Blocking S-adenosylmethionine synthesis in yeastallows selenomethionine incorporation andmultiwavelength anomalous dispersion phasingMichael G. Malkowski*†, Erin Quartley‡, Alan E. Friedman§, Julie Babulski‡, Yoshiko Kon‡¶, Jennifer Wolfley*,Meriem Said*, Joseph R. Luft*†, Eric M. Phizicky‡¶, George T. DeTitta*†, and Elizabeth J. Grayhack‡¶�

‡Center for Pediatric Biomedical Research and Departments of §Environmental Medicine and ¶Biochemistry and Biophysics, University of Rochester MedicalSchool, Rochester, NY 14642; *Hauptman-Woodward Medical Research Institute, 700 Ellicott Street, Buffalo, NY 14203; and †Department of StructuralBiology, State University of New York, 700 Ellicott Street, Buffalo, NY 14203

Edited by Wayne A. Hendrickson, Columbia University, New York, NY, and approved February 27, 2007 (received for review November 21, 2006)

Saccharomyces cerevisiae is an ideal host from which to obtain highlevels of posttranslationally modified eukaryotic proteins for x-raycrystallography. However, extensive replacement of methionineby selenomethionine for anomalous dispersion phasing has provenintractable in yeast. We report a general method to incorporateselenomethionine into proteins expressed in yeast based on ma-nipulation of the appropriate metabolic pathways. sam1� sam2�

mutants, in which the conversion of methionine to S-adenosylme-thionine is blocked, exhibit reduced selenomethionine toxicitycompared with wild-type yeast, increased production of proteinduring growth in selenomethionine, and efficient replacement ofmethionine by selenomethionine, based on quantitative massspectrometry and x-ray crystallography. The structure of yeasttryptophanyl-tRNA synthetase was solved to 1.8 Å by using mul-tiwavelength anomalous dispersion phasing with protein that wasexpressed and purified from the sam1� sam2� strain grown inselenomethionine. Six of eight selenium residues were located inthe structure.

x-ray crystallography � methionine � Saccharomyces cerevisiae

S tructural analysis of proteins by x-ray crystallography is apowerful tool to illuminate protein function. Expression of

recombinant proteins in Escherichia coli has been used exten-sively to obtain the large amounts of highly purified proteinrequired for structural analysis. However, many eukaryoticproteins are not soluble in E. coli (1), and many eukaryoticproteins bear posttranslational modifications that are not addedin E. coli because of the lack of requisite modification enzymes.Thus, expression of many proteins may require a eukaryotic host.

The yeast Saccharomyces cerevisiae, which is simple and inex-pensive to culture, is attractive as a eukaryotic expression host.High-level protein expression and affinity purification, yielding2 mg of purified protein per liter of culture (2, 3), has beendemonstrated in appropriate strains. The use of S. cerevisiae asboth a homologous and heterologous expression system for x-raycrystallography is established. The structures of a number ofproteins have been determined with protein expressed in yeast,including yeast Rad6 (4), and Ade1 (5), as well as the catalyticdomain of human ADAR2 (6) and the rabbit Ca2�-ATPase (7),a membrane protein.

A major impediment to using S. cerevisiae as a routineexpression host for x-ray crystallography has been the inabilityto efficiently replace methionine residues with selenomethioninefor anomalous dispersion phasing. In contrast to either multipleisomorphous replacement or molecular replacement (8), multi-wavelength anomalous dispersion (MAD) and single-wavelengthanomalous dispersion (SAD) phasing of selenomethionine-containing crystals provide a solution to the problem of phasedetermination that can be applied to many proteins (9, 10). Thestructural and functional identity of methionine and selenome-thionine derivatives is well established because selenomethi-

onine replacement is easily effected in E. coli. Furthermore, theuse of MAD and SAD phasing has increased with advances insynchrotron beam lines and automated methods for structuresolution, such as SOLVE (11), RESOLVE (12), SnB (13), andBnP (14), to locate the anomalous scattering atoms.

Efficient replacement of methionine with selenomethionine ismuch more difficult in organisms other than E. coli, in whichnearly 100% replacement is routine. The problem with sel-enomethionine incorporation in S. cerevisiae is likely due to thetoxicity of selenomethionine because growth ceases within min-utes after addition of 10 �g/ml selenomethionine (15). Previousattempts to obtain selenomethionine resistant strains in S.cerevisiae have not been helpful either for elucidating the basisof toxicity or for improving incorporation of selenomethionineinto proteins. Selenomethionine resistant mutants that havebeen isolated either increase the intracellular methionine poolsor block the uptake of selenomethionine (16).

We considered it likely that selenomethionine is fully com-petent to replace methionine in protein synthesis. Investigationsin the 1970s had suggested that selenomethionine competedefficiently with methionine in charging tRNA (17) and inincorporation into proteins in vivo (15). Furthermore, wethought it unlikely that replacement of a particular methioninewith selenomethionine in an essential protein is the primarycause of its toxicity, because Singer et al. (15) found that RNAsynthesis is inhibited within minutes after addition of selenome-thionine, whereas protein synthesis continues for �40 min. Somerecent attempts to incorporate selenomethionine in yeasts havemet with limited success, but in most cases have resulted in onlypartial substitution. Bushnell et al. (18) obtained �50% sel-enomethionine incorporation into RNA polymerase II, whichwas insufficient for phasing, but the selenomethionine peaks didserve as markers to guide model building. Larsson et al. (19)reported 40–50% selenomethionine incorporation for two pro-

Author contributions: M.G.M. and E.Q. contributed equally to this work; M.G.M., E.Q.,E.M.P., G.T.D., and E.J.G. designed research; M.G.M., E.Q., and E.J.G. performed research;A.E.F., J.B., Y.K., J.W., M.S., and J.R.L. contributed new reagents/analytic tools; M.G.M.,A.E.F., E.M.P., and E.J.G. analyzed data; and M.G.M. and E.J.G. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.

Abbreviations: AdoMet, S-adenosylmethionine; Se-AdoMet, Se-adenosylselenomethi-onine; MAD, multiwavelength anomalous dispersion; SAD, single-wavelength anomalousdispersion.

Data deposition: The crystallographic coordinates and structure factors have been depos-ited in the Protein Data Bank, www.pdb.org (PDB ID code 2IP1).

�To whom correspondence should be addressed. E-mail: elizabeth�grayhack@urmc.rochester.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0610337104/DC1.

© 2007 by The National Academy of Sciences of the USA

6678–6683 � PNAS � April 17, 2007 � vol. 104 � no. 16 www.pnas.org�cgi�doi�10.1073�pnas.0610337104

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

2, 2

020

teins expressed in Pichia pastoris, in one case, the selenomethi-onine was used for MAD phasing. Laurila et al. (20) obtainedfully selenomethionyl substituted RNA-dependent RNA poly-merase in S. cerevisiae, but induction of expression and sel-enomethionine incorporation occurred in the absence of cellgrowth, in which many proteins are poorly expressed.

We hypothesized that the toxicity of selenomethionine is dueto its conversion to the seleno derivative of S-adenosylmethi-onine (Se-AdoMet). Either Se-AdoMet itself or one of itsmetabolic products could be toxic. S-adenosylmethionine(AdoMet), the primary metabolic product of methionine, isinvolved in a number of important processes, including methyl-ation of proteins, RNAs, and lipids (21) and the biosynthesis ofbiotin (22) and polyamines (Fig. 1A) (23, 24). It has beenestimated that only ATP participates in more reactions thanAdoMet (21). In addition, a product of AdoMet metabolism,such as S-adenosylhomocysteine, homocysteine, cysteine, orglutathione (21), could have toxic effects as selenium derivatives;for example, S-adenosylhomocysteine, itself a potent inhibitor ofmany methyltransferases (21), could be more inhibitory as aseleno derivative. Furthermore, there are two AdoMet syntheta-ses in yeast (encoded by SAM1 and SAM2), either of whichsuffices for synthesis of AdoMet. Thus, one would not be able toobtain an AdoMet� mutant in a simple genetic screen forselenomethionine resistance, if Se-AdoMet were the cause of thetoxicity. Moreover, because AdoMet is required for life, a sam1�

sam2� mutant would only grow if AdoMet was supplied in themedium (21).

We report here that a mutant strain of S. cerevisiae deleted forboth SAM genes exhibits significantly enhanced resistance toselenomethionine, and allows efficient incorporation of sel-enomethionine into proteins. Replacement of methionine withselenomethionine is nearly complete based on quantitative massspectrometry, and permitted structural determination of S.cerevisiae tryptophanyl-tRNA synthetase, Wrs1 protein, usingMAD phasing techniques. Thus, this manipulation of metabo-

lism provides a general solution to the problem of selenomethi-onine incorporation in S. cerevisiae.

ResultsSelenomethionine Resistance Is Conferred by Blocking Conversion ofMethionine to S-Adenosylmethionine. If Se-AdoMet, or one of itsmetabolites, is the primary toxic compound inhibiting growth ofyeast cells in selenomethionine, then preventing conversion ofselenomethionine to Se-AdoMet should allow yeast to grow onselenomethionine. Because two highly homologous yeast genes(SAM1 and SAM2) each encode an AdoMet synthetase, theenzyme responsible for this reaction (21, 25), we created adouble mutant strain in which both of these genes were preciselydeleted, and replaced by drug resistance cassettes derived fromthe yeast knockout collection (26). The resulting sam1� sam2�double mutant strain should block conversion of selenomethi-onine to Se-AdoMet (see Fig. 1 A) and require AdoMet forgrowth (21).

As shown in Fig. 1B, the sam1� sam2� double mutant straingrows well on media containing as much as 120 �M selenome-thionine, whereas growth of the wild-type strain is severelyinhibited on media containing as little as 30 �M selenomethi-onine. Both single mutants have intermediate effects; the sam2�mutant is slightly more resistant to 30 �M selenomethioninethan the wild-type strain, whereas the sam1� mutant is nearly asresistant as the double mutant strain at 60 �M selenomethioninebut sensitive at 120 �M. Resistance to selenomethionine was alsoobserved in a sam1� sam2� mutant created in a different strainbackground (data not shown).

There are two possible explanations for the selenomethionineresistance of the mutants blocked in AdoMet synthesis: eitherfailure to convert selenomethionine to Se-AdoMet could resultin the absence of the toxic compound, or failure to consumemethionine in AdoMet synthesis could result in an increase inthe intracellular methionine concentration, which in turn com-petes with selenomethionine (see Fig. 1 A). To distinguish be-tween these possibilities, we tested the importance of the intra-cellular methionine pool in selenomethionine resistance bydeleting the met6 gene, which blocks the conversion of homo-cysteine to methionine (Fig. 1 A). We find that the sam1� sam2�met6� triple mutant grows better on selenomethionine com-pared with the SAM1� SAM2� met6� strain (Fig. 1C). Thus weinfer that the selenomethionine resistance of the sam1� sam2�double mutant is primarily due to blocking conversion of sel-enomethionine to a toxic compound.

sam1� sam2� Mutants Efficiently Produce Protein in the Presence ofSelenomethionine. To assess the usefulness of the selenomethi-onine resistant mutant strains for the preparation of selenome-thionine-substituted proteins, we examined the yield of ORFfusion proteins expressed in the wild-type and mutant strainsduring growth in selenomethionine. We first generated a plas-mid for expression of ORFs (under control of the GAL1promoter) that were fused at their C terminus to a tripartite tagcontaining a site recognized by 3C protease, His6, an HA epitopeand the ZZ domain of protein A, similar to that used in theMORF collection (2). Protein expression was induced by addi-tion of galactose with or without simultaneous addition ofselenomethionine, and the amount of protein produced wasassessed after affinity purification on IgG resin and SDS/PAGE.

The amount of Ncl1-fusion protein expressed after 24 hinduction in either 0.1 mM or 0.5 mM selenomethionine wassubstantially greater in the sam1� sam2� strain than in thewild-type strain (Fig. 2A, compare lanes h and i with d and e).The sam1� sam2� mutant also grew better than the wild-typestrain in both conditions, based on OD600. In media lackingselenomethionine, Ncl1 protein is produced in good yield in all

0 30 µM 60 µM15 µM 120 µM[Selenomethionine]

SAM1+ SAM2+

sam1- SAM2+

SAM1+ sam2-

sam1- sam2-

SAM1+ SAM2+

0 30 µM 60 µM15 µM 120 µM[Selenomethionine]

SAM1+ SAM2+

sam1- SAM2+

SAM1+ sam2-

sam1- sam2-

SAM1+ SAM2+

B

met6- SAM1+ SAM2+

met6- sam1- SAM2+

met6- SAM1+ sam2-

met6- sam1- sam2-

[Selenomethionine] 0 30 µM 60 µM15 µM

met6- SAM1+ SAM2+

met6- sam1- SAM2+

met6- SAM1+ sam2-

met6- sam1- sam2-

[Selenomethionine] 0 30 µM 60 µM15 µMC

Polyamines &

biotin

methionine

homocysteine

S-adenosylmethionine

S-adenosylhomocysteine

MET6

SA

M1

SAM2

SAH1*

CH3

methyltransferases

methylthioadenosine

MEU1

cysteine

Polyamines &

biotin

A

methionine

homocysteine

S-adenosylmethionine

S-adenosylhomocysteine

MET6

SA

M1

SAM2

SAH1*

CH3

methyltransferases

methylthioadenosine

MEU1

cysteine

Fig. 1. Selenomethionine toxicity and S-adenosylmethionine synthesis. (A)Diagram of methionine and AdoMet metabolic pathways in S. cerevisiae. (B)Selenomethionine resistance of sam1� and sam2� strains. Yeast strains bear-ing the indicated mutations were serially diluted and spotted on SD-methionine minimal media containing 60 �M AdoMet and increasing con-centrations of selenomethionine and then incubated at 30°C. (C)Selenomethionine resistance of sam1� and sam2� strains bearing a met6�mutation on SD complete media with 60 �M AdoMet.

Malkowski et al. PNAS � April 17, 2007 � vol. 104 � no. 16 � 6679

BIO

PHYS

ICS

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

2, 2

020

strains except the sam1� sam2� met6� triple mutant (Fig. 2 anddata not shown).

Analysis of Wrs1 expression confirms that growth (data notshown) and protein production in selenomethionine are signif-icantly better in the sam1� sam2� mutant compared with thewild-type strain (Fig. 2B). Examination of five other proteins(Rad6p, Dps1p, Dus1p, Mes1p and Gus1p) confirms that thismethod is generally applicable for expression in selenomethi-onine. In each case, we observed efficient protein production inthe sam1� sam2� mutant in selenomethionine at �30–60% ofthe amount obtained from cells grown in sulfate (data notshown). We note that there are likely some protein-specificdifferences in expression in selenomethionine in that addition of0.5 mM selenomethionine has a more inhibitory effect on Wrs1protein production relative to that observed with Ncl1.

We also examined protein expression in the sam1� SAM2�

met6� strain after transient repression of SAM2 by addition of1 mM choline and 50 �M myoinositol (27, 28), coincident withinduction of ORF fusion expression. Incorporation of selenome-thionine could in principle be more efficient in a met6� back-ground which should contain reduced levels of intracellularmethionine. However, we find that production of proteins in 0.2mM selenomethionine is poor in this strain and is only marginallybetter in conditions in which SAM2 is repressed (Fig. 2). Thus,despite the fact that this strain is quite resistant to selenome-thionine (Fig. 1C), it is not useful for protein production in theseconditions.

Selenomethionine Incorporation Is Nearly Complete, Based on MassSpectrometry. The usefulness of protein purified from the sam1�sam2� mutant for MAD/SAD phasing depends primarily upon thefraction of methionine that is replaced by selenomethionine. Tojudge the efficiency of selenomethionine substitution, we usedquantitative mass spectrometry, based on differential isotope la-beling of two samples, to measure the specific reduction in methi-onine-containing tryptic peptides in Ncl1 protein that was prepared

from yeast grown in selenomethionine. Differential isotope labelingof Ncl1 protein (Fig. 3A) was accomplished by digestion with trypsinin either H2

16O or H218O (29, 30); tryptic peptides will then be

labeled at their newly generated C termini with one or two 18O(16O) molecules, resulting in a m/z difference of 2 or 4 (31). Mixturesof H2

16O trypsin-digested and H218O trypsin-digested Ncl1 protein

were cospotted and analyzed by MALDI-TOF, allowing directcomparison of signal intensities of chemically identical but differ-entially labeled peptides (29). We first identified 6 easily detectabletryptic peptides derived from Ncl1 protein. We established that the� 2 and � 4 signals from these peptides can be attributed solely tothe Ncl1 from the 18O sample, and that the protein digested in 18Ocontributes little (if any) signal to the 16O bands (data not shown).Four of these peptides do not contain methionine and are expectedto exhibit equal representation in Ncl1 protein whether or not themedia contains selenomethionine; two peptides contain methio-nine and are expected to show reduced representation if theirmethionine is replaced by selenomethionine.

As shown in Fig. 3B, the two methionine-containing peptidesare severely depleted in Ncl1 protein derived from yeast grownin 0.5 mM selenomethionine compared with Ncl1 protein frommedia lacking selenomethionine. The mixture of equal amountsof trypsin-digested Ncl1 protein made from cultures grown insulfate and in 0.5 mM selenomethionine yields similar signals forthe four peptides that do not contain methionine (Fig. 3B). In

Harvest OD: 5.1 2.6 1.4 8.8 3.8 4.3 2.1 4.1 5.2 0.9 2.3 4.1Harvest OD: 5.1 2.6 1.4 8.8 3.8 4.3 2.1 4.1 5.2 0.9 2.3 4.1

a b c d e f g h i j k l m na b c d e f g h i j k l m n

Wrs1p

B WT sam1∆sam2∆ sam1∆met6∆Strain:

Ncl1p

WTStrain:

a b c d e f g h i j k l m na b c d e f g h i j k l m n

[Selmet] 0 0 00 0 00.1 0.1 0.20.5 0.5 0.2[Selmet] 0 0 00 0 00.1 0.1 0.20.5 0.5 0.2

A

Induction OD: 1.8 1.4 1.7 0.8 1.2 1.4 0.81.2 1.3 0.8 1.5 1.3Induction OD: 1.8 1.4 1.7 0.8 1.2 1.4 0.81.2 1.3 0.8 1.5 1.3

sam1∆sam2∆ sam1∆met6∆

Fig. 2. Evaluation of protein expression in sam1� sam2� and sam1� met6�mutants during growth in media containing selenomethionine. (A) Analysis ofNcl1 protein expression. Ncl1-fusion protein was induced with galactose in thepresence and absence of selenomethionine as indicated, and expressed proteinwas evaluated after binding to IgG Sepharose, followed by SDS/PAGE and Coo-massiestaining.Lanes:a,Bio-Radbroad-rangestandards;b,GST-3Cprotease;c–n,protein from 7 ml of culture. The concentration of selenomethionine (millimolar)and OD600 at induction and at harvest are indicated. Cells in lanes f, j, and n wereinduced in rich media; others were induced in synthetic media. Cells in lane mwere treated with 1 mM choline and 50 �M myoinositol at induction. (B) Analysisof Wrs1 protein expression. Samples were treated as in A.

[Selenomethionine] 0 0.2 0.375 0.50 0.125b c da e f g

Met peptide

SeMet peptide

Met Peptide 2073

LNSANLMVVNHDAQFFPR

Re

lative

ab

un

da

nce

1206 1412 1417 1716 1248 2073

[Selenomethionine] mM

Re

lative

ab

un

da

nce

B

A

C

[Selenomethionine] 0 0.2 0.375 0.50 0.125[Selenomethionine] 0 0.2 0.375 0.50 0.125

b c da e f gb c da e f g

Met peptide

SeMet peptide

Met peptide

SeMet peptide

Met Peptide 2073

LNSANLMVVNHDAQFFPR

Re

lative

ab

un

da

nce

0

100

40

140

1206 1412 1417 1716 12481206 1412 1417 1716 1248 2073

[Selenomethionine] mM

Re

lative

ab

un

da

nce

0 0.125 0.25 0.50.375

Met peptides

Ncl1p tryptic peptides (m/z)

16O peptide (lane b)

O peptide (lane g)18

0

20

70

50

Fig. 3. Quantification of individual peptides from purified Ncl1 after growthin selenomethionine. (A) SDS/PAGE of purified Ncl1 protein. Lanes: a, Bio-Radmarkers; b, 15 �g of Ncl1 protein from culture grown in sulfate; c–g, 3 �g ofNcl1 protein each, from cultures grown with selenomethionine (millimolar) asindicated. Samples of Ncl1 protein were subsequently excised, dialyzed, andtreated with trypsin in buffer containing H2

16O (lane b) or H218O (lanes c–g).

(B) Comparison by mass spectrometry of yield of individual peptides contain-ing or lacking methionine. Samples of trypsin-treated Ncl1 protein in H2

16O (A,lane b, sulfate) and H2

18O (A, lane g, 0.5 mM selenomethionine) were mixedin equal amounts and subjected to MALDI mass spectrometry, and relativeheights of peaks were compared. Tryptic peptides that lack methionine (16O)are m/z 1206.70, 1412.88, 1417.89, and 1716.00, and those that containmethionine are m/z 1248.77 and 2073.31. (C) Analysis of the abundance of themethionine-containing peptide LNSANLMVVNHDAQFFPR and its correspond-ing selenomethionine-containing peptide (m/z �48 for selenomethionineand �2/4 for 18O) as a function of the concentration of selenomethionine inthe media. Both peptides were normalized based on their correspondingmethionine-lacking peptides.

6680 � www.pnas.org�cgi�doi�10.1073�pnas.0610337104 Malkowski et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

2, 2

020

each case, there is slightly more signal for the non-methioninecontaining peptide from the selenomethionine-grown samplethan from the corresponding peptide from the sulfate grownsample (which is set at 100%). By contrast, there is dramaticallyless signal (2% and 5% respectively) for the two methionine-containing peptides from the selenomethionine-grown samplecompared with the corresponding peptides from the sulfate-grown sample. This result demonstrates that �95% of themethionine is replaced by selenomethionine in the Ncl1 proteingrown in 0.5 mM selenomethionine. For one of these peptides(LNSANLMVVNHDAQFFPR of m/z 2073.31), we also observean increase in the corresponding selenomethionine-containingpeptide (Fig. 3C) with increasing selenomethionine in the media,although this peptide cannot be quantified relative to thecorresponding methionine peptide because they are chemicallydifferent.

We also examined the effect of selenomethionine concentra-tion in the media on the efficiency of selenomethionine incor-poration. We observed 83% incorporation at 0.125 mM sel-enomethionine, 90% at 0.25 mM, and 95% at 0.5 mM (Fig. 3C).

Structure of Yeast Wrs1. To further establish that the sam1�sam2� S. cerevisiae strain efficiently incorporates selenomethi-onine into proteins, we determined the crystal structure of S.cerevisiae Wrs1 protein (eight Met residues), produced in thisstrain, using MAD phasing techniques. Selenium incorporationwas verified based on fluorescence scans at the beamline and bythe anomalous difference electron density shown in Fig. 4. Theselenium substructure of Wrs1 was solved by using BnP (14),with the positions of six of the eight Se atoms identified in thetop 8 peaks, producing phases with an initial mean figure of meritof 0.57. Subsequent phase refinement in BnP and RESOLVE(32) resulted in a mean figure of merit increase to 0.78. Fouriermaps generated at this stage were easily interpretable, allowingfor facile building of the initial protein model (Fig. 4).

The final model of yeast Wrs1 consists of 371 residues,corresponding to residues 99–477 based on sequence alignmentto the human enzyme (33). The N-terminal 46 residues (residues54 through 98) and the region containing the conserved KMSKSmotif (residues 345–352) are disordered and not included in themodel. SDS/PAGE analysis of dissolved crystals results in a

single band of protein migrating at a molecular mass of �45 kDa,indicating a loss of �5 kDa from the theoretically calculatedmolecular mass (factoring in the additional 8 residues left at theC terminus after 3C protease cleavage) and consistent with theloss of the first �45 residues of the protein (data not shown).Hence, the structure of Wrs1 reported here represents thecatalytic fragment of full length Wrs1, generated through pro-teolysis of the N terminus during purification and similar to thatreported for human Wrs1 protein (33). Taken together, theseobservations explain why only six of the possible eight seleniumsites were located by BnP. Analysis of an anomalous differenceelectron density map, calculated by using the anomalous differ-ences from the peak data set and the phases derived from BnPand RESOLVE clearly define the positions of the seleniumatoms for the six ordered methionine residues in the proteinmodel (Fig. 4). Detailed analysis of Wrs1 will be reportedelsewhere.

DiscussionWe have demonstrated that preventing the conversion of sel-enomethionine to Se-AdoMet dramatically reduces the toxicityof selenomethionine in S. cerevisiae, allowing both better growthin the presence of selenomethionine and nearly complete re-placement of methionine with selenomethionine in expressedprotein. Based on quantitative mass spectrometry, �90% of themethionine is replaced by selenomethionine in the presence of0.25–0.5 mM selenomethionine. These growth conditions arelikely to be generally applicable for incorporation of selenome-thionine because we have produced seven proteins in the sam1�sam2� mutant grown in selenomethionine. Furthermore, thestructure of Wrs1 protein was solved by using MAD phasing ofprotein purified from the sam1� sam2� mutant grown inselenomethionine. Thus it is now feasible to consider yeast as asource of proteins for structural genomics, because phasingissues are greatly eased by efficient selenomethionine incorpo-ration. We speculate that these results may provide a generalsolution for production of selenomethionine containing proteinsin other organisms.

MethodsStrain Construction and Growth. Yeast strains [supporting infor-mation (SI) Table 2] derived from strain Y258 (MATa, pep4-3,

Fig. 4. Representative electron densityfor yeast Wrs1. A MAD experimentalelectron density map, calculated at 3.0 Åand contoured at 1� (blue), is shown fora region containing Met-169, Met-174,and Met-360 of yeast Wrs1. The map wasgenerated by using data from BnP andRESOLVE, with the final coordinates re-fined at 1.8 Å represented as a stickmodel. An anomalous difference elec-tron density map, calculated at 3.0 Å andcontoured at 10� (red), is also shown forthe same region. This map was calculatedby using the anomalous differences fromthe peak data set and the phases fromBnP and RESOLVE. The three seleniumatoms within the Met side chains areclearly defined in the map. The remain-ing three ordered Met residues in thestructure (Met-242, Met-429, and Met-460) show equivalent peaks (data notshown). This figure was made by usingthe molecular graphics program PYMOL(www.pymol.org).

Malkowski et al. PNAS � April 17, 2007 � vol. 104 � no. 16 � 6681

BIO

PHYS

ICS

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

2, 2

020

his4-580, ura3-52, leu2-3, 112) (2) were constructed by transfor-mation with kanMX and nat1 cassettes as described (26, 34, 35).Oligonucleotides for amplification and confirmation are listed inSI Table 3.

Selenomethionine resistance was assayed by growth of cul-tures overnight in YPD (with 60 �M AdoMet as required),followed by dilution in YPD to OD600 � 0.5, growth for 4 h,dilution to OD600 � 1.25, spotting 2.5 �l of 5-fold serial dilutionson synthetic (S) dropout media (lacking methionine or completeas indicated) containing 60 �M AdoMet and selenomethionineas indicated, and incubation at 30°C.

Protein Affinity Purification. ORFs were cloned under PGAL con-trol into vector BG2483, a 2� URA3 vector derived from BG1766(2), which expresses ORFs fused at their C terminus to amultipart tag (3C site-HA epitope-His6-ZZ domain of proteinA) and which has been modified for ligation independentcloning. Yeast transformants were grown overnight in SD lack-ing uracil containing 64 �M AdoMet, diluted 20-fold in Sdropout media lacking uracil and methionine, containing 2%raffinose (36) and 64 �M AdoMet, and grown for 8 h at 30°C,followed by dilution to OD600 of 0.025, and overnight growth. AtOD600 �1, cultures were induced by addition of galactose to 2%(and selenomethionine and other additions, as indicated), andgrowth was continued for 24 h, followed by harvesting of cells.Expressed protein was evaluated from crude extracts by bindingto IgG Sepharose (2), elution of bound protein by boiling thebeads in loading buffer, and SDS/PAGE. Proteins subjected tomass spectrometry were removed from IgG Sepharose by 3Cprotease cleavage (2), subjected to SDS/PAGE, and eluted asdescribed (37). Purification of Wrs1 protein for crystallizationinvolved growth of 18 liters of culture, IgG Sepharose chroma-tography, 3C protease cleavage, and gel filtration on a HiLoad16-60 Superdex 200 prep grade (Amersham, Piscataway, NJ)column.

Quantitative MALDI Mass Spectrometry. Samples were prepared formass spectrometry analysis by mixing two portions of trypticpeptides extracted from gel samples and digested as described (37),one from the H2

16O and the other from 95% H218O (Isotec)

digestions. MALDI spectra, obtained on a Voyager DE STRMALDI-TOF instrument (Applied Biosystems, Framingham,MA), were acquired in positive ion reflectron mode as described(38), with an accelerating voltage of 20 kV and a 150-ns delay time.Spectra were obtained by averaging 200 acquisitions from singlelaser pulse (20 Hz, Voyager, N2 337-nm laser). Proteins wereidentified by using ion searches performed on the processed spectraagainst the NCBI NR database using the MASCOT search engine(Matrix Science, Inc., Boston, MA) with carbamidomethyl cysteineand oxidized methionine as variable modifications (www.ebi.ac.uk/IPI/IPIhuman.html), to determine possible sequence correlationsof known proteins with variable carbamidomethyl cysteine, oxi-dized methionine and selenomethionine modifications.

Crystallization and Data Collection. Wrs1 [4.8 mg/ml in 20 mMHepes (pH 7.5), 5% (vol/vol) glycerol, 500 mM sodium chloride,2 mM DTT, 0.025% (wt/vol) sodium azide] was screened forcrystallization by using a high throughput microbatch-under oiltechnique (39). After 1 week at 4°C, 19 of 1,536 cocktailsproduced conditions suitable for optimization. One mixture[80% (vol/vol) polyethylene glycol 400 (PEG 400), 100 mMammonium bromide, 100 mM Mops (pH 7.0)] was optimized forcrystallization by volumetric codilution of the protein and mix-ture solutions, using a modified version of a batch protocol (40).Crystals of the Se-Met derivative of Wrs1 (5.0 mg/ml) wereproduced by using the same technique.

Crystals of the native and Se-Met derivative of Wrs1 weref lash frozen in liquid nitrogen using the high concentration ofPEG 400 from the crystallization mixture as the cryopro-tectant. Multiwavelength anomalous x-ray diffraction data to3.0 Å were collected at beamline 9-1 at the Stanford Synchro-tron Research Laboratory (SSRL) by using remote data

Table 1. Data collection, phasing, and refinement statistics for Wrs1p

Peak Inflection High remote Native

Data collection statistics for WRS1 from SCALA*Wavelength, Å 0.97935 0.97952 0.96109 0.97945Resolution, Å 3.0 (3.16–3.0) 3.0 (3.16–3.0) 3.0 (3.16–3.0) 1.8 (1.90–1.8)Space group P43212 P43212 P43212 P43212Unit cell (a � b), Å 56.35 56.35 56.35 56.21Unit cell (c), Å 313.64 313.64 313.64 313.25Unique reflections 11,002 (1,558) 11,026 (1,558) 11,000 (1,554) 45,655 (4,942)Rmerge, %† 5.0 (12.3) 5.0 (12.4) 6.2 (15.4) 7.0 (38.2)Completeness, % 99.9 (100) 99.9 (100) 99.9 (100) 94.0 (73)Average I/�(I) 26.9 (15.5) 26.7 (15.1) 22.2 (11.9) 16.9 (2.3)

Final refinement statistics from REFMAC5*Resolution range, Å* 40–1.8Residues 371Waters 358Ligands (PEG 400) 4Rcryst (43,259 reflections), %‡ 18.1 (27.5)Rfree (2,288 reflections), %‡ 21.4 (33.4)Average B factor, protein (all, main, side), Å2 25.08, 24.32, 25.84Average B factor, solvent, Å2 38.87Average B factor, PEG 400, Å2 44.75rmsd, bond lengths, Å 0.013rmsd, bond angles, ° 1.343

*Values in parentheses are for last shell of data.†Rmerge � � I � �I /�I, where I is the observed intensity of an individual reflection, and �I is the mean intensity of that reflection.‡Rcryst and Rfree � � Fo � Fc /�Fo, where Fo and Fc are observed and calculated structure factors, respectively. Rfree is the cross-validationR factor computed for the test set of reflections (5% of the total were used), which are omitted during the refinement process.

6682 � www.pnas.org�cgi�doi�10.1073�pnas.0610337104 Malkowski et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

2, 2

020

collection techniques (41, 43, **). Data sets were measured atthe selenium K-edge peak (0.97935 Å), inf lection point(0.97952 Å), and remote (0.96109 Å) energy wavelengths.Native crystals of Wrs1, which diffracted to 1.8 Å, were alsocollected remotely by using beamline 11-1 at SSRL. All datawere processed by using MOSFLM (44) combined withSCALA from the CCP4 suite of programs (45). The datacollection statistics are summarized in Table 1.

Structure Solution and Refinement. The structure of Wrs1 wassolved by using MAD phasing techniques. The heavy-atomsubstructure was solved by using BnP (14). Wrs1 protein wascrystallized in the tetragonal space group P4x212. As such, theenantiomorph determination feature of BnP was used to deter-mine the correct space group. By using unrefined protein phases,the ratio of the standard deviations of the electron density in theprotein and solvent regions was 1.11 for P41212 and 1.81 forP43212, indicating that the latter was the true space group. Therefined phases from BnP were input into RESOLVE (32) fordensity modification and automated chain tracing. RESOLVEwas able to position 196 of 432 residues. Fourier maps weregenerated from the refined and solvent-f lattened phases outputfrom RESOLVE and used to adjust the initial protein model andmanually build an additional 175 residues into the electrondensity using COOT (46).

The model was initially refined with CNS (47). Rigid bodyrefinement, followed by cycles of positional and group B factor

refinement (main chain/side chain) resulted in R and free Rvalues of 27.1% and 31.1%, respectively. Iterative cycles ofmodel building in COOT followed by positional and group Bfactor refinement were carried out to adjust f lexible loop regionsand build side chain residues. The resulting atomic model wasthen used as a starting point for refinement of the isomorphoushigh-resolution data set. Iterative cycles of refinement (posi-tional and individual B factor refinement) followed by modelbuilding in COOT dropped the R and free R factors to 25.4% and28.0%, respectively. Strong electron density was observed in the2Fo � Fc and Fo � Fc electron density maps calculated at thisstage for four PEG 400 molecules, which were built in. A totalof 358 water molecules were also added. In the final stages ofrefinement, the TLS feature of REFMAC 5.0 (42, 48) wasapplied. Residues 99–153 (N-terminal), 154–362 and 453–471(catalytic), and 363–452 (C-terminal) were defined as distinctdomains with TLS to model anisotropic displacements of eachdomain as a rigid-body. Final refinement statistics are summa-rized in Table 1.

We thank M. Dumont for advice and comments and Dr. Eddie Snell foradvice and assistance during remote data collections at the StanfordSynchrotion Radiation Laboratory (SSRL). This work was supported byNational Institutes of Health (NIH) Grant NIH 1 U54 GM074899establishing the Center for High Throughput Structural Biology. Por-tions of this research were carried out at SSRL, a national user facilityoperated by Stanford University on behalf of the U.S. Department ofEnergy, Office of Basic Energy Sciences. The SSRL Structural Molec-ular Biology Program is supported by the Department of Energy, Officeof Biological and Environmental Research and by the NIH, NationalCenter for Research Resources, Biomedical Technology Program, andthe National Institute of General Medical Sciences.

1. Phizicky EM, Grayhack EJ (2006) Crit Rev Biochem Mol Biol 41:315–327.2. Gelperin DM, White MA, Wilkinson ML, Kon Y, Kung LA, Wise KJ,

Lopez-Hoyo N, Jiang L, Piccirillo S, Yu H, et al. (2005) Genes Dev 19:2816–2826.

3. Martzen MR, McCraith SM, Spinelli SL, Torres FM, Fields S, Grayhack EJ,Phizicky EM (1999) Science 286:1153–1155.

4. Worthylake DK, Prakash S, Prakash L, Hill CP (1998) J Biol Chem 273:6271–6276.

5. Levdikov VM, Barynin VV, Grebenko AI, Melik-Adamyan WR, Lamzin VS,Wilson KS (1998) Structure (London) 6:363–376.

6. Macbeth MR, Schubert HL, Vandemark AP, Lingam AT, Hill CP, Bass BL(2005) Science 309:1534–1539.

7. Jidenko M, Nielsen RC, Sorensen TL, Moller JV, le Maire M, Nissen P, JaxelC (2005) Proc Natl Acad Sci USA 102:11687–11691.

8. Ealick SE (2000) Curr Opin Chem Biol 4:495–499.9. Hendrickson WA (2000) Trends Biochem Sci 25:637–643.

10. Hendrickson WA, Horton JR, LeMaster DM (1990) EMBO J 9:1665–1672.11. Terwilliger TC, Berendzen J (1999) Acta Crystallogr D Biol Crystallogr 55:849–

861.12. Terwilliger TC (2000) Acta Crystallogr D Biol Crystallogr 56:965–972.13. DeTitta GT, Weeks CM, Thuman P, Miller R, Hauptman HA (1994) Acta

Crystallogr A 50:03–210.14. Weeks CM, Blessing RH, Miller R, Mungee R, Potter SA, Rappleye J, Smith

GD, Xu H, Furey W (2002) Z Kristallogr 217:686–693.15. Singer RA, Johnston GC, Bedard D (1978) Proc Natl Acad Sci USA 75:6083–

6087.16. Cherest H, Surdin-Kerjan Y, De Robichon-Szulmajster H (1975) J Bacteriol

123:428–435.17. Colombani F, Cherest H, de Robichon-Szulmajster H (1975) J Bacteriol

122:375–384.18. Bushnell DA, Cramer P, Kornberg RD (2001) Structure (London) 9:R11–14.19. Larsson AM, Stahlberg J, Jones TA (2002) Acta Crystallogr D Biol Crystallogr

58:346–348.20. Laurila MR, Salgado PS, Makeyev EV, Nettelship J, Stuart DI, Grimes JM,

Bamford DH (2005) J Struct Biol 149:111–115.21. Thomas D, Surdin-Kerjan Y (1997) Microbiol Mol Biol Rev 61:503–532.22. Phalip V, Kuhn I, Lemoine Y, Jeltsch JM (1999) Gene 232:43–51.23. Chattopadhyay MK, Tabor CW, Tabor H (2006) Biochem Biophys Res Commun

343:203–207.24. Subhi AL, Diegelman P, Porter CW, Tang B, Lu ZJ, Markham GD, Kruger

WD (2003) J Biol Chem 278:49868–49873.

25. Cherest H, Surdin-Kerjan Y (1978) Mol Gen Genet 163:153–167.26. Winzeler EA, Shoemaker DD, Astromoff A, Liang H, Anderson K, Andre B,

Bangham R, Benito R, Boeke JD, Bussey H, et al. (1999) Science 285:901–906.27. Santiago TC, Mamoun CB (2003) J Biol Chem 278:38723–38730.28. Kodaki T, Tsuji S, Otani N, Yamamoto D, Rao KS, Watanabe S, Tsukatsune

M, Makino K (2003) Nucleic Acids Symp Ser 3:303–304.29. Domon B, Aebersold R (2006) Science 312:212–217.30. Yao X, Freas A, Ramirez J, Demirev PA, Fenselau C (2001) Anal Chem

73:2836–2842.31. Schnolzer M, Jedrzejewski P, Lehmann WD (1996) Electrophoresis 17:945–953.32. Terwilliger TC (2001) Acta Crystallogr D Biol Crystallogr 57:1755–1762.33. Yu Y, Liu Y, Shen N, Xu X, Xu F, Jia J, Jin Y, Arnold E, Ding J (2004) J Biol

Chem 279:8378–8388.34. Tong AH, Evangelista M, Parsons AB, Xu H, Bader GD, Page N, Robinson

M, Raghibizadeh S, Hogue CW, Bussey H, et al. (2001) Science 294, 2364–2368.35. Alexandrov A, Grayhack EJ, Phizicky EM (2005) RNA 11:821–830.36. Sherman F, Fink G, Hicks JB (1986) in Methods in Yeast Genetics (Cold Spring

Harbor Lab Press, Cold Spring Harbor, NY), pp 145–149.37. Shevchenko A, Wilm M, Vorm O, Mann M (1996) Anal Chem 68:850–858.38. Parker KC, Patterson D, Williamson B, Marchese J, Graber A, He F, Jacobson

A, Juhasz P, Martin S (2004) Mol Cell Proteomics 3:625–659.39. Luft JR, Collins RJ, Fehrman NA, Lauricella AM, Veatch CK, DeTitta GT

(2003) J Struct Biol 142:170–179.40. Rayment I (2002) Structure (London) 10:147–151.41. Cohen AE, Ellis PJ, Miller MD, Deacon AM, Phizackerley RP (2002) J Appl

Crystallogr 35:720–726.42. Winn MD, Isupov MN, Murshudov GN (2001) Acta Crystallogr D Biol

Crystallogr 57:122–133.43. McPhillips TM, McPhillips SE, Chiu HJ, Cohen AE, Deacon AM, Ellis PJ,

Garman E, Gonzalez A, Sauter NK, Phizackerley RP, et al. (2002) J SynchrotronRadiat 9:401–406.

44. Leslie AG (1999) Acta Crystallogr D Biol Crystallogr 55:1696–1702.45. Collaborative Computational Project, Number 4 (1994) Acta Crystallogr D

50:760–763.46. Emsley P, Cowtan K (2004) Acta Crystallogr D 60:2126–2132.47. Brunger AT, Adams PD, Clore GM, DeLano WL, Gros P, Grosse-Kunstleve

RW, Jiang JS, Kuszewski J, Nilges M, et al. (1998) Acta Crystallogr D BiolCrystallogr 54:905–921.

48. Murshudov GN, Vagin AA, Dodson EJ (1997) Acta Crystallogr D BiolCrystallogr 53:240–255.

**Gonzales, A., Moorhead, P., McPhillips, S., Sauter, N. K. (2005) Acta Crystallogr A 61:C486(abstr. P.25.03.1).

Malkowski et al. PNAS � April 17, 2007 � vol. 104 � no. 16 � 6683

BIO

PHYS

ICS

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

2, 2

020