8238.
- 5. Slot JC, Hibbett DS (2007) Horizontal transfer of a nitrate assimilation gene cluster and ecological transitions in fungi: a phylogenetic study. PLoS One 2: e1097.
- 6. Slot JC, Rokas A (2010) Multiple GAL pathway gene clusters evolved independently and by different mechanisms in fungi. Proc Natl Acad Sci U S A 107: 10136–10141.
- 7. Slot JC, Rokas A (2011) Horizontal transfer of a large and highly toxic secondary metabolic gene cluster between fungi. Curr Biol 21: 134–139.
- 8. Ward TJ, Bielawski JP, Kistler HC, Sullivan E, O'Donnell K (2002) Ancestral polymorphism and adaptive evolution in the trichothecene mycotoxin gene cluster of phytopathogenic Fusarium. Proc Natl Acad Sci U S A 99: 9278–9283.
- 9. Wong S, Wolfe KH (2005) Birth of a metabolic gene cluster in yeast by adaptive gene relocation. Nat Genet 37: 777–782.
- 10. Price MN, Huang KH, Arkin AP, Alm EJ (2005) Operon formation is driven by co-regulation and not by horizontal gene transfer. Genome Res 15: 809–819.
- 11. Lawrence J (1999) Selfish operons: the evolutionary impact of gene clustering in prokaryotes and eukaryotes. Curr Opin Genet Dev 9: 642–648.
- 12. Lawrence JG, Roth JR (1996) Selfish operons: horizontal transfer may drive the evolution of gene clusters. Genetics 143: 1843–1860.
- 13. Stahl FW, Murray NE (1966) The evolution of gene clusters and genetic circularity in microorganisms. Genetics 53: 569–576.
- 14. Walton JD (2000) Horizontal gene transfer and the evolution of secondary metabolite gene clusters in fungi: an hypothesis. Fungal Genet Biol 30: 167–171.
- 15. Lang GI, Murray AW, Botstein D (2009) The cost of gene expression underlies a fitness trade-off in yeast. Proc Natl Acad Sci U S A 106: 5755–5760.
- 16. Tsai IJ, Bensasson D, Burt A, Koufopanou V (2008) Population genomics of the wild yeast Saccharomyces paradoxus: Quantifying the life cycle. Proc Natl Acad Sci U S A 105: 4957–4962.
- 17. Lynch M, Conery JS (2003) The origins of genome complexity. Science 302: 1401–1404.
- 18. Greger IH, Aranda A, Proudfoot N (2000) Balancing transcriptional interference and initiation on the GAL7 promoter of Saccharomyces cerevisiae. Proc Natl Acad Sci U S A 97: 8415–8420.
- 19. Greger IH, Proudfoot NJ (1998) Poly(A) signals control both transcriptional termination and initiation between the tandem GAL10 and GAL7 genes of Saccharomyces cerevisiae. EMBO J 17: 4771–4779.
- 20. Hittinger CT, Goncalves P, Sampaio JP, Dover J, Johnston M, et al. (2010) Remarkably ancient balanced polymorphisms in a multi-locus gene network. Nature 464: 54–58.
- 21. Hittinger CT, Rokas A, Carroll SB (2004) Parallel inactivation of multiple GAL pathway genes and ecological diversification in yeasts. Proc Natl Acad Sci U S A 101: 14144–14149.
- 22. Khaldi N, Collemare J, Lebrun MH, Wolfe KH (2008) Evidence for horizontal transfer of a secondary metabolite gene cluster between fungi. Genome Biol 9: R18.
- 23. Khaldi N, Wolfe KH (2008) Elusive origins of the extra genes in Aspergillus oryzae. PLoS One 3: e3036.
- 24. Patron NJ, Waller RF, Cozijnsen AJ, Straney DC, Gardiner DM, et al. (2007) Origin and distribution of epipolythiodioxopiperazine (ETP) gene clusters in filamentous ascomycetes. BMC Evol Biol 7: 174.
- 25. Ronen M, Botstein D (2006) Transcriptional response of steady-state yeast cultures to transient perturbations in carbon source. Proc Natl Acad Sci U S A 103: 389–394.
- 26. Hartl D (2000) A primer of population genetics. Sunderland, MA: Sinauer Associates.