0147-61© 2002 ElsAll rights re

Plasmid 47,94–107 (2002)doi:10.1006/plas.2001.1557, available online at http://www.academicpress.com on

Selective Fitness of Four Episomal Shuttle-Vectors Carrying HIS3, LEU2,TRP1, and URA3 Selectable Markers in Saccharomyces cerevisiae

Simone Ugolini,* Valentina Tosato,† and Carlo V. Bruschi*,1

*Microbiology Group, International Centre for Genetic Engineering and Biotechnology, AREA Science Park,Padriciano 99, I-34012 Trieste, Italy; and†Genomics Group, Laboratory of Genetics,

Department of Biology, University of Trieste, P. le Valmaura 9, I-34148 Trieste, Italy

Received December 18, 2000, revised November 23, 2001

nesate

lowerfor

mongee ofe withigated.

1To w139-40-

A comparison of the slective fitness of four 2-mm-based shuttle-plasmids carrying the yeast geHIS3, LEU2, TRP1, andURA3was performed. The effect of each marker on long-term growth rand plasmid maintenance was measured. In selective medium, theLEU2 andURA3plasmids weremaintained at the lowest and the highest levels, respectively, while theHIS3 and TRP1plasmidswere maintained at an intermediate level. In synthetic complete medium, plasmid loss rate wasfor the genesTRP1andURA3than for the other two markers, and a similar pattern was observedcells growing in rich medium. These results were confirmed by competition experiments atransformants with different plasmids in complete and rich media, indicating a different degrfitness for the markers used. A potential correlation of the energy cost of plasmid maintenancthe secondary DNA structure and the level of expression of the selective markers is also invest© 2002 Elsevier Science (USA)

Key Words:plasmid maintenance; selective pressure; metabolism; Saccharomyces cerevisiae;secondary DNA structure.

e ofla-

shet toec-ctte-ce

t oftiveheker

g-te-toofltsr-

ltoehrrn

no

l ol

einee

Selection for a gene product in the biosynthsis of an amino acid or nucleotide is generaapplied when using plasmid DNA vectors transform the yeast Saccharomyces cerevisia.These vectors provide an invaluable tool for tuse of this organism as a paradigm for eukaotic molecular biology, as well as a cell factofor the expression of biotechnologically relevaheterologous proteins (Romanos et al., 1992).

Starvation for any one of 11 amino acids (icluding histidine, leucine, and tryptophan) nonly induces an increase in the level of the ezymes necessary for their biosynthesis, but aincreases from 2- to 10-fold the transcriptiongenes under the GCN4-dependent generaamino acid control system (Hinnebusch, 199Jones and Fink, 1982). Therefore, under seltive conditions of starvation for an amino acunder this cross-pathway control, i.e., histidior leucine, the cell undergoes metabolic strdue to the parallel induction of a large numb

9X/02 $35.00evier Science (USA)served.

and

94

hom correspondence should be addressed. F375-7343. E-mail: bruschi@icgeb.trieste.it.

of coregulated genes. Conversely, in the casselection for a gene not under Gcn4p regution, i.e.,TRP1or URA3, the number of geneturned on would be limited only to those of tpathway under stress. Hence, it is importanoptimize the use of biosynthetic markers, seltion for which should result in the least impaon the cellular metabolism and optimal mainnance of the vector carrying the DNA sequenof interest.

In this report we describe the use of a sefour episomal plasmids to measure the selecadvantage conferred to the cell by four of tmost commonly used yeast biosynthetic margenes, two under the control of Gcn4,HIS3andLEU2, and two independent of it,TRP1 andURA3. We focused our attention on the lonterm effects on cell growth and plasmid mainnance, using multicopy plasmid constructs amplify the differences in the energy cost plasmid maintenance to the cell. Our resuclearly indicate that, in long-term growth expeiments, the markers independent of the GCN4system are most successfully selected for

e-ly

ey-yt

-tn-sof

2;c-

dessr

ax:

M

ot u

ut

g

COMPARATIVE PLASMID

promote a more vigorous growth of their hcells. A correlation between the energy cosplasmid maintenance and the different struct

and functional features of the DNA of the fou

f

,

1

n

cns

in

l

o

ofed

R.e

s)

-,

-sendut-aa

rsas

A

genes studied, including their transcription eciency, is discussed.

MATERIALS AND METHODS

Strains, Media, and DNA Manipulations

Escherichia coli bacterial strains used foroutine work were DH5a (endA1, hsdR17supE44, thi1, recA1, gyrA96, relA1,DlacU169(phi80lacZDM15)), (Hanahan, 1983); HB10(hsdS20, supE44, ara14, galK2, lacY1, proArpsL20, xyl-5, mtl-1, recA13, mcrB, mcrA, mrr),(Boyer and Roulland-Dussoix, 1969); and KC(hsdR17, hisB463, leuB6, pyrF::Tn5 KmR,trpC9830,D(lacZYA), strA, galU, GalK) (kindlyprovided by K. Struhl). Saccharomyces cerevisiae yeast strain CBU1-31 (Mat a [cir 1]ade8-18 his3-D1 leu2-3,112 trp1-289a ura3-52can1R) was constructed in this work using stadard genetic techniques (Rose et al., 1990). Pro-totrophic revertant colonies were observed oin the case of the revertible amber trp1-289a

mutation (one revertant out of 3.6 3 106 cfu),but not for the other nonreverting marke(ade8-18, his3-D1, leu2-3,112, and ura3-52).

Standard Luria–Bertani and minimal seletive media were used for bacterial cultivatioaccording to Miller (1992). Semidefined yeaextract–peptone–dextrose (YPD),2 selectivedrop-outs, and sporulation and drug-containmedia were as described in Roseet al. (1990).The final concentration of the relevant suppments for the synthetic complete (COMmedium is as follows: 20mg/ml L-histidineHCl 100 mg/ml L-leucine, 20mg/ml L-trypto-phan, and 20mg/ml uracil. Double and tripleconcentrations were also used for enrichCOM medium. Bacterial plasmid DNAminipreps were prepared by a modificationthe alkaline lysis procedure, as described

Zhou et al. (1990). Total yeast DNA was pre-pared according to Roseet al. (1990), while

to aith

2Abbreviations used: COM, complete; SEL; selectivYPD, yeast extract–peptone–dextrose.

AINTENANCE IN YEAST 95

stofralr

fi-

r

2,

8

-

n-

ly

rs

-,t

g

e-)

ed

fby

yeast plasmid DNA minipreps were carried oaccording to Hoffman and Winston (1987).E.coli-competent cells were prepared followinthe RbCl method based on the experimentsHanahan (1983). Yeast cells were transformaccording to the method of Bruschiet al.(1987).

Quantitative RT-PCR

Expression ofHIS3, LEU2, TRP1, andURA3has been analyzed by quantitative RT-PCYeast transformant were grown in selectivmedium up to 107 cells/ml and spheroplastwere obtained after incubation (40 min at 30°Cwith 50 units of lyticase. Total RNA was isolated using the RNeasy Mini Kit from Qiagenand approximately 1mg was utilized as tem-plate for RT-PCR. The resulting cDNA was obtained using the AMV Reverse Transcriptaenzyme and RNase Inhibitor from Promega aincubating at 42°C for 1 h. PCR was carried owith Taq DNA Polymerase (Sigma) and standard quantitative amplification protocols inSprint Hybaid Thermocycler machine usinglimiting number of cycles ranging from 18 to24. The DNA sequences of the prime(Sigma–Genosys) used for amplification arefollows (from 58 to 38 end):HIS 5: GGGTTTCTGGACCATATGATHIS 3: CGAAAGATCTACCACCGCTCLEU 5: TTTGCTAAAGGTACTGACTTLEU 3: CGTTCTTGATGGTTTCCTCCTRP 5: TACGTGATTAAGCACACAAAGGTRP 3: GTTGATTACGAAACACGCCAAURA 5: TCCATGGAGGGCACAGTTAAURA 3: TGTCATAATCAACCAATCGTAMP 5: TGGTGAAAGTAAAAGATGCTGAAMP 3: CCTCCATCCAGTCTATTAATTGADH 5: GTTAGCTCTAACGTATCTGGTAADH 3: GCCAACGAATTGTTGATCAACG

These primer pairs produced amplified DNfragments of the following sizes: 232 nt (HIS3),263 nt (LEU2), 399 nt (URA3), 274 nt (TRP1),549 nt (beta-lactamase), and 699 nt (ADH1).One-tenth of the PCR product was loaded on1.5% agarose gel that was further stained w

ethidium bromide. The ratio of the specificgene(s) band with the episomal b-lactamaseband indicates the relative expression of the

e,

O

tf

ao

p

oiiti

Dec.2

n

ra

e

th

1S

eh

t-byro

P--

-d

a-.a

e

-

e

-ryan

P

y-Po-c-g.

csr-n,

ty.it

96 UGOLINI, TOSAT

gene without the copy number factor. Thvalue, compared with the expression of chromosomal gene ADH1, indicates the level oexpression of the specific gene(s) with respecsingle-copy gene expression. The experimewere repeated at least three times, and negcontrols were also added to exclude DNA ctamination.

Plasmid Constructions

We constructed a series of yeast multicoplasmids carrying theHIS3, LEU2, TRP1,orURA3 gene, on an otherwise identical vectmolecule. To this end, the pBH15 plasm(Bruschi and Howe, 1988) was restricted wHpaI and NruI and religated, and the recombnant clone lacking the wholeLEU2gene (YpA)was identified by restriction analysis. The potion of YpA carrying the 2-mm origin of repli-cation was ligated to theADE8-2mm-FRT por-tion deriving from the pBHD plasmid (Bruschand Howe, 1988). The resulting vector, pBUis similar to pBHD, but it lacks the complet3.3-kbHpaI–NruI LEU2 gene and its genomiflanking sequences (Ugolini, 1995; see Figfor details). This plasmid also carries the twomm FRTs cloned in the direct orientation, aupon activity of the FLP recombinase the plamid undergoes site-specific resolution, geneing an integrative (pIA) and an episom(pRAP) plasmid without any yeast selectabmarker and with a uniqueBamHI cloning site.To recover this episomal form, pBUD was rsolvedin vitro by cutting withXbaI within theFRT site and religating the purified 3.73-kb epsomal fragment. The pRAP construct wasrecipient carrier for cloning in itsBamHI site ofdifferent yeast and bacterial markers (Fig.Thus, the 5.49-kb episomal vector pRAP-HIcarries the 1.76-kbHIS3/BamHI fragment fromplasmid pYAC3 (Burkeet al.,1987). Similarly,the pRK plasmid carries the 1.26-kAPH/BamHI gene from plasmid pUC-4K(Pharmacia, Uppsala, Sweden). This genecodes the enzyme aminoglycoside 3´-phosptransferase conferring kanamycin, neomycand G418 resistance (Taylor and Rose, 198

The pRK plasmid was restricted withSalI, the3.74-kb vector DNA purified and ligated to th

, AND BRUSCHI

ishe

t tontstiven-

y

rdh-

r-

i,

1-

ds-at-lle

-

i-e

).3

b

n-o-

in,8).

2.2-kbLEU2/SalI/XhoI DNA from pBH15. Theresulting 5.93-kb pRAP-LEU2 recombinanclone carrying theLEU2 gene in the same orientation as the unique FRT was identifiedrestriction analysis. The pRAP-LEU2 vectowas used for the construction of the last twepisomal plasmids, pRAP-TRP1 and pRAURA3. TheTRP1gene was obtained as a 1.3kb FspI/SspI blunt-ended fragment from plasmid YIplac204 (Gietz and Sugino, 1988) ancloned into theSmaI site of plasmid pUC19(Yanisch-Perronet al., 1985). A recombinantclone—with the insert in the opposite orienttion to lacZ—was identified and named pUC-TThe TRP1gene was moved from pUC-T as1.42-kb PvuII/SalI fragment and cloned intopRAP-LEU2/HpaI/SalI, generating the 5.41-kbpRAP-TRP1 vector. With a similar strategy thURA3 gene was isolated as a 1.6-kbURA3/SspI/PvuII blunt-ended fragment from plasmidYIplac211 (Gietz and Sugino, 1988). This fragment was inserted into theSmaI cloning site ofpUC19, generating plasmid pUC-U. Thmarker was cloned as a 1.72-kbURA3/PvuII/SalI fragment into pRAP-LEU2/HpaI/SalI,thus resulting in the 5.71-kb pRAP-URA3 vector. These four episomal vectors have a vesimilar size and carry yeast markers that ceasily be isolated by restriction withBamHI. Asummary of the cloning strategy for the pRAfamily of vectors is given in Fig. 1.

Bioinformatic Analysis of the Four DifferentInsert DNA Sequences

The DNA sequences of the four inserts carring the different selectable markers in the pRAplasmid series were analyzed for their thermdynamic propensity to form secondary strutures, in particular hairpins and DNA bendinThe probability of formation of DNA hairpinswas determined by the StemLoop informatiprogram of the GCG Wisconsin package, vesion 10.0 (Genetics Computer Group; MadisoWI) (Devereuxet al., 1984). The predicted in-trinsic curvature propensity and bendabilivalues were calculated with the bendserver http://www2.icgeb.trieste.it/,dna/bend),

eusing the DNase I-based bendability parametersof Brukneret al. (1995) and the related consen-

COMPARATIVE PLASMID MAINTENANCE IN YEAST 97

FIG. 1. Construction of the pRAP family of yeast episomal vectors used in the growth curve, plasmid stability, and com-petition experiments. Maps of the final plasmid constructs are shown with main restriction sites.

O

d

ie a

latees

l

h

8

a

ht at

acms

t

mh,ner--l-,

edyedhen

fforer

dtofd.d

-c

il-3

aor.dofekb

-

e

-

o-p-, in

98 UGOLINI, TOSAT

sus bendability scale (Gabrielian and Pong1996). Individual DNA sequences were submted in raw format and the predicted curvatudata, expressed as degrees/helical turn (10helical turn 5 1°/basepair), were collectethrough e-mail in ASCII text table format.

Growth Curves under Different ExperimentalConditions

CBU1-31 yeast cells were transformed wthe pRAP series of episomal vectors; the cwere plated onto appropriate selective mediaincubated at 30°C for 3 days. Three independtransformant colonies, 1, 2, and 3 (for each pmid type), were picked from the drop-out plaand purified by streaking for single isolatcolonies onto selective plates. Two single ilated colonies for each independent transformwere then used to inoculate two 10-ml paracultures of selective medium (His2, Leu2, Trp2,and Ura2), complete medium (COM), and ric(YPD) medium, at a cell density of 1 3 105

cells/ml in 20-ml glass vials. The resulting cultures (72 transformant cultures 1 12 non-transformed control cultures) were grown inrotary shaker at 30°C up to stationary phase,cell growth was measured as absorbanceOD600 after 9, 15, 21, 32, 38 and 44 h.

Plasmid Stability Experiments

As for the growth curve experiments, tCBU1-31 yeast strain was transformed with different episomal plasmids, and an aliquotthe cells from three independent transformclones (1, 2, and 3) was plated onto selecmedium and incubated at 30°C for 3 days. Sgle isolated colonies were used to inocultwelve 10-ml parallel cultures (three for eatype of marker) into selective (SEL) mediuCells were grown overnight to stationary phaand stability experiments were carried outpreviously described (Ludwig et al., 1993), con-sidering the stability value of these culturesthe starting (G0) value. Briefly, an aliquot fromthese 12 G0 SEL cultures was inoculated innew cultures of selective (SEL), comple(COM), and rich (YPD) media, for a total of 3

parallel cultures. After overnight growth for 1generations to stationary phase, aliquots fro

, AND BRUSCHI

or,it-re.5°/

thllsnd

ents-sdo-antlel

4

and

at

eheofnt

ivein-teh.e,as

as

ote6

the cultures were inoculated into fresh mediuand grown again by another round of growtand the process was repeated for up to 50 geations (G50). At the end of each round of 10 generations of growth, one aliquot from each cuture was plated onto rich YPD mediumincubated at 30°C for 3 days, and replicatonto diagnostic medium for plasmid stabilitdetermination. Plasmid stability was determinas the number of prototrophic colonies over ttotal number of colonies. The remaining portioof each culture was combined with 1 ml of 1 Msodium azide and stored at 280°C, for lateranalysis of plasmid copy number. A total o62,000 single colonies were scored, 15,500 each marker, approximately 2580 per markevery 10 generations.

Plasmid Copy Number

The frozen cells from the end of each rounof growth of the plasmid stability experimenwere thawed, and 3.3-ml aliquots from eachthe three cultures, 1, 2, and 3, were pooleYeast total DNA was prepared from the poolemixture and aliquots of this DNA were restricted overnight with an excess of a specifirestriction enzyme, according to the site avaable on each plasmid type. Plasmid pRAP-HISwas restricted with an excess ofXbaI, whichcuts only once in the plasmid, thus generating5.49-kb band corresponding to the linear vectPlasmids pRAP-LEU2, pRAP-TRP1, anpRAP-URA3 were digested with an excessBamHI, which cuts in all cases at the end of thyeast gene. This restriction generated a 2.2-LEU2 band, a 1.65-kbTRP1band, and a 1.95-kb URA3 band, respectively. We could not restrict pRAP-HIS3 withBamHI because this en-zyme cuts exactly in the same place in thcorresponding chromosomalHIS3 gene, thusgenerating two indistinguishable 1.78-kb fragments. In the chromosomalLEU2 and URA3genes, BamHI digestion created fragmentsmuch longer than those deriving from the epismal plasmids, and the two bands were well searated on the agarose gel. On the other handthe case of the chromosomalTRP1, theBamHI

0mrestriction created a shorter fragment than theplasmidic fragment. Dilutions 1:10 and 1:100 of

M

m

m

gu

c

hoie

ito-arnant(h

e

mfoo

)

us-1

u

as-ereor-

mla

send

, ase

ionsi-one

-e-

quency of appearance of each plasmid type was-

gnu--l-

iveia. asp-hetictasia,D,

p-

yMe-

COMPARATIVE PLASMID

the digestions, in addition to the undiluted saple, were run on a 0.7% agarose gel that wblotted to a nylon Hybond-N1 membrane(Amersham) as described by the suppliers. Hbridization was carried out according to Sabrook et al. (1989) with specific, randomlyprimed [a32P]dCTP-labeled probes that reconize both the wild-type sequence of the focorresponding genes on the plasmid and thmutated allele on the chromosome. In partilar, the same 2.2-kbLEU2, 1.65-kbTRP1, 1.95-kb URA3, and 1.78-kbHIS3/BamHI fragmentsdescribed above were used as probes. The cmosomal band was used as an internal cnumber control for the quantitation of plasmcopy number. Multiple exposures were devoped to measure the intensity of each bawithin the linear range of sensitivity of the filmfor at least one exposure (not shown). Densmetric quantification of the autoradiographs wdone with GelScan XL evaluation softwa(version 2.1) in conjunction with the UltroScaXL laser densitometer (both from PharmaciThe value of the area of each hybridization bawas determined, and the ratio of the area ofplasmid band(s) to the chromosomal bandwas assumed to reflect the copy number of tplasmid in the haploid cell population. Sincsome or most of the cells have lost their plamid, the ratio plasmid/chromosome band undestimates the real copy number of the plasmspecies in the cells carrying them. To overcothis effect, the copy number was normalizedthe stability of the plasmid, dividing this ratiby the percentage of plasmid stability.

Competition Experiments in NonselectiveComplete and Rich Media (COM and YPD

Yeast strain CBU1-31 was transformed indpendently with all four plasmid species, thobtaining four different transformant cloneCBU1-31 (pRAP-HIS3), CBU1-31 (pRAPLEU2), CBU1-31 (pRAP-TRP1), and CBU1-3(pRAP-URA3). Three independent transfomant clones from each transformation were fther purified and then grown to stationary pha

in 5 ml of the proper liquid selective mediumFinally, three cocultures, 1, 2, and 3, were prpared by combining equal volumes of the fou

AINTENANCE IN YEAST 99

-as

y--

-r

eiru-

ro-pydl-nd

stationary phase cultures with the different plmids. In all cases, these three cocultures wprepared with purified independent transfmants. The three initial (G0) combinations wereused to inoculated three parallel culture in 5of complete (COM) and rich media (YPD) atcell density of 13 105 cells/ml (2.5 3 104

cells/ml for each type of transformant). Thecocultures were grown for 10 generations, anew cultures were restarted by serial inoculafor the plasmid stability experiments. Thprocess was repeated up to 50 generat(G50). At the end of each round of growth a dlution of the cell suspensions was platedYPD medium. After 3 days of growth thesplates were replicated onto His2, Leu2, Trp2,and Ura2 plates, and the number of prototrophic colonies was determined. The fr

ter

se

).d

hes)at

es-r-ider

e-s,

r-r-se.

determined, by setting to 100% the total number of prototrophs.

RESULTS

Growth in Selective and Nonselective Media

The series of 2-mm-based plasmids carryingenes for the synthesis of amino acids or a cleotide were individually introduced in the recipient yeast strain CBU1-31 and parallel cutures were grown for up to 30 h in both selectand nonselective defined and rich medGrowth was measured at 5-h time intervalsabsorbance at OD600and the data showed no apreciable difference in the growth rate of ttransformed cultures in both selective, synthecomplete (COM) and rich (YPD) media (noshown). The optical density reached in YPD whigher than that attained in the other two medwith a value of more than 2.5 OD versus an Oof 2.0. Accordingly, cell density in this mediumapproximately 1.3 3 108 cells/ml, was twofoldhigher than in selective and COM media (aproximately 6.0 3 107 cells/ml), a phenomenonknown in literature. In YPD, the stationarphase was reached much later than in COmedium, probably because of the lack of undfined nutrients, like trace elements, in the lat

e-r(Werner-Washburne et al., 1993). To testwhether the concentration of the three amino

O

loeth srohuan-e

e

tasee

u

a

a

e

ie

ite oa)

ne

eas-lts

ereenda

nds-hedSer

esll; it.yr

s-rd

tsul-di-asstss

llngre-

-elyndndfarthedeic

100 UGOLINI, TOSAT

acids and uracil in the COM medium could fulcompensate for the auxotrophic mutations csidered, we grew the various strains in the prence of double and triple concentrations of supplements. The cell biomass, measuredcell density and dry weight, showed no increaproving that the standard concentrations ported under Materials and Methods are nlimiting. It is interesting to note that in botcomplete and rich media the transformed ctures showed a 2–3% lower growth profile ththe untransformed controls, hence confirmithat plasmid maintenance is a sizable energypensive task. This phenomenon has been scribed also for the cryptic 2-mm circle, whosemaintenance imposes upon to the host cell alective disadvantage of about 1% in terms growth rate, when compared to plasmid-frcells (Futcher and Cox, 1983; Walmsley et al.,1983; Mead et al., 1986).

Plasmid Stability Experiments

Plasmid maintenance was monitored over course of 50 generations, under selective nonselective conditions. Cells harboring theplasmids showed a different level of maintnance when kept under selection (Fig. 2) whthe plasmid carrying the LEU2gene was clearlymaintained at the lowest level of all in the poplation, on average 51.3 6 4.2%, while that withthe URA3gene was kept at the highest level,average of 67.4 6 3.5%. Vectors with HIS3andTRP1genes were maintained at an intermedilevel, on average 59.2 6 3.0% and 62.9 6 3.9%,respectively. When cells were grown without slective pressure in synthetic medium (COMplasmid loss was lower for the genes encodtryptophan and uracil biosynthetic enzym(11.5 and 4.6% maintenance after 50 genetions) and higher for the other two markers, htidine and leucine (2.2 and 1.7% of mainnance), as shown in Fig. 2. A similar patternplasmid loss after 50 generation of growth wobserved for cells growing in rich (YPDmedium, again with TRP1 and URA3 genesbeing maintained at a higher level (14.9 a8.8%, respectively) than the other two gen

HIS3 (1.8%) and LEU2 (2.5%). Since plasmidcopy number might play a major role in the se

, AND BRUSCHI

yn-s-ebye,e-n-

l-ngx-

de-

se-ofe

hende-re

-

n

te

-)ngsra-s--fs

ds,

lective impact of these different plasmids, wdetermined the exact copy number of each plmid in all the experiments described. The resuare reported in Fig. 2. As can be seen, thseems to be a differential correlation betweplasmid stability under selective conditions ancopy number. In the case of SEL medium,high stability of plasmids in the cell populatiocorresponds to a low copy number of plasmiper cell. In COM medium, overall stability decreases progressively while copy number of tvarious plasmids diversifies. pRAP-LEU anpRAP-TRP remain consistently low, pRAP-HIshows a nearly constant level threefold highthan in SEL medium, and pRAP-URA increassteadily to more than 80 copies per cenevetheless, its stability does not benefit fromFinally, the situation in YPD shows a tendencof copy number to increase up to twofold fopRAP-HIS, pRAP-TRP, and pRAP-URA, whileremaining consistently low for pRAP-LEU.

Competition Experiments in NonselectiveComplete (COM) and Rich (YPD) Media

To confirm the results obtained with the plamid stability experiments—namely, the poostability of the pRAP-LEU2 plasmid compareto that of pRAP-URA3, pRAP-TRP1, andpRAP-HIS3—we set up coculture experimenin which all these plasmids were present simtaneously under the same environmental contions. The expectation for this experiment wthat, given an appropriate lag time, the yeastrain harboring the most stable, and thus lecostly, plasmid form would compete out aother less stable, more costly plasmid-carryistrains. Results from these experiments are ported in Fig. 3.

At the beginning of the experiment all plasmid species are represented in approximatthe same proportions. Right after the first rouof growth, the relative proportions changed, athe pRAP-LEU2 species appears to drop below the level of the other constructs, in bocomplete and rich media. This decline persistfor the remaining part of the growth, with thplasmid completely disappearing in synthet

-complete (COM) medium, while reaching thelowest level (7.3%) in rich (YPD) medium. The

COMPARATIVE PLASMID MAINTENANCE IN YEAST 101

FIG. 2. Stability (left) and related copy number (right) of different episomal plasmids in the yeast strain CBU1-31 moni-tored during the course of 50 generations of growth, under selective (SEL) and nonselective (COM and YPD) conditions. Forplasmid copy number, the first determination in COM and YPD media was performed after the first round of growth for 10generations (G10), since the initial inoculum was made only in SEL medium. This ensures having the largest possible initialnumber of cells harboring the plasmid.

ie

rinti

Peeni

ophs

exl

Diu

pt sr

mr u

h

fte

hee-s.ndisn-asteta-inp-ionsst,

-s.

o-l-e

a

alre,

FIG. 3. Relative percentage of plasmid species after coculture experiments in complete (COM) medium and in rich (YPD)g 1 1 1 1

pRAP-HIS3 plasmid is also competed out complete medium, though at a much slowrate, dropping to 3.5% at the end of the segrowth. Nevertheless, in rich medium this costruct proves to be more stable, still represen17.0% of the total population of prototrophs.

The other two plasmid species, pRAP-TRand pRAP-URA3, are very stable in both typof media, never dropping below the input levIn fact, plasmid pRAP-TRP1 remains arou25% in complete medium and, interestingly,rich medium represents the majority of the pulation, attaining 48.6% of all prototrophs at tend of the experiment. The pRAP-URA3 plamid shows the most striking behavior, being tremely stable in complete medium, largecompeting out all other forms. In rich YPmedium this species is also persistent, berepresented at the same level of the initial cture, but it does not seem capable of outcoming the other plasmid species. It is importanstress the fact that these values do not repreabsolute stability values, which instead are ported in Fig. 2. In fact, to score for the sanumber of prototrophs, an increasing numbecells had to be plated at each successive roof growth due to the continuous dilution of tplasmid-bearing cells (data not shown).

The above results clearly indicate that a

medium. Graphic representation of the relative percentalated on the total number of prototrophs (100%).

the coculture experiments, in both complete anrich media, the CBU1-31 strain carrying th

nr

al-

ng

1sl.dn-

e--

y

ngl-

et-toente-eofnd

e

r

pRAP-LEU2 plasmid is underrepresented in tpopulation, hence strongly confirming the rsults obtained with the stability experimentThe high permanence of the pRAP-TRP1 apRAP-URA3 plasmids in long-term cultures also a valid confirmation that these two costructs are preferentially maintained in the yestrain CBU1-31. It is quite interesting to notthat, as previously observed for the plasmid sbility experiments, there is a clear difference kinetics in the two nonselective media. It apears, in fact, that while the release of selectin COM medium results in the progressive loof the plasmid, in rich YPD medium it does noat least not with the same rapidity.

Analysis of Gene Expression of the DifferentMarkers

RT-PCR was performed on total RNA isolated from the four yeast transformant strainThe ADH1 band (Figs. 4f–4i) typically repre-sents the basal level of expression of a chromsomal gene in the wild-type strain CBU1-31, alowing the evaluation of the expression of thchromosomal copies of the LEU2, TRP1, URA3,and HIS3genes. The intensity of their bands in24-cycle amplification, relative to ADH1, showsthat the expression level of these chromosomgenes is identical (data not shown). Therefo

e of the cells with phenotypes His, Leu , Trp , and Ura calcu-

102 UGOLINI, TOSATO, AND BRUSCHI

dethe ratio between the band intensity of the spe-cific gene and the b-lactamase gene carried on

npu

th

–

ya

fere toactan

tionle-

re-thepti-

each yeast gene (G) and one for the b-lactamase control episomal gene (Bla). In each sample the intensity of the band for theo srn le.

the same plasmid represents the expressiothe episomal gene without the bias of its conumber factor. When the amplification was rfor 24 cycles the LEU2and TRP1genes showedgood levels of expression if compared with episomal b-lactamase and the chromosomADH1 genes in the same sample (Figs. 4aand 4f–4i, respectively). The URA3gene showsthe lowest level of expression, indicating a hpothetical correlation with the strong second

specific gene (bottom) is compared with that of the contrused simultaneously, generating the previous band patte

COMPARATIVE PLASMID MAINTENANCE IN YEAST 103

FIG. 4. Quantitative RT-PCR for the expression of the genes LEU1, TRP1, URA3, and HIS3 (in order from left to right).Lanese and j , 100-bp ladder marker (Gibco). Lanes a, b, c, and d, amplification with two sets of primers, one specific for

structures which characterize its DNA codino

awuive

tor-e-

ta-theidllhete--

rst

sequence (see Discussion) and an inverse clation with its high stability and copy number.

DISCUSSION

Maintenance of Vectors Carrying DifferentSelectable Markers

Analysis of the growth kinetics indicates ththere are no significant differences in the grorate of the transformed cultures with the fodifferent plasmid types, neither when selectpressure is applied (SEL) nor in the absenc

selection (COM and YPD) (data not shownThus, the presence, replication, and express

ofyn

eal4d

-rygrre-

tthre of

of any of the HIS3, LEU2, TRP1, or URA3genes in a multicopy plasmid seem not to cona particular growth advantage or disadvantagthe host cell under selective conditions. The fthat the untransformed cultures grow faster ththe transformed cultures strengthens the nothat the maintenance of extrachromosomal ements inside the cell per se is a costly eventquiring more energy than the maintenance of genome alone, hence, the importance of omizing the construction of artificial plasmids,avoid titration of cellular resources and interfeence with the normal host growth rate and mtabolism (Brownlie et al., 1990; Summers,1991; Bergstrom et al., 2000). Significant differ-ences are observed for the level of plasmid sbility when these different genes are used in same genetic background. In particular, plasmpRAP-LEU2 is lost at the highest rate in atypes of media. Under selective conditions, tdifference between the percentage of mainnance of pRAP-URA3 and pRAP-TRP1 compared to pRAP-LEU2—the best and the wo

l (top). Lanes f, g, h, and i, amplification with three sets of primer plus the ADH1endogenous control gene (ADH1) for each samp

).ionperformers—is quite consistent, reaching valuesof approximately 16 and 12%, respectively. This

O

ma

atdUh

be

eni-

re

oen

n ihhel

b

in

th

ovn

v

ed

howlo-ac-ll,e,r

ws:

lyis

la-,he6sb-fes

ntac-

rix-est-ti-leInly

i-es

g,fores

ri-r

ithe

104 UGOLINI, TOSAT

difference is also evident in complete mediuwith pRAP-URA3 being roughly 10% more stble than pRAP-LEU2; the best performepRAP-TRP1, is even more stable, by an avervalue of 14%. Such a situation holds for the sbility in rich medium, where pRAP-URA3 anpRAP-TRP1 are more stable than pRAP-LEby about 12 and 14%, respectively. Clearly, tLEU2 gene seems not to be the best candidfor the construction of episomal plasmids to stably maintained in the host cell and this expimental observation is corroborated by our labratory experience. The pRAP-HIS3 vector bhaves quite well only under selective conditioshowing a stability level of roughly 59%, simlar to that of pRAP-TRP1. However, in complete and rich medium, its rate of loss corsponds to that of the pRAP-LEU2 plasmid.

These results are reinforced by the data tained with the competition experiments undnonselective conditions, clearly demonstratithat the plasmid carrying the LEU2 gene isquickly competed out of the population. Agaithe URA3and TRP1genes are the most stablecomplete medium and perform well in ricmedium (Fig. 3). It is interesting to note that tcomplete supplement (COM) appears to bmedium more “stressing” than YPD, being abto enhance subtle differences in the stability tween plasmids carrying various genes. In fathe HIS3 gene is more stable in COM than YPD, possibly due to the fact that complete suplement is missing some still undefined growpromoting substances, like trace elemenwhich are available in YPD medium (Rose etal., 1990; Difco, 1994) and are fundamental foptimal growth rates. This phenomenon is edenced also by the growth curve experime(not shown) and the same holds true for conumber, the differences in which are more edent in COM than in YPD medium.

Involvement of Codon Usage and Transcriptio

A clear positive correlation between the dgree of codon bias, expressed as the coadaptation index value (CAI, Sharp and L1987) and/or codon bias index value (CBI, Be

netzen and Hall, 1982), and the level of gene epression has been well established, showing t

, AND BRUSCHI

,-r,gea-

2e

ateer-

o--

s,

--

b-rg

,n

e aee-ct,

p--

ts,

ri-ts

pyi-

n

-oni,n-

genes which are transcribed at a high rate sa strong bias for those codons highly homogous to the anticodons of the major yeast isoceptor tRNA species (Bennetzen and Ha1982; Sharp and Li, 1986; Sharp and Cow1991). In this view, the CAI and CBI values fothe genes examined in this study are as folloCAI CBI GC content GC content

Of ORF (%) Of insert (%)

HIS3 0.09 0.01 43.63 43.7LEU2 0.44 0.60 38.95 38.8TRP1 0.12 0.05 40.40 45.8URA3 0.18 0.21 39.79 43.0

For comparison, the CAI value of the highexpressed yeast ribosomal protein genes0.529–0.915, while the weakly expressed regutory geneGAL4 has a CAI value of 0.116, andinterestingly, the CAI values for the genes on tvery stable 2-mm plasmid are 0.099–0.10(Sharp and Li, 1987). The CAI and CBI valueof our genes of interest correlate with their oserved low level of transcription, typical oamino acid- or nucleotide-biosynthesis enzym(HIS3, Struhl and Davis 1980;LEU2, Andreadiset al., 1984; TRP1, Kim et al., 1986; URA3,Rose and Botstein, 1983). Interestingly, differeauthors have documented the association oftively transcribed genes with the nuclear mat(Robinsonet al., 1983; Hutchinson and Weintraub, 1985; Jackson and Cook, 1985), sugging that transcription may lead to unequal partioning of a gene located on a plasmid molecudue to its association with the mother nucleus.this regard, it is relevant to note that a weaktranscribedTRP1-ARS1plasmid can be destablized by the insertion of pBR322 sequenc(Stinchcombet al., 1979), which in yeast areheavily transcribed (Marczynski and Jaehnin1985). A similar effect has also been reported2-mm derivatives containing pBR322 sequenc(Dobsonet al., 1980; Toh-eet al., 1980; Futcherand Cox, 1984). In the context of our expements, the lower CAI values, indicative of lowelevels of transcription, appear to correlate wthe higher plasmid stability values. In fact, thlower the value of the CAI of the gene (URA3

x-

hatandTRP1), the higher the stability of the plas-mid carrying that same gene (pRAP-URA3 and

M

iv

oy

lieNishn

-neca

l

ne

nthnab

it

lic ft s

P

rip-o sicar-ent, ofng

othe-F,rly

thislatee.ndceur

antemrin-nlyd a

ly,l

r-

inse-rti-on

rted

B.,fene

n in

COMPARATIVE PLASMID

pRAP-TRP1), in both selective and nonselectmedia. The highest CAI value is that of thLEU2 gene, which is always lost at the higherate of all. However, theHIS3gene seems not tbehave in accordance with the CAI value hpothesis, being poorly maintained in the stabiexperiments, although its highest GC contmay represent an expensive burden of its Dreplication. It is important to mention that thgene was cloned in the pRAP vector togetwith its genomic flanking sequences, includipart of thePET56 to the left (58) and the un-known ORFYOR303wand theDED1 genes tothe right (38), both including their promoters firing in divergent orientation. Improper initiatioof transcription from these two promoter squences into the neighboring plasmid sequen(Ugolini, 1995) may cause instability in the finconstruct pRAP-HIS3, leading to increased loof the plasmid in the cell population, particularwhen selective pressure is released.

It is possible that, when the pool of amiacids and overall protein synthesis is reduccodon usage becomes less important sincehigh rate of protein synthesis associated whigh growth rate is not needed for survival, aalso more time is available for selection of correct charged tRNA. Thus, the use of gewhich are generally expressed at lower rate with higher versatility for codon usage may favored when the pool of charged tRNA is dcreasing. In addition, the difference in stabilobserved under nonselective conditions (COand YPD) may simply reflect the difference acumulated during the phase of culture ampcation in selective medium, a condition in whithere was coupling between maintenance ofmarker and its transcription and translation the production of the necessary gene producfact, as indicated above, plasmids are lost atsame relative rate to each other in both typemedia, as becomes evident by comparing difference in the level of plasmid maintenanbetween plasmids pRAP-URA3 and/or pRATRP1 with pRAP-LEU2.

Possible Role of Secondary DNA Structure

Recently, the presence of secondary DNstructures in the genomic DNA of different or

AINTENANCE IN YEAST 105

eest

-tyntA

erg

-eslssy

od,theithdeesnde

e-yM

ganisms has been associated with low transction rates and genetic recombination (Tosatetal., 2001). In the present work, some intrinproperty of the DNA sequence of the insert crying these genes, such as the GC contseems to correlate with their different levelsstability, since the insert fragment containithe most stable marker (URA3) has a GC valuehigher than the least stable marker (LEU2). Onthe other hand, analysis of DNA bending (nshown) indicates that the most rigid DNA is tinsert containing the TRP1gene, with the exception of a strong curvature within its ORwhile the other three genes show a similamoderate propensity to bending. Therefore,structural feature does not seem to correwith the efficiency of plasmid maintenancOther DNA secondary structures, like stem aloop hairpins, may contribute to the differenin stability. In this regard, the results of obioinformatic analysis indicate that the URA3insert has more potential to form hairpins ththe others. Using the parameters of 20 nt in slength, and 40 hydrogen bonds per stem (stgency), a potential stem formation is found oin this gene. With a stem length of 15 nt, anstringency of 30,URA3 and LEU2may formfour and two significant hairpins, respectivewhile TRP1 and HIS3 revealed no potentiahairpin. Given the higher affinity of DNA haipins for histone proteins (Wang et al., 1994),one hypothesis regarding their involvementplasmid segregation could be that DNA

c-fi-hthe

quences with such structures would be pationed more or less efficiently depending their binding to the nuclear matrix.

or. Inthe ofthece

-

ACKNOWLEDGEMENTS.U. was a ICGEB-ISAS predoctoral student suppo

by ICGEB.

REFERENCESAndreadis, A., Hsu, Y. P., Hermodson, M., Kohlhaw, G.

and Schimmel, P. (1984) Yeast LEU2. Repression oRNA levels by leucine and primary structure of the gproduct. J. Biol. Chem.259,8059–8062.

Bennetzen, J. L., and Hall, B. D. (1982) Codon selectioyeast. J. Biol. Chem.257,3026–3031.

A-

Bergstrom, C. T., Lipsitch, M., and Levin, B. R. (2000) Nat-ural selection, infectious transfer and the existence condi-tions for bacterial plasmids.Genetics15,51505–51519.

O

le

E

c

c

e

n

s

f-

-

s

in

e

t

p.ld

Aus

of in-

oftive

urs

no

mandng

.

-

ip-ted

heh

s:

ry

dllyn

2)

ofne-

inldY.e--

106 UGOLINI, TOSAT

Boyer, H. W., and Roulland-Dussoix, D. (1969) A compmentation analysis of the restriction and modificationDNA in Escherichia coli. J. Mol. Biol.41,459–472.

Brownlie, L., Stephenson, J. R., and Cole, J. A. (1990) fect of growth rate on plasmid maintenance by Es-cherichia coliHB101(pAT153). J. Gen. Microbiol.136,2471–2480.

Brukner, I., Sanchez, R., Suck, D., and Pongor S. (1995)quence dependent bending propensity of DNA as vealed by DNase I: Parameters for trinucleotides. EMBOJ. 14,1812–1818.

Bruschi, C. V., Comer, A. R., and Howe, G. A. (1987) Speficity of DNA uptake during whole cell transformation oS. cerevisiae. Yeast3, 131–137.

Bruschi, C. V., and Howe, G. A. (1988) High frequenFLP-independent homologous recombination of 2m plas-mid in the yeast Saccharomyces cerevisiae. Curr. Gen14,191–199.

Burke, D. T., Carle, G. F., and Olson, M. W. (1987) Cloniof large segments of exogenous DNA into yeast means of artificial chromosome vectors. Science236,806–812.

Caulcott, C. A. (1984) Competition between plasmid-potive and plasmid-negative cells. Biochem. Soc. Trans.12,1140–1142.

Devereux, J., Haeberli, P., and Smithies, O. (1984) A coprehensive set of sequence analysis programs forVAX. Nucleic Acids Res.12,387–395.

Difco (1994) “Dehydrated Culture Media and Reagents Microbiology,” Difco Manual, 10th ed. Difco Laboratories, Detroit, MI.

Dobson, M. J., Futcher, A. B., and Cox, B. S. (1980) Loss2 mm DNA from Saccharomyces cerevisiaetransformedwith the chimaeric plasmid pJDB219. Curr. Genet.2,201–205.

Futcher, A. B., and Cox, B. S. (1983) Maintenance of 2mmcircle plasmid in populations of Saccharomyces cerevisiae. J. Bacteriol.154,612–622.

Futcher, A. B., and Cox, B. S. (1984) Copy number and bility of 2-mm circle-based artificial plasmids of Saccha-romyces cerevisiae. J. Bacteriol.157,283–290.

Gabrielian, A., and Pongor, S. (1996) Correlation of intrsic DNA curvature with DNA property periodicity. FEBSLett.393,65–68.

Gietz, R. D., and Sugino, A. (1988) New yeast–Escherichiacoli shuttle vectors constructed with in vitro mutagenizyeast genes lacking six-base pair restriction sites. Gene74,527–534.

Godwin, D., and Slater, J. H. (1979) The influence of growth environment on the stability of a drug resistanplasmid in Escherichia coliK12. J. Gen. Microbiol.111,201–210.

Hanahan, D. (1983) Studies on transformation of Es-cherichia coliwith plasmids. J. Mol. Biol.166,557–580.

Hinnebusch, A. G. (1992) General and pathway-specregulatory mechanisms controlling the synthesis

amino acids biosynthetic enzymes in Saccharomycescerevisiae. In“The Molecular and Cellular Biology ofthe Yeast Saccharomyces: Gene Expression” (E.

, AND BRUSCHI

-of

f-

Se-re-

i-f

y

t.

gby

i-

m-the

or

of

ta-

-

d

hece

ificof

Jones, J. R. Pringle, and J. R. Broach, Eds.), Vol. II, p319–414. Cold Spring Harbor Laboratory Press, CoSpring Harbor, NY.

Hoffman, C. S., and Winston, F. (1987) A ten minute DNpreparation from yeast efficiently releases autonomoplasmids for transformation of Escherichia coli. Gene57,267–272.

Hutchinson, N., and Weintraub, H. (1985) Localization DNase I-sensitive sequences to specific to regions ofterphase nuclei. Cell 43,471–482.

Ikemura, T. (1982) Correlation between the abundanceyeast transfer RNAs and the occurrence of the respeccodons in protein genes. J. Mol. Biol.158,573–597.

Jackson, D. A., and Cook, P. R. (1985) Transcription occat a nucleoskeleton. EMBO J.4, 919–925.

Jones, E. W., and Fink, G. R. (1982) Regulation of amiacid and nucleotide biosynthesis in yeast. In “The molec-ular Biology of the Yeast Saccharomyces: Metabolisand Gene Expression” (J. N. Strathern, E. W., Jones,J. R. Broach, Eds.) Volume II, pp. 181–299. Cold SpriHarbor Laboratory Press, Cold Spring Harbor, NY.

Kim, S., Mellor, J., Kingsman, A. J., and Kingsman, S. M(1986) Multiple control elements in the TRP1promoterof Saccharomyces cerevisiae. Mol. Cell. Biol.6,4251–4258.

Ludwig, D. L., Ugolini, S., and Bruschi, C. V. (1993) Highlevel heterologous gene expression in Saccharomycescerevisiaefrom a stable 2mm plasmid system. Gene132,33–40.

Marczynski, G. T., and Jaehning, J. A. (1985) A transcrtion map of a yeast centromere plasmid: Unexpectranscripts and altered gene expression. Nucleic AcidsRes.13,8487–8506.

Mead, D. J., Gardner, D. C. J., and Oliver, S. G. (1986) Tyeast 2m plasmid: Strategies for the survival of a selfisDNA. Mol. Gen. Genet.205,417–421.

Miller, J. H. (1992) “A Short Course in Bacterial GeneticA Laboratory Manual and Handbook for Escherichia coliand Related Bacteria.” Cold Spring Harbor LaboratoPress, Cold Spring Harbor, NY.

Robinson, S. I., Small, D., Idzerda, R., McKnight, G. S., anVogelstein, B. (1983) The association of transcriptionaactive genes with the nuclear matrix of the chickeoviduct. Nucleic Acids Res.11,5113–5130.

Romanos, M. A., Scorer, C. A., and Clare, J. J. (199Foreign gene expression in yeast: A review. Yeast8,423–488.

Rose, M., and Botstein, D. (1983) Structure and functionthe yeast URA3gene. Differentially regulated expressioof hybrid b-galactosidase from overlapping coding squences in yeast. J. Mol. Biol.170,883–904.

Rose, M. D., Winston, F., and Hieter, P. (1990) “MethodsYeast Genetics: A Laboratory Course Manual.” CoSpring Harbor Laboratory Press, Cold Spring Harbor, N

Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) “Molcular Cloning: A Laboratory Manual.” Cold Spring Har

W.

bor Laboratory Press, Cold Spring Harbor, NY.Sharp, P. M., and Cowe, E. (1991) Synonymous codon

usage in Saccharomyces cerevisiae. Yeast7, 657–678.

COMPARATIVE PLASMID MAINTENANCE IN YEAST 107

Sharp, P. M., and Li, W.-H. (1986) An evolutionary perspec-tive on synonymous codon usage in unicellular organ-

ioo

op

n

ss

un

A

.A

s

five Leishmania majorFriedlin chromosomes. Curr.Genet.40,186–194.

a-

er-

)tat

- ate.

nd

-s:c-

isms. J. Mol. Evol.24,28–38.Sharp, P. M., and Li, W.-H. (1987) The codon adaptat

index—A measure of directional synonymous codusage bias, and its potential applications. Nucleic AcidsRes.15,1281–1295.

Stinchcomb, D. T., Struhl, K., and Davis, R. W. (1979) Islation and characterization of a yeast chromosomal recator. Nature282,39–43.

Struhl, K., and Davis, R. W. (1980) A physical, genetic, atranscriptional map of the yeast his3 gene of Saccha-romyces cerevisiae. J. Mol. Biol.136,309–332.

Summers, D. K. (1991) The kinetics of plasmid loTibTech.9, 273–278.

Taylor, L. A., and Rose, R. E. (1988) A correction in the ncleotide sequence of the Tn903 kanamycin resistadeterminant in pUC-4K.Nucleic Acids Res.16, 358.

Toh-E, A., Guerry-Kopecko, P., and Wickner, R. (1980)stable plasmid carrying the yeast LEU2 gene and con-taining only yeast deoxyribonucleic acid. J. Bacteriol.141,413–416.

Tosato, V., Ciarloni, L., Ivens, A. C., Rajandream, M. ABarrell, B. G., and Bruschi, C. V. (2001) Secondary DNstructure analysis of the coding strand switch region

nn

-li-

d

.

-ce

,

of

Ugolini, S. (1995) Plasmid DNA maintenance as an indiction of metabolic stress in Saccharomyces cerevisiae: Fit-ness of different selectable markers. Ph.D. thesis. Intnational School for Advanced Studies, Trieste, Italy.

Walmsley, R. M., Gardner, D. C., and Oliver, S. G. (1983Stability of a cloned gene in yeast grown in chemosculture. Mol. Gen. Genet.194,361–365.

Wang, Y. H., Amirhaeri, S., Kang, S., Wells, R. D., and Griffith, J. D. (1994) Preferential nucleosome assemblyDNA triplet repeats from the myotonic dystrophy genScience265,669–671.

Werner-Washburne, M., Braun, E., Johnston, G. C., aSinger, R. A. (1993) Stationary phase in the yeast Sac-charomyces cerevisiae. Microbiol. Rev.57,383–401.

Yanisch-Perron, C., Vieira, J., and Messing, J. (1985) Improved M13 phage cloning vectors and host strainNucleotide sequences of the M13mp18 and pUC19 vetors.Gene33,103–119.

Zhou, C., Yujun, Y., and Jong, A. Y. (1990) Mini-prep in tenminutes.BioTechniques8, 172–173.

Communicated by P. Farabaugh