SWI/SNF is required for transcriptional memory at the yeast GAL gene cluster

Abstract

Post-translational modification of nucleosomal histones has been suggested to contribute to epigenetic transcriptional memory. We describe a case of transcriptional memory in yeast where the rate of transcriptional induction of GAL1 is regulated by the prior expression state. This epigenetic state is inherited by daughter cells, but does not require the histone acetyltransferase, Gcn5p, the histone ubiquitinylating enzyme, Rad6p, or the histone methylases, Dot1p, Set1p, or Set2p. In contrast, we show that the ATP-dependent chromatin remodeling enzyme, SWI/SNF, is essential for transcriptional memory at GAL1. Genetic studies indicate that SWI/SNF controls transcriptional memory by antagonizing ISWI-like chromatin remodeling enzymes.

Figure 1.

Transcriptional memory at the GAL1 gene. (A) Northern analysis of GAL1 RNA levels. Schematic at top depicts regimen of growth in different carbon sources. (Raf) 2% raffinose; (Gal) 2% galactose; (Glc) 2% glucose. Initial induction of GAL1 occurs with slower kinetics than when GAL1 is reinduced following glucose repression. (B) Graph comparing kinetics of GAL1 induction and reinduction, averaged over three experiments performed as described in A. Error bars represent the standard deviation at each point. Slightly different time points were taken in different experiments, so in these cases no error bars are shown. (C) Cells were grown overnight in raffinose prior to addition of galactose. After 1 h, cells were transferred to glucose media. At indicated times, an aliquot of cells was washed into fresh galactose media and GAL1 reinduction was observed. The “memory” state is maintained through at least 4 h of repression. (D) Glucose-grown daughter cells retain transcriptional memory. Cells were grown overnight in galactose media and then arrested at the G1/S boundary with α factor (lane labeled “Gal”). Arrested cells were then released from α factor into glucose medium to repress GAL1 and simultaneously undergo one synchronous division (lane labeled “Gal + Glc”). An aliquot of these cells was washed into galactose media to monitor reinduction kinetics (lanes labeled “no elutriation”). The remainder of the cells were elutriated to isolate daughter cells that had undergone mitosis in glucose media. Daughter cells were washed into galactose media to follow kinetics of GAL1 reinduction. The bottom panel represents an ACT1 loading control for total RNA levels. The numbers indicate fold induction of GAL1 transcripts normalized to ACT1 transcripts, with the maximally induced state set to a value of 1.

Figure 2.

Rapid dissociation of the transcription machinery during glucose repression. (A) Schematic representation of the GAL1–10 regulatory region. UAS_GAL marks the Gal4p-binding sites. URS_GAL contains the binding sites for the glucose repressor, Mig1p. Regions covered by primers for ChIP are shown as horizontal lines. TATA represents the TBP-binding sites and +1 represents the transcription start sites. (B) Gal4p ChIP in wild-type and gal4Δ strains. (C) ChIP for TBP, RNA Polymerase II, Mediator (α-Srb4-13myc), SWI/SNF (α-Snf6), and SAGA (α-Spt3-13myc). For all factors, appropriate strains were grown in raffinose media until mid-log phase, and GAL1 was induced for 1 h by adding galactose. Cells were then washed into media with glucose to repress GAL1 transcription for the indicated times. 3′ GAL1 ORF, telomere (Chr VI-70 bp from the right end), and ACT1 PCR primer sets were included as nonspecific controls. Numbers indicate ratio of percentage of immunoprecipitation values to corresponding nonspecific control. (D) RNA Polymerase II ChIP in wild-type strain grown for the reinduction regimen shown above the panel. Right panel shows quantitation of left panel.

Figure 3.

Histone-modifying enzymes are not essential for rapid GAL1 reinduction. Northern blot analyses are as in Figure 1. (A) Schematic at top illustrates growth media regimen. RNA was isolated from set1Δ, set2Δ, or dot1Δ strains. (B) Identical analysis as in A, but with a rad6Δ strain. (C) Identical analysis as in A, but with a gcn5Δ strain. All Northerns were subsequently probed for ACT1 as a loading control.

Figure 4.

SWI/SNF is essential for GAL1 memory. (A) Northern analyses. Schematic at top illustrates the growth regimen for swi2Δ cells. (B) Comparison of GAL1 induction and reinduction kinetics, averaged over three experiments performed as in A. Error bars represent the standard deviation at each point. Slightly different time points were taken in different experiments, so in these cases no error bars are shown. (C) Isogenic wild-type and swi2Δ strains were grown overnight in glucose and then cells were transferred to galactose media. Both strains showed similar GAL1 induction kinetics after overcoming long-term glucose repression. (D) An intact Swi2p ATPase domain is required for rapid GAL1 reinduction. Northern analysis of RNA isolated from cells harboring a swi2K798A allele, which inactivates the ATPase activity of SWI/SNF.

Figure 5.

SWI/SNF antagonizes ISWI-like remodeling complexes. (A) Northern blot analysis for RNA isolated from wild-type (WT), isw1Δ, and isw1Δ swi2Δ strains. (B) Northern analysis for RNA isolated from isw2Δ and isw2Δ swi2Δ strains. (C) Inactivation of Set1p does not restore rapid reinduction kinetics in the absence of SWI/SNF. (D) Analysis of INO1 expression. The indicated isogenic strains were grown in minimal media containing high (100 μM) or low (10 μM) concentrations of myo-inositol. Blots were reprobed for ACT1 as a control.