A novel mechanism for target gene-specific SWI/SNF recruitment via the Snf2p N-terminus.

Chromatin-remodeling complexes regulate the expression of genes in all eukaryotic genomes. The SWI/SNF complex of Saccharomyces cerevisiae is recruited to its target promoters via interactions with selected transcription factors. Here, we show that the N-terminus of Snf2p, the chromatin remodeling core unit of the SWI/SNF complex, is essential for the expression of VHT1, the gene of the plasma membrane H(+)/biotin symporter, and of BIO5, the gene of a 7-keto-8-aminopelargonic acid transporter, biotin biosynthetic precursor. chromatin immunoprecipitation (ChIP) analyses demonstrate that Vhr1p, the transcriptional regulator of VHT1 and BIO5 expression, is responsible for the targeting of Snf2p to the VHT1 promoter at low biotin. We identified an Snf2p mutant, Snf2p-R(15)C, that specifically abolishes the induction of VHT1 and BIO5 but not of other Snf2p-regulated genes, such as GAL1, SUC2 or INO1. We present a novel mechanism of target gene-specific SWI/SNF recruitment via Vhr1p and a conserved N-terminal Snf2p domain.

INTRODUCTION

The Snf2p protein (synonym Swi2p) from baker’s yeast (Saccharomyces cerevisiae) is the founding member of a huge family of nuclear regulators that are ubiquitous in eukaryotes, but found also in bacteria and archaea (1). The SNF2 gene was originally identified in screens for S. cerevisiae mutants defective in the expression of the HO gene involved in mating type switching [SWI = switching defective (2)] or for mutants affected in the expression of the SUC2 gene encoding the yeast invertase [SNF = sucrose non-fermenting (3)]. Snf2p family proteins are DNA-dependent ATPases that share a conserved region of helicase-related motifs (4). Most of these proteins act as ATP-dependent DNA translocases that can distort DNA and disrupt or remodel the structure of chromatin (RSC) that otherwise blocks the access of transcriptional activators or the basal transcriptional machinery to recognition sequences within target promoters (1).

In the cell, chromatin remodeling factors like Snf2p work in conjunction with other proteins, and the resulting complexes perform critical functions in the maintenance and expression of the genome (5–7). Baker’s yeast has two SWI/SNF-type remodeling complexes, ySWI/SNF (yeast SWI/SNF complex) and RSC, that regulate different, largely non-overlapping sets of genes (6). Based on comparative expression analyses of yeast Δsnf2 and Δswi1 mutants, it can be estimated that ySWI/SNF regulates roughly 6% of all S. cerevisiae genes (8). In contrast to RSC, ySWI/SNF is not essential for the viability of yeast cells (9). These all demonstrates that despite their common chromatin remodeling functions, these complexes are highly specific for different promoters. In human cells, which contain multiple SWI/SNF-like complexes, the distribution of regulatory functions between these complexes is even more elaborate (10).

A central question is, how the specificity of different SWI/SNF complexes for certain sets of promoters is achieved. In vitro analyses revealed that interaction of ySWI/SNF with the activation domain of Gcn4p, a transcriptional activator of amino acid biosynthetic genes, is necessary for recruitment of ySWI/SNF to Gcn4p-regulated promoters (11,12). Similarly, ySWI/SNF was shown to interact with the acidic activation domains of Swi5p known to recruit ySWI/SNF to the yeast HO promoter, or of Hap4p that regulates respiratory functions (12). Interaction of ySWI/SNF was even observed with the acidic activation domain of VP16, a transcriptional regulator of Herpes simplex virus (12,13), or with Gal4p-AH, a fusion of the Gal4p DNA-binding domain and a synthetic acidic helix (AH) (13). This led to the model that activation domain-mediated targeting of ySWI/SNF to promoters within nucleosome arrays is responsible for transcriptional activation.

In screens for ySWI/SNF components directly involved in this physical interaction with activation domains, the subunits Swi1p, Snf2p and Snf5p from S. cerevisiae could be characterized as interactors (14,15). Further analyses with in vitro-translated fragments of these three subunits identified two specific contact sites, one in the N-terminal domain of Snf5p and one in the second quarter of Swi1p (16,17).

No contact-specific interaction site could be determined for Snf2p from S. cerevisiae. However, interaction was demonstrated for one of two Snf2p orthologs from the SWI/SNF complex from man. These orthologs are human Brahma [hBrm (18)] and Brahma/SWI2-related gene 1 [BRG1 (19)]. BRG1 was shown to interact via its N-terminal helicase/SANT-associated (HAS) domain with the nuclear glucocorticoid receptor (20). This receptor targets the mammalian SWI/SNF complex to promoters containing glucocorticoid response elements (21).

SWI/SNF complexes can also repress transcription (5,10). The mammalian NuRD (nucleosome remodeling and histone deacetylation) complex was the first SWI/SNF-related complex shown to contain a histone deacetylase activity that is thought to be responsible for repression (22). In S. cerevisiae, ySWI/SNF is involved in the repression of the SER3 gene that encodes a phosphoglycerate dehydrogenase catalyzing a step in serine biosynthesis (23). However, SER3 repression depends on the enhanced expression of the non-coding SRG1 gene (SER3 regulatory gene 1) 5′ from SER3 (24). ySWI/SNF is necessary for the induction of SRG1 expression and, thus, only indirectly for the repression of SER3 (25).

We previously characterized the product of the S. cerevisiae VHT1 gene, Vitamin H Transporter 1, which encodes an H+/biotin symporter of the yeast plasma membrane (26). VHT1 is expressed only at low biotin concentrations. We demonstrated that the expression levels of VHT1 and BIO5 (Biotin biosynthesis 5), the gene of a plasma membrane transporter for the biotin biosynthetic precursor 7-keto-8-aminopelargonic acid (27), are regulated by Vhr1p [VHT1 regulator 1 (28)] that binds to a vitamin H-responsive element (VHRE) in the VHT1 and BIO5 5′-untranslated regions (28,29).

Here, we show that ySWI/SNF is essential for the induction of VHT1 and BIO5 expression, and that Snf2p is a central player in this regulation. An snf2 mutant was identified in a complementation screen for S. cerevisiae mutants that had lost the capacity to express the gene of the green fluorescent protein (GFP) from the VHT1 promoter. Unexpectedly, this snf2 mutation turned out to specifically affect only the induction of VHT1 and BIO5, but not the induction of SUC2, INO1 and GAL1 or the repression of SER3. Sequencing of the snf2 mutant allele identified a point mutation near the N-terminus of Snf2p that converts a conserved arginine at position 15 into a cysteine (R15C). Our data reveal a novel mechanism for target gene-specific ySWI/SNF recruitment in yeast via the N-terminus of Snf2p.

MATERIALS AND METHODS

Yeast strains and plasmids

Yeast strains used in this study are listed in Table 1. Strains FY3 and YBC3010 were gifts from Bradley Cairns (Salt Lake City, University of Utah, UT, USA).

Table 1.

Yeast strains used in this study

Strain	Genotype	Reference/Source
JS91.15-23	MATα, his3-Δ1, leu2-3, trp1-289, ura3-52, can1	(51)
JSYΔvht1	Isogenic to JS91.15-23; vht1::HIS3	(26)
MWY760gc	Isogenic to JS91.15-23; vht1p::ADH1p	(28)
AMYmut153	EMS mutant of MWY760gc; snf2-153	This study
BY4741	MATa, his3-Δ1, leu2-Δ0, met15-Δ0, ura3-Δ0	EUROSCARF
Y01586	Isogenic to BY4741; snf2::kanMX4	EUROSCARF
Y01250	Isogenic to BY4741; swi3::kanMX4	EUROSCARF
FY3	MATa, ura3-52	(52)
YBC3010	Isogenic to FY3; SNF2-13myc::kanMX	Bradley Cairns
MWY8003	Isogenic to YBC3010; vhr1::klURA3	This study

For deletion of VHR1 in YBC3010, the Klyveromyces lactis URA3 disruption cassette of pUG72 (30) was amplified by PCR with the primers YIL056W-ko-5′ and YIL056W-ko-3′. Disruption of VHR1 was confirmed by PCR with appropriate primers. The resulting strain was called MWY8003.

The mutant strain AMYmut153 (snf2-R) was complemented with two plasmids from a YEp24 (2-µ plasmid)-based yeast genomic library (31) that contained SNF2 as the only full-length open reading frame (ORF), including 3475 bp and 191 bp, or 3466 bp and 205 bp of upstream and downstream sequence, respectively. The SNF2 gene, together with 2881 bp of upstream and 191 bp of downstream sequence, was isolated by restriction digest of one of the complementing plasmids with XmaI and was cloned into the XmaI-linearized single copy vector YCplac33 (32) yielding plasmid pMW814.

The mutation of the snf2-R allele was introduced into pMW814 by site-directed mutagenesis with the primer ScSNF2+26f-mut153+T (Table 2) using the QuikChange Multi Kit (Stratagene, La Jolla, CA, USA). This primer also introduced a second, silent mutation into the SNF2 ORF that destroyed a HincII to allow identification of mutated plasmids by restriction digest analysis. Sequence analysis of the resulting ORF confirmed that the encoded Snf2 protein differed only in the desired Arg to Cys exchange from the WT protein encoded by the original plasmid. The resulting plasmid was called pMW817.

Table 2.

Oligonucleotides used in this study

Oligonucleotide	Sequence
ScACT1g-4f	AACAATGGATTCTGAGGTTGCTGC
ScACT1g+498r	TCTCTTGGATTGAGCTTCATCACC
ScACT1g+611f	CCCTGTTCTTTTGACTGAAGC
ScACT1g+813r	AACCAGCGTAAATTGGAACG
ScBIO5g+1040f	AAACGGCAGTCGTTGAGTCT
ScBIO5g+1244r	TTTCCCAAGGAACCCTCTTT
ScGAL1c+1379	CATATGGTTCCCGTTTGACC
ScGAL1c+1576r	ATAGACAGCTGCCCAATGCT
ScINO1c+1316f	CTTTACTGGCTACGCCCTTG
ScINO1c+1531r	CTAAGGCGGTTCTTTGCTTG
ScSER3c+1084f	GAAGTGGCTACCGCATTGTC
ScSER3c+1286r	TCGGAAAATTGCTTTTCGAT
ScSNF2+26f-mut153+T	GCAACGAAGAGGTCAATTG CTGCTATTTAAGATGG
ScSNF2c+5047f	ACTTCAAGCGTGGCTGAATC
ScSUC2g-1f	GATGCTTTTGCAAGCTTTCCTT
ScSUC2g+207r	ACCCCATACGGTGTCATTTG
ScVHT1g+607f	GGCTGGACTTGGTTCACATT
ScVHT1g+808r	CTGAAGTCACGGAACCCAAT
ScVHT1g-361f	TGATTGCCTCCATTGCTACA
ScVHT1g-144r	GCCTTGCATCTCATATTTCG
S2-SNF2	GTCTACGTATAAACGAATAAGT ACTTATATTGCTTTAGGAAG
	GTAATCGATGAATTCGAGCTCG
S3-SNF2	TCTTTCACAGATGAAGCGGACTCGAGCATGACAGAAGCGAG
	TGTACGTACGCTGCAGGTCGAC
TRP primer	GCTATTCATCCAGCAGGCCTC
YIL056Wg+1411f	CCCACGGTGTCTGAAGTTTT
YIL056Wg+1617r	GGGCACTGGTATTTTTGCAT
YIL056W-ko-5′	TTCTACCTATTCTTCTTCGTTGC TTATATCCTATACACATCAG
	CTGAAGCTTCGTACGC
YIL056W-ko-3′	AATGTGTATTAGGCAGTTATGTA TGTAGACAATTTTGTTTGC
	ATAGGCCACTAGTGGATCTG

A partial genomic sequence for an Snf2p-myc tag was amplified with the primers S3-SNF2 and S2-SNF2 from plasmid pYM6 (33) and inserted via homologous recombination into the SNF2 locus of MWY760gc yielding strain MWY760gc-SNF2-myc. The 3′-end of this modified SNF2 sequence was amplified by PCR (ScSNF2c+5047f and TRP primer), cloned into pJet1.2, sequenced and ligated via the BglII and XhoI restriction sites of this fragment into the BamHI/XhoI-cut yeast single copy vectors pMW814 or pMW817 that contain WT SNF2 or mutated snf2-R genomic sequences, respectively. The resulting single copy plasmids with the BglII/XhoI insertions encoded Snf2p-myc or snf2p-R15C-myc proteins.

Media, EMS mutagenesis, growth tests and induction tests

Ethyl methanesulfonate (EMS) mutagenesis and the complementation screening procedure were carried out as described (28). The stability of the identified mutants (= no revertants) and the absence of mutations in the GFP sequence were tested as published (28).

Rich medium and synthetic media with defined vitamin concentrations were prepared as described earlier (28). For growth tests on petri plates, yeast cells were suspended to an OD600 of 0.2 in ice cold water, diluted 1 : 8, 1 : 64 and 1 : 512, and 3 µl of each dilution were spotted onto the agar. For the growth tests with defined biotin concentrations, 1% agarose was used instead of agar agar. Media for induction tests of relevant genes were prepared as described elsewhere (28,34–36). Where rich medium could not be used for the induction tests (e.g. maintenance of plasmids), synthetic medium with 0.67% yeast nitrogen base and 2% casamino acids (Difco, Augsburg, Germany) was used. For the induction of GAL1, SUC2 or INO1, yeasts were grown for one generation time in the respective medium. Induction of VHT1 was performed as described (28).

Western blot analyses

Western blots were made after 8% sodium dodecylsulfate (SDS)–polyacrylamide gel electrophoresis with a monoclonal primary antibody against myc (9E10; Sigma, Taufkirchen, Germany) and a polyclonal rabbit antibody against purified S. cerevisiae alcohol dehydrogenase (gift from Prof. Christian Koch, University of Erlangen-Nuremberg, Germany). Secondary horse-radish peroxidase coupled anti-mouse antibody and anti-rabbit antibody were from Sigma (Taufkirchen, Germany). Antibody-binding was visualized with the ECL Lumilight-Kit from Roche (Mannheim, Germany).

Chromatin immunoprecipitation

Chromatin immunoprecipitation (ChIP) was performed essentially as described (37) after a protocol adapted from (38) and (39). Antibodies used for ChIP assays were anti-myc 9E11 (Dianova, Hamburg, Germany) and αGFP MAB3580 (Chemicon/Millipore, Schwalbach, Germany). Dynabeads Protein G (Dynal/Invitrogen, Karlsruhe, Germany) were used as matrix. Primers used for amplification were ScVHT1g-361f and ScVHT1g-144r for the VHT1 promoter, and ScACT1g+611f and ScACT1g+813r for normalization to the ACT1 ORF. Real-time PCR was carried out as described (28).

RNA isolation and real-time reverse transcription PCR

RNA isolation and RT-PCRs were performed as described earlier (28). Oligonucleotides used for the amplification were ScACT1g-4f and ScACT1g+498r for ACT1, ScVHT1g+607f and ScVHT1g+808r for VHT1, ScBIO5g+1040f and ScBIO5g+1244r for BIO5, YIL056Wg+1411f and YIL056Wg+1617r for VHR1, ScSUC2g-1f and ScSUC2g+207r for SUC2, ScGAL1c+ 1379 and ScGAL1c+1576r for GAL1, ScINO1c+1316f and ScINO1c+1531r for INO1 and ScSER3c+1084f and ScSER3c+1286r for SER3.

RESULTS

Identification of Snf2p as a factor involved in VHT1 induction at low biotin concentrations

In a previous study, we identified the S. cerevisiae transcriptional regulator Vhr1p, the prototype of a new family of transcription factors, and showed that it is essential for induction of VHT1 and BIO5 at low biotin concentrations (28). VHT1 and BIO5 encode plasma membrane-localized transporters for biotin (vitamin H; Vht1p) or for its biosynthetic precursor 7-keto-8-aminopelargonic acid (Bio5p). The screen was performed with a strain (MWY760gc) that expressed VHT1 under the control of the constitutive ADH1 promoter allowing growth on low and high biotin. In addition, MWY760gc harbored a plasmid with a VHT1 promoter (pVHT1)::GFP reporter cassette that drove expression of GFP only on low biotin (28). For the identification of VHR1, 150 000 EMS-mutagenized cells were analyzed.

To identify additional components of the biotin-dependent signaling cascade, we performed a new EMS mutagenesis on MWY760gc cells and screened additional 600 000 clones for loss of GFP fluorescence on low biotin. Fifteen newly obtained mutants were transformed with an S. cerevisiae genomic library (31) and a minimum of 200 000 transformants of each line was screened for recovery GFP fluorescence on low biotin. With 10 lines the complementation was successful, and for 8 lines the complementing sequences could be determined and assigned to complementation groups. The largest complementation group with 6 lines contained fragments carrying the VHR1 gene (data not shown) confirming our previous results (28). One line was complemented by the ECM1 gene (40) encoding a poorly characterized protein involved in cell wall biosynthesis (data not shown). This gene was not further analyzed. The last line, AMYmut153, was complemented by two different library inserts, each containing SNF2 as the only full-length gene (Figure 1A).

Figure 1.

Identification of the ySWI/SNF complex as an essential component of VHT1 induction. (A) SNF2 complements the non-fluorescence phenotype of the EMS mutant AMYmut153. Images show the GFP fluorescence resulting from a pVHT1::GFP construct in the WT (MWY760gc) or in the AMYmut153 mutant. Strains were complemented with one of the SNF2-containing multicopy (mc) plasmids identified in the complementation screen or with a single copy (sc) plasmid (pMW814) harboring SNF2. Vector controls (A) carried the plasmid YEp24. Numbers show insert lengths relative to the SNF2 start ATG. Plates were photographed in white light and under GFP excitation light. (B) Real-time RT-PCR analysis of VHT1 mRNA levels in the WT and in mutants lacking subunits of the ySWI/SNF complex (Δsnf2 or Δswi3). RT-PCRs were performed with total RNA from yeasts grown on SD medium containing the indicated biotin concentrations. Results were standardized to ACT1 mRNA levels (n = 3; ±SE).

To independently prove or disprove the identity of SNF2 as the complementing gene, the SNF2 coding sequence plus 2881 bp upstream and 191 bp downstream sequences was cloned into the single copy vector YCplac33 (32). The resulting plasmid, pMW814, was used to transform AMYmut153. GFP fluorescence was recovered only in the pMW814-transformed strain but not in the vector control (Figure 1A), which confirmed the importance of Snf2p for VHT1 expression. We also analyzed expression of VHT1 in the Δsnf2 deletion mutant Y01586 (EUROSCARF, Frankfurt/Main, Germany) by quantitative RT-PCR (qRT-PCR; Figure 1B). In agreement with the previous data, the Δsnf2 strain failed to induce VHT1 under low biotin conditions. We finally tested if a mutation in another subunit of the ySWI/SNF complex will also abolish VHT1 induction. We, therefore, analyzed VHT1 expression in the Δswi3 mutant Y01250 (EUROSCARF) on low and high extracellular biotin by qRT-PCR (Figure 1B). In fact, also the Δswi3 mutant was unable to induce VHT1 under low biotin conditions, underlining the essential role of the ySWI/SNF complex for induction of VHT1.

A single amino acid exchange in the N-terminus of Snf2p is responsible for the lack of VHT1 induction

When we sequenced the snf2 mutant allele of the EMS-mutant strain, AMYmut153 from −830 bp upstream from the start-ATG to 52 bp downstream of the stop-codon, we found a single point mutation in the codon for the arginine residue at position 15 in Snf2p. This mutation changed CGC into TGC, which replaced the arginine at position 15 by a cysteine yielding snf2p-R15C. A comparison of the Snf2p sequences from different fungi demonstrated that this amino acid is highly conserved in Snf2 proteins from other species of the order Saccharomycetales (Figure 2A). Occasionally, this arginine can be replaced by a lysine residue (see the sequence of the Candida dubliniensis Snf2 protein in Figure 2A).

Figure 2.

Alignment of N-terminal Snf2 protein sequences and comparative expression analyses. (A) The first 70 amino acids of S. cerevisiae Snf2p and the corresponding residues of Snf2 proteins from other members of the order Saccharomycetales. Amino acids identical in all proteins are highlighted. The arginine at position 15 in Snf2p is highlighted in black, predicted α-helices are underlined. (B) Model of the predicted secondary structure for Snf2p EVA domains. The position of the conserved arginine/lysine is indicated. (C) Western blot analysis of the relative amount of Snf2p-myc and snf2p-R15C-myc in extracts of cells expressing the corresponding genes. An extract of cells expressing SNF2 without a C-terminal fusion is shown to demonstrate the specificity of the anti-myc antiserum (anti-myc). Each lane was loaded with 5 µg of protein extract. Signals obtained from the same blot with an anti-Adh-antiserum (anti-Adh) are shown as loading controls.

Sequence alignments revealed that this arginine is embedded in a conserved domain of about 50 amino acids at the very N-terminus of these Snf2 proteins (Figure 2A). This conserved N-terminal domain is found exclusively in yeast Snf2 proteins, and is followed by a 100–250 amino acid region of low sequence similarity. To find possibly conserved structural motifs within this region, we performed in silico analyses on the first 100 amino acid of S. cerevisiae Snf2p and of 9 other Snf2 proteins with the Protein Homology/analogY Recognition Engine [PHYRE; http://www.sbg.bio.ic.ac.uk/phyre/ (41); output files are presented in Supplementary Figure S1]. In fact, all analyses revealed an identical structure for each of these sequences (Figure 2B). An α-helix of about 17 amino acids (α1) is predicted to start 6–10 amino acids downstream from the start methionine, and a short stretch of 6–7 disordered residues separates this first helix from a second, slightly longer α-helix (α2) of 18–25 amino acids. This α2-helix is followed by a longer disordered region (Figure 2B and Supplementary Figure S1).

The helices and the disordered sequences are predicted with maximal probability. In all predictions, the arginine residue modified in our EMS mutant and the conserved arginine or lysine residues in the other Snf2 proteins are located in the first half of the α1-helix. Based on the functional characterization of the snf2p-R15C protein (see below), we refer to this conserved structural motif in Snf2p as EVA (expression of VHT1 activating) domain.

To exclude the possibility that the observed snf2-R mutation results in a reduced amount of snf2p-R15C protein, e.g. as a result of reduced stability of the mutant protein, we compared the amounts of Snf2p and snf2p-R15C proteins in yeast cells. To this end, we generated single copy plasmids encoding modified Snf2p and snf2p-R15C proteins with C-terminal myc tags (Snf2p-myc and snf2p-R15C-myc) and expressed these constructs in the Δsnf2 mutant strain Y01586. The amounts of Snf2p-myc and snf2p-R15C-myc were compared on western blots treated with anti-myc antiserum (anti-myc-AB). Figure 2C shows that the amount of snf2p-R15C-myc proteins is not decreased compared with the Snf2p-myc control.

The ySWI/SNF complex is recruited to pVHT1 under inducing conditions

To test, if the observed Snf2p mutation reduces the expression of VHT1 indirectly by affecting the expression of VHR1, the gene for the transcriptional regulator of VHT1, we compared VHR1 mRNA levels in the Δsnf2 strain Y01586 and in the corresponding wild type (WT) strain BY4741 (EUROSCARF) on low biotin. qRT-PCR reactions revealed no differences in the expression of VHR1 (Figure 3) suggesting that the snf2-R mutation affects VHT1 expression directly, and that WT ySWI/SNF may bind directly to pVHT1.

Figure 3.

qRT-PCR of VHR1 mRNA levels. VHR1 mRNA levels were determined in the Δsnf2 deletion mutant and in the corresponding WT on low biotin medium. Results were standardized to ACT1 mRNA (n = 3; ± SE).

We tested this hypothesis in ChIP analyses with strain YBC3010 that contains a myc-tagged version of Snf2p (Snf2p-myc). We assayed binding of Snf2p-myc to the vitamin H-responsive element (VHRE) relative to its binding to the ACT1 ORF. We could confirm that Snf2p-myc is present at the VHRE at low, VHT1-inducing biotin concentrations (Figure 4). The signal obtained at non-inducing, high biotin concentrations was significantly lower and comparable with the signals obtained in two negative controls (Snf2p with no myc-tag or Snf2p-myc with a control antiserum against GFP, αGFP; Figure 4A). We also assayed binding of Snf2p-myc in a Δvhr1 deletion strain. In this strain, Snf2p-myc was not recruited to the VHRE, even when the cells were grown on low biotin. In summary, these data demonstrate (i) that Snf2p occupies the pVHT1, (ii) that this recruitment occurs at or near the VHRE element, (iii) that the presence of Vhr1p is essential to recruit Snf2p to the VHRE and (iv) that Vhr1p targets Snf2p to the VHT1 promoter only at low biotin concentrations, i.e. under conditions, when VHT1 expression is induced.

Figure 4.

ChIP analyses with Snf2p-myc and snf2p-R15C-myc. (A) ChIPs of Snf2p-myc were performed in a VHR1 WT (YBC3010), a Δvhr1 mutant (MWY8003) and in a VHR1 control strain (FY3) expressing untagged SNF2. ChIP was also performed in the SNF2-myc, VHR1 strain (YBC3010) with a control antibody (αGFP). Yeast cells were grown on SD medium containing the indicated biotin concentrations. Binding of Snf2p-myc to the VHT1 promoter was assayed relative to its binding at the ACT1 ORF. Input control and immunoprecipitated DNA (IP) were amplified with primers spanning a region containing the VHT1 promoter vitamin H-responsive element (VHRE). Snf2p occupancy was calculated as ratio of the VHT1 signal in the IP-to-input samples and normalized for the corresponding ratio calculated for ACT1. The mean value for Snf2p-myc under inducing conditions was set to 100% (n = 3; ±SE). (B) ChIPs of Snf2p (control), Snf2p-myc and snf2p-R15C-myc were performed in the Δsnf2 mutant strain Y01586 expressing SNF2 (control), SNF2-myc or snf2-R from a single copy plasmid under the control of the SNF2 promoter. Binding of Snf2p-myc and snf2p-R15C-myc to the VHT1 promoter was assayed as described in (A). The mean signal obtained with the control strain expressing untagged SNF2 was set to 1 (n = 3; ±SE).

snf2p-R15C cannot occupy pVHT1

The finding that Snf2p occupies the pVHT1 (Figure 4A) suggested that this recruitment of Snf2p might be specifically affected by the mutation in the snf2p-R15C protein. To test this hypothesis, we performed additional ChIP analyses with three yeast strains expressing SNF2, SNF2-myc or snf2-R in the background of the Δsnf2 mutant strain Y01586 at low, VHT1-inducing biotin concentrations. Figure 4B demonstrates that as shown in Figure 4A Snf2p-myc occupies the pVHT1. Most importantly, however, Figure 4B also demonstrates that snf2p-R15C is not recruited to the VHRE. Together with the data shown in Figure 4A, this result shows that snf2p-R15C fails to induce VHT1 expression, because its R15C mutation abolishes the Vhr1p-mediated recruitment of Snf2p to the VHRE element in pVHT1.

ySWI/SNF mutants cannot grow on low biotin concentrations due to the lack of VHT1 expression

Based on the data presented so far (Figures 1B, 3 and 4), we expected that snf2 mutants should be severely affected in their capacity to grow on low biotin. We, therefore, compared the growth of the Δsnf2 mutant, of the corresponding SNF2 WT and of a Δvht1 deletion mutant [JSYΔvht1 (26)] on SD media with decreasing biotin concentrations. As expected, the Δvht1 mutant did grow only on high biotin (2 mg biotin l−1; Figure 5). Also the Δsnf2 cells failed to grow on medium with low biotin (0.02 µg biotin l−1; Figure 5). The growth difference between Δvht1 and Δsnf2 cells on medium biotin concentrations (200 µg biotin l−1) results from a weak basal activity of pVHT1 in Δsnf2 cells and the complete absence of Vht1p activity in Δvht1 cells.

Figure 5.

Growth test on SD medium with varying biotin concentrations. The Δsnf2 mutant was transformed with plasmid pVHT1oe, expressing VHT1 from the constitutive ADH1 promoter (Δsnf2 + VHT1oe). To allow growth on the same medium, the deletion mutants Δvht1 and Δsnf2 and the WT were transformed with the empty vector NEV-E. Cells were spotted on SD medium containing the indicted biotin concentrations after serial 1:8 dilutions, grown for 3 days and photographed.

At high and medium biotin concentrations, the Δsnf2 cells grow slower than the WT (Figure 5), a well-known behavior of ySWI/SNF mutants on glucose media (42). Transformation of the Δsnf2 mutant with the pVHT1oe plasmid that expresses VHT1 under the control of the constitutive ADH1 promoter (26) complemented only the growth defect on low biotin, but not the slower growth of the Δsnf2 mutant at higher biotin concentrations (Figure 5). The same results were obtained in experiments performed with the Δswi3 deletion mutant Y01250 (not shown). In summary, these results confirmed that mutations in ySWI/SNF subunits result in a lack of biotin-dependent VHT1 induction (Figure 1B) and consequently in a reduced capacity to grow on low biotin (Figure 5).

The snf2-R mutant has an unusual phenotype

Compared with WT cells, Δsnf2 mutants exhibit slow growth on glucose [Figure 5 and (42,43)], fail to grow on galactose and raffinose (42,43), and are auxotrophs for inositol (44). This was re-confirmed in comparative growth analyses on media with galactose (YPGal) or raffinose (YPRaf) as sole carbon source, on medium w/o inositol (SD w/o inositol), or on glucose-containing yeast peptone dextrose (YPD) medium (Figure 6). Surprisingly, however, the snf2-R mutant did not show any of these phenotypes and behaved like the WT on all media (Figure 6). Moreover, this growth could not be further improved by expressing WT SNF2 from one of the multicopy plasmids (snf2-R + SNF2mc) identified in the complementation screening (Figure 1A) or from the single copy plasmid (snf2-R + SNF2sc) used to express SNF2 in the AMYmut153 strain (Figure 1A). This demonstrated that the snf2-R mutation does not exhibit any of the previously published growth phenotypes of a Δsnf2 deletion mutant. It rather suggested that the snf2-R mutation is hypomorphic allele that specifically affects the regulation by Vhr1p and the targeting of ySWI/SNF to the VHRE.

Figure 6.

Growth test to assay phenotypes of ySWI/SNF mutants. Yeast strains were grown on the indicated media to assay the following phenotypes: slow growth (YPD), inability to use the sugars galactose or raffinose as carbon sources (YPGal and YPRaf), and inositol auxotrophy (SD w/o inositol). The SNF2 WT (top) and the Δsnf2 mutant (bottom) were used as controls to assay the growth behavior of the snf2-R mutant. In addition, the effects of SNF2 expression from one of the complementing multicopy plasmids (snf2-R + SNF2mc; Figure 1A) or from an SNF2-containing single copy plasmid (snf2-R + SNF2sc) were analyzed. To allow growth on the same medium, control strains were transformed with the empty plasmid YEp24. Cells were spotted on plates after serial 1:8 dilutions.

snf2p-R15C fails to induce VHT1 and BIO5, but mediates WT-like expression of the ySWI/SNF-dependent genes SUC2, GAL1, INO1 and SER3

The unexpected WT-like growth of the snf2-R mutant on YPGal and YPRaf medium or on SD medium w/o inositol pointed toward a normal, WT-like expression of the responsible ySWI/SNF target genes, GAL1, SUC2 and INO1. We tested the expression of these genes by RT-PCR in the SNF2 WT, the Δsnf2 strain and the snf2-R mutant that were grown on media inducing the expression of SUC2 (YPRaf), GAL1 (YPGal) or INO1 (SD medium w/o inositol). Moreover, we analyzed the expression of SER3, a gene of the serine biosynthetic pathway known to be de-repressed in Δsnf2 mutants on rich medium (25). As expected, the mRNA levels of all genes were comparable in the SNF2 WT strain and in the snf2-R mutant (Figure 7), whereas the Δsnf2 deletion mutant failed to induce expression of GAL1, SUC2 and INO1, and did not repress SER3.

Figure 7.

Comparative qRT-PCR analysis of the expression of typical ySWI/SNF target genes. RT-PCR analysis of mRNA levels of SUC2, GAL1 and INO1 under inducing conditions, and of SER3 under repressing conditions. The snf2-R mutant strain, the Δsnf2 deletion mutant and the WT strain were first grown in YPD medium, cultures were split and incubated in the indicated media.

We could not include analyses of VHT1 expression in Figure 7, as the snf2-R mutant used in these analyses (AMYmut153; see Figure 1) expressed VHT1 from the constitutive ADH1 promoter. For a direct comparison of the snf2-R-dependent induction of VHT1 with the induction/repression of other ySWI/SNF target genes in WT and Δsnf2 cells, we introduced the snf2-R mutation into the single copy plasmid pMW814 used in Figure 1 by site-directed mutagenesis. We next transformed the Δsnf2 strain Y01586 either with this plasmid to obtain the snf2-R genotype, with the SNF2 WT plasmid to regain the WT gene, or with the empty vector to maintain the Δsnf2 genotype. To study the expression of VHT1, GAL1 and SER3 by quantitative real-time RT-PCR, these three strains were grown either on SD medium with low biotin (0.2 µg l−1; Figure 8A), on galactose medium (Figure 8C), or on CAA medium (Figure 8D). Moreover, we included analyses of BIO5 expression, another gene induced by the Vhr1p transcription factor on low biotin [Figure 8B and ref. (28)].

Figure 8.

qRT-PCR analysis of mRNA levels of VHT1, BIO5 and GAL1 under inducing conditions and of SER3 under repressing conditions. The Δsnf2 deletion mutant was transformed with the single copy plasmids harboring the SNF2 gene, with the snf2-R mutant allele, or with the YEp24 vector (maintaining the Δsnf2 genotype). After growth in CAA liquid medium, cultures were split and incubated in the following media: (A) and (B) in low biotin medium, (C) in CAA medium supplemented with 2% galactose, (D) CAA medium with 2% glucose. Results were standardized to ACT1 mRNA levels; expression levels in SNF2 complemented cells were set to 100% (n = 3; ±SE).

As expected, expression of VHT1 was very high in the SNF2-complemented Δsnf2 mutant, but negligible in the vector control and in the snf2-R complemented strain (Figure 8A), which confirmed our previous experiments (Figures 1, 4 and 5). A very similar result was obtained for BIO5 (Figure 8B), which demonstrated that the snf2-R mutation affects the expression of other Vhr1p-regulated genes in the same way as VHT1. In contrast, and in agreement with the data shown in Figures 6 and 7, snf2-R and Δsnf2 differed in their capacity to induce GAL1 on galactose medium (Figure 8C), and in their capacity to repress SER3 (Figure 8D). The expression of both genes was identical in snf2-R complemented and SNF2 complemented cells. Again this confirmed that the snf2p-R allele encodes a protein capable to regulate well-known ySWI/SNF target genes, such as GAL1 or SER3. In contrast, its capacity to induce Vhr1p-regulated genes, such as VHT1 and BIO5, is completely lost.

DISCUSSION

The nucleosome remodeling ATPase Snf2p represents the central catalytic subunit of the ySWI/SNF complex that regulates a large number of S. cerevisiae genes (6,7,45). In a complementation screen for transcriptional regulators of VHT1, the gene of the biotin/H+ symporter, we identified a novel snf2 mutant allele (Figure 1A). We verified the importance of Snf2p and the ySWI/SNF complex for VHT1 induction in additional analyses of Δsnf2 and Δswi3 mutants (Figure 1B) and characterized VHT1 and BIO5 as previously unidentified target genes of ySWI/SNF. They were not identified in former large-scale searches for ySWI/SNF target genes (5), as VHT1 and BIO5 stay repressed under the high biotin conditions applied in these experiments.

The R15C mutation in the newly identified snf2p mutant allele is located in the first of two α-helices in the structurally conserved EVA domain found in the Snf2p proteins of all Saccharomycetes (Figure 2A and Supplementary Figure S1). Occasionally, e.g. in Snf2p of Debaryomyces hansenii (accession number XP_461680) or Candida dubliniensis (accession number CAX43468), this conserved arginine can be replaced by a lysine residue. Both, this conserved basic residue and the EVA domain, are absent from other chromatin remodeling ATPases of S. cerevisiae including the less closely related Isw1p, Isw2p, Chd1p, Ino80p and Swr1p proteins, and also the closely related Sth1p protein of the RSC complex (46). This points toward a conserved function of Snf2p N-termini.

Binding of Snf2p to the VHT1 promoter depends on the presence of Vhr1p

ChIP analyses confirmed the presence of Snf2p at the VHT1 promoter in a 200-bp region (Figure 4) containing the VHRE essential for biotin-dependent VHT1 induction (28,29). This binding of Snf2p was detected only under inducing conditions, i.e. at low biotin. The binding of Snf2p also depended strictly on the presence of Vhr1p, the transcriptional regulator. No binding was detected in a Δvhr1 deletion mutant (Figure 4) that is unable to induce VHT1 (28). This demonstrated that Vhr1p is essential for Snf2p recruitment to the VHT1 promoter. It is, however, not yet possible to distinguish whether there is a direct or indirect interaction between Vhr1p and the N-terminus of Snf2p.

The snf2-R mutant represents a hypomorphic allele

Surprisingly, the snf2-R mutant did not show the slow growth phenotype on glucose, the growth defects on galactose or raffinose (42), and the inositol auxotrophy (44) characteristic for Δsnf2 strains (Figure 6). Moreover, expression of genes (GAL1, INO1 and SUC2) that are miss-regulated in the Δsnf2 mutants and responsible for these growth phenotypes was not affected in the snf2-R mutant (Figure 7). Similarly, expression of SER3, a gene repressed in SNF2 WT strains but not in the Δsnf2 mutant (23) was not affected in the snf2-R mutant. In contrast, both the snf2-R mutation and the Δsnf2 deletion resulted in a loss of VHT1 and BIO5 induction (Figure 8). This characterized snf2-R as a hypomorphic allele that affects Snf2p function only for Vhr1p-regulated target promoters.

Induction of Vhr1p target genes by Snf2p reveals a novel mechanism of ySWI/SNF recruitment

In summary, the protein encoded by the snf2-R allele acts like WT Snf2p on most ySWI/SNF target promoters, but it is unable to induce VHT1, BIO5 and putatively also other Vhr1p-regulated genes. Thus, the R15C mutation and the conserved EVA domain (Figure 2 and Supplementary Figure S1) seem to be crucial for VHT1 and BIO5 induction, but are dispensable for the regulation of most other ySWI/SNF target genes. This suggests that the transcriptional regulator Vhr1p recruits the ySWI/SNF complex to the VHT1 promoter via its N-terminal EVA domain.

This is in line with a report that suggested multiple functional domains for Snf2p, especially in its N-terminus (47). In fact, the N-terminal region preceding the Snf2p ATPase domain is accessible for protein/protein interactions, as a 50-amino acid stretch (also named domain 1 = residues 239–289) was shown to interact with Snf11p, another ySWI/SNF subunit (48). Moreover, the crystal structure of the bacterial Snf2p homolog RapA, the only available crystal structure of a full-length Snf2 protein, demonstrates that the N-terminus of Snf2p is freely accessible and probably points away from the DNA helix (49).

In summary, the characterization of the mutant allele snf2-R reveals a novel mechanism of ySWI/SNF recruitment to its target promoters. Interactions between other subunits of the ySWI/SNF complex and transcriptional regulators have been described. The Gal4p transcription factor, for example, recruits ySWI/SNF to the GAL1 promoter by interactions with the subunits Swi1p and Snf5p (17), and Gcn4p recruits ySWI/SNF via the subunits Snf5p, Swi3p and Snf6p (50). Interestingly, Snf2p was shown to be important for the repression of SER3 whereas neither Swi1p nor Swi3p is required for this repression (23). For Snf2p a direct contact to or at least a close proximity with transcriptional activators has also been suggested (15); however, a specific interaction site has so far not been identified and a target gene-specific role of the ySWI/SNF ATPase unit has never been demonstrated. Obviously, the N-terminal site of Snf2p shown to be important for the induction of VHT1 or BIO5 is not involved in the Snf2p-dependent repression of SER3 (Figure 8).

To our knowledge, a hypomorphic allele affecting only a subset of ySWI/SNF target genes has never been reported. The identification and characterization of the snf2-R allele provides new insights into the mechanisms allowing gene-specific action of a machinery that is involved in the global regulation of gene expression.

SUPPLEMENTARY DATA

Supplementary Data is available at NAR Online.

FUNDING

This work was funded by the Deutsche Forschungsgemeinschaft (grant SA 382/17-1 to N.S.). Funding for open access charge: German Research Foundation (DFG).

Conflict of interest statement. None declared.