Genetic Dissection of Transcriptional Regulation in Budding Yeast

DeRisi J. L., Iyer V. R., Brown P. O., Science 278, 680 (1997).

Wodicka L., Dong H., Mittmann M., Ho M. H., Lockhart D. J., Nature Biotechnol. 15, 1359 (1997).

Jansen R. C., Nap J. P., Trends Genet. 17, 388 (2001).

Sandberg R., et al., Proc. Natl. Acad. Sci. U.S.A. 97, 11038 (2000).

Primig M., et al., Nature Genet. 26, 415 (2000).

Jin W., et al., Nature Genet. 29, 389 (2001).

Cavalieri D., Townsend J. P., Hartl D. L., Proc. Natl. Acad. Sci. U.S.A. 97, 12369 (2000).

Parent strains were BY4716, an S288c derivative (MATa lys2Δ0) (27), and RM11-1a (MATa leu2Δ0 ura3Δ0 HO:kan), a haploid derived from Bb32(3), a natural isolate collected by Robert Mortimer, as in (28). Auxotrophic deletions were used to select diploids during mating and as positive controls for linkage. Both strains were gifts from Lee Hartwell, FHCRC, Seattle, WA.

Cultures were grown to 1 × 10⁷ cells/mL in synthetic C medium at 30°C and shaken at 175 rpm in a shaker bath. RNA was isolated with hot acid phenol and chloroform (29), followed by the RNA cleanup protocol for RNeasy columns (Qiagen, Valencia, CA). Samples were labeled and hybridized to cDNA microarrays, containing 6215 open reading frames, as described (30). Images were analyzed with GenePix software. Each two-color experiment involved one sample and one reference, with the same BY4716 RNA reference being used for all experiments. Two hybridizations were carried out for each sample, one with the sample labeled with Cy3 and the reference with Cy5, and one with the fluors reversed; for each gene, the two log ratios were averaged.

Differences in the expression between parent strains were assessed by comparing the six RM values to the six BY values with the Wilcoxon-Mann-Whitney test (31), and nominal P values are reported. To estimate the number of differences expected by chance at a given P-value threshold, we randomly assigned 6 of the 12 measurements to one group and 6 to the other, without regard for strain identity. We then compared the two groups, using the same statistical test we applied in comparing the two strains, and recorded how many genes had P values below the threshold. The randomization procedure was repeated 100 times, and the results were averaged.

Heritability was estimated as (segregant variance – parent variance)/segregant variance.

Winzeler E. A., et al., Science 281, 1194 (1998).

GeneChip Yeast Genome S98 microarrays were purchased from Affymetrix (Santa Clara, CA). Genomic DNA was isolated using Qiagen columns, digested with DnaseI and end-labeled as described (12), and hybridized following Affymetrix instructions. We developed algorithms to select oligos as reliable genetic markers and to genotype them in the segregants. These are available as supplementary information (32).

Cherry J. M., et al., Nature 387, 67 (1997).

For each pair of markers with all flocculent segregants sharing one parental genotype at one marker and the opposite parental genotype at the other, a Chi-square test of independence was performed on a 2 × 4 contingency table of all 40 segregants according to their flocculation phenotype (+ or −) and genotype (BY/BY, BY/RM, RM/BY, or RM/RM) at the two markers. A single pair showed linkage with P <10⁻⁷.

Kobayashi O., Yoshimoto H., Sone H., Curr. Genet. 36, 256 (1999).

Kron S. J., Trends Microbiol. 5, 450 (1997).

We tested for linkage between a marker and a message by partitioning the segregants into two groups according to marker genotype and comparing the message levels between the groups with the Wilcoxon-Mann-Whitney test (31). The number of linkages expected by chance was estimated by empirical permutation tests (33).

deVicente M. C., Tanksley S. D., Genetics 134, 585 (1993).

Rieseberg L. H., Archer M. A., Wayne R. K., Heredity 83, 363 (1999).

For each message differentially expressed between the parents at P <0.005, we computed the expression difference, x, in units of standard deviation. We simulated a locus explaining a fraction f of this difference by randomly assigning parent genotypes at the locus for 40 segregants and drawing the expression level of each segregant from one of two Gaussian distributions of unit variance, one with mean 0 and the other with mean fx conditional on the genotype. We recorded the proportion q of messages for which linkage was detected at P <5 × 10⁻⁵ in the simulated data, averaged over 100 simulations. In the text, q = 97% for f = 1, and q = 29% for f = 1/3. For n loci, each explaining a fraction f = 1/n of the difference x, we computed the rate of detecting at least one locus as 1 − (1 − q)ⁿ.

The genome was divided into 611 bins of 20 kb each (the bins at ends of chromosomes were smaller). Excluding self-linkages, we found 385 linkages at P <5 × 10⁻⁵. If these were randomly distributed across the genome, the number of linkages in any one bin would follow a Poisson distribution with a mean of 0.63. The probability that the bin with the most linkages would contain >5 is <0.03.

Gaisne M., Becam A. M., Verdiere J., Herbert C. J., Curr. Genet. 36, 195 (1999).

Colman-Lerner A., Chin T. E., Brent R., Cell 107, 739 (2001).

Hughes T. R., et al., Cell 102, 109 (2000).

Causton H. C., et al., Mol. Biol. Cell 12, 323 (2001).

Brachmann C. B., et al., Yeast 14, 115 (1998).

Mortimer R. K., Romano P., Suzzi G., Polsinelli M., Yeast 10, 1543 (1994).

The protocol for RNA isolation from yeast, developed by P. O. Brown's lab at Stanford University, can be found online at.

Fazzio T. G., et al., Mol. Cell. Biol. 21, 6450 (2001).

B. W. Lindgren, Statistical Theory (Macmillan, New York, ed. 2, 1968).

Supplemental material is available at www.sciencemag.org/cgi/content/full/1069516/DC1.

Churchill G. A., Doerge R. W., Genetics 138, 963 (1994).

We thank J. Delrow, M. Eberle, I. Furman, B. Garvik, J. Hartman, L. Iwasaki, C. Neal, J. Pincus, E. Ramos, C. Saunders, and P. Tata for technical assistance; and H. Coller, A. Colman-Lerner, E. Giniger, D. Gottschling, L. Hartwell, K. Markianos, J. Olson, and G. Smith for helpful discussions and comments on the manuscript. Supported by the Howard Hughes Medical Institute (of which L.K. is an associate investigator) and NIH postdoctoral fellowship GM64268 (R.B.). L.K. is a James S. McDonnell Centennial Fellow.