The Swi-Snf chromatin remodeling complex mediates gene repression through metabolic control

Abstract In eukaryotes, ATP-dependent chromatin remodelers regulate gene expression in response to nutritional and metabolic stimuli. However, altered transcription of metabolic genes may have significant indirect consequences which are currently poorly understood. In this study, we use genetic and molecular approaches to uncover a role for the remodeler Swi-Snf as a critical regulator of metabolism. We find that snfΔ mutants display a cysteine-deficient phenotype, despite growth in nutrient-rich media. This correlates with widespread perturbations in sulfur metabolic gene transcription, including global redistribution of the sulfur-sensing transcription factor Met4. Our findings show how a chromatin remodeler can have a significant impact on a whole metabolic pathway by directly regulating an important gene subset and demonstrate an emerging role for chromatin remodeling complexes as decisive factors in metabolic control.


INTRODUCTION
Metabolic perturbations are commonly found in cancer.The tendency for tumor cells to favor aerobic gl ycol ysis over respiration (the 'Warburg effect') was an early clue that metabolic reprogramming occurs in transformed cells ( 1 , 2 ).Impair ed r espiration can be due in part to alter ed transcription of gl ycol ytic enzymes under the control of the oncoprotein c-Myc ( 3 ).This biases gl ycol ysis towards lactate production and thus can reduce aerobic respiration in cancer cells.Howe v er, alter ed r espiration is not the only metabolic perturbation found in cancer as many tumor cells also show altered amino acid biosynthesis, in particular relating to methionine metabolism ( 4 ).Such cells are said to be 'addicted to methionine', as they r equir e large amounts of the amino acid to synthesize S-Adenosyl Methionine (SAM / AdoMet) which is a cofactor r equir ed to carry out methylation reactions in the cell ( 5 ).Another metabolic signature of many tumor cells is that they r equir e cysteine as an essential amino acid, in contrast to healthy cells which can synthesize cysteine from dietary methionine ( 6 ).
In addition to metabolic abnormalities, transcriptional patterns ar e alter ed in cancer cells and many transcriptional r egulators ar e known to be oncogenes or tumor suppr essors.One example is SWI / SNF, a conserved chromatin remodeling complex found to be mutated in > 20% of all cancers ( 7 ).Chromatin itself consists of repeating arrays of nucleosomes, which are a DNA-protein complex generally considered to be an impediment to gene activation ( 8 ).However, nucleosomes can be enzymatically modified or remodeled to regulate gene transcription.SWI / SNF displaces / evicts nucleosomes in an ATP-dependent manner and was first described in Sacchar om y ces cer evisiae (Swi-Snf in S. cerevisiae ) where it was found to remodel chromatin and facilitate binding of transcription factors to nucleosomal DNA (9)(10)(11)(12)(13)(14).By promoting a more open chromatin conformation upstream of transcription start sites (TSSs) and facilitating transcription factor binding, SWI / SNF can positi v ely influence gene transcription.Yeast and human SWI / SNF are large complexes that contain multiple subunits, including the catalytic ATPase (Snf2 / SMAR CA2 / SMAR CA4) and other subunits which are essential for the structure / activity of the complex (15)(16)(17)(18)(19). Several human SWI / SNF subunits have been shown to be involved in cancer, with mutations in some being strongly correlated with specific malignancies, such as is the case with INI1 / SMARCB1 / SNF5 and rhabdoid tumors ( 17-18 , 20 ).In yeast, the complex is closely associated with activation of str ess-r esponse genes, particularly those involved in mating type switching and carbon source utilization ( 21 , 22 ).
In contrast to its association with gene activation, loss of Swi-Snf in S. cerevisiae leads to activation of several metabolic genes under r epr essing conditions, particularly those involved in synthesis of the amino acids methionine and cysteine ( 16 , 23 ).These genes (known as MET genes) ar e ordinarily r epr essed during growth in the pr esence of methionine and other sulfur metabolites but are activated by the transcription factor Met4 upon sulfur starvation (24)(25)(26)(27).MET gene-containing pathways are involved in amino acid biosynthesis, redox homeostasis, cell cycle progression and cell signaling ( 28 ).As transcription of metabolism genes responds to cellular nutrition, the activation of MET genes in snf Δ mutants could hint at a metabolic imbalance in these cells caused by loss of Swi-Snf.Ther efor e, Swi-Snf may have a novel, direct role in maintaining le v els of important metabolites via an unknown mechanism.Evidence for a conserved role for Swi-Snf in regulating genes involved in sulfur amino acid metabolism has been found in ovarian cancer cell lines, where the human SWI / SNF complex was shown to be essential for activation of a gene encoding a cysteine transporter ( 29 ).
Here, we identify Swi-Snf as a regulator of a key point in sulfur metabolism, the loss of which leads to impaired cysteine biosynthesis during growth in rich media.This defect is sensed by the transcription factor Met4, whose global recruitment becomes altered in response to loss of Swi-Snf, in a manner resembling that seen in starved cells.Met4 is activated and recruited to se v eral sulfur metabolism genes, leading to their activation under r epr essi v e conditions.Together, this demonstra tes tha t Swi-Snf has a more pr ominent r ole in metabolic regulation than previously appreciated, both through its dir ect r egulation of genes r equir ed for cysteine biosynthesis and the indirect consequences of Swi-Snf inactivity on metabolic transcription.

Biological r esour ces
All strains and plasmids used are listed in Supplementary Tables S1 and S2, respectivel y.For anal ysis of amino acid starvation, cells were grown to mid-log phase (OD 600 = 0.6-0.8) at 30 • C in media containing yeast extract with peptone supplemented with 2% dextrose (YPD), subjected to centrifugation and cell pellets were washed twice in sterile ddH 2 O. Pellets were then resuspended in 1 ml sterile ddH 2 O and split between Complete Synthetic Medium (CSM) with / without methionine / cysteine (CSM-Met) or any amino acids (SD).CSM refers to BD Difco Yeast Nitrogen Base without amino acids (BD 291920) with CSM / CSM-Met mix (Sunrise Science 1001-100) / (Sunrise Science 1019-010) and 2% glucose unless otherwise stated.

Statistical analyses
Details of high-throughput data analysis can be found in Supplementary Materials and Methods.Details of statistical analysis of data can be found in figure legends.In all cases error bars r epr esent standard deviation (SD).Significance was defined based on a Student's T-test in each case.W hen comparing dif fer ent strains, unpair ed parametric tests were used.When comparing trea ted / untrea ted conditions within the same sample, paired parametric tests were used.Graphpad Prism (9.3.1) was used for statistical calculations concerning qPCR and cell viability data.

Western blot
W hole cell lysa te was pr epar ed as described pr eviously ( 30 ) and applied to a NuPAGE Bis-Tris 3-8% gradient gel (Thermo Fisher Scientific EA0375BOX) and run at 150 V for 1 h.Proteins wer e transferr ed to a Immobilon ®-P PVDF Membrane (Millipore IPVH00010) in 1 × NuPAGE transfer buffer (Thermo Fisher Scientific NP0006) with 20% methanol for 90 min at 350 mA.Membranes were blocked for 30 min in 5% dried skimmed milk in TBST before being incubated with primary antibodies overnight at 4 • C. For non-HRP-conjugated antibodies, membranes were washed in TBST and incubated with a secondary, HRP-conjugated antibody diluted 1:10 000 in 5% dried skimmed milk in TBST for 45 min at room temperature.Following incuba tion with HRP-conjuga ted antibodies, membranes were washed in TBST and TBS before being incubated for 5 min in Pierce ™ ECL Plus Western Blotting Substrate (Thermo Fisher Scientific 32132 × 3) and de v elopment.

Metabolite assays
SAM and methionine wer e measur ed, based on a pre viously-pub lished protocol ( 31 ).Briefly, yeast were grown to mid-log phase (OD 600 ∼0.6-0.8), and 20 OD units were collected and pelleted.Cells were washed twice in ddH 2 O and then resuspended in 100 l 0.2% perchloric acid prior to incubation at room temperature for 1 hour.Samples were then centrifuged at 10 621 rcf for 5 min and supernatant was transferred to new tubes.10 l (SAM) or 5 l (Met) sample ( ∼1-2 OD units) was used for each r eplicate.Fluor escence was measured using a Tecan Infinite 200 Pro plate reader.

Anchor a w ay
Proteins of interest were tagged using an FRB-GFP construct based on that used by Haruki et al. in the HHY221 background (( 32 ), Supplementary Table S1).To deplete tagged pr oteins fr om nuclei, cells were gr own to log phase (OD 600 ∼0.5) at 30 • C and incubated with rapamycin (TSZ Chem R1017) at a final concentration of 1 g / ml for times indicated.To verify nuclear export of proteins, cells were analyzed by confocal microscopy using a Leica LSM-780 DS.To measure transcription, samples were taken before and after rapamycin treatment and processed for RT-qPCR or sequenced as described below.

RNA pr epar ation
Cells were grown to mid-log phase (OD 600 ∼0.6-0.8) at 30 • C and pelleted by centrifugation, washed once with DEPC-H 2 O, and either stored at −80 • C or processed as follows: For samples analyzed by RT-qPCR, RNA was prepared by hot phenol extraction as described in ( 33 ). 10 g RNA was DNase-treated using RQ1 RNase-Free DNase (Promega).1 g of DNase-treated RNA was then used to generate cDNA using the Applied Biosystems High-Ca pacity RN A-to-cDN A Kit (ThermoFisher Scientific).This was analyzed using PerfeCTa SYBR Green Fastmix, Low ROX (QuantaBio 95074-05K) in an Applied Biosystems QuantStudio 5 ther mocycler (Ther mo Fisher Scientific).Two to six biological (as outlined in figure legends) and 3 technical r eplicates wer e analyzed for each experiment.Error bars for RT-qPCR data r epr esent standard deviation of biological replicates.P -values were calculated using a paired (for comparing treatments within strains) or unpaired (when comparing different strains) Student's t-test using GraphPad Prism 9.0.All RT-qPCR primers used are listed in Supplementary Table S3.
For samples to be analyzed by sequencing, 3 OD units of cells were processed using the Qiagen RNeasy Mini Kit according to manufacturer's instructions.An on-column DNase I (Zymo Research E1010) digestion was carried out for 15 min at room temperature prior to elution and sequencing.

Chromatin immunoprecipitation (ChIP)
ChIP was performed as described previously ( 33 ).Briefly, cells were grown to mid-log phase (OD 600 ∼0.5-0.8), and crosslinked by incubating in 1% formaldehyde (Sigma Aldrich 252549-500ML) at room temperature for 15 min, before quenching in 125 mM glycine.Cell pellets were lysed using 400 l acid-washed glass beads in lysis buffer (50 mM HEPES, 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxychola te).Lysa te was sonica ted (8-10 pulses at 20% power, 10s on, 30s on ice) using a Branson Sonicator to yield DNA fragments between 100 and 500 bp, before being clarified by centrifugation at 18000 rcf, 30 min at 4 • C. Clarified lysate was divided into 10 or 40 OD 600 unit aliquots, to be used for ChIP-qPCR or ChIP-seq, respecti v el y.Clarified l ysa tes were incuba ted with a t 4 • C after reserving a portion as an Input, which was treated with Pronase (Sigma 10165921001), and crosslinks re v ersed.Following antibody incubation, lysates were incubated with 30 l magnetic Dynabeads (Thermo Fisher Scientific 10003D) for 2 h and washed before elution, Pronase treatment and re v ersal of crosslinks.DNA was purified using a QIAquick PCR Purification kit (Qiagen).IPs / Inputs were then either analyzed by qPCR or sequenced.

High throughput sequencing
Sequencing libraries were prepared using High Throughput Library Prep Kit (KAPA Biosystems) or NEBNext Ultra II DNA Library Prep Kit for Illumina following the manufacturer's instructions.The library was sequenced on an Illumina HiSeq platform with pair ed-end r eads of 75 bp for both ChIP-seq and RNA-seq.Whole-genome RNA-seq comparisons between mutants / wild type and Met4 ChIP cpm data, in addition to MET -regulon specific RNA-seq and Rpb1 ChIP cpm data in Met4-AA strains can be found in Supplementary File S2.

Loss of Swi-Snf results in MET gene activation in rich medium
Swi-Snf is well known as an activator of gene transcription, hence the fact that loss of Swi-Snf led to activation of sulfur metabolism ( MET ) genes in S. cerevisiae was puzzling.Additionally, SWI / SNF is an important factor in human disease and its role as a metabolic regulator deserves further study.S. cerevisiae possesses most sulfur metabolic enzymes / pathwa ys f ound in mammals, including the enzymes cystathionine beta synthase (CBS; Cys4 in yeast) and glutathione synthetase (GSS; Gsh2 in yeast), making it an attracti v e model to study these pathways (Figures 1 A, B, Supplementary Figure S1C) ( 34 , 35 ).
To monitor the effect of Swi-Snf loss on MET gene transcription, we performed RNA sequencing (RNA-seq) in rich medium (YPD)-grown snf2 Δ and snf5 Δ mutants, comparing total transcript le v els to wild type cells (Supplementary Figure S1A) ( 16 , 23 ).An unbiased pathway enrichment analysis re v ealed tha t MET gene-containing pa thways are some of the most highly enriched among upregulated transcripts in snf Δ mutants vs wild type cells (Figure 1 C and Supplementary figure S1B) ( 16 , 23 ).Se v eral genes involved in sulfur metabolism share a common regulatory network known as the MET regulon, and this overlaps with genes in the pathways shown in Figures 1 C and S1B, though the genes SAH1 , STR2 and GSH2 are not included in the MET r egulon ( 26 ).MET genes ar e generally activated in response to sulfur metabolite starvation, and repressed during growth in rich media (Figure 1 D, E) ( 24 ).Remar kab ly, when the MET regulon gene transcription was specifically analyzed, it was found that many MET genes were activated in YPD-grown snf Δ mutants, compared to wild type (Figure 1 F).These genes were selected for analysis as they had been previously found to share a common activator ( 26 ).Interestingly, the most highly-induced MET genes in the snf Δ mutants also tended to be highly induced during amino acid starvation in wild type cells (wt SD) suggesting that these mutants were exhibiting a starvation phenotype (Figure 1 Because snf Δ mutants had elevated MET transcription, we hypothesized that these cells could have reduced levels of some sulfur metabolites compared to wild type cells.One possibility was that loss of Swi-Snf resulted in a methionine uptake defect leading to low intracellular methionine, which would explain the transcriptional effect.To test this, methionine was measured in wild type vs snf2 Δ and snf5 Δ cells (Supplementary Figure S1D).snf Δ mutants exhibited a noticeably elevated methionine concentration vs wt, demonstrating that these mutants do not have a defect in methionine biosynthesis or uptake in YPD.To confirm that our method could detect reduced le v els of methionine in our strains, we also starved cells of amino acids (SD medium) and measured methionine (Supplementary Figure S1D).We next measured S-adenosyl methionine (SAM / AdoMet), whose availability can be sensed by the cell as a means of discerning sulfur status ( 24 ).Alterations in SAM le v els can affect histone methylation and MET gene expression ( 36 , 37 ).To investigate whether snf Δ mutants had a SAM deficiency, we measured SAM levels in wild type, snf2 Δ, and snf5 Δ mutant cells (Supplementary Figure S1E).It was found that snf Δ mutants had wild-type SAM le v els, and ther efor e SAM deficiency cannot explain the observed MET gene induction in snf Δ mutants.
These data show that surprisingly, loss of Swi-Snf in YPD-grown cells correlates with a transcriptional phenotype resembling starvation, suggesting altered sulfur metabolism in these cells.Howe v er, this cannot be explained by a failure of mutants to synthesize SAM, or to use methionine available in growth media.

Exogenous cysteine cures the transcriptional defect in snf Δ mutants
The transcription data for snf Δ mutants grown in rich medium were similar to what would be expected in cells starved of sulfur.This was a puzzling result but hinted that Swi-Snf was r equir ed for synthesis of an important sulfurcontaining metabolite.We hypothesized that if we could provide the correct metabolite exo genousl y, it may compensate for the deficiency and rescue MET transcription in YPD-grown snf Δ mutants.Growth in the absence or presence of 3 mM methionine, GSH or cysteine re v ealed that exogenous methionine and GSH led to a significant reduction in transcription of se v eral MET genes in snf Δ mutants compared to wild type (Figures 2 A-C and S2B-D).Howe v er, under these conditions MET transcription was still substantially higher than in wild type cells.Strikingly, incubation in the presence of exogenous cysteine rescued MET transcription to a gr eater degr ee than either methionine or GSH, resulting in wild type-le v el transcription at MET5, MET14 and MET16 (Figures 2 A-C and S2B-D).
Due to its instability, it is difficult to accurately measure intracellular cysteine, but we hypothesized that if snf Δ mutants had a cysteine deficiency, most of the MET regulon should be affected.We ther efor e gr ew cells with (+Cys) or without ( −Cys, data from Figure 1 F) exogenous cysteine and measured transcription by RNA-seq.We found that MET regulon transcript le v els in snf Δ mutants were almost entir ely r estor ed to wild type le v els after 30 min' growth in the presence of cysteine (Figure 2 D).
We next wished to determine whether cysteine supplementa tion was af fecting MET transcription specifically, or altering global transcription generall y.Surprisingl y, there wer e r elati v ely fe w genes affected in wild type cells in +Cys samples compared to −Cys (Figure 2 E).Howe v er, specific r epr ession of many MET genes in a snf2 Δ mutant was apparent in +Cys conditions compared to −Cys (Figure 2 F).GO term analysis also showed that all pathways significantly r epr essed in snf Δ mutants during growth in 3 mM cysteine are involved in sulfur metabolism (Supplementary Figure S2A).This showed that cysteine could specifically resolve the transcriptional defect in MET genes caused by loss of Swi-Snf and was not resulting in widespread transcription effects in wild type cells.This also showed that unexpectedl y, snf Δ m utants had a cysteine starvation phenotype during growth in rich media and that this was likely related to the elevated MET gene transcription in this background.Howe v er, we were still unable to determine why these cells had a phenotype similar to cysteine-starved cells.

Defective cysteine biosynthesis is the cause of elevated MET transcription in snf Δ mutants
We had established that snf Δ mutants exhibited a cysteine starvation phenotype, and that this correlated with activation of the MET regulon under repressing conditions, but we had not identified the cause of this apparent starvation.We reasoned that the most likely cause was r epr ession of one or more genes required for cysteine biosynthesis in snf Δ mutants.Biosynthesis of cysteine in S. cerevisiae first involves synthesis of SAM from methionine by Sam1 / Sam2.SAM is used as a cofactor in methylation r eactions, wher e its methyl group is removed yielding S-adenosyl homocysteine (SAH), which is used to synthesize homocysteine.Homocysteine is then used to synthesize cysteine via cystathionine by Cys4 and Cys3, respecti v ely (Figure 1 B) ( 24 , 35 ).If Swi-Snf dir ectly r egulated genes involved in cysteine biosynthesis, then it should be detectable at such gene promoters under nutrient-rich conditions.We ther efor e used pr eviously published Snf2 ChIP data from YPD-grown cells, and found that Snf2 was detected at ∼40% of MET gene promoters, and this occupancy correlates with higher transcription in YPD-grown cells vs promoters not bound by Swi-Snf (Figures 3 A and S3B) ( 38 ).Moreov er, se v eral Snf2bound genes were also r epr essed in snf Δ mutants, based on our RNA-seq data (Figure 3 A, blue genes), indicating that Swi-Snf is a direct activator of these genes.A number of the Swi-Snf-activated genes could be r equir ed for cysteine biosynthesis.To determine whether d ysregula tion of any of these genes could explain the snf Δ mutant phenotype, we measured MET gene transcription by RT-qPCR in mutants (Figure 3 B-E).We found that while most deletions had no effect on MET gene transcription, loss of CYS4 or SAM1 led to noticeable de-r epr ession of all genes studied .cys4 Δ and sam1 Δ mutants have cysteine synthesis defects, impairing the ability of cells to synthesize cysteine from methionine and ther efor e these genes were good candidates for the cause of cysteine deficiency in snf Δ mutants ( 36 , 39 , 40 ).
As S. cerevisiae contains two SAM synthetases ( SAM1 and SAM2 ), we wanted to test whether loss of SAM2 would lead to a similar phenotype to loss of SAM1 , and so we monitored MET transcription in both sam1 Δ and sam2 Δ m utants.Strikingl y, w hile a sam1 Δ m utant showed de-r epr ession of MET transcription (Supplementary Figure S3G-J), loss of SAM2 did not lead to a similar phenomenon.Loss of either gene also differentially affected transcription of the other, in a manner similar to what was described previously (Supplementary Figure S3L, M) ( 41 ).Interestingly, loss of either SAM1 or SAM2 led to r epr ession of CYS4 , to different extents (Supplementary Figure S3K).
Our data confirmed that snf Δ mutants shared a MET de-r epr ession phenotype with cys4 Δ and sam1 Δ mutants, but to determine the direct contribution of CYS4 or SAM1 loss in a snf Δ mutant, we constructed vectors containing a V5-or HA-tagged Cys4 or Sam1, respecti v ely, both under the control of a TEF1 promoter for a high le v el of constituti v e e xpr ession.These constructs wer e transformed into wild type and snf2 Δ mutant cells, and their expression was confirmed before MET transcription was measured (Figures 3 F-I, S3A, C, D).Ov ere xpression of either gene had little effect on wild type cells, but ov ere xpression of SAM1 in a snf2 Δ mutant significantly reduced the le v el of MET3 , MET5 , MET14 and MET16 transcription, (Figure 3 F-I, compare snf2 Δ + vector to snf2 Δ + SAM1 ).Over expr ession of CYS4 was found to significantly reduce MET14 transcription in a snf2 Δ mutant, and MET3 and MET5 transcript le v els wer e r eproducibly r educed compar ed to a vector control (Figure 3 F-H, compare snf2 Δ + vector to snf2 Δ + CYS4 ).In the case of MET16 , despite a significant rescue in snf2 Δ mutants ov ere xpressing SAM1 , ov ere xpression of CYS4 did not affect transcript le v els compared to the vector (Figure 3 I).This suggests that in snf Δ mutants, loss of SAM1 transcription may be the dominant factor in impairing cysteine biosynthesis, with CYS4 also contributing.
Importantly, these data support a role for Swi-Snf as a dir ect r egula tor of genes tha t ar e r equir ed for cysteine biosynthesis.We have shown that restoring transcription of these genes is sufficient to reduce MET transcription in snf Δ mutants.Ther efor e, we have identified the cause of the cysteine starvation phenotype characteristic of snf Δ m utants, w hich ultimately leads to widespread perturbations in metabolic gene transcription.

Loss of Swi-Snf leads to activation and altered recruitment of Met4 in rich media
Met4 is the transcription factor responsible for activation of the MET regulon and hence many genes involved in sulfur metabolism in S. cerevisiae , with roles in heavy metal / oxidati v e stress protection and cell cycle progression ( 42 , 43 ).In the presence of sulfur metabolites, Met4 is normally ubiquitinated (Met4-ub) by the SCF Met30 ligase, potentially inhibiting the Met4 transactivation domains, whereas sulfur starvation results in deubiquitination of Met4 (Figure 4 A, ( 25 , 44-46 )).This 'acti v e' species of Met4 can then be recruited to its target promotors via interaction with DNA-binding partners where it promotes MET gene transcription ( 26 , 47 ).We hypothesized that if loss of Swi-Snf resulted in cysteine deficiency in rich media, then this could result in altered Met4 ubiquitination.To test this, we grew wild type and snf Δ mutant cells in YPD and analyzed 3XHA-tagged Met4 (Met4-HA) from whole cell lysate by western blot in each background.In wild type cells grown in YPD, the majority of Met4 protein is of the higher molecular weight species, as would be expected for Met4-ub (Figure S4A lane 1) ( 44 ).Howe v er, loss of SNF2 or SNF5 resulted in a dramatic increase in the le v el of lower molecular weight Met4, which corresponds to the acti v e species present in starved cells (Figure 4 B lanes 2&3 versus lanes 4-6).To confirm that Met4 was ubiquitinated in wild type cells, ubiquitin was GST-tagged in a Met4-HA background and a coimmunoprecipitation was performed (Figure S4B).
We next wanted to determine whether this alteration in Met4 modification correlated with alteration of its genomic recruitment.We therefore performed ChIP-seq on 13XMyc-tagged Met4 (Met4-Myc) in wild type & snf Δ mutant cells.Comparison of pathway enrichment analysis data from wild type and snf Δ mutants grown in YPD showed that while Met4-Myc peaks are overrepresented at genes involved in translation & growth in wild type cells, these terms were not found in the snf2 Δ mutant analysis, where instead pathwa ys in volved in sulfur / methionine / cysteine metabolism were enriched (Figure 4 C, D).Interestingly, loss of Swi-Snf and activation of Met4 resulted in far fewer peaks in both snf Δ mutants compared to wild type cells (Figure 4 E).This is in agreement with pre vious wor k that showed that during growth in the presence of methionine, Met4 occupied many loci despite remaining in its ubiquitina ted sta te ( 26 , 48 ).Howe v er, increased Met4 binding was observed at several MET genes (including MET2, MET5 & MET6 ) in snf Δ mutants compared to wild type cells (Figure 4 F-H).In agreement with data shown above, incubation in the presence of exogenous cysteine both reduced the amount of deubiquitinated Met4 and eliminated the in-creased MET promoter binding observed in a snf2 Δ background (Supplementary Figures S4E and 4F-H).
Although these results supported a role for Swi-Snf in maintaining sulfur homeostasis, it was possible that snf Δ mutants were simply defecti v e for acti vation of genes encoding the SCF Met30 ligase responsible for inactivating Met4 during growth in rich media.To test this, transcription of the components of SCF Met30 was analyzed (Supplementary Figure S4G).It was found that loss of Swi-Snf did not reduce SCF Met30 gene transcription, and in the case of two genes ( MET30 and CDC34 ), a snf2 Δ mutant displayed significantly higher transcript le v els compared to wild type.The fact that snf Δ mutants show reduced Met4 ubiquitination despite higher MET30 transcription may be due to post-transla tional regula tion of Met30 activity ( 49 ).MET4 transcript le v els were slightly ele vated in the snf2 Δ mutant compared to wild type, though less than MET30 (Supplementary Figure S4G).Transcript le v els of the gene encoding the Met4-interacting transcription factor Cbf1 were also unchanged in mutant cells (Supplementary Figure S4F).These data show that upon loss of Swi-Snf, Met4 is activated in a manner consistent with sulfur starvation.This also results in global redistribution of Met4 in snf Δ mutants, mor e specifically occup ying promoters of genes involved in sulfur metabolism.Furthermore, snf Δ mutants can sense sulfur a ppropriatel y, and in these strains Met4 appears to be responding to the cysteine biosynthesis defect described above, resulting in its altered activity.

Met4 occupancy changes in snf Δ mutants correlate with MET gene activation and a starvation response
Our da ta indica ted tha t upon loss of Swi-Snf, Met4 was modified in a manner consistent with sulfur starvation (Figure 4 B) and that its recruitment patterns were altered compared to wild type cells (Figure 4 C-H).We next chose to specifically investigate Met4 occupancy at MET gene promoters, and found that in snf Δ mutants, Met4 le v els increased at many promoters compared to wild type cells (Figure 5 A, Supplementary File S1).Howe v er, there were several genes where Met4 le v els remained constant or were reduced compared to wild type cells (Figure 5 A, 'Mid' and 'Low').Throughout the MET regulon, changes in Met4 occupancy correlated to transcriptional changes (Figures 5 B, S5A and S5D), howe v er this correlation was much stronger at genes where Met4 le v els increased in snf Δ mutants compared to wild type (Figure 5 B).
As transcription patterns in snf Δ mutants resembled a wild type strain starved of amino acids (Figure 1 F), we next wanted to compare Met4 recruitment in snf Δ mutants to starved cells, and so wild type cells were grown under these conditions (SD medium) for 30 min and Met4-13Myc occupancy was monitored by ChIP-seq (Figure 5 C).It was found that starvation significantly reduced the number of Met4 peaks identified compared to YPD-grown cells (compar e Figur es 4 E to 5 C).GO-term analysis also showed that in starved wild type cells, Met4 peaks were overrepresented at genes involved in sulfur metabolism in contrast to what was seen in YPD-grown cells (compare Figures 4 C and  S5B).This more narrowly-focused binding to promoters of genes involved in metabolism is similar to what was observed in snf Δ mutants, and Met4 peaks in mutants overlapped a higher percentage of starved wild type peaks compared to YPD-grown cells (compare Figures 4 E to 5 C).The Met4 occupancy patterns at genes in the MET regulon observed in YPD-grown cells wer e alter ed similarly in starved wild type cells and snf Δ mutants grown in YPD, though the former showed stronger Met4 recruitment and did not show Met4 loss at some promoters as seen in snf Δ mutant cells (Supplementary Figure S5C).
As the many Met4-Myc peaks in YPD-grown wild type cells were unexpected because these cells had largely ubiquitinated Met4, we next wished to know where these peaks were actually located relati v e to the TSS.To test this, we monitored the occupancy profile of Met4-Myc bound to the TSS ±1 kb of all genes with verified peaks.Surprisingly, we found that YPD-grown wild type cells showed relati v ely low le v els of enrichment at gene promoters on average, whereas SD-grown wild type and snf Δ mutant cells displayed defined peaks upstream of the TSS (Figure 5 D-F).By measuring the ratio of Met4 occupancy upstream of the TSS to that found in the ORF, we determined that in YPD-grown wild type cells Met4 was more likely to localize to the gene body compared to starved or snf Δ mutant cells (Supplementary Figure S5E).This showed that not only were Met4-Myc peaks more prominently detected at genes involved in sulfur metabolism, but Met4 was more likely to be recruited to promoters in starved or snf Δ mutant cells compared to YPD-grown wild type.
These data suggest that in snf Δ mutants, Met4 occupancy increases at promoters of MET regulon genes compared to wild type cells and this recruitment is moderately correlated with transcriptional activation.Additionally, snf Δ mutants show similar patterns of Met4 recruitment to SD-grown wild type cells, both within the MET regulon and globally whereas in wild type cells grown in YPD, Met4 binding is not as strictly limited to gene promoter regions.

Met4 is responsible for increased expression of MET transcription in the absence of Swi-Snf
We observed altered Met4 modification and occupancy at se v eral MET gene promoters in snf Δ mutants (Figures 4  and 5 ).Howe v er, while Met4 was found to be recruited to MET genes in YPD-grown snf Δ mutants, this did not necessarily mean that this recruitment was the cause of elevated MET transcription.This is an important distinction, as it has been shown that Met4 may occupy some MET genes e v en in rich media ( 26 , 48 ) (Supplementary Figure S4D).To test whether Met4 was responsible for MET gene activation in snf Δ mutants, we generated a strain that could conditionally deplete nuclear Met4 via the anchor away method ( 32 ).This results in nuclear depletion of an FRB-tagged target protein upon incubation with rapamycin.Conditional Met4 loss is preferable to a stead y-sta te met4 Δ mutant, as stable MET4 loss is known to fr equently r esult in suppr essor muta tions tha t may interfer e with gene r egulation ( 39 ).
To verify that Met4 was depleted from cell nuclei upon rapamycin treatment, a Met4-FRB-GFP (Met4-AA) strain was used and monitored by confocal microscopy in both snf2 Δ mutant and wild type backgrounds following 60 min' incubation in YPD with either rapamycin or DMSO (Figure 6 A).While Met4-FRB-GFP is fully nuclear in the presence of DMSO (DMSO+), in rapamycin-treated samples (Rap+), Met4-FRB-GFP is lost from the nucleus both in wild type and snf2 Δ mutant backgrounds (Figure 6 A).A Met4-GFP strain lacking the FRB epitope was also monitored, and in these strains rapamycin addition did not affect Met4-GFP localization (Supplementary Figure S6A) .Addition of cysteine to cell cultures using the anchor away strain background also affected MET gene transcription in a manner similar to other strains used (Supplementary Figure S6C, D).
Having established that Met4 could be depleted from nuclei, we wished to know if loss of Met4 in a snf2 Δ mutant could affect MET gene de-r epr ession in this strain.If MET genes were activated in a snf2 Δ mutant due to increased Met4 activity in this background compared to wild type, we would expect that MET genes with elevated transcription in the mutant should show reduced transcription upon Met4 loss.We ther efor e measur ed transcription of the MET regulon by RNA-seq before and after 90 min' incubation in the presence of rapamycin in both Met4-AA wild type and snf2 Δ mutant backgrounds (Figure 6 B).These data showed that loss of Met4 in wild type cells grown in rich media had little impact on transcription of the MET genes de-r epr essed in snf Δ mutants.In contrast to the wild type Met4-AA cells, loss of Met4 in the absence of SNF2 (Met4-AA snf2 Δ) decreased the de-r epr ession of many MET genes, particularly those most highly activated in snf2 Δ mutants vs wild type cells (Figure 6 B).This striking result shows that Met4 activity is critical for the elevated transcription of many MET genes in snf Δ mutants.
To confirm that the observed effects were due to direct influence of Met4 on tr anscription r ather than a posttranscriptional effect, occupancy of an RN A pol ymerase II (Pol II) subunit (Rpb1 / Rpo21) was monitored by ChIPseq at MET gene ORFs (Supplementary Figure S6B).It was found that Pol II occupancy increased at many MET genes in the snf Δ mutants, and that loss of Met4 reduced this occupancy in a manner similar to what was observed in the RN A seq data.Additionall y, the subset of genes that were most sensiti v e to Met4 loss overlapped with those with increased Met4 recruitment in snf2 Δ mutants based on Met4-Myc ChIP data shown in Figure 5 A (Supplementary Figures S6B and S6E).Similarly, the group of genes with reduced Pol II occupancy in snf2 Δ mutants overlapped with genes that showed Met4 loss in this strain compared to wild type cells (Supplementary Figures S6B  and S6F).
These data demonstrate that upon loss of Swi-Snf, Met4 is recruited to genes in the MET regulon specifically in response to percei v ed cysteine deficiency, though the relationship between recruitment and activation is complex.Howe v er, once acti vated / recruited, Met4 is indispensable for der epr ession of the majority of MET genes whose elevated expression is characteristic of snf Δ mutants grown under repressing conditions.This implies that as Met4 is crucial for the activation of much of the MET regulon in snf Δ mutants, controlling the le v el of cysteine biosynthesis appears to be the key role of Swi-Snf in exerting an effect on the MET regulon.

DISCUSSION
We aimed to investigate a potential relationship between chromatin remodelers and sulfur metabolism based on findings which showed that loss of the remodeler Swi-Snf resulted in aberrant transcription of metabolic genes under r epr essing conditions ( 16 , 23 ).Both sulfur metabolism and the SWI / SNF complex ar e fr equently alter ed in tumor cells, and we wished to explore the relationship between these factors.We used S. cerevisiae as a model to establish the basic interactions between Swi-Snf and sulfur metabolic pathways (the MET regulon) as a whole, rather than initially focusing on individual genes.Here, we have demonstrated that by dir ectly r egulating a crucial point in sulfur metabolism, the chromatin remodeler Swi-Snf can indirectly control a wider body of metabolic gene expression.This is a similar concept to what was described for cancer cells with respiration defects.c-Myc was found to activate LDH-A , which encodes an enzyme responsible for the conversion of pyruvate to lactate, and it was found that reducing LDH-A expression in a c-Myc-transformed cancer model could impair tumor cell growth ( 3 ).Overproduction of LDH-A in transformed cells was associated with a bias towards lactate production from pyruvate.To generate ATP from glucose, cells must ultimately commit their stores of pyruvate either to the TCA cycle, or to produce lactate.By ov ere xpressing the LDH-A gene, cancer cells can bias this decision in favor of lactate production.Similarly, in a snf Δ mutant, reduced transcription of SAM1 and CYS4 biases sulfur metabolism towards methionine and away from cysteine synthesis.
Initially, we found many MET genes were activated in snf Δ mutants under r epr essing conditions (Figur e 1 F).Al-though initial experiments used the methionine auxotroph BY4741 strain, the most important findings were repeated in methionine pr ototr ophs, supporting the original results (Figur es 6 , S6, S7).Pr e vious studies had noted that se veral MET genes wer e de-r epr essed in snf Δ mutants, but the cause was unknown ( 16 , 23 ).We found that snf Δ mutants resemb le starv ed cells, and incubation in the presence of exogenous cysteine cured the elevated MET transcription phenotype (Figure 2 A-D).High le v els of methionine and GSH also affected MET transcription (Figures 2 A-C and S2B-D).Our data showed that additional loss of CYS4 in a snf2 Δ mutant abolished the ability of methionine to restore wild type le v els of MET transcription, confirming that cysteine is the ultimate metabolite affecting MET transcription in these strains (Supplementary Figure S2E-G) ( 27 ).Mounting evidence suggests that cysteine is the primary signal sensed by Met4, which is modified in a manner similar to starved cells in snf Δ mutants (Figures 4 B and S4A) ( 25 , 27 , 44 ).
The mechanism by which cysteine starvation occurs was found to be via r epr ession of genes r equir ed for cysteine biosynthesis, namely SAM1 and CYS4 .Mutation of either gene led to elevated MET expression, similar to snf Δ mutants (Figure 3 B-E) ( 36 ).While SAM1 ov ere xpression led to a more dramatic decrease in MET transcription in a snf2 Δ m utant, CY S4 was also shown to be a contributor to the phenotype (Figure 3 F-I).These experiments were performed under nutrient-rich conditions, but analyzing SAM1 and CYS4 transcription in starved cells revealed a role for Swi-Snf in activating these genes under MET der epr essing conditions (Supplementary Figure S3E, F).The relati v e importance of SAM1 compared to SAM2 in this context is not fully known, though SAM1 transcript levels appeared to be much higher under these conditions and incr eased expr ession of SAM2 may not compensate in a sam1 Δ mutant, despite the apparent redundancy of these enzymes (Supplementary Figure S3L, M).
Loss of Swi-Snf was also found to be correlated with altered modification of the activator of MET transcription, Met4 (Figures 4 B, S4A).We propose that Met4 reacted to a nutritional deficiency rather than a signaling defect upon loss of Swi-Snf, as Met4 PTM patterns in snf Δ mutants responded to loss of methionine / cysteine in a manner similar to wild type cells (Figure 4 B).In agreement with this result, ChIP data showed that Met4 occupancy was altered globally in snf Δ mutants, resulting in fewer peaks compared to YPD-grown wild type cells, but being more prominently recruited to genes involved in sulfur amino acid metabolism (Figure 4 C-E).This reduction in peak number compared to a YPD-grown wild type was unexpected, but wild type cells grown in the absence of amino acids also display a reduction in peak number (compare Figure 5 C to 4 E).It was found that in both snf Δ mutants and SD-grown wild type cells, Met4-Myc was not only detected more prominently at metabolic genes, but was more focused upstream of TSS regions globally, with snf Δ mutants showing similar profiles to starved cells, at a slightly reduced scale (Figures 5 D-G, S5E).In fact, in the case of both MET gene transcription and Met4 occupancy at MET promoters, snf Δ mutants tend to show an intermediate phenotype between YPD-gr own and SD-gr own wild type cells, displaying elevated transcription, altered Met4-ub levels and generally incr eased r ecruitment of Met4 to MET promoters (Figures 1 F, 4 B and 5 A).Howe v er, in contrast to YPD or SD-grown wild type cells, snf Δ mutants show reduced Met4 binding both globally (Figure 5 C) and specifically at se v eral genes in the MET r egulon (Figur es 5 A and S5C).This may suggest that Swi-Snf affects Met4 activity both indirectly through its influence on cysteine biosynthesis and directly by facilitating Met4 binding at some promoters.
Met4 occupancy specifically increased over the MET regulon in snf Δ m utants, and particularl y over genes whose expr ession incr eased in these strains (Figur e 5 A,B).Howe v er, the presence of Met4 at se v eral MET genes in wild type cells under r epr essing conditions may complicate these findings and suggests that this mechanism is more complex than currently understood ( 26 , 48 ).Similar le v els of dif ferentially-ubiquitina ted Met4 bound to a promoter may have dif ferent ef fects on gene transcription, and the ability of these different species of protein to be recruited to target sites deserves further study.Even so, conditional depletion of Met4 via the anchor away technique significantly reduced activation of many MET genes in a snf2 Δ mutant (Figure 6 B).Howe v er, there is a subset of MET genes which are activated in response to loss of Swi-Snf, but do not appear to respond to Met4 loss.These genes ar e r epr essible by incubation in the presence of cysteine (Figure 2 D), so we may specula te tha t they are redundantly activated in the absence of Swi-Snf in response to cysteine biosynthesis defects.It has been shown that se v eral genes in the MET regulon are activated by transcription factors such as Yap1 and Gcn4, which may contribute to this phenomenon ( 50 , 51 ).
Our findings indica te tha t the elevated MET transcription in snf Δ mutants occurs following activation of Met4, which is not conserved in other non-yeast eukaryotes.Howe v er, wor k carried out on Caenorhabditis elegans has identified a factor similar in function to Met4, and also described differential regulation of genes involved in cysteine and methionine biosynthesis ( 52 ).This may indicate that while the proteins involved may not be direct orthologs, similar mechanisms have evolved to regulate sulfur metabolism transcription in many species.Mammalian SWI / SNF has been shown to be frequentl y m utated in cancer ( 7 ).Cancer cells with SWI / SNF mutations have also been shown to be especially sensiti v e to inhibition of GSH biosynthesis, demonstra ting tha t SWI / SNF has a role in glutathione metabolism in humans ( 29 ).Recent work done in yeast has similarly suggested that maintaining redox balance, potentially through regulation of GSH le v els is a conserved role of Swi-Snf ( 53 ).
These data allow us to construct a model whereby in the presence of abundant nutrients, the MET regulon may be divided into Swi-Snf-dependent and independent genes.Based on Snf2 ChIP data from Dutta et al , Swi-Snf tends to be bound to highly expressed genes in both categories (Figures 7 A, S5B) ( 38 ).Our own data suggest that one difference between Swi-Snf dependent and independent genes is Met4 activity during growth in rich media, with the latter showing no reduction in transcription in wild type cells following Met4-AA (Figure 5 B).Upon loss of Swi-Snf, transcription of these Swi-Snf-dependent genes is reduced.Some of these genes ( SAM1 / CYS4 ) are critical to maintaining cysteine biosynthesis (Figure 7 B).This reduction in cysteine biosynthesis is sensed by the cellular machinery, pre v enting ubiquitination of Met4, and acti v e Met4 is recruited to its MET regulon targets, where it may activate the Swi-Snf independent genes (Figure 7 C).This group includes genes that are dependent on Met4 alone for elevated transcription, and those which may be activated redundantly.Met4 may also be recruited to Swi-Snf dependent genes, but in the absence of Swi-Snf these genes are still repressed.

DA T A A V AILABILITY
Original data underlying this manuscript can be accessed from the Stowers Original Data Repository at http://www.sto wers.org/r esear ch/publica tions/libpb-1735 .The da tasets are available in the Gene Expression Omnibus (GEO) database under the accession number GSE197919.

SUPPLEMENT ARY DA T A
Supplementary Data are available at NAR Online.

Figure 1 .
Figure 1.Loss of Swi-Snf results in MET gene activation in rich medium.( A ) The human methionine biosynthetic pathway, adapted from ( 34 ).( B ) The S. cerevisiae methionine biosynthetic pathway, adapted from ( 35 ).( C ) GO term analysis of all DE-upregulated transcripts in YPD-grown snf2 Δ mutant vs wild type (wt) cells, analyzed using ShinyGO ( 54 ).( D , E ) Model for sulfur-mediated r epr ession of MET gene transcription in yeast, showing increased production of sulfur metabolism enzymes (Cys4) during starva tion.( F ) Hea tmap r epr esenting log 2 fold-change vs wild type of MET r egulon gene transcript le v els in SD-grown wild type (wt SD) and YPD-grown snf2 Δ and snf5 Δ mutants from an RNA-seq experiment.

Figur e 2 .
Figur e 2. Exo genous cysteine cures the transcriptional defect in snf Δ mutants.RT-qPCR data showing ( A ) MET5, ( B ) MET14 and ( C ) MET16 transcription in wild type (wt) and snf2 Δ strains after incubation in YPD, YPD containing 3 mM methionine (YPD + Met), 3 mM GSH (YPD + GSH) or 3 mM cysteine (YPD + Cys).Transcript le v els normalized to TAF10 .Error bars r epr esent standard deviation from three independent experiments.P values indicated by asterisks, with a P value ≤0.05 being considered statistically significant (1 asterisk) and P ≤ 0.01 being r epr esented by two asterisks.( D ) Heatmap r epr esenting log 2 fold-change versus wild type transcript le v els in YPD-grown snf2 Δ and snf5 Δ mutants without ( −Cys) or with (+Cys) the addition of 3 mM cysteine from an RNA-seq experiment.The -Cys condition is the same RNA seq data shown in Figure 1 F. ( E ) Volcano plot using RNA-seq data, analyzing total transcription in wild type cells, comparing -Cys and + Cys conditions within a single strain.( F ) Volcano plot using RNA-seq data, analyzing total transcription in snf2 Δ cells, comparing −Cys and +Cys conditions within a single strain.Both (E) and (F), use a 1.5-fold -log 10 FDR cutoff for differential expression.

Figure 3 .
Figure 3. Defecti v e cysteine biosynthesis is the cause of elevated MET transcription in snf Δ mutants.( A ) ChIP data from Dutta et al. showing Snf2 occupancy at MET regulon promoters (18 out of 1372 total promoters identified ( 38 )).Genes in blue are significantly r epr essed in snf2 Δ mutants grown in YPD compared to wild type cells.RT-qPCR data showing transcription of ( B ) MET3, ( C ) MET5 , ( D ) MET14 and ( E ) MET16 in wild type (wt) cells and strains lacking MET genes whose transcription is r epr essed vs wild type in YPD-grown snf Δ mutants, based on RNA-seq data ( F-I ) RT-qPCR data showing MET transcription in CSM-Ura grown wild type and snf2 Δ mutants containing an empty vector (pRS416), or ov ere xpressing SAM1 or CYS4 .Transcript le v els normalized to TAF10 .Error bars r epr esent standar d de viation of 2-6 (A-D) or 4 (F-I) independent experiments.P v alues indicated b y asterisks, with a P value ≤0.05 being considered statistically significant (one asterisk) and P ≤0.0001 being represented by four asterisks.

Figure 4 .
Figure 4. Loss of Swi-Snf leads to activation and global redistribution of Met4 in rich media.( A ) Model depicting sulfur-mediated control of Met4 activity.( B ) Western blot of HA-tagged Met4 in wild type (wt), snf2 Δ and snf5 Δ mutant cells grown in media with (CSM) and without (CSM-Met) methionine.Acti v e (Met4) and inacti v e (Met4-ub) Met4 species ar e indicated.Repr esentati v e b lot shown from three independent experiments.( C ) GO term analysis (Shin yGO) f or genes occupied by Met4-13Myc in YPD-grown wild type cells, as determined by IDR analysis.( D ) GO term analysis for genes occupied by Met4-13Myc in YPD-grown snf2 Δ mutant cells, as determined by IDR analysis.( E ) Venn diagram of all Met4-13Myc peaks in wild type, snf2 Δ and snf5 Δ mutants ( 55 , 56 ).ChIP data shown r epr esent two independent e xperiments.(F-H) Met4-My c ChIP carried out in wild type and snf2 Δ mutant cells grown in YPD with or without exogenous cysteine addition at the promoters of ( F ) MET2, ( G ) MET5 and ( H ) MET6.Error bars r epr esent standar d de viation of three independent experiments.P values indicated by asterisks, with a P value ≤0.05 being considered statistically significant (one asterisk).

Figure 5 .
Figure 5. Met4 occupancy changes in snf Δ mutants correlate with MET gene activation and a starva tion response.( A ) Hea tmap / profiles of Met4-Myc binding upstream of MET regulon genes in YPD-grown snf Δ mutants compared to wild type, separated into groups based on increased (High) unchanged (Mid) or decreased (Low) binding compared to wild type cells.( B ) RNA-seq data comparing MET gene transcription in a snf2 Δ mutant versus wild type cells, grouped by whether Met4-Myc occupancy incr eased, decr eased or r emained unchanged at the respecti v e gene promoters based on ChIP-seq data.( C ) Venn diagram of Met4-Myc peaks as determined by IDR in SD-grown wild type (wt SD) and YPD-grown snf2 Δ and snf5 Δ mutants ( 55 ) (D-G) profile of Met4-Myc occupancy ±1 kb relative to TSS in ( D ) YPD-grown wild type, ( E ) SD-grown wild type, ( F ) a YPD-grown snf2 Δ mutant and ( G ) a YPD-grown snf5 Δ mutant.In each case profiles were generated based on genes with Met4-Myc peaks as determined by IDR.P values indicated by asterisks, with a P value ≤0.01 being considered statistically significant (two asterisks) and P ≤ 0.0001 being represented by four asterisks.

Figure 6 .
Figure 6.Acti v e Met4 is responsib le for increased e xpression of MET transcription in the absence of Swi-Snf.( A ) Comparison of Met4-FRB-GFP localization between wild type and snf2 Δ strains following incubation for 1 h in 1 g / ml Rapamycin (+Rap) or an equivalent volume of DMSO (+DMSO).DRAQ5 staining of double-stranded DNA was used to visualize nuclei.( B ) RNA-seq data showing MET transcription in wild type (Met4-AA) or snf2 Δ (Met4-AA snf2 Δ) strains.

Figure 7 .
Figure 7. Model for Swi-Snf-mediated control of MET transcription in rich media.( A ) During growth in rich media, Swi-Snf occupies both Swi-Snfdependent and independent MET genes, with the former requiring both Swi-Snf and Met4 for full activity.( B ) Loss of Swi-Snf results in r epr ession of Swi-Snf-dependent MET genes, leading to a cysteine biosynthesis defect and reduced cysteine biosynthesis.( C ) Cysteine starvation caused by Swi-Snf loss is sensed by cells, whereby Met4 is no longer ubiquitinated by SCF Met30 , and can activate transcription of the MET regulon in cooperation with other transcriptional activators.