RSC2, Encoding a Component of the RSC Nucleosome Remodeling Complex, Is Essential for 2{micro}m Plasmid Maintenance in Saccharomyces cerevisiae

The stable maintenance of the 2µm circle plasmid dependson its ability to overcome intrinsic maternal inheritance bias,which in yeast normally results in the failure to transmit DNAmolecules efficiently to daughter cells. In addition to theplasmid proteins Rep1 and Rep2 acting on the plasmid DNA locusSTB, it is likely that other chromosomally encoded yeast proteinsare required. We have isolated mutants of yeast unable to maintain2µm and found that RSC2 is essential for 2µm toovercome maternal inheritance bias. Rsc2 is part of a multisubunitRSC chromatin remodeling complex, and we show that in the absenceof Rsc2 the chromatin structure of the STB region is significantlyaltered and the Rep1 protein loses its normal localization tosubnuclear foci. Rsc1, a closely related homolog of Rsc2 presentin an alternative form of the RSC complex, is not required for2µm maintenance and does not replace the requirement forRsc2 when overexpressed. This represents the first specificrole for Rsc2 that has been related to a change in chromatinstructure, as well as the first direct evidence linking chromatinstructure to 2µm segregation.

INTRODUCTION

Top

Introduction

The 2µm circle is a high-copy-number plasmid widespreadin laboratory and industrial strains of Saccharomyces cerevisiaethat has been used as a molecular model for aspects of cellularfunction such as DNA replication, gene transcription, and recombination,as well as the basis of yeast cloning vectors. High-copy-numberplasmids carrying only replication origins show maternal inheritancebias (MIB) and are poorly segregated to daughter cells (31),whereas 2µm and its derivatives are effectively segregatedbetween mother and daughter cells at division (22).

However, the mechanism by which 2µm escapes MIB resultingin equal segregation is not fully understood, nor have all ofthe host components involved been identified. Of the four genescarried by the plasmid, REP1 and REP2 are necessary for MIBto be overcome, and their products act at the plasmid locusSTB to confer plasmid stability (2, 10, 18, 22, 33, 36). Withinthe nucleus, Rep1 and Rep2 proteins are localized into discretefoci (39) in which the plasmid is also colocalized (39, 52).These plasmid and protein foci are segregated in a manner similarto that of the chromosomes themselves, a finding consistentwith the observation that the 2µm plasmid is packagedinto nucleosomes (50). Thus, the plasmid may be exploiting thechromosomal segregation machinery indirectly by associatingwith specific or nonspecific chromosomal sites. It has beentherefore suggested that the segregation of 2µm, and henceits overcoming of MIB, likely requires host (non-plasmid-encoded)genes (18). One candidate is the product of the SHF1 (CST6)gene, which has been shown to bind to STB in vivo and in vitro(51).

Nucleosomes regulate and modify the ability of proteins to interactwith DNA, and the remodeling of chromatin and repositioningof nucleosomes is important in gene regulation and other functionsof DNA. There is evidence that specific chromatin structuresare important in 2µm functions, since the STB region hasbeen reported to be relatively free of nucleosomes (15, 25),and changes occur in nucleosome positioning in the presenceof 2µm gene products (50). One class of chromatin remodelingcomplex is typified by the SWI/SNF and RSC complexes, whichcontain 8 to 15 subunits (6). SWI/SNF is required for the regulatedexpression of about 200 yeast genes (35, 44) but is nonessentialand of relatively low abundance.

Components of the RSC complex were first isolated as homologsof the SWI/SNF chromatin remodeling complex but, unlike thoseof the SWI/SNF complex, most of the components of RSC are essentialfor growth (7, 9, 24, 47). The complex is 10-fold more abundantthan the SWI/SNF complex (7) and comprises at least 15 subunits.Sth1 is an essential ATPase in RSC that is required for alterationof the chromatin structure at the centromeres (13, 48), repressionof the CHA1 gene (30), and maximal expression of early meioticgenes (54). In addition, both Sth1 and another component, Sfh1,are required for mitotic cell cycle progression (9, 24, 49).

At least two forms of the RSC complex have been identified thathave both shared core components and unique subunits. Both Sth1and Sfh1 are core components, but two proteins found in separatecomplexes are Rsc1 and Rsc2 (8). Rsc1 and Rsc2 are 45% identicaland share the same domain structure, and both contain two bromodomains.Although the rsc1 ${Delta}$ rsc2 ${Delta}$ double null mutant is nonviable, thesingle nulls have relatively mild growth phenotypes, suggestingthat there is a high level of redundancy between the two complexesas well as potential unique roles (8).

We have used a genetic screen to isolate mutants deficient inthe maintenance of the 2µm plasmid. We show that, surprisingly,five of six mutants isolated have mutations in RSC2. Loss ofRSC2 function results in a failure to effectively overcome MIB,and hence cells are defective in 2µm plasmid maintenance.This plasmid instability correlates with significant changesin the chromatin structure at the STB locus and a loss of thenormal localization of the Rep proteins. In contrast, loss ofthe RSC2 homolog RSC1 does not affect 2µm plasmid maintenance.This identifies the 2µm plasmid as uniquely requiringRsc2 and provides the first evidence for a specific moleculartarget for the Rsc2-containing complex.

MATERIALS AND METHODS

Top

Materials and Methods

Strains, growth procedures, and mutagenesis. S. cerevisiae GY59 MAT ${alpha}$ ade1 his3 leu2 trp1 ura3 [cir⁰] and GY18MAT ${alpha}$ ade1-101 leu2-3,112 his3-11,15 [cir⁺] (derived from LL20[16]), A363A MATa ade1 ade2 ura1 his7 lys2 tyr1 gal1 [cir⁺](19), W303 MATa ade2 his3 leu2 trp1 ura3 can1-100 (45), andmutant derivatives were grown as previously described at 30°Cunless otherwise stated (39). We found only a marginal growthdisadvantage for rsc2 mutants at this temperature. YPD medium(1% yeast extract, 2% Bacto Peptone, and 2% D-glucose) was supplementedwith Geneticin G418 (Gibco) at 200 µg/ml for selectionof KanMX gene disruptants (1). YPGal has glucose replaced by2% D-galactose. GY59 carrying plasmid 2µm-ADE1 was mutagenizedwith ethyl methanesulfonate as described previously (46). Mutantswere screened for plasmid loss by the qualitative colony sectoringassay described below.

Plasmids. Plasmid pZ10 carries one complete copy of the 2µm genome,plus a duplication of one of the inverted repeat sequences,and a 1.4-kb fragment spanning the ADE1 gene inserted withinthe unique 2µm sequences at the SnaBI site which doesnot disrupt 2µm plasmid maintenance significantly (4,11). In addition, pZ10 carries sequences for propagation inEscherichia coli, positioned within the plasmid such that theaction of the 2µm Flp recombinase is to excise the bacterialsequences, leaving the plasmid 2µm-ADE1. 2µm-ADE1is identical to wild-type 2µm plasmid except for the ADE1insertion and carries no bacterium-derived or other duplicatedsequences (Fig. 1A ). pCEN-ADE1 carries a 1.6-kb fragment spanningADE1 cloned between the XhoI-HindIII sites of pRS313 (41). pASF60(43) contains multiple copies of the lac operator cloned intothe KpnI-SphI sites of plasmid YEplac181a, and pASF135 (43)carries the Lac^r lac repressor encoding gene fused to greenfluorescent protein (GFP) inserted between the NotI-XhoI sitesof pRS303. The series of centromeric plasmids expressing RSC2with truncated or mutated domains were a gift of B. R. Cairns(8). pCM5 and pCM13 carry RSC2 and RSC1 inserted in the galactose-controlledexpression vector pYX143 (R&D Systems, Abingdon, UnitedKingdom).

View larger version (21K):
[in this window]
[in a new window]
FIG. 1. Characterization of mutants unable to stably maintain 2µm-type plasmids. (A) pZ10 carries the complete 2µm plasmid genome plus a duplication of one inverted repeat containing the Flp recognition target (FRT) site inserted in the E. coli vector pUC7. The ADE1 gene is inserted in 2µm sequences between STB and Ori at the unique SnaBI site. FLP-mediated recombination in yeast excises 2µm- ADE1 carrying a wild-type 2µm plasmid except for the ADE1 insertion. (B) Colony phenotypes of dpm19 (left) and wild-type (right) GY59 colonies containing 2µm-ADE1. (C) dpm mutants show a much higher frequency of pink (ade^-) sectors and colonies with 2µm-ADE1 (left) than with pCEN-ADE1, as shown by the overall pink color developed on plates. (D) Plasmid stability assay of dpm mutants carrying 2µm-ADE1 and the centromeric plasmid pCEN-ADE1.

Colony-sectoring (plasmid-loss) assay. The loss of 2µm-ADE1 was monitored in an ade1 geneticbackground by accumulation of red pigment due to the absenceof ADE1 in plasmid-free cells. Yeast cells were grown on YPD,and the red color was allowed to develop at 4°C for at least1 week. Qualitative assays (as used for mutant screening) werebased on examination of the frequency of sectoring. For quantitativeanalysis of loss rates, half-sector analysis of loss rate of2µm-ADE1 was carried out as described previously (20).This measures the loss of plasmid in one generation, since sectorsencompassing half of a complete colony arise from loss in thefirst division that occurs in the formation of a colony froma single cell.

Loss of native 2µm plasmid was monitored by PCR. Plasmid2µm was detected by the REP1 amplifying primers (forwardprimer, ACAGCGCTGATATACAATG; reverse primer, CTGTCGGCTATTATCTCCG)and the presence of the disruption (KanMX) sequence by usinginternal primers (forward, TAGCCCATACATCCCCATGT; reverse, ACAGGAATCGAATGCAACCG).Single colonies were picked as a template for the amplificationreactions, resuspended in a PCR mixture containing 2.5 U ofTaq DNA polymerase (Roche), Taq buffer 1, 0.2 µM concentrationsof each deoxynucleoside triphosphate, and a 1 µM concentrationof each primer, and then cycled as follows: 94°C (10 min),followed by 30 cycles of 94°C (30 s), 55°C (1 min),and 72°C (5 min), with a final 7 min at 72°C.

Complementation of the dpm19 mutant. The dpm19 mutant was complemented with a yeast genomic libraryin the centromere-containing plasmid YCp50 (37). Since dpm19loses 2µm-ADE1 rapidly, pZ10 and the library plasmid wereused to cotransform dpm19 as described previously (17). Transformantswere plated on minimal medium lacking adenine and uracil toselect for both 2µm-ADE1 and YCp50. A complementing fragmentof 9.3 kb was isolated and designated pMW1. Deletion derivativesidentified the complementing gene as RSC2 (Fig. 2A).

View larger version (52K):
[in this window]
[in a new window]
FIG. 2. Characterization of cloned genomic library fragments. (A) Plasmids pMW4 and pMW5 were derived from the complementing clone pMW1, which carries the 9.3-kb fragment (shown at top). The "+" indicates subclones complementing the dpm19 mutant phenotype; the "-" indicates subclones unable to complement dpm19. (B) The subclones derived from pMW1 were tested for their ability to reestablish plasmid stability in dpm19. Subpanels: a, wild-type GY59; b, dpm19 mutant; c, dpm19 mutant with complementing pMW1 clone (all carrying 2µm-ADE1).

PCR-mediated gene disruption. PCR-mediated gene disruption was as described previously (3).RSC2 was disrupted with the KanMX gene (conferring Geneticinresistance) or TRP1. Details of primers and PCR conditions usedcan be obtained upon request. PCR fragments were used to generateyeast transformants, which were selected on YPD containing Geneticinor synthetic dextrose dropout (1) medium lacking tryptophan.Disruptants were confirmed with primers external and internalto the disruption cassette and both boundaries.

Chromatin analysis. Micrococcal nuclease (MNase) analysis of plasmid DNA was carriedout as described previously (29, 50) with minor modifications.Yeast cells were washed once with cold water, resuspended in5 ml of lysis buffer (1 M sorbitol plus 5 mM 2-mercaptoethanolcontaining 2 mg of 100T Zymolyase [Seikagaku Corp., Tokyo, Japan])per 1 g (wet weight) of cells, and incubated at 30°C for20 min. Pelleted cells were washed and resuspended in 7 ml ofFicoll solution (18 [wt/vol] Ficoll type 400 [Sigma], 20 mMKH₂PO₄ [pH 6.8], 1 mM MgCl₂, 0.25 mM EGTA, 0.25 mM EDTA) per1 g (wet weight) of cells. Cells were centrifuged at 16,000rpm in a Sorvall SS34 rotor (30 min at 5°C). The nuclearpellet was digested in 1.2 ml of digestion buffer (15 mM Tris-HCl[pH 7.5], 75 mM NaCl, 3 mM MgCl₂, 1.5 mM CaCl₂, 1 mM 2-mercaptoethanol).Then, 200-µl aliquots were transferred to microfuge tubes,digested with 0.125 to 1 U of MNase (Worthington BiochemicalCorp. catalog no. 4797; stock solution dissolved in MNase buffer[50 mM Tris-HCl, pH 8.0; 0.5 mM CaCl₂; 20% glycerol] to a concentrationof 100 U/µl), and digested for 10 min at 37°C. Digestionwas terminated with 30 µl of stop solution (1% sodiumdodecyl sulfate [SDS] and 5 mM EDTA, final concentration) and30 µl of proteinase K (0.5 mg/ml) at 55°C for 2 h.Purified DNA was resuspended in 100 to 200 µl of Tris-EDTAbuffer and digested to completion with an appropriate restrictionenzyme that cleaves DNA at a site adjacent to the region tobe mapped. DNA was electrophoresed on a 1.5% SeaKem LE agaroseat 45 V overnight in Tris-borate-EDTA buffer. DNA was transferredto Genescreen (Du Pont) by using the method of Sambrook et al.(38). DNA probes immediately proximal to the chosen restrictionenzyme site were used (see Fig. 5 legend). Then, 25 ng of probewas radiolabeled by using [ ${alpha}$ -³²P]dCTP with the Rediprime II kit(Amersham Pharmacia Biotech, Amersham, United Kingdom). Filterswere hybridized in 0.5 M NaH₂PO₄-0.5 M NaH₂PO₄-7% SDS (pH 7.2)(12) for 48 h at 60°C and washed twice with 2x SSC (1x SSCis 0.15 M NaCl plus 0.015 M sodium citrate) at 60°C for5 min and once with 2x SSC-0.1% SDS at 60°C for 30 min.

View larger version (71K):
[in this window]
[in a new window]
FIG. 5. Alterations in STB chromatin structure but not in other 2µm plasmid regions in rsc2 cells as analyzed by MNase digestion and indirect end labeling. Chromatin was isolated from GY59 (right lanes) and GY59 rsc2 ${Delta}$ (left lanes), digested with MNase, deproteinized, blotted, and probed as indicated (see Materials and Methods). (A) Chromatin structure of the STB region. Main differences between wild-type and rsc2 cells are indicated by white arrows. Chromatin was digested to completion with EcoRI and probed with the EcoRI/PstI fragment from the RAF gene (positions 2407 to 2652 as numbered from the EcoRI site in the FLP gene of the 2µm-A form) (probe 1). The concentrations of MNase in both rsc2 and RSC2 were 0.125 U (left lane of each) and 0.25 U (right lanes). (B) Purified naked control DNA subjected to MNase digestion and analyzed with probe 1. Left lane, 0.125 U of MNase; right lane, 0.25 U of MNase. (C) Chromatin structure of FLP/REP2 promoter (intergenic) region in rsc2 and RSC2 cells. Chromatin was digested with HincII and probed with a PCR product from bases 2651 to 2960 at the end of RAF (probe 2). The left lane of rsc2 and the right lane of RSC2 samples were digested with 0.125 U of MNase, and the right lane of rsc2 and the left lane of RSC2 were digested with 0.5 U. (D) Chromatin structure of the RAF/REP1 promoter (intergenic) region. Chromatin was digested with EcoRI and probed with a PCR product from positions 5910 to 6308 of FLP (probe 3). The concentrations of MNase from left to right of both rsc2 and RSC2 lanes were 0.5, 0.25, and 0.125 U. Marker lanes were derived from double digestion of 2µm DNA with EcoRI and the enzymes indicated. Marker sizes are shown in base pairs.

Immunostaining and GFP localization of plasmids. Strains were grown in selective SD media (1) with 6.5 g of sodiumcitrate/liter for yeast strains expressing GFP. Strains containingpAFS60 and pAFS135 were induced with 15 mM 3-amino-1,2,4-triazoleadded to the medium to a final concentration of 10 mM for 15min prior to live imaging or fixation in 3.7% (vol/vol) formaldehydein 0.1 M KH₂PO₄ solution.

Immunostaining for Rep1 and Rep2 protein was as described previously(39). Images were captured by using a Nikon Optiphot2 with aHamamatsu Orca2 digital camera and then processed by using Openlabsoftware (Improvision, Ltd., Warwick, United Kingdom).

RESULTS

Top

Results

Identification of chromosomal mutants causing 2µm plasmid loss. Replicating plasmids in yeast are normally subject to strongMIB, resulting in frequent production of plasmid-free cells.The 2µm plasmid is able to overcome MIB by a mechanismthat requires the plasmid proteins Rep1 and Rep2 and the DNAsequence STB, together with unknown host cell functions. A colony-sectoringassay was devised for identification of mutations in host genes,resulting in the loss of plasmid stability and MIB of 2µmplasmids. pZ10 (see Materials and Methods) disintegrates inyeast, resulting in the plasmid 2µm-ADE1, which is identicalto wild-type 2µm plasmid except for insertion of the ADE1gene into a nonessential region. Between the STB and ORI loci(Fig. 1A), 2µm-ADE1 exhibits stability characteristicssimilar to that of native 2µm plasmid. Since ade1 cellsaccumulate a pink pigment, the loss of 2µm-ADE1 from theade1 strain GY59 can be readily observed as pink ade1 sectorsin otherwise white (i.e., 2µm-ADE1-containing) colonies(Fig. 1B).

A total of 225,000 colonies were examined after ethyl methanesulfonatemutagenesis of strain GY59 (2µm-ADE1), and six independentcolonies exhibited consistent frequent sectoring. 2µm-ADE1was reintroduced into plasmid-free segregants of each putativemutant, and formation of sectored colonies was observed withall six mutants after the fresh transformation, confirming thata genomic mutation was responsible for the defect in plasmidmaintenance (Fig. 1C). These mutants were named dpm (defectivein plasmid maintenance) mutants, i.e., the dpm1, dpm3, dpm12,dpm16, dpm18, and dpm19 mutants.

dpm mutants are defective in escaping MIB. Plasmid loss from dpm mutant cells could be caused either bydefective plasmid replication or by plasmid missegregation asa consequence of failure to escape MIB. In order to distinguishbetween these possibilities, we introduced into the dpm mutantsthe minichromosome (centromeric) plasmid pCEN-ADE1, which replicatesby using a 2µm plasmid replication origin but is stablymaintained at one to two copies per cell as a consequence ofcarrying the centromere of chromosome IV. If dpm mutants wereaffected in plasmid replication, a more severe loss phenotypewould be predicted for pCEN-ADE1 than 2µm-ADE1 due tothe lower copy number of pCEN-ADE1.

Introduction of pCEN-ADE1 into all six dpm mutants resultedin a mild plasmid loss phenotype with reduced sectoring of colonies(Fig. 1C). A quantitative plasmid stability assay comparingthe stability of the centromeric and 2µm plasmids in dpmand wild-type cells showed that all dpm mutants lose 2µm-ADE1at frequencies of between 15 and 26% per cell per generation(Fig. 1D). In contrast, the loss rate of the centromeric plasmidpCEN-ADE1 was ca. 5% or less per cell per generation. The parentstrain GY59 loses both plasmids at ca. 1% per generation.

The results indicate that the stability of a low-copy-numbercentromeric minichromosome was mildly affected, whereas thestability of the high-copy-number 2µm-ADE1 plasmid carryingthe same replication origin was severely affected. Therefore,elevated loss of 2µm-ADE1 in dpm mutants was likely dueto a failure of normal 2µm plasmid segregation.

RSC2 complements the mutant phenotype of dpm19. The dpm19 mutant was cotransformed with a yeast genomic libraryin the centromeric vector YCp50 (37) and plasmid pZ10, whichdisintegrates to produce 2µm-ADE1 in yeast cells. Transformantswere plated on medium lacking adenine and uracil to select forthe presence of both 2µm-ADE1 and YCp50. From about 500transformed colonies, eight nonsectoring white colonies wereobtained. The library plasmids from two of these colonies werefound on retransformation to confer a stable ADE⁺ phenotypeon the dpm19 mutant only in the presence of 2µm-ADE1,confirming that these clones reestablished plasmid stabilityrather than carrying ADE1 genomic sequences. Restriction mappingindicated that both plasmids carried the same fragment of $~$ 10kb, and they were designated pMW1. We note that the frequencyof rescuing library clones is high (2/500), which may reflectthe extreme instability of 2µm-ADE1 in dpm19 cells, resultingin a low frequency of colony development of noncomplementedtransformants.

Partial sequencing of the genomic insert in pMW1 showed thatit comprises 9.3 kb of genomic sequence from the right arm ofchromosome XII, encompassing five complete (TAL1, ILV5, YLR356W,RSC2, and YLR358C) and two incomplete (BUD8 and ADE13) openreading frames. Deletion and subcloning analysis showed thatplasmid pMW4, containing RSC2 and YLR358c complemented the mutantphenotype, whereas pMW5, which contains YLR358c and a partialdeletion of RSC2, failed to complement (Fig. 2). Sequencingof the rsc2 allele of the dpm19 mutant showed a single nucleotidedeletion at position 871, causing a frameshift mutation. Theseresults indicate that RSC2 is required for normal segregationof 2µm-ADE1.

All dpm mutants except the dpm1 mutant carry alleles of rsc2. Plasmid pMW4, carrying RSC2, was introduced into the five otherdpm mutants, together with pZ10. pMW4 was found to complementthe mutant phenotype of all of the dpm mutants except the dpm1mutant, resulting in stable maintenance of the 2µm-ADE1plasmid. Sequencing of the RSC2 alleles from other dpm mutantsconfirmed that all mutants except the dpm1 mutant have mutationsin RSC2. The dpm16 and dpm19 mutants both have a single basedeletion at position 871, but the dpm16 mutant carries an additionalA-G base substitution at position 260. The dpm3 mutant has asingle base substitution from G to A at position 1402, causinga glutamate residue to be replaced by lysine. The dpm12 mutanthas a single C-T base substitution at position 235, convertinga glutamine codon to a premature stop. The dpm18 mutant hasa single G-to-A base substitution at position 1802, changingthe glycine residue to a glutamate.

These results indicate that RSC2 plays a significant role inovercoming MIB in 2µm plasmid maintenance, since fiveindependent rsc2 alleles were identified in our screen. dpm1was not complemented by RSC2, indicating that at least one othergenomic gene is required for 2µm plasmid maintenance.

RSC2 is essential for native 2µm plasmid maintenance. To confirm that RSC2 is involved in plasmid maintenance, RSC2was deleted in the haploid strain GY59 carrying 2µm-ADE1and in strains A364A [cir⁺] and GY18 [cir⁺], which harbor thenative 2µm plasmid. The GY59 rsc2 ${Delta}$ (2µm-ADE1) nullmutant formed highly sectored colonies identical to dpm19 colonies,indicating high-frequency loss of 2µm-ADE1 (not shown).Since native 2µm plasmid does not confer a phenotype,colonies of A364A rsc2 ${Delta}$ [cir⁺] and GY18 rsc2 ${Delta}$ [cir⁺] mutants werescreened for the presence of native 2µm plasmid by PCR.Immediately after deletion of RSC2, most colonies still containedcells carrying 2µm, a finding consistent with a defectin segregation. After several subcultures, four of six rsc2 ${Delta}$ colonies contained no plasmid-bearing cells, while all of thecolonies of the control RSC2 strain maintained plasmid (Fig.3). Similar results were obtained with W303 rsc2 ${Delta}$ and with sporesgerminated from a diploid GY59 rsc2 ${Delta}$ x W303 cross (not shown).We conclude that RSC2 is essential for normal 2µm plasmidmaintenance.

View larger version (71K):
[in this window]
[in a new window]
FIG. 3. RSC2 is required for wild-type 2µm plasmid maintenance. RSC2 disruption and the presence of the 2µm plasmid was tested in six colonies each of GY18 rsc2 ${Delta}$ (A and C) and GY18 RSC2 (B and D). Colonies were analyzed by using KanMX primers (specific for rsc2 ${Delta}$ ) and 2µm primers (detecting 2µm plasmid). Panels A and B show analysis directly after the detection of rsc2 ${Delta}$ ; panels C and D show the results six subcultures later. An amplification product with the KanMX primers confirms the presence of the rsc2 ${Delta}$ allele, and product with 2µm primers indicates the presence of plasmid in that colony.

RSC2 is required for escape from MIB and for normal Rep protein localization. DNA molecules carrying lac operator-binding sites can be visualizedin yeast in the presence of a GFP-Lac^r protein fusion (43, 52).To confirm the involvement of RSC2 in the escape from MIB, weobserved the behavior of a 2µm plasmid derivative carryingthe lac operator repeats in RSC2 and rsc2 ${Delta}$ strains.

In GY59 (RSC2), the plasmids were detected in ca. 60 to 80%of cells, and the fluorescent intensity indicated relativelyeven distribution of the copy number between cells. Plasmidswere observed to be concentrated in clusters within the nucleus,and plasmid segregation into daughter buds was generally observedconcomitantly with bulk chromatin, as previously reported (Fig.4A and B) (39, 52). In contrast, in GY59 rsc2 ${Delta}$ plasmids werefound in ca. 5% of cells, which predominantly showed intensefluorescent signals (Fig. 4C and D). Some mother-bud pairs entermitosis failing to transmit plasmids to the daughter bud (Fig.4E and F). This is indicative of a high plasmid copy numberaccumulating in a subset of cells, a finding consistent withthe appearance of MIB in rsc2 ${Delta}$ cells. These observations confirmthat RSC2 is required for 2µm to overcome MIB.

View larger version (67K):
[in this window]
[in a new window]
FIG. 4. MIB is associated with loss of normal Rep1 localization in rsc2 ${Delta}$ mutants. (A) Localization of 2µm-based plasmid pAFS60, which carries lac operator repeats that bind GFP-Lac^r fusion protein. Discrete plasmid clusters (A) that bind coincident with bulk nuclear DNA (B) are observed. (C) 2µm is found at a high copy number in a small subset of GY59 rsc2 ${Delta}$ cells. (D) Same as for panel C, but visualized with DAPI. (E) GY59 rsc2 ${Delta}$ mother-daughter cell pair exhibiting MIB (failure to transmit plasmid to daughter bud). (F) Same as for panel E, but visualized with DAPI, showing nuclear DNA in both mother and daughter cells. (G) Immunolocalization of Rep1 in GY59 showing discrete foci of plasmid proteins. (H) Merged image as in panel G, showing Rep1 (green), DAPI staining of nuclear DNA (blue), and spindle microtubules detected by using Texas red-conjugated anti-tubulin (red). Note that the Rep1 protein lies entirely within the domain of DAPI staining. (I) Immunolocalization of Rep1 in GY59 rsc2 ${Delta}$ showing the absence of foci of plasmid proteins and the presence of Rep1 throughout the nucleus. Note in merged image J (same colors as for panel H) that Rep1 immunostaining lies substantially outside the region of DAPI staining, indicating that Rep1 has lost its association with chromatin. (K) Wild-type GY18 shows multiple plasmid foci (green). (L) Immunolocalization of Rep1 shows nuclear foci. (N) Merged image of panels K, L, and M (DAPI stained) shows that the majority of 2µm is colocalized with the Rep1 protein and the chromatin. (O) Image shows that GY18 rsc2 ${Delta}$ 2µm plasmid is subject to MIB. (P) Immunolocalization of Rep1 shows some loss of nuclear localization. (R) Merged images of O, P, and Q (DAPI) show that Rep1 and some 2µm plasmid lose association. (A, C, E, K, and O) Localization of 2µm-based plasmids by using GFP-Lac^r bound to pAFS60; (B, D, F, M, and Q) bulk nuclear DNA visualized by using DAPI; (G and I) indirect immunofluorescent detection of Rep1 (FITC); (L and P) indirect immunofluorescent detection of Rep1 (Texas red); (H and J) merged images of Rep1 (green), bulk nuclear DNA (blue), and spindle microtubules (red); (N and R) merged images of 2µm-based plasmid pAFS60 made with GFP-Lac^r (green), Rep1 (red), and DNA (blue).

The 2µm plasmid segregation proteins Rep1 and Rep2 havepreviously been shown to localize to discrete nuclear foci (39).These foci are believed to contain plasmid-protein complex assemblysites essential to allow 2µm plasmid to overcome MIB (39a,52).

Formation of Rep foci was examined in GY59 rsc2 ${Delta}$ by indirectimmunofluorescence (39). In wild-type GY59 cells, Rep1 immunofluorescencewas observed in the majority of cells and is clearly presentin discrete foci (Fig. 4G), which in merged images overliesthe bulk of DAPI (4',6'-diamidino-2-phenylindole)-stained chromatin(Fig. 4H). In GY59 rsc2 ${Delta}$ , Rep1 and Rep2 immunofluorescence wasobserved in ca. 5% of cells, a finding consistent with the numberpreviously found to contain plasmid, indicating that the lossof segregation of 2µm plasmids in rsc2 mutants is notdue to a failure to express the REP genes. However, Rep1 wasno longer predominantly observed in foci (Fig. 4I), and thedistribution of Rep1 is no longer coincident with the bulk DAPI-stainedchromatin (Fig. 4J). Similar observations were made for Rep2(not shown).

The extreme instability of 2µm-ADE1 in GY59 rsc2 ${Delta}$ makesit technically difficult to assess colocalization of GFP-labeledplasmid with Rep protein foci. Strain GY18 rsc2 ${Delta}$ has a less-severeloss rate of native 2µm plasmid than GY59 rsc2 ${Delta}$ (see above).As expected from previously published data (39, 52), the majorityof GFP foci colocalize with Rep1 protein foci in GY18 cells(Fig. 4K to N). In GY18 rsc2 ${Delta}$ the proportion of cells carryingplasmid is 25% lower than in GY18 and, although the Rep proteinhas a similar overall staining intensity to GY18 cells, it appearsto be less concentrated in foci. In addition, MIB can be observedin some division events (Fig. 4O to R).

The MIB and lack of Rep foci seen in rsc2 ${Delta}$ mutant cells suggestthat Rep proteins may be unable to form productive complexeswith STB in the absence of Rsc2. This could be explained bya direct interaction of Rsc2 with the Rep proteins. Two-hybridanalysis between Rep1, Rep2, and Rsc2 confirmed the Rep1-Rep2interactions previously reported (39, 51), but no interactionbetween Rsc2 and Rep1 or Rep2 was detected in this assay (datanot shown).

Altered STB chromatin structure in rsc2 mutants. Our previous results have shown that Rep protein focus formationrequires the presence of plasmids carrying the STB locus (39).An alternative explanation to the MIB and failure to form normalRep complexes is therefore that Rsc2 is required to create achromatin structure amenable to Rep protein interaction withplasmid DNA.

The RSC complex perturbs nucleosome structure in vitro (7),and mutation of the RSC component Nps1/Sth1 alters the chromatinstructure of centromeres (48), suggesting a possible role forRsc2 in establishing specific chromatin structures. The chromatinstructure of the STB locus, Flp target sites (FRT), and thepromoter regions of FLP, REP1, REP2, and RAF were examined byMNase digestion and indirect end labeling in wild-type and rsc2 ${Delta}$ mutant GY18 [cir⁺] cells. No changes in the chromatin structurein the promoter regions of the genes (Fig. 5C and D) or theFRT sites (not shown) were observed, demonstrating that Rsc2is not required for the general nucleosomal structure of 2µmplasmid.

In contrast, a significant difference in the chromatin patternat the STB region was observed (Fig. 5A). The essential regionof STB consists of two repeats of 124 bp superimposed on fiverepeats of 62 bp located between the plasmid HpaI and AvaI sites(32). In wild-type cells, strong MNase cleavage sites are observedat the borders of STB and a weaker site is seen at the center,together with a general background smear over STB, indicativeof partial protection by two protein complexes. In contrast,in rsc2 ${Delta}$ cells no bands were observed over the STB region, andthe central and border cleavage sites were absent or very stronglydiminished. The cleavage pattern in rsc2 ${Delta}$ cells was distinctfrom that observed with naked DNA (Fig. 5B), indicating thatSTB continues to interact with proteins in rsc2 ${Delta}$ cells. We concludethat RSC2 is needed for normal protein-DNA interactions at STBbut not for the general chromatin structure of the remainderof the 2µm plasmid.

RSC1 is not required for 2µm plasmid segregation. RSC2 has previously been shown to be a nonessential componentof the RSC complex due to the existence of RSC1, a closely relatedRSC complex member with 45% identity (62% similarity) to RSC2.Deletion of either RSC1 or RSC2 creates only mild growth phenotypesunder specific stress conditions, whereas the deletion of bothis lethal. Rsc1 and Rsc2 have been shown to be present in separatesubcomplexes that have separate and common targets (7, 8). RSC1was disrupted in strain GY59 (2µm-ADE1), and the stabilityof 2µm-ADE1 was assessed in rsc1 ${Delta}$ and rsc2 ${Delta}$ strains. 2µm-ADE1MIB was only observed in the rsc2 ${Delta}$ strain and not in the rsc1 ${Delta}$ mutant (Fig. 6A). GAL1-directed expression of RSC2 was ableto complement the rsc2 ${Delta}$ plasmid instability phenotype (Fig. 6B),but GAL1-directed expression of RSC1, which resulted in itsaccumulation to the same level as that observed with GAL1-RSC2expression, was unable to complement the MIB phenotype of rsc2 ${Delta}$ and stabilize 2µm-ADE1. We conclude that Rsc1 is not requiredfor 2µm plasmid segregation, and thus Rsc2 has a uniquerole in this process.

View larger version (54K):
[in this window]
[in a new window]
FIG. 6. Overexpression of RSC1 cannot compensate for loss of RSC2 function. (A) GY59 rsc1 ${Delta}$ maintains 2µm-ADE1 with no loss of stability. (B) GY59 rsc2 ${Delta}$ stably maintain 2µm-ADE1 on galactose medium when cells are cotransformed with pCM5 (GAL1:RSC2). (C) GY59 rsc2 ${Delta}$ cells do not maintain 2µm-ADE1 on galactose medium when cells are cotransformed with pCM13 (GAL1:RSC1). All plates contain YPGal.

The bromodomains and the C-terminal (CT) domain of Rsc2 are essential for 2µm plasmid maintenance. The region of similarity between Rsc1 and Rsc2 is divided intofive sections (8), bromodomains BD#1 and BD#2, the AT-hook motif,a bromodomain-adjacent homology (BAH), and a CT domain (Fig.7A). BD#2, BAH, and a CT domain were previously found to beabsolutely required for any of the tested functions in bothRsc1 and Rsc2, but BD#1 was required only for certain Rsc2 functions(8). The AT-hook motif of Rsc2 was found to be required onlyfor growth in an rsc1 rsc2 background. As discussed above, twoof the originally isolated dpm mutants have substitution mutationsin RSC2. dpm3 causes an E467K mutation in the BAH, and dpm18results in the substitution G600E within the CT domain (Fig.7C). Both E467 and G600 are conserved in Rsc1 and Rsc2 and liewithin regions of very high identity between the two proteins.Their phenotype in dpm3 and dpm18 strains suggests that theseresidues are essential for Rsc2 function. We tested the abilityof the domain deletions of Rsc2 to complement 2µm-ADE1plasmid stability in GY59 rsc2 ${Delta}$ by using plasmids described byCairns et al. (8), which have previously been shown to resultin the approximately equal accumulation of each protein. Wefound that all domains of Rsc2 are important in its functionin 2µm maintenance. Rsc2 derivatives lacking both bromodomainsor the CT domain were nonfunctional, indicating that the presenceof a BD and the CT domain is essential. Loss of either BD#2or the BAH had severe effects but the protein retained limitedfunction, whereas Rsc2 with a mutated AT-hook motif or BD#1resulted in intermediate stability of 2µm-ADE1 (Fig. 7B).

View larger version (50K):
[in this window]
[in a new window]
FIG. 7. Domain analysis of Rsc2 function in 2µm maintenance. (A) Five domains of Rsc2: BD#1, AT-hook, BD#2, BAH, and CT as defined previously (8). (B) Centromeric plasmids expressing truncated or deleted Rsc2 domains as indicated were assessed for complementation of 2µm-ADE1 stability in rsc2 ${Delta}$ cells. Complementation is indicated from full (+) to none (-) as shown in panel B. (C) Positions of dpm mutations within RSC2. (D) Plates illustrating the phenotypes in the complementation experiment described for panels B and C.

DISCUSSION

Top

Discussion

Replicating extrachromosomal DNA circles are inherently lostat a high rate from yeast cells. This is due to MIB that resultsin the frequent failure to transmit such molecules to daughtercells (31, 42). The molecular explanation of MIB is unclear,since the nuclear volume is shared approximately equally betweenmother cell and daughter bud. However, it may result from thefailure of DNA circles to be liberated from replication sites,since linear plasmids do not exhibit MIB (31) and since a mutationthat reduces MIB of some plasmids was found to lie in DNA polymerase ${delta}$ (21). Mechanisms that allow circular DNA molecules to be efficientlytransmitted generally involve their association with segregatednuclear structures (5, 14, 23, 27, 28).

The 2µm plasmid of S. cerevisiae is a rare example ofa widespread high-copy-number eukaryotic nuclear extrachromosomalelement. In order to be maintained in yeast populations, 2µmmust therefore possess a mechanism allowing it to overcome MIB.Although the 2µm plasmid contains only four open readingframes, the molecular details of how this is achieved are notclear. The plasmid proteins Rep1 and Rep2 associate with theSTB locus (18, 51) and result in the localization of plasmidand Rep proteins to discrete subnuclear foci which are segregatedat mitosis and appear to be associated with bulk nuclear chromatin(39, 52).

Here we show that chromatin structure is central to plasmidstability and demonstrate that 2µm plasmid is a noveltarget for Rsc2, which contains two bromodomains (34, 53) andis a component of the RSC complex (7, 8). The related proteinRsc1 is unable to substitute for Rsc2. This is the first exampleof a phenotype attributed to RSC2 that is correlated with specificchanges in chromatin structure in vivo.

Six chromosomal mutants were isolated that maintain the 2µm-likeplasmid 2µm-ADE1 poorly and, by complementation with agenomic library, RSC2 was found to rescue the mutant phenotypeof the dpm19 mutant and four other dpm mutants. RSC2 disruptionin other strains revealed a phenotype similar to that of thedpm19 mutant, affecting not only 2µm-ADE1 but also thenative 2µm plasmid. A failure of RSC2 to complement dpm1indicates that there is at least one other gene that is requiredfor 2µm plasmid maintenance.

RSC1, a homologue of RSC2, has 45% identity (62% similarity),and both RSC1 and RSC2 have been shown to be present in differentsubcomplexes that have both separate and common targets (8).The combined loss of both RSC1 and RSC2 is lethal, but lossof either results in mild growth defects (8). We have shownthat the role of RSC2 in 2µm plasmid stability is unique,since loss of RSC1 has no effect on 2µm maintenance.

A high plasmid loss rate could be due to a failure either inplasmid replication or in the segregation of the replicatedmolecules. Plasmid replication failure would be predicted toproduce a low average plasmid copy number across the cell population,whereas segregation failure would lead to a high plasmid copynumber in a small population of cells. We believe that it isunlikely that the failure of plasmid maintenance in rsc2 mutantsis due to a failure in DNA replication, since the loss rateof the high-copy-number 2µm-ADE1 is substantially higherthan that of a low-copy-number centromere-containing plasmid,pCEN-ADE1. This is supported by the visualization of plasmidswith a GFP-Lac^r fusion that shows accumulation of plasmid ata high copy number in a small proportion of rsc2 cells. In certaininstances plasmids were observed failing to segregate, despitethe presence of a significant plasmid signal. This demonstratesthat plasmid segregation rather than replication is primarilyaffected in rsc2 mutants.

2µm plasmid is condensed into 21 to 34 nucleosomes perplasmid that closely resemble those of chromosomal DNA (26).The protection of 2µm chromatin from nuclease digestionis not random within the plasmid (25), and in particular theSTB region has an abnormal structure (15, 50). Our analysisof the 2µm chromatin structure in rsc2 mutants suggeststhe following. First, the chromatin structure is only alteredwithin STB, and no changes were observed in the promoter regionsof the four open reading frames. This indicates that loss ofRSC2 does not perturb the overall chromatin structure of 2µmplasmid but rather has a specific effect on STB. Second, RSC2does not seem to form a stable complex with the STB region,since there is no increase in sensitivity at STB when RSC2 isabsent. Finally, we found no interaction between either Rep1or Rep2 and Rsc2 in two-hybrid assays (not shown).

STB has been divided into two domains, lying on either sideof the unique HpaI site. STB-distal is the region further fromthe replication origin (Ori) and corresponds to the transcriptiontermination region of the RAF gene. STB-proximal is made upof five direct tandem repeats of an AT-rich 62-bp consensussequence which can also be aligned as two tandem repeats of124 bp (32) and is the binding site of Rep1-Rep2 complexes (51).Our results suggest that, in wild-type cells, STB-proximal isprotected by two protein complexes that bind to the tandem repeatsequences and are separated by a nuclease-sensitive spacer sequenceand flanked by hypersensitive sites. These may represent complexeson each of the 124-bp repeats. In the absence of RSC2, a moreextensive complex is formed that is able to bind the spacerregion and protects the whole of STB-proximal from nucleasedigestion. The two sensitive sites at the border of the STB-proximalare displaced and have reduced nuclease sensitivity. This changein chromatin structure is coincident with the failure of plasmidsegregation.

rsc2 mutant cells therefore exhibit MIB of 2µm despitethe presence of Rep1 and Rep2. We propose that the altered STBchromatin structure observed in rsc2 mutants is due to the failureof productive Rep protein-STB complexes to form. In rsc2 mutantcells, Rep1 is nuclear but is less distinctly localized intodiscrete foci. We have previously shown that the ability ofRep1 to form normal subnuclear foci requires the presence ofplasmids carrying STB (39). The absence of such Rep1 foci inrsc2 mutants is therefore consistent with a failure of Rep1to interact correctly with STB. We therefore propose that inwild-type cells RSC2 functions to maintain the correct nucleosomepositioning at the STB so that a productive segregation proteincomplex can bind the STB and effect plasmid segregation. Inthe absence of Rsc2, STB chromatin is not maintained in thecorrect form, and thus the protein complexes may bind, randomlyproducing an unproductive complex. Alternatively, a differentcomplex may be able to bind to the STB sequence, preventingefficient plasmid segregation.

Recently, Sengupta et al. (40) have shown that specific domainsof Rep1 and Rep2 mediate dimerization, interaction, and DNAbinding. They proposed that competition between Rep1 and Rep2for association with Rep1 determines the formation or disassemblyof the segregation complex. Changes in the chromatin structureat STB may influence this competition, promoting disassemblyor otherwise changing the segregation complex and leading tosegregation failure.

Surprisingly, five of six of the dpm mutants that we generatedcontain a mutation in RSC2. dpm12, dpm16, and dpm19 mutantshave mutations which cause frameshifts of premature terminationand hence abolish the function of the C terminus of the RSC2gene and produce a phenotype similar to that of an rsc2 nullmutant. dpm3 and dpm18 mutants have point mutations within theBAH and C terminus, respectively, and identify E467 and G600as key residues in Rsc2 function.

Overall, the structure of 2µm-derived plasmid constructshas long been suspected to be important in influencing theirstability. For example, Futcher and Cox (16) found that plasmidsthat carry very similar regions of the 2µm genome, butin different DNA contexts, can differ substantially in stability.Moreover, the plasmid chromatin structure is dependent on theplasmid proteins present (50). Here we present strong evidencethat the chromatin structure of STB is of key importance inallowing the plasmid to escape MIB and show that Rsc2 is essentialfor maintaining that chromatin structure. Remaining challengesinclude the characterization of the complexes that bind STBin both wild-type and rsc2 backgrounds.

ADDENDUM IN PROOF

Top

Addendum in proof

In contrast to the results we report here indicating a specificrole for Rsc2 in 2µm plasmid maintenance, Ng et al. (GenesDev. 16:806-819, 2002) have recently shown that the global bindingprofiles of the Rsc1 and Rsc2 isoforms of the RSC complex areindistinguishable.

ACKNOWLEDGMENTS

This work was funded by Biotechnology and Biological SciencesResearch Council (BBSRC) grant 8/G10977. M.J.H. was supportedby a BBSRC studentship, and M.C.V.L.W. was supported by a Ph.D.studentship from the Ministry of Science Technology and Environmentof Malaysia.

FOOTNOTES

* Corresponding author. Mailing address: Institute of Biotechnology, University of Cambridge, Tennis Court Rd., Cambridge CB2 1QT, United Kingdom. Phone: 01223-334166. Fax: 01223-334167. E-mail: j.murray{at}biotech.cam.ac.uk.

Present address: Department of Biotechnology, Faculty of FoodScience and Biotechnology, Universiti Putra Malaysia, 43400Serdang, Selangor, Malaysia.

Present address: Department of Cell Biology, Institute of Ophthalmology,University College London, London EC1V 9EL, United Kingdom.

REFERENCES

Top