| INVESTIGATION

Mutations in the S-Adenosylmethionine SynthetaseGenes SAM1 and SAM2 Differentially Affect Genome

Stability in Saccharomyces cerevisiaeKellyn M. Hoffert,* Kathryn S. P. Higginbotham,* Justin T. Gibson,* Stuart Oehrle,† and Erin D. Strome*,1

*Department of Biological Sciences and †Waters Field Laboratory, Chemistry Department, Northern Kentucky University, HighlandHeights, Kentucky 41099

ORCID IDs: 0000-0002-0191-458X (K.M.H.); 0000-0003-2967-5338 (J.T.G.); 0000-0002-6334-9619 (E.D.S.)

ABSTRACT Maintenance of genome integrity is a crucial cellular focus that involves a wide variety of proteins functioning in multipleprocesses. Defects in many different pathways can result in genome instability, a hallmark of cancer. Utilizing a diploid Saccharomycescerevisiae model, we previously reported a collection of gene mutations that affect genome stability in a haploinsufficient state. In thiswork we explore the effect of gene dosage on genome instability for one of these genes and its paralog; SAM1 and SAM2. Thesegenes encode S-Adenosylmethionine (AdoMet) synthetases, responsible for the creation of AdoMet from methionine and ATP.AdoMet is the universal methyl donor for methylation reactions and is essential for cell viability. It is the second most used cellularenzyme substrate and is exceptionally well-conserved through evolution. Mammalian cells express three genes, MAT1A, MAT2A, andMAT2B, with distinct expression profiles and functions. Alterations to these AdoMet synthetase genes, and AdoMet levels, are found inmany cancers, making them a popular target for therapeutic intervention. However, significant variance in these alterations are foundin different tumor types, with the cellular consequences of the variation still unknown. By studying this pathway in the yeast system, wedemonstrate that losses of SAM1 and SAM2 have different effects on genome stability through distinctive effects on gene expressionand AdoMet levels, and ultimately separate effects on the methyl cycle. Thus, this study provides insight into the mechanisms by whichdifferential expression of the SAM genes have cellular consequences that affect genome instability.

KEYWORDS cancer biology; genomic instability; S-adenosylmethionine (AdoMet); Saccharomyces cerevisiae; yeast genetics

CHROMOSOMAL instability was originally proposed toplay a role in tumor development more than a century

ago (Boveri 2008). Since that time aneuploidy, characterizedby deviation from the euploid chromosome number, has beenobserved in a majority of human cancer cells (Sansregretand Swanton 2017). Studies utilizing the budding yeastSaccharomyces cerevisiae have established that missegrega-tion of even a single chromosome is sufficient to induce fur-ther genomic instability, resulting in additional chromosomalinstability, mutagenesis, and sensitivity to genotoxic stress

(Sheltzer et al. 2011). Cells must balance the dual needs ofgenome maintenance and environmental adaptation. In-creases in genome instability can enable accumulation of fa-vorable genotypes but also allow premalignant cells to morerapidly acquire the biological hallmarks of cancer (Hanahanand Weinberg 2011). The cellular processes that ensure ge-nome stability are highly conserved from yeast to humans(Skoneczna et al. 2015), allowing chromosomal instabilityand genome instability findings in yeast to be directly appliedto hypotheses about human malignancy and predictions ofnovel therapeutic targets.

Previously, we used diploid S. cerevisiae to screen for het-erozygous mutations able to modify genome instability. Astrain deleted for one of the S-adenosylmethionine (AdoMet)synthetase genes demonstrated a haploinsufficient effect ongenome instability, indicating the human homolog could be apotential cancer predisposition gene (Strome et al. 2008).Sam1 and its paralog Sam2 play roles in the methyl cycle

Copyright © 2019 by the Genetics Society of Americadoi: https://doi.org/10.1534/genetics.119.302435Manuscript received June 14, 2019; accepted for publication July 15, 2019; publishedEarly Online July 18, 2019.Supplemental material available at FigShare: https://doi.org/10.25386/genetics.8949467.1Corresponding author: Department of Biological Sciences, Northern KentuckyUniversity, FH-359G, Nunn Drive, Highland Heights, KY 41099. E-mail: stromee1@nku.edu

Genetics, Vol. 213, 97–112 September 2019 97

(Figure 1); catalyzing the biosynthesis of AdoMet by transferof the adenosyl moiety of ATP to the sulfur atom of methio-nine (Chiang and Cantoni 1977). The two AdoMet synthe-tase genes, SAM1 and SAM2, in S. cerevisiae are paralogsarising from the whole-genome duplication (Cherest andSurdin-Kerjan 1978). While cells remain viable after the de-letion of either SAM1 or SAM2, the double homozygous de-letion of both genes is lethal unless growth medium issupplemented with AdoMet (Thomas and Surdin-Kerjan1997). These two genes share 83% identity between theiropen reading frames and 92% identity between protein se-quences (Thomas et al. 1988). Despite this high level of ho-mology and findings that GFP-tagged versions of bothproteins localize to the cytoplasm (Huh et al. 2003), differ-ences in the abundance of each protein (Ghaemmaghamiet al. 2003) and in the regulation of expression have beenfound. SAM2 is subject to inositol-choline regulation (Kodakiet al. 2003) and induced by the addition of excess methionine(Thomas et al. 1988); conversely, SAM1 is unresponsive toinositol-choline and repressed by excess methionine. Further,proteome studies on post-translational modifications of theSam1 and Sam2 proteins indicate these proteins vary in thenumber of sites that are modified and the types of modifica-tions that occur (ubiquitin, succinyl, acetyl, and phosphategroups) (Peng et al. 2003; Holt et al. 2009; Henriksen et al.2012; Swaney et al. 2013; Weinert et al. 2013). These find-ings speak to the differential regulation and use of theseproteins by the cell.

As a cellular enzyme substrate, the use of AdoMet issecond only to ATP, and it is the methyl donor for thepredominance of methylation reactions in all organisms(Cantoni 1975; Chiang et al. 1996) In S. cerevisiae this in-cludes methylation of proteins, RNAs, lipids, and othersmall molecules. Beyond this role in transmethylation,AdoMet’s extreme versatility allows it to function in addi-tional metabolic pathways such as transsulfuration and ami-nopropylation (Figure 1). One of the metabolic products ofAdoMet is homocysteine from which glutathione (GSH) canbe generated via the transsulfuration pathway and addi-tional reactions (Tehlivets et al. 2013). GSH is used as anelectrophilic acceptor by glutathione-S-transferases (GSTs),which are important for preventing cellular damage causedby reactive oxygen species [reviewed in Hayes and Pulford(1995), Whalen and Boyer (1998), Strange et al. (2001)].AdoMet is also used in the synthesis of polyamines such asspermidine and spermine, which are involved in cell growth(Bottiglieri 2002). Additionally, AdoMet functions as aregulator of sulfur amino acid metabolism (Blaiseau et al.1997) and as a donor of other constituents such as aminogroups (in the formation of biotin), ribosyl groups, and59 deoxyadenosyl radicals (Carman and Henry 1989; Slanyet al. 1993; Thomas and Surdin-Kerjan 1997; Phalip et al.1999; Chattopadhyay et al. 2006; Tehlivets et al. 2013).

As in all organisms studied to date, humans have genesencoding methionine adenosyl transferases (MATs), alsoknown as AdoMet synthetases. In humans, however, the

formation of these AdoMet synthetase isozymes occursdifferently. Three genes, MAT1A, MAT2A, and MAT2B, eachencode a catalytic or regulatory subunit used in formation oftheMATI (homotetramer), MATII (heterotrimer), andMATIII(homodimer) isozymes (Martínez-Chantar et al. 2002). TheSAM1 and SAM2 genes in S. cerevisiae are homologous to theMAT2A gene in Homo sapiens (Mato and Lu 2007). MAT2Aand SAM1 share 63.5% nucleotide sequence and 68.2%protein sequence similarity, while MAT2A and SAM2 share64.1% nucleotide sequence and 67.8% protein sequence sim-ilarity. MAT1A is expressed only in adult liver, while MAT2Aand MAT2B are expressed in fetal liver and nonhepatic tis-sues. These genes, and their product AdoMet, have been im-plicated in multiple cancer types, but the mechanism ofaction is not well understood, and upregulation is found insome cancers while downregulation is found in others(Martínez-Chantar et al. 2002; Kodaki et al. 2003; Chenet al. 2007; Greenberg et al. 2007; Mato and Lu 2007; Liuet al. 2011; Zhang et al. 2013; Wang et al. 2014; Ilisso et al.2016; Phuong et al. 2016).

Our studies of SAM gene dosage add to this area of in-vestigation by documenting the effects of changes in AdoMetsynthetase genes on genome stability. These findings help usunderstand the differences in the roles of the unlinked SAM1and SAM2 genes in a diploid S. cerevisiae system and howaltered expression of the homologous genes in humans mayaffect cancer development.

Materials and Methods

Strains

Our parental strain (hereafter referred to as wild type) geno-type is: MAT a/a, leu2-3/leu2-3 his3-D200/his3-D200trp1-D1/trp1-D1 lys2-801/LYS2 ura3-52/ura3-52 can1-100/CAN1 ade2-101/ade2-101 23 [CF:(ura3::TRP1,SUP11, CEN4, D8B)]. A rad9-deficient strain, rad9D/rad9D,was created by the insertion of a HIS3 cassette into bothRAD9 loci of the wild-type strain. The SAM gene deletionswere created utilizing the homologous recombination switch-out method (Wach et al. 1994). The sam1::KANMX cassettewas PCR-amplified from the yeast heterozygous gene dele-tion collection (YSC1055; Dharmacon) with primers 500–800 bp upstream and downstream from the SAM1 openreading frame. The sam2::LEU2 cassette was created usinga two-step PCR reaction with primers 300–800 bp upstreamand downstream, replacing the SAM2 open reading framewith the LEU2 gene. The PCR generated products weretransformed into our wild-type and rad9D/rad9D diploidsto create SAM heterozygotes. Appropriate haploids, derivedfrom the wild-type and rad9D/rad9D diploids, of oppositemating type, were transformed with SAM knockout cas-settes and mated to create SAM1/SAM2 deletion combina-tions and homozygous deletion strains. Transformants wereselected on appropriate media and cassette integration atthe correct location was verified via PCR. Later addition of a

98 K. M. Hoffert et al.

HPHMX cassette to appropriate haploids to ensure additionto the opposite arm of chromosome V from the location ofthe CAN1 gene allowed for further genome instability as-says. A complete listing of strains and their genotypes can befound in Supplemental Material, Table S1.

Chromosome transmission fidelity assay for sectoring

Each strain was struck for individual colony formation onlow-adenine concentration (6 mg/ml) synthetic complete(SC) plates and allowed to grow at 30� for 7 days, followed byovernight incubation at 4� for color development. Colonies

were examined for the appearance of pink/red sectors. Twotrials were completed for each strain with a minimum of600 individual colonies examined per trial. The chromosomefragment and assay are described in more detail in Spenceret al. (1990), Strome et al. (2008), Duffy and Hieter (2018).Fold change in the chromosome transmission fidelity (CTF)rate of each mutant compared to the appropriate-parentalstrain was calculated. A Student’s t-test was performed, in-dividually comparing each mutant strain to the parental, toidentify mutant strains with significantly different CTF rates.Assays for CTF rates in the presence of AdoMet supplemen-tation were carried out as above, with the addition of 60 mMof AdoMet (NEB B9003S) to the SC-low adenine plates.

Chromosome V instability rate assays

Cells were grown at 30� on appropriate selection media,allowing genome instability events to occur, until individualcolonies reached �3 mm. Twenty-four individual coloniesper strain were each dispersed in 200 ml of water in a96-well plate. Absorbance was measured at 562 nm(ELx800; BioTek) and the 15 colonies with the most similaroptical densities (reflecting similar population size) .0.8were used for analysis. The numbers of viable and canava-nine-resistant cells were determined by plating dilutions onnonselective (YPD) and SC-Arg2 plus canavanine (60mg/ml)(C9758; Sigma-Aldrich) plates, respectively. Plates weregrown for 3–5 days at 30�, followed by colony counting.The fluctuation analysis-based chromosome V instability rateand 95% confidence intervals (95% CIs) were calculatedutilizing the R advanced calculation package Salvador(rSalvador), taking plating dilution into account (Zheng 2002,2008, 2016). The 95% CI overlap method mimics a two-tailed,two-population t-test at the conventional P, 0.05 level with animprovement in type 1 error rate and statistical power whencompared to a t-test, which has been found unsuitable forfluctuation data analysis (Zheng 2015). The placement of thehygromycin (HPHMX) resistance cassette on the oppositearm of chromosome V, but on the same parental chromosomeas the CAN1 allele, allows for the differentiation of localevents from full chromosome events. Cells from colonies thatgrew on canavanine plates were struck to hygromycin plates(300 mg/ml) (VWR K547) and scored for growth. Chromo-some V instability rates and confidence intervals were mea-sured for a minimum of two biological replicates for eachstrain. Assays for chromosome V instability rates in the pres-ence of AdoMet supplementation were carried out as above,with the addition of 60 mM of AdoMet to the media duringthe initial growth phase when instability events would occur.Assay for CAN1 mutation rate in haploids was again calcu-lated using fluctuation analysis methods as above and therSalvador package was utilized for mutation rate estimationand 95% CI calculations.

RT-PCR: RNA extraction: Three-milliliter culturesweregrownshaking at 30� overnight, and then 1.5 ml of the culture waspelleted at 20,000 rpm for 1 min, and the supernatant was

Figure 1 The methyl cycle in the yeast Saccharomyces cerevisiae.AdoMet functions in three main metabolic pathways from themethyl cycle: transmethylation, transsulfuration, and aminopropylation.Transmethylation reactions are catalyzed by methyltransferases, whichtransfer the methyl group from AdoMet to a wide range of acceptormolecules. The transfer of the methyl group from AdoMet leavesS-adenosylhomocysteine (AdoHcy) as a byproduct. AdoHcy is convertedto adenosine and homocysteine (Hcy) via the enzyme adenosylhomocys-teinase. This reaction occurs only if adenosine and Hcy levels are low,which the cell must balance to prevent buildup of AdoHcy and thereforethe prevention of AdoMet-dependent methylation reactions. Methioninecan be regenerated by methylation of Hcy via methionine synthase (MS).Remethylation via MS requires products of the folate cycle, specifically5-methyltetrahydrofolate (5-MTHF), which is then converted to THF whenit loses its methyl group. The folate cycle is required for dNTP production,specifically dTTPs, and therefore correct balance of dNTPs. Furthermore,dNTP pool size is monitored by dATP/ATP ratios. As AdoMet, producedfrom ATP and methionine, is the second most used enzyme substrate(behind ATP); deviations in its production could alter ATP ratios (if ATPis not converted to AdoMet due to lack of AdoMet synthetases) andfurther disrupt appropriate dNTP production. The conversion of Hcy tocysteine is a two-step reaction with the production of cystathionine as anintermediate. This requires two enzymes cystathionine b-synthase, andcystathionine g-lyase. Cysteine is the rate-limiting precursor to glutathi-one (GSH) synthesis. GSH, used with glutathione-S-transferases (GSTs),are a major oxidative species sink used by cells to prevent DNA and pro-tein reaction with reactive oxygen species. AdoMet decarboxylase, decar-boxylates AdoMet to then enter the polyamine synthesis pathway.Polyamines are low-molecular-weight, positively charged moleculesthat include spermidine and spermine, with known roles in growth andapoptosis.

AdoMet and Genome Instability in Yeast 99

discarded. Next, 750 ml TRIzol (Ambion) and �200 ml glassmicrobeads (425–600 mm; Sigma-Aldrich) were added andcells were homogenized for cycles of 15, 25, and 15 sec(BioSpec Mini-BeadBeater-8) with a 5 min rest on ice be-tween each homogenization. Samples were incubated atroom temperature (RT) for 5min, 150ml chloroform (FisherScientific) was added and vigorous shaking to mix was car-ried out for 15 sec, followed by RT incubation for 3 min andcentrifugation at 12,000 3 g for 15 min at 4�. Next, 375 mlisopropanol was added to the aqueous phase and mixed byinversion, followed by a RT incubation for 10 min and cen-trifugation at 12,000 3 g for 10 min at 4�. The supernatantwas removed, and the RNA pellet was washed with 750 ml75% ethanol, followed by centrifugation at 7500 3 g for5 min at 4�. The supernatant was again removed, the pelletwas air dried at RT for 10 min, resuspended in 20 ml RNase-free water, and then incubated at 55� for 10 min. The RNAwas used immediately for complementary DNA (cDNA)synthesis.

cDNA synthesis: cDNA was synthesized using the VersocDNA synthesis kit (Thermo Scientific) following the manu-facturer’s instructions for 20 ml reactions, using 1 mg RNAand 1 ml of a 3:1 (v/v) primer blend of random hexamer:-anchored oligo(dT). cDNAwas stored on ice and then addedinto the RT-PCR set-up within the same day.

RT-PCR: RT-PCR was performed using the DyNAmo FlashSYBR Green qPCR kit (Thermo Fisher). Each 20 ml reactioncontained 1.2 ml of 5 mM forward primer, 1.2 ml of 5 mMreverse primer, 10 ml of 23 master mix, 0.4 ml of ROX, 2 mlof cDNA template, and 5.2 ml of nuclease-free water. Eachreaction had a technical replicate in the same 96-well plate.A minimum of three biological replicates were tested perstrain. The reaction was cycled 40 times as directed by themanufacturer with a 15 sec denaturation and 1min extensionperiods (7300 RT-PCR system; Applied Biosystems). Ct datawere analyzed using the 2009 REST software (http://rest-2009.gene-quantification.info) in standard mode. For eachstrain the data were normalized to TUB1 and ACT1 and ex-pression was compared to the parental strain. The followingprimers were used for amplification in each strain: SAM1FW

AATTACTACCAAGGCACAGT, SAM1RV ATCCTTCTCCTCGTGGACAC, SAM2FW AATTACCACCAAAGCTAGAC, SAM2RVGCTCTTTTCATAGTGCAGAC, TUB1FW CCAAGGGCTATTTACGTGGA, TUB1RV GGTGTAATGGCCTCTTGCAT, ACT1FWTTCAACGTTCCAGCCTTCTAC, and ACT1RV ACCAGCGTAAATTGGAACGAC.

Ultraperformance liquid chromatography mass spectrome-try analysis: Strains were grown to saturation at 30� in 3 mlselective media followed by a 1:1000 dilution into 200 ml ofthe same media. Cells were then grown shaking at 30� to logphase as measured by an OD600 of 0.5–0.8 (Genesys20;Thermo Scientific), centrifuged at 2500 rpm for 2 min topellet cells, then stored at 280�. Cell pellets were thawedon ice and resuspended in residual liquid. A portion of thepellet (300 ml) was moved to a clean microcentrifuge tubeand spun at 20,000 rpm for 1 min. Any excess liquid wasremoved and the pellet was weighed. Cells were added orremoved until the cell pellets weighed between 100 and160 mg. Then, 350 ml of freshly prepared, cold 25:75 auto-claved, distilled H2O:Acetonitrile (HPLC grade; Fisher Scien-tific) and �200 ml glass microbeads were added and cellswere homogenized (BioSpec Mini-BeadBeater-8) for 30 secfollowed by a 5 min rest on ice. The homogenization and iceincubation were repeated two additional times. The micro-centrifuge tube was then pierced on the bottom, stacked intoa fresh microcentrifuge tube, and both were centrifuged at1000 rpm for 1 min. The upper tube containing glass beadswas discarded while the lower tube was centrifuged at20,000 rpm for 5 min to pellet any residual debris. The su-pernatant was syringe filtered (0.22 mmPVDF; Restek) into aLC/GC certified glass vial (186000384c; Waters, Milford,MA) and stored at 4�, then analyzed within 24 hr. Sampleswere diluted into the 25:75 H2O:Acetonitrile diluent and 1mlinjected for analysis. Analysis of the extract was performed byultraperformance liquid chromatographymass spectrometry;the system consisted of an Acquity HClass and QDa MassDetector (Waters Corporation). The sample compartmentwas maintained at 10�. The mass detector (QDa) was oper-ated in electrospray positive mode with a capillary voltage of0.8 V and probe temperature of 500�. Analysis of the AdoMet

Table 1 Chromosome transmission fidelity frequency in SAM mutants alone and with AdoMet supplementation

Strain alone Strain + AdoMet

Strain SectorsTotal

coloniesSectoring

frequency (%) Strain SectorsTotal

coloniesSectoring

frequency (%)

Wildtype 1 2077 0.05 Wildtype 4 1964 0.20sam1D/SAM1 4 1973 0.11 sam1D/SAM1 0 1437 0.00sam1D/sam1D 3 1973 0.15 sam1D/sam1D 2 2533 0.08sam2D/SAM2 9 2259 0.40 sam2D/SAM2 3 2401 0.13sam2D/sam2D 17 1886 0.90** sam2D/sam2D 4 1911 0.20sam1D/SAM1 sam2D/SAM2 11 2071 0.58 sam1D/SAM1 sam2D/SAM2 2 2150 0.11sam1D/SAM1 sam2D/sam2D 87 1707 5.07** sam1D/SAM1 sam2D/sam2D 2 2073 0.08sam1D/sam1D sam2D/SAM2 12 1942 0.61 sam1D/sam1D sam2D/SAM2 1 1441 0.06

Chromosome transmission fidelity frequency due to SAM gene deletions alone and with exogenous AdoMet. A minimum of 600 cells were scored in each trial. Data shownrepresents the average of two independent experiments on loss of SAM genes in a wildtype background. Pairwise comparisons were conducted between each mutant andthe parental and a two-tailed Student’s t-test was performed (** P , 0.05).

100 K. M. Hoffert et al.

was done by monitoring the protonated molecular ion(M+H+) at mass 399 in selected ion recording mode. Sepa-ration of the compound was done using an Amide column(BEH Amide column, 2.1 3 50 mm; Waters Corporation) at45�. A hydrophilic interaction liquid chromatography(HILIC) separation method used a gradient of ammoniumacetate, formic acid, and acetonitrile at a flow rate of0.5 ml/min was used. AdoMet eluted at 2.5 min in the5min analysis. A calibration curve was generated by injecting1 ml of eight AdoMet standards (32 mM; New EnglandBiolabs) in 25:75 H2O:Acetonitrile from a concentrationof 2–100 mg/ml, resulting in a correlation coefficient (r2) of0.994 on an eight-point curve. AdoMet concentrations foreach strain were individually compared to the appropriateparental strain and a Student’s t-test was performed to iden-tify mutant strains with significantly different AdoMet pools.

Morphology: Strains were grown overnight to log phasein appropriate selection media. Aliquots were sonicated toremove cell clumping and counted on a hemocytometer.Images of at least five separate fields of cells were taken.Length and width measurements were analyzed for 100 cellsper strain using Image Quant to determine mother cell size,bud size, and elongation. One-way ANOVA for comparison ofthe sizes and elongations between strains was performed andresidual plots were checked for goodness of fit, indicatingANOVA was the correct model to use. Pairwise comparisonsbetween mutant strains and the parental were done usingthe Tukey–Kramer adjustment for multiple comparisons. Inassessing bud size, a bud is scored as small if it is one-third thesize of the mother, or smaller; medium if it is larger thanone-third and smaller than two-thirds the size of the mother;and large if it is two-thirds the size of the mother, or larger. Achi-square test for the difference of the distribution of budsize was conducted and the Pearson P-value is reported.

Genotoxic stress assays: Cells were grown overnight inselection media in a 96-well plate at 30�, diluted to 0.2 OD,and allowed to grow back for 3–4 hr to log phase range (0.5–0.8 OD). Absorbance was measured at 562 nm (ELx800;

BioTek). Cells were then washed and used to create a five-fold serial dilution in water across six wells and stampedin duplicate on the genotoxic stress plates. Plates wereincubated at 30� for 3 days. Genotoxic stressors testedwere as follows: ultraviolet light (UV) at 17.5, 35, and70 J/m2; hydroxyurea (HU) at 50, 75, and 100 mM; phleo-mycin at 0.5, 1, and 6 mg/ml; and benomyl at 10, 20, and30 mg/ml.

Data availability

Strains are available upon request. File S1 includes all straingenotype information. The authors affirm that all data nec-essary for confirming the conclusions of the article are presentwithin the article, figures, and tables. Supplemental materialavailable at FigShare: https://doi.org/10.25386/genetics.8949467.

Results

SAM1 and SAM2 deletions have different dosageeffects on genome instability

To determine the effect of SAM gene dosage on one form ofgenome instability, we performed the CTF assay (Spenceret al. 1990), which monitors the inheritance of an artificialchromosome fragment. Briefly, our diploid strains are homo-zygous for the ade2-101 allele, which prevents completion ofthe adenine biosynthesis pathway and leads to the develop-ment of a red pigment. Our strains also carry two chromo-some fragments that harbor the SUP11 ochre suppressor,enabling full adenine synthesis and normal coloration. If cellslose one or both of their chromosome fragments during thegrowth of a colony, the cells develop a pink or red color, re-spectively, and they and their progeny can be visualized as apie-shaped sector portion of the colony. This assay identifiedthat the complete loss of sam2 (sam2D/sam2D and sam1D/SAM1 sam2D/sam2D strains) significantly increases CTFrates (Table 1). Further, when this assay is performed inthe presence of exogenous AdoMet in the media, these

Figure 2 Chromosome V instability rates inSAM mutants alone and with AdoMet Sup-plement. The data shown represents a com-bination of a minimum of two independentexperiments. (A) The black circle depicts themean instability rate (instability events/cell/generation) with the tails showing the ex-perimental 95% confidence intervals (95%CI). The gray bar represents the 95% CIof the parental strain. Those SAM mutantstrains with a non-overlapping 95% CI, tothe parental strain, are considered signifi-cantly different from the parental. (B) Meaninstability rates (pink square) of all strainsgrown with exogenous AdoMet. Tails repre-sent the experimental 95% CIs. The pink barrepresents the 95% CI of the parental straingrown with AdoMet supplementation. Blackcircles and tails are the same as in (A) andare overlaid to aid in viewing changes.

AdoMet and Genome Instability in Yeast 101

significant increases in loss of the chromosome fragment aresuppressed.

Our second genomic instability assay monitors the CAN1locus on chromosome V and uses fluctuation analysis (Luriaand Delbrück 1943; Lea and Coulson 1949) to estimate agenome instability rate. Our diploid strains contain a wild-type CAN1 gene, conferring sensitivity to canavanine, and arecessive allele can1-100, conferring resistance. This assayquantitates conversion to canavanine resistance (CanR; asinstability events per cell per generation), which could occurthrough a variety of genome instability mechanisms such asdeletions, chromosome loss events, mutations, or recombina-tion. This assay expands our measure of genome instabilityfrom the previous CTF assay as it both measures events oc-curring on an endogenous chromosome and also accounts fora broad range of the types of instability that have the capacityto affect cancer development. Chromosome V instability wasmeasured using a modification of the method described pre-viously (Klein 2001; Strome et al. 2008). The 95% CI wascalculated using rSalvador. Strains, with nonoverlappingCIs with parental chromosome V instability rates, harbordeletions that result in statistically significant changes ingenome instability compared to the parental strain. Fullsam2 loss, as well as heterozygous loss alone, results instatistically significant increases in chromosome V instability[strains sam2D/SAM2, sam2D/sam2D, and sam1D/SAM1sam2D/sam2D (Figure 2A)], whereas loss of sam1, either heter-ozygously or homozygously, while SAM2 is intact, confers a pro-tective effect on the genome, significantly decreasing instability.Because of sam1 and sam2 mutations inducing opposite effectson instability, it is not surprising to see that strains that harbormutations in both genes have intermediate effects. Strains het-erozygous for one of these genes and homozygous for the other,show chromosome V instability rates that trend toward the effectobserved for the homozygously mutated gene. That is, in thesam1D/SAM1 sam2D/sam2D strain, instability is increased butnot to the same level as sam2 mutation alone, whereas thesam1D/sam1D sam2D/SAM2 strain displays a near wild-typelevel of instability with only a trend toward stability, likely dueto the full loss of sam1.

When this assay is performed in the presence of exogenousAdoMet, many but not all, of these instability effects aresuppressed (Figure 2B). The stability conferred due to theheterozygous sam1D/SAM1 deletion alone is fully sup-pressed with instability rates returning to wild-type levels.However, the stabilizing effect of full homozygous deletionof sam1 is not repressed and stays at the same level as un-treated. This indicates that the Sam1 protein (Sam1p) is re-quired for the suppression effects of supplemental AdoMetaddition.

As the chromosome V instability assay to CanR measuresmultiple types of genome instability without distinguishingbetween mechanisms we created additional strains to differ-entiate and provide a more complete analysis. First, haploidCAN1 mutation analysis was assessed in wild-type, sam1D,and sam2D strains to understand the role of the loss of these

genes on spontaneous mutation rate. Previous work hasshown that the mutations responsible for change to CanR inhaploid yeast most frequently occur via point mutations:transversions, transitions, frameshift changes, and smallscale (, �100 bp) duplications and deletions (Holbeck andStrathern 1997; Tishkoff et al. 1997; Ohnishi et al. 2004). Nochange in the rate of conversion to CanR was observed, com-pared to the parental strain, in either mutant (Table 2). Thisindicates the solo loss of either of these genes does not alterthe rate of point mutations, or their repair, in the haploidsystem. Second, to better distinguish among the loss of het-erozygosity mechanisms that could account for the changesto CanR in our diploid strains, we recreated all of our strainswith a hygromycin resistance marker on the opposite armfrom CAN1, on chromosome V. When we then assay for con-version to CanR, colonies are further assessed for theirhygromycin sensitivity (HphS) or resistance (HphR). Coloniesthat are CanR HphR represent instability events affecting onlyone arm of chromosome V, such as point mutations, geneconversion, and mitotic recombination events. CanR HphS

colonies most likely arise from full chromosome V loss events,with or without reduplication. For almost all CanR coloniestested, the HphR cassette was maintained, indicating insta-bility occurred via partial chromosome affecting events,which in yeast most frequently occur due to allelic mitoticrecombination (Klein 2001) (Table 3). No statistically signif-icant difference between strains and the parental were found.These two additional assays taken together indicate thatsam1 and sam2 mutations alter genome stability primarilythrough their alteration of mitotic recombination eventsand not via point mutations or full chromosome loss events.Although inconsistency is noted in full sam2D/sam2D dele-tant strains, which increased CTF loss rates, indicating thatfor a smaller, artificial chromosomal fragment, loss events doincrease due to loss of sam2.

Genome instability increases due to loss ofSAM2 are absent in an RAD9 DNA damagecheckpoint-deficient background

To gain further insight into the relationship between variousSAM gene deletions and genome instability, we performed ourchromosome V instability experiments in strains lacking the

Table 2 CAN1 mutation rate in haploid strains

Strain CAN1 mutation rate

95% CI

Lower Upper

Wildtype haploid 1.68E207 1.06E207 2.49E207sam1D 1.74E207 1.06E207 2.64E207sam2D 1.80E207 1.16E207 2.62E207rad9D 2.07E207 1.31E207 3.06E207sam1D rad9D 2.08E207 1.41E207 2.89E207sam2D rad9D 2.37E207 1.60E207 3.32E207

Haploid strains were assessed for rate of conversion to canavanine resistance(events/cell/generation). Data represent estimated mutation rates from two in-dependent trials for each strain. Mutation rates and 95% CIs were calculated usingrSalvador. No statistical difference was found between any mutant strain and itsparental.

102 K. M. Hoffert et al.

RAD9-dependent DNA damage checkpoint (Figure 3A, Table2, and Table 3). Rad9 is involved in sensing and respondingto DNA damage, is required for checkpoint-induced cell cyclearrest due to damage in all phases of the cell cycle, and loss issufficient to increase genome instability on its own (Weinertand Hartwell 1988, 1989, 1990; Al-Moghrabi et al. 2001; Tohand Lowndes 2003). Full loss of SAM1 continues to confer aprotective effect and reduces the level of instability seen due tothe rad9D/rad9D deletion alone (Figure 3A). Addition of ex-ogenous AdoMet again fails to have a substantial effect, andstrains remain with decreased genome instability (Figure 3B).However, in strains lacking SAM2 and this checkpoint, no in-crease in instability is observed beyond the level induced bythe rad9D/rad9D deletion alone. In fact it appears that thegenome-stabilizing effects due to SAM1 deletions are morereadily seen in this background as strains harboring mutationsin both SAM1 and SAM2 now show lower rates of instabilitycompared to the rad9D/rad9D parental. The effects due to lossof SAM1 may be more apparent as the effects of SAM2 muta-tion on genome stability are already absorbed in theRAD9 loss.

We again sought to more completely characterize thetypes of instability contributing to the rates of conversionto CanR. Haploid CAN1 mutation analysis was assessed inrad9D, sam1D rad9D, and sam2D rad9D strains to charac-terize the effects of combination of these mutants on pointmutation rate. No change in the rate of conversion to CanR

was observed in the double mutants, compared to the rad9Dstrain (Table 2), indicating that again loss of neither of thesegenes alters the rate of point mutations, or their repair, inthe haploid system. We then characterized the portion ofCanR that occurred via loss or mitotic recombination eventsin the rad9-deficient background. While a slight increase inloss events was noted due to the rad9D/rad9D alone, addingSAM gene mutations did not statistically significantly alterthe rate, with most events still occurring through mitoticrecombination (Table 3). Therefore while the instabilityrates change in the presence of the rad9-deficiency, thetypes of instability events that make up that rate are notsignificantly different.

SAM1 and SAM2 demonstrate dosage-sensitiveexpression profiles, and differentially affect expressionof each other

Genome instability data indicates different roles for the SAM1and SAM2 genes (with further alterations seen in a rad9-deficient background), and previous work has shown thesegenes have different inducers and repressors (Thomas et al.1988; Kodaki et al. 2003). We therefore wanted to character-ize how our gene deletions affected SAM gene expression lev-els by performing quantitative RT-PCR for both SAM1 andSAM2 in each strain. Levels of expression in the wild-typeand rad9D/rad9D strains were used as the control parentallevels for the respective set of strains and expression is dis-played as the fold increase or decrease compared to that level(Figure 4, A and B). As expected, the heterozygous deletion ofone gene resulted in reduced expression of that genewhile thehomozygous deletion of the gene resulted in no expression ofthat gene. In a wild-type background in the absence of sam1,SAM2 expression is at its highest measured value in any strain(Figure 4A). These results are in line with previous workreporting that SAM2 expression increases in response to excessmethionine (Holbeck and Strathern 1997; Tishkoff et al. 1997;Ohnishi et al. 2004). The sam1 deletions, resulting in de-creased AdoMet synthetase production from this locus, likelyresult in an increase in methionine. This increase in methio-nine could then be enough to induce increased expressionfrom the SAM2 locus (Thomas et al. 1988), resulting in theincreased mRNA expression detected. Indeed, in the sam1D/sam1D sam2D/SAM2 strain the expression of SAM2 is not re-duced to the level seen in the sam2D/SAM2 single mutant(Figure 4A). This is likely the effect of the sam1 homozygousdeletion leading to excess methionine and increased SAM2expression. Conversely, those strains homozygously deletedfor sam2, either alone or in combination with sam1 deletions,resulted in a significant reduction in the expression of SAM1(Figure 4A). In this case, the increase in methionine due to thereduction in AdoMet synthetase expression, leads to signifi-cant repression of SAM1 expression, as previously described(Thomas et al. 1988).

Table 3 Chromosome V instability mechanisms

Strain CanR colonies HphR (%) Strain CanR colonies HphR (%)

Wildtype 137 100 rad9D/rad9D 286 90sam1D/SAM1 400 100 sam1D/SAM1 rad9D/rad9D 57 100sam1D/sam1D 339 100 sam1D/sam1D rad9D/rad9D 107 100sam2D/SAM2 99 96 sam2D/SAM2 rad9D/rad9D 153 100sam2D/sam2D 271 100 sam2D/sam2D rad9D/rad9D 57 98sam1D/SAM1 sam2D/SAM2 46 100 sam1D/SAM1 sam2D/SAM2 rad9D/rad9D 44 100sam1D/SAM1 sam2D/sam2D 289 100 sam1D/SAM1 sam2D/sam2D rad9D/rad9D 40 100sam1D/sam1D sam2D/SAM2 239 100 sam1D/sam1D sam2D/SAM2 rad9D/rad9D 429 100

Canavanine resistant colonies that grew from two independent estimates of chromosome V instability rate, via fluctuation analysis assays, were struck to hygromycincontaining media (300 mg/ml). The hygromycin resistance cassette was placed on the opposite arm of chromosome V from the CAN1 gene. Instability leading to change tocanavanine resistance while maintaining hygromycin resistance is interpreted as a local event not affecting both arms of the chromosome. Instability leading to change tocanavanine resistance and change to hygromycin sensitivity is interpreted as a full chromosome loss event. Rates are compared to the appropriate parental strain and a two-tailed Student’s t-test was performed to identify mutant strains with significantly different rates. No strains were found with significantly different mechanisms of canavanineresistance.

AdoMet and Genome Instability in Yeast 103

In strains harboring rad9D/rad9D deletions, both hetero-zygous and homozygous deletion of sam2 resulted in signif-icant decreases in SAM1 expression (Figure 4B). This is in linewith observations in the wild-type background strains. How-ever, where the average expression level of SAM2 in sam1deletion strains in a wild-type background was.1.0 relativeto the parental, this increase is lost in the rad9-deficient back-ground. SAM2 expression appears unchanged due to the het-erozygous loss of sam1 and decreased due to the homozygousloss of sam1.

AdoMet levels are differentially altered in SAM1 andSAM2 knockout strains

We next hypothesized that if altering SAM gene expressionaffects genome stability via a mechanism involving changesto AdoMet concentration, we should be able to detectchanges in the amount of AdoMet between the strains thatdisplay instability and those that do not. In order to directlyquantify AdoMet pools, cells were homogenized and sub-jected to ultraperformance liquid chromatographymass spec-trometry analysis. Total pool concentrations of AdoMetare shown in Table 4. In a wild-type background we seeincreased AdoMet levels due to the sam1 deletion alone:sam1D/SAM1 and sam1D/sam1D strains. This correlateswith our expression data, as reduced sam1 copy number re-sults in increases in SAM2 expression, which could then beresponsible for increases in overall AdoMet production. Sta-tistically significant decreases in AdoMet concentrations areseen due to the complete loss of sam2, (sam2D/ sam2D andsam1D/SAM1 sam2D/sam2D strains). This again correlateswith expression data; in the absence of sam2D/sam2D, SAM2is not expressed and SAM1 expression is repressed. This leadsto the lowest overall amount of total SAM gene expression,which then leads to substantially reduced AdoMet levels.

In the rad9-deficient background we again see decreasesin AdoMet levels due to the loss of SAM2. This decreasedconcentration, however, returns AdoMet levels to that seen

in wild-type cells, as the rad9-deficiency elevates AdoMetconcentrations on its own relative to wild type (Table 4).Three strains in this category show these decreases,sam2D/sam2D rad9D/rad9D, sam1D/SAM1 sam2D/SAM2rad9D/rad9D, and sam1D/SAM1 sam2D/sam2D rad9D/rad9D. This aligns with our results showing deletion ofsam2 results in significant decreases in SAM1 expression(Figure 4B).

SAM gene mutations affect overall cell size withoutdisplaying consistent differences that would point tocell cycle delay phenotypes

Alterations in S. cerevisiae cell morphology have previouslybeen associated with deficiencies in the cell cycle. Bud sizehas been shown to correlate with cell cycle phase and over-abundance of cells with a particular bud-to-mother size ratiocan therefore indicate a cell cycle halt or delay (Pringle andHartwell 1981; Weinert and Hartwell 1988). Multibuddedphenotypes are associated with failed progression throughG1 in cell cycle mutants, and have been seen due to deletionsin cyclins as well as checkpoint control genes (Snyder et al.1991; Schwob et al. 1994). Further, observations on colonyand cell size/area have been measured to assess relativehealth of a colony and cells, with smaller colonies oftendenoting smaller individual cells or slower progressionthrough the cell cycle and thus growth rate of cells withinthe colony. Therefore we assessed our strains to determine ifSAM gene mutations result in altered morphologies thatcould indicate cell cycle progression defects (Table 5). Onehundred cells for each strain were measured for the lengthand width of both the mother and the bud (if present), andthe area of each was then calculated as (p 3 radius oflength 3 radius of width). Calculations of the average areasof the mother cell for each strain, with 95% CIs, indicate thatfull deletion of sam1 as well as sam2 mutations on theirown result in cells of decreased size (sam1D/sam1D,sam1D/sam1D sam2D/SAM2, sam2D/SAM2, and sam2D/

Figure 3 Chromosome V instability rates inrad9-deficient SAM mutants alone and withAdoMet Supplement. The data shown rep-resents a combination of a minimum of twoindependent experiments. (A) The black cir-cle depicts the mean instability rate (events/cell/generation) with the tails showing theexperimental 95% confidence intervals(CI). The gray bar represents the 95% CIof the parental strain. Those SAM mutantstrains with a non-overlapping 95% CI, tothe rad9-deficient parental strain, are con-sidered significantly different. (B) Mean in-stability rates (pink squares) of all strainsgrown with exogenous AdoMet. Tails repre-sent the experimental 95% CIs. The pink barrepresents the 95% CI of the rad9-deficientparental strain grown with AdoMet supple-mentation. Black circles and tails are thesame as in part (A) and are overlaid to aidin viewing changes.

104 K. M. Hoffert et al.

sam2D strains). Interestingly strains heterozygous forsam1 and homozygously deletant for sam2, sam1D/SAM1sam2D/sam2D, are larger than wild-type cells. Further, ho-mozygous sam1 mutation alone results in cell that are elon-gated; thus, these cells are both small and less round thanwild-type cells. We investigated all strains for bud size distri-bution to determine if this phenotype might correlate withhaving more cells that failed to enter the cell cycle or froze atparticular points. However, no changes were observed in thedistribution of cells with no buds, small buds, medium buds,or large buds in any of our mutant strains. Investigationof these mutations in a rad9-deficient background showedthat all mutant strains containing any SAM gene mutationwere smaller than the rad9D/rad9D parental; mutation tosam1D/SAM1 sam2D/sam2D no longer leads to an enlargedphenotype in the rad9D/rad9D background. No additionalabnormalities, in bud size distribution or elongation, werenoted.

Strains mutated for SAM1 and SAM2, in differentcombinations, demonstrate alternate responses toHU-induced stress

Many deletions linked to increases in cancer incidence exerttheir functions by increasing the instability rates of cells, weak-ening defenses against exogenous stress, or both. Therefore, wesought to characterize our strains for alterations in response toexogenous stressors. To this end, we tested our strains forsensitivity or resistance to a range of insults, including HU,UV, phleomycin, and benomyl (Table 6). In strains in thewild-type background no significant change in the response toagents that cause direct DNA damage was seen due to anycombination of SAM1 or SAM2 deletions; response to UV expo-sure–induced thymine-dimers and phleomycin-induced ad-ducts via direct intercalation were unchanged. Exposure tobenomyl-induced microtubule blockage resulted in a slight in-crease in sensitivity in both of the strainsmutant for three of thefour copies of the SAM genes: sam1D/sam1D sam2D/SAM2and sam1D/SAM1 sam2D/sam2D. Themost interesting results,

however, were the growth patterns in response to HU, whereheterozygous loss of SAM2 in the sam1D/SAM1 sam2D/SAM2and sam2D/SAM2 strains showed a resistance to this inhibitorof ribonucleotide reductase (RNR). However, homozygousloss of SAM1, in the sam1D/sam1D and sam1D/sam1Dsam2D/SAM2 strains, showed increased sensitivity to thesame treatment.

In strains already lacking a Rad9-dependent DNA damagecheckpoint, we again saw no increased response (over paren-tal) due to direct DNA damage from UV or phleomycinexposure (Table 6). We also saw no changes in growth dueto the benomyl inhibition of microtubule dynamics. However,HU treatment once again resulted in an altered response.Here, we again observed sensitivity in the two strains homo-zygously deleted for SAM1: sam1D/sam1D rad9D/rad9D andsam1D/sam1D sam2D/SAM2 rad9D/rad9D.

Discussion

AdoMet is themainmethyl donor in the cell andalso feeds intoother pathways andproduct syntheses via themethyl cycle. Asmany of these pathways have potential effects on genomestability, it is unsurprising that AdoMet synthetase disruptionand altered AdoMet levels have been linked to a variety ofcancer types. Interestingly, both increases and decreases ingene expression, as well as in AdoMet concentration itself,have been found across these various cancer types.

Previously we reported a strain deleted for one of theAdoMet synthetase genes demonstrated a haploinsufficienteffect on genome instability, but the underlyingmechanism ofaction remained unclear (Strome et al. 2008). We presentdata here that the SAM1 and SAM2 genes have differenteffects on genome stability. By studying these AdoMet syn-thetase genes in yeast and creating the full complement ofviable mutant strain combinations, we have been able tocontribute to the field by generating sets of strains that dem-onstrate different phenotypes for further study. Additionally,by including studies of morphology and effects on growth

Figure 4 Changes in SAM1 and SAM2 ex-pression. A quantitative RT-PCR analysis ofSAM1 (gray) and SAM2 (pink) gene expres-sion normalized to ACT1 and TUB1. Datarepresent average expression levels over pa-rental with SE represented in error bars froma minimum of three independent experi-ments in (A) a wildtype background and (B)a rad9-deficient background. REST (Relativeexpression software tool) was used to com-pare expression data from each mutantstrain to its parental and generate relativeexpression profiles where ** P , 0.05.

AdoMet and Genome Instability in Yeast 105

due to exogenous stressors, we are able to further categorizeour mutants and propose possible mechanisms by whichthese gene mutations cause their effects on genome instabil-ity. While we began this study in an attempt to identify onemechanism of action for SAM mutational effects on genomestability, we clearly have two distinct mechanisms dependenton having functional copies of SAM1 vs. SAM2.

Loss of SAM2

Strains homozygously deleted for SAM2, sam2D/sam2D andsam1D/SAM1 sam2D/sam2D, share the characteristics ofhaving increased genome instability (Figure 2A and Table1), decreased AdoMet levels (Table 4), and no altered reac-tion to HU (Table 6), in a wild-type background. Thesestrains have significant increases in CTF rates, indicating thatthe increases in genome instability reflect, at least in part,increases in chromosome loss events. SAM2 mutations alsoincrease conversion to CAN resistance at an elevated rate andthese events occur primarily through mitotic recombinationmechanisms. These strains have the lowest total cumulativeexpression from the SAM1 and SAM2 loci (Figure 4A), as wellas the lowest AdoMet levels (Table 4). Work by others hasfound that SAM2 loss results in a significant increase in me-thionine levels at 4.5 mM compared to 0.13 mM in wild-typecells (P-value = 7.39 3 10285) (Mülleder et al. 2016), andthat excess methionine represses SAM1 expression (Thomaset al. 1988). Inclusion of AdoMet in the media fully sup-presses genome instability increases seen due to loss ofSAM2. The suppression back to wild-type levels of instabilitywithout being further stabilizing, as well as the observationthat adding AdoMet to wild-type cells does not confer a sta-bilizing effect, leads to the conclusion that in the presence ofexogenous AdoMet, cells maintain normal AdoMet levelswithout acquiring excess concentrations. Thus, genome in-stability in these strains likely results from lowAdoMet levels,a necessary compound for survival, as demonstrated by thelethality of a double sam1D sam2D full deletant. As the sec-ond most highly utilized enzyme substrate in any cell, lowlevels of AdoMet could bring about genome instabilitythrough a variety of different mechanisms (for model seeFigure 5), the resolution of which will require further study.For example, AdoMet has also been implicated in G1 progression

delay, which could be disrupted in strains with low levels of thiscompound, allowing cells to cycle when they are ill equipped todo so without mistakes (Mizunuma et al. 2004). AdoMet is alsothought to suppress the production of other methylation com-pounds (Bawa and Xiao 1999). In these cells, reduced AdoMetlevels could lead to the cellular production of more mutagenicmethyl donors that interact more frequently, resulting in in-creased alkylation and increased instability. Also, reductions inAdoMet likely impede production of other components of themethyl cycle. In one branch, the transsulphuration pathwayfeeds out of the methyl cycle where AdoMet conversion toS-adenosylhomocysteine leads to production of homocysteineand then GSH. GSH, used with GSTs, are a major oxidativespecies sink used by cells to prevent DNA and protein reactionswith reactive oxygen species, thereby protecting the genomefromdamage. At another point, homocysteine feeds into tetrahy-drofolate pools, which are a necessary deoxyribonucleotide tri-phosphate (dNTP) production cofactor. Reduced dNTP levelshave been shown to decrease DNA synthesis, thus decreasinga cell’s ability to repair DNA and perform recombination(Paulovich et al. 1997; Zhao et al. 1998; Chabes et al. 2003). Afinalmechanismmight relate to recentwork that has identified anovel protein complex named SESAME (SErine-responsive SAM-containing Metabolic Enzyme complex), which contains Pyk1,serine metabolic enzymes, Sam1, Sam2, and acetyl-CoA synthe-tase. Both H3K4methylation by the Set1methyltransferase com-plex as well as H3T11 phosphorylation require this complex.Deletions of either sam1D or sam2D resulted in a global reduc-tion of bothH3K4me3 andH3pT11 (Li et al. 2015), indicating adirect role for AdoMet synthetase genes in histone methylationand phosphorylation events. These alterations in histone regula-tionmay cause global or local gene expression changes that resultin increased genome instability through a variety of pathways.

Work in our rad9-deficient strains adds additional infor-mation to the model. In these strains the instability due toloss of SAM2 is not seen (Figure 3A). We propose this likelycomes about in one of two ways. First, AdoMet levels aredecreased in these strains but are measured at levels weobserved in wild-type cells (Table 4). Perhaps this reductionis not low enough to perturb the system and AdoMet levelsare sufficient to suppress production of more mutagenicmethyl donors and fully serve in all methylation reactions

Table 4 S-Adenosylmethionine (AdoMet) concentrations in SAM mutant strains

Strain AdoMet nmol/mg cells SD Strain AdoMet nmol/mg cells SD

Wildtype 353 6 74 rad9D/rad9D 517 6 8sam1D/SAM1 968** 6 40 sam1D/SAM1 rad9D/rad9D 458 6 94sam1D/sam1D 896* 6 238 sam1D/sam1D rad9D/rad9D 672 6 118sam2D/SAM2 425 6 105 sam2D/SAM2 rad9D/rad9D 462 6 23sam2D/sam2D 188* 6 97 sam2D/sam2D rad9D/rad9D 332** 6 21sam1D/SAM1 sam2D/SAM2 611 6 45 sam1D/SAM1 sam2D/SAM2 rad9D/rad9D 318** 6 29sam1D/SAM1 sam2D/sam2D 148** 6 47 sam1D/SAM1 sam2D/sam2D rad9D/rad9D 270** 6 12sam1D/sam1D sam2D/SAM2 574 6 98 sam1D/sam1D sam2D/SAM2 rad9D/rad9D 454 6 32

Data collected by Hydrophilic interaction ultra performance liquid chromatography mass spectrometry (HILIC UPLC-MS), n = 2 biological replicates. AdoMet concentrationsare measured in nanomole/milligram of cells with SD reported. Concentrations are compared to the appropriate parental strain and a two-tailed Student’s t-test wasperformed to identify mutant strains with significantly different AdoMet pools (** P , 0.05, * P , 0.1).

106 K. M. Hoffert et al.

including those mediated by SESAME. However, the additionof exogenous AdoMetmarginally lowers the instability rate inthese strains (Figure 3B). This could indicate the strains arenot quite maintaining an adequate level of AdoMet for fullfunctionality. A second mechanism could involve Rad9’sknown positioning upstream of the RNR2 and RNR3 genes,which are involved in the ribonucleotide reductase produc-tion of dNTPs (Navas et al. 1996). If the induction of insta-bility due to loss of SAM2 comes through lowered dNTPproduction, this pathway is already suppressed due to lossof RAD9 and thus no further increase in genome instability isobserved. A combination of these mechanisms also cannot beruled out.

Loss of SAM1

Conversely, harboring sam1 mutations alone, either hetero-zygous or homozygous, results in statistically significant de-creases in genome instability (Figure 2A), and increases inAdoMet levels (Table 4). SAM2 expression is not reduced inthese sam1 mutant strains, and appears to trend toward in-creased expression (Figure 4). Work by others has found thesame increase in methionine levels due to SAM1 loss, at4.6 mM compared to 0.13 mM in wild-type cells (P-value =5.483 10285) (Mülleder et al. 2016), and that excess methi-onine induces SAM2 expression (Thomas et al. 1988). Inthese strains, the observed genome protection is likely linkedto increased AdoMet (for models, see Figure 5). The additionof AdoMet to the media returns the instability rate to awild-type level in sam1D/SAM1 strains but not in thesam1D/sam1D strains (Figure 2B). This could indicate thatin heterozygous SAM1mutant strains the alterations leadingto accumulation of AdoMet are suppressed, as the cell doesnot rely on the mutated SAM1/methyl cycle pathway, insteadimporting and using exogenous AdoMet and maintaining

normal AdoMet levels. The lack of action of this mechanismin the full SAM1 deletant cells points to a role for SAM1 inthe sensing or use of exogenous AdoMet to accomplish thesuppression. Further the increased levels of AdoMet seen inSAM1 mutant cells points to a role of SAM1 in appropriatesensing of AdoMet levels or usage of AdoMet once pro-duced. Two models therefore emerge from these observa-tions: that AdoMet accumulation is tied to impediment ofthe methyl cycle if the AdoMet cannot be appropriately usedin the absence of SAM1, or alternatively, that AdoMet accu-mulation is tied to overactive movement through the methylcycle if the AdoMet cannot be appropriately sensed to reg-ulate cycling (Figure 5). In the first model of cycle impedi-ment, genome stability is likely due to the protective effectof AdoMet that has previously been reported in yeast (Bawaand Xiao 1999), specific to O-methyl lesions, and may bedue to the suppression of production of other compoundswith reactive methyl groups. In addition, several studieshave described the protective effects of AdoMet in hepato-carcinoma. The mechanism of this protective effect remainsunclear, potentially involving DNA methylation effects(Adams and Burdon 1987) or a reduction of DNA synthesisand cell loss in the cancerous tissues (Taguchi and Chanarin1978; Pascale et al. 1995). In the second model where in-creased AdoMet results in increases across the methyl cycle,genome stability could come from multiple branches. Asdiscussed above, the methyl cycle creates precursors forGSH production used in Reactive Oxygen Species (ROS)scavenging as well as for dNTP production critical for repli-cation and repair. Increases in either or both of which couldresult in a more stable genome due to less ROS insults orincreased repair capacity with increased dNTP levels. Theobservation that homozygous deletion of sam1 (sam1D/sam1Dand sam1D/sam1D sam2D/SAM2) makes these cells more

Table 5 Morphological characteristics of SAM mutants

Strain Mother size (mm2) 95% CI (mm2)

Bud size

ElongationUnbudded Small Medium Large

Wildtype 27 25.7–28.3 43 20 26 11 1.281sam1D/SAM1 28 26.4–30.0 46 16 31 7 1.295sam1D/sam1D 20** 19.0–21.6 41 21 30 8 1.401**sam2D/SAM2 22** 21.2–23.8 42 25 25 8 1.336sam2D/sam2D 17** 15.2–17.8 58 19 18 5 1.245sam1D/SAM1 sam2D/SAM2 25 23.6–26.2 41 23 21 15 1.326sam1D/SAM1 sam2D/sam2D 32** 31.1–33.7 50 23 20 7 1.259sam1D/sam1D sam2D/SAM2 21** 20.0–21.9 47 20 26 7 1.309rad9D/rad9D 28 26.9–30.0 48 17 26 9 1.402sam1D/SAM1 rad9D/rad9D 23** 21.8–24.0 41 26 19 14 1.474sam1D/sam1D rad9D/rad9D 24** 23.4–25.6 42 29 18 11 1.419sam2D/SAM2 rad9D/rad9D 22** 21.4–23.5 47 20 19 14 1.456sam2D/sam2D rad9D/rad9D 22** 20.0–23.1 41 23 19 17 1.493sam1D/SAM1 sam2D/SAM2 rad9D/rad9D 23** 21.8–23.9 43 20 23 14 1.469sam1D/SAM1 sam2D/sam2D rad9D/rad9D 23** 21.9–24.1 41 21 26 12 1.474sam1D/sam1D sam2D/SAM2 rad9D/rad9D 22** 20.0–22.9 49 16 26 9 1.444

Cells were grown to log phase, sonicated, diluted and observed at 4003 magnification on a hemocytometer. Images of five separate fields of cells were captured for eachstrain and 100 cells were analyzed for size using Image Quant software. A one-way ANOVA, with a Tukey–Kramer adjustment for multiple comparisons, was conducted tocompare the morphologies of the parental strain to the mutant strain of that background. Size, bud size distribution, and cell elongation patterns were assessed, strains withsignificant differences in morphology are denoted with ** P , 0.05.

AdoMet and Genome Instability in Yeast 107

sensitive to HU, may support decreased conversion (Figure5, model A) over increased cycling (Figure 5, model B). Thecytotoxic and cytostatic effects of HU have primarily beenlinked to its reduction of dNTP pools and/or its effects inraising the levels of ROS. To be more sensitive to HU, wewould therefore expect these cells to already have lowerdNTP pools and/or higher ROS levels, both of which areaffected by the portions of the methyl cycle downstreamof AdoMet. If SAM1 mutation results in increased AdoMetdue to a lack of proper utilization of this compound, then thenecessary downstream components needed for reducingROS (GSH) and producing dNTPs (tetrahydrofolate), wouldalso be reduced.

In the rad9-deficient strains we still see the protectiveeffects of harboring SAM1 mutations (Figure 3A), but wedo not see the same level of elevation of AdoMet concentra-tion as in wild-type cells (Table 4). While this does not pointto one model over another it does facilitate our future abilityto ask whether the observed AdoMet levels in these strainsare sufficient to still suppress production of other more re-active methyl species.

Strains harboring mutations in both SAM1 andSAM2 may speak to methyl cycle importance ingenome stability

The differing effects on instability, with SAM2 mutationsincreasing genome instability and SAM1 mutations confer-ring a protective effect, make interpreting the data in strainsthat harbor mutations in both genes difficult. However theintermediary phenotypes may be showing us more informa-tion about the importance of steady state regulation of themethyl cycle in genome stability, as well as the mechanismsin place to maintain this steady state. For example, strainswith heterozygous deletions in SAM2, sam2D/SAM2 andsam1D/SAM1 sam2D/SAM2, are characterized by increases

in genome instability (Figure 2A), normal AdoMet levels (Ta-ble 4), and resistance to HU (Table 6), in a wild-type back-ground. While the total AdoMet pool level is not statisticallyaltered due to these particular SAM gene mutations (Table4), this is likely due to a vacillation between the effects ofpartial loss of the genes and not due to lack of alteration tothe system. Therefore, we can see in these strains that Ado-Met level alone is not sufficient to predict genome instabilityas these mutants, and several others, have normal AdoMetconcentrations but altered stability rates. Likely cellularchanges to the methyl cycle, occurring in response to theAdoMet vacillation or in an attempt to mitigate these fluctu-ations, contribute to instability increases. This demonstratesthe balancing act the cell must accomplish in the regulation ofthe methyl cycle used tomaintain these AdoMet levels. Thesemethyl cycle alterations could then play roles in the observedinstability. The altered growth patterns in the presence of HUcould indicate some of these differences. As mentioned abovethe effects of HU are linked to its reduction of dNTP pools andits effects in raising the levels of ROS. To see resistance to HUbased on these mechanisms, it could be hypothesized thatthese strainmutations result in alterations to themethyl cyclethat lead to increased dNTP pools (thus strains are able towithstand the dNTP reductions resulting fromHU treatment)or have increased resistance to the HU-induced rise in ROS(possibly through higher GST/GSH levels making ROS levelsreduced before HU treatment in these cells). While we arenot able to distinguish the mechanism within the scope of theresults presented here, the more likely model would be thatthese strains are showing HU resistance due to increases indNTP pools, which have been linked (unlike decreases inROS) to increases in genome instability. Previous work dem-onstrated a mutator phenotype when dNTP levels exist athigher than normal amounts in both yeast (Chabes et al.2003; Davidson et al. 2012; Fleck et al. 2013) andmammalian

Table 6 Genotoxic stress response in SAM mutants

Strains

UV, J/m2 Hydroxyurea, mM Phleomycin, mg/ml Benomyl, mg/ml

17.5 35 70 50 75 100 0.5 1 6 10 20 30

Wildtype + + + + + — + + — + + +sam1D/SAM1 + + + + + — + + — + + +sam1D/sam1D + + + + 2 2 + + — + + +sam2D/SAM2 + + + + + + + + — + + +sam2D/sam2D + + + + + — + + — + + +sam1D/SAM1 sam2D/SAM2 + + + + + + + + — + + +sam1D/SAM1 sam2D/sam2D + + + + + — + + — + + 2

sam1D/sam1D sam2D/SAM2 + + + + 2 2 + + — + + 2

rad9D/rad9D + + — + + — + + — + + +sam1D/SAM1 rad9D/rad9D + + — + + — + + — + + +sam1D/sam1D rad9D/rad9D + + — + 2 2 + + — + + +sam2D/SAM2 rad9D/rad9D + + — + + — + + — + + +sam2D/sam2D rad9D/rad9D + + — + + — + + — + + +sam1D/SAM1 sam2D/SAM2 rad9D/rad9D + + — + + — + + — + + +sam1D/SAM1 sam2D/sam2D rad9D/rad9D + + — + + — + + — + + +sam1D/sam1D sam2D/SAM2 rad9D/rad9D + + — + 2 2 + + — + + +

Strain response to genotoxic stressors. All strains were assayed twice (biological replicates) with each assay containing technical replicates. + denotes growth under the givenconditions. 2 denotes lack of growth under these conditions. UV, ultraviolet light.

108 K. M. Hoffert et al.

cells (Weinberg et al. 1981; Caras and Martin 1988) [for indepth review see Pai and Kearsey (2017)]. This effect has beenattributed to increased dNTP availability speeding up theS-phase (Kunkel et al. 1987; Stodola and Burgers 2016), in-creasing DNA polymerase binding and extension from an in-accurate primer-template pairing, and reduced proofreadingefficiency (Beckman and Loeb 1993; Kunkel and Bebenek2000). This effect, seen in yeast and mammalian cells, couldbe evenmore dramatic in mammalian cells where high dNTPslevels have also been shown to inhibit apoptosome formation(Chandra et al. 2006). Thus, in cancer cells this pathway could

be doubly important as it affects two phenotypic hallmarks ofcancer, increased genome instability and decreased apoptosis.

Conclusions

Our group has conducted studies of SAM gene dosage anddetermined the effects of changes in AdoMet synthetasegenes on genome stability. SAM1 and SAM2 clearly operateby two distinct mechanisms to impart different effects ongenome stability. The findings reported here provide evi-dence that can aid in the interpretation of how differentcancer types are associated with increases or decreases in

Figure 5 Models of SAM1 vs. SAM2 perturbations to the methyl cycle resulting in genome stability effects. Top left panel: Known methyl cycle pathwayin S. cerevisiae. Bullet points in blue at the top are known points about cells in steady state and the portions of the methyl cycle that might affectgenome stability. Top right panel: Effects observed in sam2-deficient cells due to the sole functionality of SAM1. Points in blue are supported fromprevious data or data presented here. Red X represents the model supported here in which presence of SAM1 alone does not sufficiently generateAdoMet from methionine and ATP, leaving a decreased AdoMet pool and an increased methionine concentration, impeding the methyl cycle leading toincreased genome instability. Bottom panel: Effects observed in sam1-deficient cells due to the sole functionality of SAM2. Points in blue are supportedfrom previous data or data presented here. Two models are proposed that might explain the decreased genome instability observed. Model A (left): RedXs represent points where lack of SAM1 impedes cellular sensing of AdoMet for AdoMet usage in downstream pathways. Decreased genome instabilitylikely occurs due to increased AdoMet levels suppressing production of more reactive methyl donor species. Model B (right): Red arrows represent pointswere cycle is overactive due to lack of SAM1 to sense and control AdoMet levels. Increased AdoMet increases methyl cycle cyclization and decreasedgenome instability may arise from increased AdoMet suppression of alternate more reactive methyl donors, increased GSH for ROS scavenging, and/orincreased dNTP that can facilitate genome repair.

AdoMet and Genome Instability in Yeast 109

expression from the MAT genes and adds to the field by dem-onstrating a link in yeast between SAM gene dosage andgenome instability. S. cerevisiae are particularly well-suitedto continue to work on more mechanistic insight into theroles of AdoMet and the methyl cycle in genome stability asthese are a unique model organism not shown to use DNAmethylation. Many current hypotheses on the effects ofAdoMet in cancer involve alteration to DNA methylationchanges, but yeast show there must be additional compo-nents contributing to the instability phenotypes.

Acknowledgments

The content is solely the responsibility of the authors anddoes not necessarily represent the official views of theNational Institutes of Health. Research was supported byan Institutional Development Award (IDeA) from theNational Institute of General Medical Sciences (NIGMS) ofthe National Institutes of Health (P20GM1234) and anNIGMS R15 Area Award (1R15GM109269-01A1). Statisti-cal support for this publication was provided by Arnold JStromberg and Jiaying Weng through the Applied StatisticsLaboratory at the University of Kentucky via NIGMS grantnumber P20GM103436.

Literature Cited

Adams, R., and R. Burdon, 1987 Molecular biology of DNA meth-ylation. J. Basic Microbiol. 27: 138.

Al-Moghrabi, N. M., I. S. Al-Sharif, and A. Aboussekhra,2001 The Saccharomyces cerevisiae RAD9 cell cycle check-point gene is required for optimal repair of UV-induced pyrim-idine dimers in both G(1) and G(2)/M phases of the cell cycle.Nucleic Acids Res. 29: 2020–2025. https://doi.org/10.1093/nar/29.10.2020

Bawa, S., and W. Xiao, 1999 Methionine reduces spontaneousand alkylation-induced mutagenesis in Saccharomyces cerevi-siae cells deficient in O6-methylguanine-DNA methyltransfer-ase. Mutat. Res. 430: 99–107. https://doi.org/10.1016/S0027-5107(99)00163-3

Beckman, R. A., and L. A. Loeb, 1993 Multi-stage proofreading inDNA replication. Q. Rev. Biophys. 26: 225–331. https://doi.org/10.1017/S0033583500002869

Blaiseau, P. L., A. D. Isnard, Y. Surdin-Kerjan, and D. Thomas,1997 Met31p and Met32p, two related zinc finger proteins,are involved in transcriptional regulation of yeast sulfur aminoacid metabolism. Mol. Cell. Biol. 17: 3640–3648. https://doi.org/10.1128/MCB.17.7.3640

Bottiglieri, T., 2002 S-Adenosyl-L-methionine (SAMe): from thebench to the bedside–molecular basis of a pleiotrophic mole-cule. Am. J. Clin. Nutr. 76: 1151S–1157S. https://doi.org/10.1093/ajcn/76.5.1151S

Boveri, T., 2008 Concerning the origin of malignant tumours bytheodor boveri. Translated and annotated by henry harris.J. Cell Sci. 121: 1–84. https://doi.org/10.1242/jcs.025742

Cantoni, G. L., 1975 Biological methylation: selected aspects.Annu. Rev. Biochem. 44: 435–451. https://doi.org/10.1146/annurev.bi.44.070175.002251

Caras, I. W., and D. W. Martin, 1988 Molecular cloning of thecDNA for a mutant mouse ribonucleotide reductase M1 thatproduces a dominant mutator phenotype in mammalian cells.

Mol. Cell. Biol. 8: 2698–2704. https://doi.org/10.1128/MCB.8.7.2698

Carman, G. M., and S. A. Henry, 1989 Phospholipid biosynthesisin yeast. Annu. Rev. Biochem. 58: 635–669. https://doi.org/10.1146/annurev.bi.58.070189.003223

Chabes, A., B. Georgieva, V. Domkin, X. Zhao, R. Rothstein et al.,2003 Survival of DNA damage in yeast directly depends onincreased dNTP levels allowed by relaxed feedback inhibitionof ribonucleotide reductase. Cell 112: 391–401. https://doi.org/10.1016/S0092-8674(03)00075-8

Chandra, D., S. B. Bratton, M. D. Person, Y. Tian, A. G. Martin et al.,2006 Intracellular nucleotides act as critical prosurvival fac-tors by binding to cytochrome C and inhibiting apoptosome. Cell125: 1333–1346. https://doi.org/10.1016/j.cell.2006.05.026

Chattopadhyay, M. K., C. W. Tabor, and H. Tabor, 2006 Methyl-thioadenosine and polyamine biosynthesis in a Saccharomy-ces cerevisiae meu1delta mutant. Biochem. Biophys. Res.Commun. 343: 203–207. https://doi.org/10.1016/j.bbrc.2006.02.144

Chen, H., M. Xia, M. Lin, H. Yang, J. Kuhlenkamp et al., 2007 Roleof methionine adenosyltransferase 2A and S-adenosylmethioninein mitogen-induced growth of human colon cancer cells. Gastro-enterology 133: 207–218. https://doi.org/10.1053/j.gastro.2007.03.114

Cherest, H., and Y. Surdin-Kerjan, 1978 S-adenosyl methioninerequiring mutants in Saccharomyces cerevisiae: evidences forthe existence of two methionine adenosyl transferases. Mol.Gen. Genet. MGG 163: 153–167. https://doi.org/10.1007/BF00267406

Chiang, P. K., and G. L. Cantoni, 1977 Activation of methioninefor transmethylation. Purification of the S-adenosylmethioninesynthetase of bakers’ yeast and its separation into two forms.J. Biol. Chem. 252: 4506–4513.

Chiang, P. K., R. K. Gordon, J. Tal, G. C. Zeng, B. P. Doctor et al.,1996 S-adenosylmethionine and methylation. FASEB J. 10:471–480. https://doi.org/10.1096/fasebj.10.4.8647346

Davidson, M. B., Y. Katou, A. Keszthelyi, T. L. Sing, T. Xia et al.,2012 Endogenous DNA replication stress results in expansionof dNTP pools and a mutator phenotype. EMBO J. 31: 895–907.https://doi.org/10.1038/emboj.2011.485

Duffy, S., and P. Hieter, 2018 The chromosome transmission fidel-ity assay for measuring chromosome loss in yeast. Methods Mol.Biol. 1672: 11–19. https://doi.org/10.1007/978-1-4939-7306-4_2

Fleck, O., R. Vejrup-Hansen, A. Watson, A. M. Carr, O. Nielsen et al.,2013 Spd1 accumulation causes genome instability indepen-dently of ribonucleotide reductase activity but functions to pro-tect the genome when deoxynucleotide pools are elevated.J. Cell Sci. 126: 4985–4994. https://doi.org/10.1242/jcs.132837

Ghaemmaghami, S., W.-K. Huh, K. Bower, R. W. Howson, A. Belleet al., 2003 Global analysis of protein expression in yeast. Na-ture 425: 737–741. https://doi.org/10.1038/nature02046

Greenberg, A. K., B. Rimal, K. Felner, S. Zafar, J. Hung et al.,2007 S-adenosylmethionine as a biomarker for the early de-tection of lung cancer. Chest 132: 1247–1252. https://doi.org/10.1378/chest.07-0622

Hanahan, D., and R. A. Weinberg, 2011 Hallmarks of cancer: thenext generation. Cell 144: 646–674. https://doi.org/10.1016/j.cell.2011.02.013

Hayes, J. D., and D. J. Pulford, 1995 The glutathione S-transferasesupergene family: regulation of GST and the contribution of theisoenzymes to cancer chemoprotection and drug resistance. Crit.Rev. Biochem. Mol. Biol. 30: 445–600. https://doi.org/10.3109/10409239509083491

Henriksen, P., S. A. Wagner, B. T. Weinert, S. Sharma, G. Bacinskajaet al., 2012 Proteome-wide analysis of lysine acetylation sug-gests its broad regulatory scope in Saccharomyces cerevisiae.

110 K. M. Hoffert et al.

Mol. Cell. Proteomics MCP 11: 1510–1522. https://doi.org/10.1074/mcp.M112.017251

Holbeck, S. L., and J. N. Strathern, 1997 A role for REV3 in mu-tagenesis during double-strand break repair in Saccharomycescerevisiae. Genetics 147: 1017–1024.

Holt, L. J., B. B. Tuch, J. Villén, A. D. Johnson, S. P. Gygi et al.,2009 Global analysis of Cdk1 substrate phosphorylation sitesprovides insights into evolution. Science 325: 1682–1686.https://doi.org/10.1126/science.1172867

Huh, W.-K., J. V. Falvo, L. C. Gerke, A. S. Carroll, R. W. Howsonet al., 2003 Global analysis of protein localization in buddingyeast. Nature 425: 686–691. https://doi.org/10.1038/nature02026

Ilisso, C. P., L. Sapio, D. Delle Cave, M. Illiano, A. Spina et al.,2016 S-adenosylmethionine affects ERK1/2 and Stat3 path-ways and induces apotosis in osteosarcoma cells. J. Cell. Physiol.231: 428–435. https://doi.org/10.1002/jcp.25089

Klein, H. L., 2001 Spontaneous chromosome loss in Saccharomy-ces cerevisiae is suppressed by DNA damage checkpoint func-tions. Genetics 159: 1501–1509.

Kodaki, T., S. Tsuji, N. Otani, D. Yamamoto, K. S. Rao et al.,2003 Differential transcriptional regulation of two distinctS-adenosylmethionine synthetase genes (SAM1 and SAM2) ofSaccharomyces cerevisiae. Nucleic Acids Res. Suppl. (3): 303–304. https://doi.org/10.1093/nass/3.1.303

Kunkel, T. A., and K. Bebenek, 2000 DNA replication fidelity.Annu. Rev. Biochem. 69: 497–529. https://doi.org/10.1146/an-nurev.biochem.69.1.497

Kunkel, T. A., R. D. Sabatino, and R. A. Bambara, 1987 Exonucleolyticproofreading by calf thymus DNA polymerase delta. Proc. Natl. Acad.Sci. USA 84: 4865–4869. https://doi.org/10.1073/pnas.84.14.4865

Lea, D. E., and C. A. Coulson, 1949 The distribution of the num-bers of mutants in bacterial populations. J. Genet. 49: 264–285.https://doi.org/10.1007/BF02986080

Li, S., S. K. Swanson, M. Gogol, L. Florens, M. P. Washburn et al.,2015 Serine and SAM responsive complex SESAME regulateshistone modification crosstalk by sensing cellular metabolism.Mol. Cell 60: 408–421. https://doi.org/10.1016/j.molcel.2015.09.024

Liu, Q., L. Liu, Y. Zhao, J. Zhang, D. Wang et al., 2011 Hypoxiainduces genomic DNA demethylation through the activation ofHIF-1a and transcriptional upregulation of MAT2A in hepatomacells. Mol. Cancer Ther. 10: 1113–1123. https://doi.org/10.1158/1535-7163.MCT-10-1010

Luria, S. E., and M. Delbrück, 1943 Mutations of bacteria fromvirus sensitivity to virus resistance. Genetics 28: 491–511.

Martínez-Chantar, M. L., F. J. Corrales, L. A. Martínez-Cruz, E. R.García-Trevijano, Z.-Z. Huang et al., 2002 Spontaneous oxida-tive stress and liver tumors in mice lacking methionine adeno-syltransferase 1A. FASEB J. Off. Publ. Fed. Am. Soc. Exp. Biol.16: 1292–1294. https://doi.org/10.1096/fj.02–0078fje

Mato, J. M., and S. C. Lu, 2007 Role of S-adenosyl-L-methioninein liver health and injury. Hepatology 45: 1306–1312. https://doi.org/10.1002/hep.21650

Mizunuma, M., K. Miyamura, D. Hirata, H. Yokoyama, and T.Miyakawa, 2004 Involvement of S-adenosylmethionine in G1cell-cycle regulation in Saccharomyces cerevisiae. Proc. Natl.Acad. Sci. USA 101: 6086–6091. https://doi.org/10.1073/pnas.0308314101

Mülleder, M., E. Calvani, M. T. Alam, R. K. Wang, F. Eckerstorferet al., 2016 Functional metabolomics describes the yeast bio-synthetic regulome. Cell 167: 553–565.e12. https://doi.org/10.1016/j.cell.2016.09.007

Navas, T. A., Y. Sanchez, and S. J. Elledge, 1996 RAD9 and DNApolymerase epsilon form parallel sensory branches for transduc-ing the DNA damage checkpoint signal in Saccharomyces cere-visiae. Genes Dev. 10: 2632–2643. https://doi.org/10.1101/gad.10.20.2632

Ohnishi, G., K. Endo, A. Doi, A. Fujita, Y. Daigaku et al.,2004 Spontaneous mutagenesis in haploid and diploid Sac-charomyces cerevisiae. Biochem. Biophys. Res. Commun. 325:928–933. https://doi.org/10.1016/j.bbrc.2004.10.120

Pai, C.-C., and S. E. Kearsey, 2017 A critical balance: dNTPs andthe maintenance of genome stability. Genes (Basel) 8: 57.https://doi.org/10.3390/genes8020057

Pascale, R. M., M. M. Simile, M. R. De Miglio, A. Nufris, L. Dainoet al., 1995 Chemoprevention by S-adenosyl-L-methionine ofrat liver carcinogenesis initiated by 1,2-dimethylhydrazine andpromoted by orotic acid. Carcinogenesis 16: 427–430. https://doi.org/10.1093/carcin/16.2.427

Paulovich, A. G., R. U. Margulies, B. M. Garvik, and L. H. Hartwell,1997 RAD9, RAD17, and RAD24 are required for S phase reg-ulation in Saccharomyces cerevisiae in response to DNA dam-age. Genetics 145: 45–62.

Peng, J., D. Schwartz, J. E. Elias, C. C. Thoreen, D. Cheng et al.,2003 A proteomics approach to understanding protein ubiqui-tination. Nat. Biotechnol. 21: 921–926. https://doi.org/10.1038/nbt849

Phalip, V., I. Kuhn, Y. Lemoine, and J. M. Jeltsch, 1999 Char-acterization of the biotin biosynthesis pathway in Saccharomycescerevisiae and evidence for a cluster containing BIO5, a novelgene involved in vitamer uptake. Gene 232: 43–51. https://doi.org/10.1016/S0378-1119(99)00117-1

Phuong, N. T., S. K. Kim, J. H. Im, J. W. Yang, M. C. Choi, et al.,2016 Induction of methionine adenosyltransferase 2A in ta-moxifen-resistant breast cancer cells. Oncotarget 7: 13902–13916. https://doi.org/10.18632/oncotarget.5298

Pringle, J. R., and L. H. Hartwell, 1981 The Molecular Biology ofthe Yeast Saccharomyces cerevisiae. Cold Spring Harbor Labora-tory, Cold Spring Harbor, NY.

Sansregret, L., and C. Swanton, 2017 The role of aneuploidy incancer evolution. Cold Spring Harb. Perspect. Med. 7: a028373.https://doi.org/10.1101/cshperspect.a028373

Schwob, E., T. Böhm, M. D. Mendenhall, and K. Nasmyth,1994 The B-type cyclin kinase inhibitor p40SIC1 controls theG1 to S transition in S. cerevisiae. Cell 79: 233–244. https://doi.org/10.1016/0092-8674(94)90193-7

Sheltzer, J. M., H. M. Blank, S. J. Pfau, Y. Tange, B. M. George et al.,2011 Aneuploidy drives genomic instability in yeast. Science333: 1026–1030. https://doi.org/10.1126/science.1206412

Skoneczna, A., A. Kaniak, and M. Skoneczny, 2015 Genetic insta-bility in budding and fission yeast-sources and mechanisms.FEMS Microbiol. Rev. 39: 917–967. https://doi.org/10.1093/femsre/fuv028

Slany, R. K., M. Bösl, P. F. Crain, and H. Kersten, 1993 A newfunction of S-adenosylmethionine: the ribosyl moiety of AdoMetis the precursor of the cyclopentenediol moiety of the tRNAwobble base queuine. Biochemistry 32: 7811–7817. https://doi.org/10.1021/bi00081a028

Snyder, M., S. Gehrung, and B. D. Page, 1991 Studies concerningthe temporal and genetic control of cell polarity in Saccharomy-ces cerevisiae. J. Cell Biol. 114: 515–532. https://doi.org/10.1083/jcb.114.3.515

Spencer, F., S. L. Gerring, C. Connelly, and P. Hieter, 1990 Mitoticchromosome transmission fidelity mutants in Saccharomycescerevisiae. Genetics 124: 237–249.

Stodola, J. L., and P. M. Burgers, 2016 Resolving individual stepsof Okazaki-fragment maturation at a millisecond timescale. Nat.Struct. Mol. Biol. 23: 402–408. https://doi.org/10.1038/nsmb.3207

Strange, R. C., M. A. Spiteri, S. Ramachandran, and A. A. Fryer,2001 Glutathione-S-transferase family of enzymes. Mutat. Res.482: 21–26. https://doi.org/10.1016/S0027-5107(01)00206-8

Strome, E. D., X. Wu, M. Kimmel, and S. E. Plon, 2008 Heterozygousscreen in Saccharomyces cerevisiae identifies dosage-sensitive

AdoMet and Genome Instability in Yeast 111

genes that affect chromosome stability. Genetics 178: 1193–1207.https://doi.org/10.1534/genetics.107.084103

Swaney, D. L., P. Beltrao, L. Starita, A. Guo, J. Rush et al.,2013 Global analysis of phosphorylation and ubiquitylationcross-talk in protein degradation. Nat. Methods 10: 676–682.https://doi.org/10.1038/nmeth.2519

Taguchi, H., and I. Chanarin, 1978 The effect of homocysteine,methionine, serine and glycine on DNA synthesis by human nor-moblastic and megaloblastic bone marrow cells. J. Nutr. Sci.Vitaminol. (Tokyo) 24: 83–89. https://doi.org/10.3177/jnsv.24.83

Tehlivets, O., N. Malanovic, M. Visram, T. Pavkov-Keller, and W.Keller, 2013 S-adenosyl-L-homocysteine hydrolase and methyl-ation disorders: yeast as a model system. Biochim. Biophys. Acta1832: 204–215. https://doi.org/10.1016/j.bbadis.2012.09.007

Thomas, D., and Y. Surdin-Kerjan, 1997 Metabolism of sulfuramino acids in Saccharomyces cerevisiae. Microbiol. Mol. Biol.Rev. 61: 503–532.

Thomas, D., R. Rothstein, N. Rosenberg, and Y. Surdin-Kerjan,1988 SAM2 encodes the second methionine S-adenosyl trans-ferase in Saccharomyces cerevisiae: physiology and regulationof both enzymes. Mol. Cell. Biol. 8: 5132–5139. https://doi.org/10.1128/MCB.8.12.5132

Tishkoff, D. X., N. Filosi, G. M. Gaida, and R. D. Kolodner, 1997 Anovel mutation avoidance mechanism dependent on S. cerevi-siae RAD27 is distinct from DNA mismatch repair. Cell 88: 253–263. https://doi.org/10.1016/S0092-8674(00)81846-2

Toh, G. W.-L., and N. F. Lowndes, 2003 Role of the Saccharomy-ces cerevisiae Rad9 protein in sensing and responding to DNAdamage. Biochem. Soc. Trans. 31: 242–246. https://doi.org/10.1042/bst0310242

Wach, A., A. Brachat, R. Pöhlmann, and P. Philippsen, 1994 Newheterologous modules for classical or PCR-based gene disruptionsin Saccharomyces cerevisiae. Yeast 10: 1793–1808. https://doi.org/10.1002/yea.320101310

Wang, X., X. Guo, W. Yu, C. Li, Y. Gui et al., 2014 Expression ofmethionine adenosyltransferase 2A in renal cell carcinomas andpotential mechanism for kidney carcinogenesis. BMC Cancer 14:196. https://doi.org/10.1186/1471-2407-14-196

Weinberg, G., B. Ullman, and D. W. Martin, 1981 Mutator phe-notypes in mammalian cell mutants with distinct biochemicaldefects and abnormal deoxyribonucleoside triphosphate pools.Proc. Natl. Acad. Sci. USA 78: 2447–2451. https://doi.org/10.1073/pnas.78.4.2447

Weinert, T., and L. Hartwell, 1989 Control of G2 delay by the rad9gene of Saccharomyces cerevisiae. J. Cell Sci. Suppl. 12: 145–148.https://doi.org/10.1242/jcs.1989.Supplement_12.12

Weinert, T. A., and L. H. Hartwell, 1988 The RAD9 gene controlsthe cell cycle response to DNA damage in Saccharomyces cer-evisiae. Science 241: 317–322. https://doi.org/10.1126/sci-ence.3291120

Weinert, T. A., and L. H. Hartwell, 1990 Characterization ofRAD9 of Saccharomyces cerevisiae and evidence that its func-tion acts posttranslationally in cell cycle arrest after DNA dam-age. Mol. Cell. Biol. 10: 6554–6564. https://doi.org/10.1128/MCB.10.12.6554

Weinert, B. T., C. Schölz, S. A. Wagner, V. Iesmantavicius, D. Suet al., 2013 Lysine succinylation is a frequently occurring mod-ification in prokaryotes and eukaryotes and extensively overlapswith acetylation. Cell Rep. 4: 842–851. https://doi.org/10.1016/j.celrep.2013.07.024

Whalen, R., and T. D. Boyer, 1998 Human glutathione S-transferases.Semin. Liver Dis. 18: 345–358. https://doi.org/10.1055/s-2007-1007169

Zhang, T., Z. Zheng, Y. Liu, J. Zhang, Y. Zhao et al., 2013 Over-expression of methionine adenosyltransferase II alpha (MAT2A)in gastric cancer and induction of cell cycle arrest and apoptosisin SGC-7901 cells by shRNA-mediated silencing of MAT2Agene. Acta Histochem. 115: 48–55. https://doi.org/10.1016/j.ac-this.2012.03.006

Zhao, X., E. G. Muller, and R. Rothstein, 1998 A suppressor of twoessential checkpoint genes identifies a novel protein that nega-tively affects dNTP pools. Mol. Cell 2: 329–340. https://doi.org/10.1016/S1097-2765(00)80277-4

Zheng, Q., 2002 Statistical and algorithmic methods for fluctuationanalysis with SALVADOR as an implementation. Math. Biosci. 176:237–252. https://doi.org/10.1016/S0025-5564(02)00087-1

Zheng, Q., 2008 A note on plating efficiency in fluctuation exper-iments. Math. Biosci. 216: 150–153. https://doi.org/10.1016/j.mbs.2008.09.002

Zheng, Q., 2015 Methods for comparing mutation rates usingfluctuation assay data. Mutat. Res. 777: 20–22. https://doi.org/10.1016/j.mrfmmm.2015.04.002

Zheng, Q., 2016 Comparing mutation rates under the Luria-Delbrück protocol. Genetica 144: 351–359. https://doi.org/10.1007/s10709-016-9904-3

Communicating editor: J. Nickoloff

112 K. M. Hoffert et al.