Saccharomyces cerevisiae Donor Preference During Mating-Type Switching Is Dependent on Chromosome Architecture and Organization

Abstract

Saccharomyces mating-type (MAT) switching occurs by gene conversion using one of two donors, HMLα and HMRa, located near the ends of the same chromosome. MATa cells preferentially choose HMLα, a decision that depends on the recombination enhancer (RE) that controls recombination along the left arm of chromosome III (III-L). When RE is inactive, the two chromosome arms constitute separate domains inaccessible to each other; thus HMRa, located on the same arm as MAT, becomes the default donor. Activation of RE increases HMLα usage, even when RE is moved 50 kb closer to the centromere. If MAT is inserted into the same domain as HML, RE plays little or no role in activating HML, thus ruling out any role for RE in remodeling the silent chromatin of HML in regulating donor preference. When the donors MAT and RE are moved to chromosome V, RE increases HML usage, but the inaccessibility of HML without RE apparently depends on other chromosome III-specific sequences. Similar conclusions were reached when RE was placed adjacent to leu2 or arg4 sequences engaged in spontaneous recombination. We propose that RE's targets are anchor sites that tether chromosome III-L in MATα cells thus reducing its mobility in the nucleus.

HAPLOID cells of Saccharomyces cerevisiae have the striking ability to change their mating type as often as every generation. Mating-type switching results from a gene conversion event at MAT induced by a double-strand break generated by the site-specific HO endonuclease. One of two loci, HMLα or HMRa, located, respectively, near the left and right ends of chromosome III, is used as a template for the repair event (Haber 2002). Both donors are maintained in heterochromatin and are therefore transcriptionally silent and resistant to HO cleavage (Loo and Rine 1994; Weiss and Simpson 1998; Ravindra et al. 1999). One of the most surprising aspects of mating-type switching is the mating type-dependent bias existing in the choice of the donor. MATa cells preferentially (80–90%) recombine with HMLα, whereas 90% of MATα switching cells recombine with HMRa (Klar et al. 1982; Weiler and Broach 1992; Wu and Haber 1995; Wu et al. 1996).

Donor preference depends on the activation of the entire left arm of chromosome III, since donors artificially placed at different loci on the left arm are always preferred in MATa cells (Wu et al. 1996). When the preferred donor HML is deleted in MATa cells, HMR is easily used instead, meaning that there is no restraint on HMR but instead that the left arm is strongly activated in these cells. However, the deletion of HMR in MATα cells leads to 30% of lethality after HO induction, showing that HML is strongly excluded (Wu and Haber 1995) even though the DNA checkpoint causes the cells to arrest prior to mitosis to have more time to repair the DSB. Therefore, donor preference is mostly controlled by changes on the left arm of chromosome III. The activation of the left arm for recombination is not specific for mating-type switching. The spontaneous rate of recombination between two leu2 alleles, one placed near MAT and the other replacing HML, is 10–30 times higher in MATa than in MATα cells (Wu and Haber 1995).

Donor preference is independent of whether the donor carries a or α information; moreover, donor selection does not depend on any sequences that uniquely define HML or HMR, or any sequences flanking or centromere-distal to HML or HMR (Weiler and Broach 1992; Wu and Haber 1995). The activation of the entire left arm depends on the recombination enhancer (RE), an ∼750-bp noncoding sequence located 17 kb centromere-proximal to HML (Wu and Haber 1996). Deletion of RE in MATa cells makes them behave like MATα cells (i.e., HML is used only 10% of the time), but has no effect in MATα cells. In MATα cells, RE is simply turned off and has no influence on the left-arm exclusion.

RE is well conserved in the group of the Saccharomyces sensu stricto [Kellis et al. (2003) and Saccharomyces genome database (http://www.yeastgenome.org)]. Comparison of RE sequences within this group showed five well-conserved regions (A, B, C, D, and E) (Wu et al. 1998; Sun et al. 2002; Coïc et al. 2006). Only B is not essential for RE activity (Wu et al. 1996; Sun et al. 2002). Region C contains a binding site for the repressor Mcm1-Matα2 that regulates RE in the same way as similar sequences control several a-specific genes (Wu et al. 1998). In MATα cells, the binding of Mcm1-Matα2 leads to the Tup1-dependent formation of a highly organized chromatin structure covering the entire 2.5-kb KAR1-SPB1 intergenic region containing RE (Weiss and Simpson 1997). In MATa cells, Mcm1 binding apparently opens this structure, allowing trans-activators to bind.

Regions A, B, D, and E contain binding sites for the Fkh1 forkhead transcription factor, which has been shown to play an essential role in HML activation in MATa cells through its interaction with RE (Sun et al. 2002). Region C contains a Swi4/Swi6 (SBF) cycling box, which binds the SBF complex, also involved in the activation of the left arm for recombination. SBF activates RE in an Fkh1-independent pathway involving also the putative helicase Chl1 and Yku80 (E. Coïc, K. Sun, C. Wu and J. E. Haber, unpublished results).

We substantiate here the observation by Bressan et al. (2004) that RE activates HML by modifying the architecture or mobility of the left arm of chromosome III and not by establishing a local chromatin organization within HML that is more suitable for strand invasion. We show that the two arms of chromosome III constitute two different “domains” not accessible to each other in MATα cells, a barrier that is suppressed by the activation of RE in MATa cells. We show that RE only partially overcomes the exclusion of a donor located in trans relative to MAT in a competition assay with a cis-located donor. Therefore, there is more constraint on interchromosomal than on intrachromosomal recombination. We also show that RE is portable on chromosome III and acts over about a 50-kb distance from a new location. When HMR, MAT, and HML are moved to chromosome V, RE stimulates usage of an adjacent donor; however, chromosome V apparently lacks sequences similar to those on chromosome III that are responsible for the tethering of the left arm in the absence of RE.

MATERIALS AND METHODS

Plasmids:

To insert MATa at different loci (HIS4 and ARG5,6), we constructed the integrative plasmid pEC2 bearing MATa and a gene coding for the resistance to hygromycin (HPH) (Goldstein and McCusker 1999). A PvuII–EcoRV fragment containing HPH isolated from the plasmid pAG32 has been introduced in the HpaI site of pJH3 (our collection), a pBR322-based plasmid containing an EcoRI–HindIII fragment containing MATa.

To move the centromere of chromosome III, we constructed the plasmid pEC10 in which the centromere of chromosome III is linked to HPH. A BglII fragment containing the centromere of chromosome III isolated from pMJ691 (a generous gift from Michael Lichten) has been introduced into the plasmid pAG32 containing the HPH gene.

To insert RE at position 74 kb on chromosome III in the FUS1-YCL026C-B intergenic region, the following construct was made. The 1.7-kb RE (Wu and Haber 1996) was ligated to complementary oligonucleotides corresponding to the CYC1 transcription terminator and cloned in the EcoRI restriction site in plasmid pSKURA3 containing the URA3 selectable marker to give plasmid pSURE1. Sequences of complementary oligonucleotides were: CYC1-T3 (5′-aat tct ttt ttt aat agt tat gtt agt att aag aac gtt att tat ccc-3′) and CYC1-T4 (5′-ggg ata aat aac gtt ctt aat act aac ata act att aaa aaa ag-3′). They correspond to the published sequence of the CYC1 terminator (Zaret and Sherman 1982). The FUS1-YCL026C-B intergenic region was PCR amplified in two parts, one with primers FUS1-1 (5′-ggg gac cga aga cta att gag ctt-3′) and FUS1-2 (5′-ccg ctc gag ctt ttc ggg cta ggg t-3′) and the other with primers CL026-1 (5′-cgg gat cct cga ggc acc cca gga tct tt-3′) and CL026-2 (5′-gaa gat ctt aat tta tct cat gaa gta at-3′). The two PCR fragments have been cloned in pSURE1 on each side of the URA3-RE fragment resulting in plasmid pRET1. The natural orientation of RE relative to the centromere has been respected.

To insert HMRα on the left arm of chromosome V, we amplified a 921-bp fragment corresponding to coordinates 8,945–9,866 with the primers AN1 (5′-aag gaa aaa agc ggc cgc caa gtt gat aag cca tga aat ca-3′) and AN2 (5′-aag gaa aaa agc ggc cgc acc ccg aag ctg ctt cac aat a-3′), which contain NotI sites. After digestion with NotI, this fragment was cloned into the NotI site of a modified pBluescript plasmid in which a KpnI–BamHI fragment was removed from the polylinker (pSKDS) to give plasmid pSP1. A SalI fragment from pXW221 (our collection), containing an HMRα-LEU2 cassette, was cloned at the SalI site inside the insert of pSP1. The resulting plasmid was named pSP1-HMR.

Primers AN3 (5′-aag gaa aaa agc ggc cgc gtg gtg cct ttg ttc ctt ggt-3′) and AN4 (5′-aag gaa aaa agc ggc cgc tgg tat cta gta agc aaa att ga-3′), which contain NotI sequences, were used to PCR amplify a 1004-bp region corresponding to coordinates 23,625–24,629 on the left arm of chromosome V. This fragment was cloned into the NotI site of pSDKDS to give pSP2. A SacI–XhoI DNA fragment from pSURE1, containing a URA3-RE cassette, was cloned in pSP2 previously digested with SacI and SalI to generate pSP2-RE.

A fragment containing the KanMX4 gene has been linked to HMRα-BamHI to insert it on the right arm of chromosome V. A PvuII–EcoRV fragment from pFA6-KanMX4 (Wach et al. 1994) has been introduced in the SnaBI site of pXW169, a pGEM3Zf(+)-derivated plasmid containing a HMRα-BamHI–HindIII fragment (Wu and Haber 1995), generating pEC3.

Strains:

Most of the strains used in this study are derivatives of DBY745 (ho MATa ade1-100 ura3-53 leu2-3,112) except as otherwise indicated. Strains used to monitor MATa donor preference carry HMRα-B, where Ya has been replaced by a Yα allele containing a single base-pair mutation that creates a BamHI site (Wu and Haber 1995), while strains used to monitor MATα donor preference carry MATα-BamHI. All these strains have a galactose-inducible HO endonuclease gene integrated at the ADE3 locus (Sandell and Zakian 1993).

All yeast transformations were done by one-step transformation (Chen et al. 1992). All constructions have been verified by colony PCR or Southern blot. To delete RE with the URA3 gene, cells were transformed with a BamHI–HindIII fragment from plasmid pXW292 (Wu and Haber 1996) carrying a 1.8-kb RE deletion marked by URA3.

The displacement of MAT from the right to the left arm of chromosome III was performed in two steps. First, the MAT locus was deleted by transformation of a PCR fragment containing the KanMX4 cassette (Wach et al. 1994) flanked by MAT distal and proximal sequences. This fragment was amplified with mixed PCR primers composed of MAT and KanMX4 sequences, AW03 (5′-agt agt gtc tga gga gag ggt tga ttg ttt atg tat ttt tgc gaa ata tac agc tga agc ttc gta cgc-3′) and AW04 (5′-tca aat agg ata gct ata ctg aca aca ttc agt act cga aag ata aac aag cat agg cca cta gtg gat ctg-3′). Then MATa was introduced at the HIS4 locus by transforming cells with a PCR fragment containing MATa linked to the HPH gene flanked by HIS4 surrounding sequences. This fragment was obtained by amplifying the MATa-HPH insert from pEC2 with mixed primers containing HIS4 and MATa sequences, his4MATU (5′-tgc gct gtg taa tag taa tac aat agt tta caa aat ttt ttt tct gaa taa atg ggc gat att ttg ata c-3′) and his4MATL (5′- ata tat aca tat tat ttt cgt tag tgt tcg gtt tcc aag tta gaa ata atg tga agc cga agg taa cta-3′).

The introduction of RE near the HMR locus was performed by the transformation of an EcoRI–HindIII fragment from pXW295 (our collection) containing the 750-bp RE linked to URA3 and flanked by sequences corresponding to coordinates 286,383–287,381.

To study the effect of the deletion of the tT(AGU)C tRNA gene linked to HMR on donor preference in MATa cells, we used strain ROY1681, kindly provided by Masaya Oki. The deletion in this strain is unmarked and we therefore had to work in this strain, which is derived from w303. We introduced HMRα-BamHI and ade3∷GAL-HO in w303 strains bearing or not bearing the tRNA deletion. To study the effect of the tRNA deletion in MATα cells, we selected strains bearing the MATα-BamHI allele after induction of the HO endonuclease in strains bearing HMRα-BamHI. We then backcrossed them to a w303 wild-type strain to recover the HMRa allele.

RE was introduced at different positions around the centromere in a strain bearing HMRα-BamHI, HMRα at the LEU2 locus, and deletions of HML and RE (XW711) (Wu and Haber 1996) to study the effect of the centromere on its function. We generated a PCR fragment containing the 750-bp RE linked to the gene coding for the resistance to nourseothricin (NAT) (Goldstein and McCusker 1999) and flanked by sequences corresponding to the centromeric position at which we wished to introduce RE. We used mixed primers containing the sequences necessary to amplify an RE–NAT fragment from pDB25 (kindly provided by Debra Bressan) and the centromeric target sequences. The following primers were used to target RE: to position 112,500, RErightRER1U (5′-aca ttc gta tct taa tta tat gta aac aaa att ata tga tag tta cag aac tcg aaa agt aaa taa aca a-3′) and RErightRER1L (5′-ttc ttc caa tat aag aga aga tga gta gaa gga gaa aat aat cac aat tag aag ctt cgt acg ctg cag g-3′); to position 115,000, CEN3RENATIB (5′-gtt agt att gtc atc gtt att ata tct caa tat ttc caa tta ttt acg gag aag ctt cgt cag ctg cag g-3′) and CEN3RENATIBII (5′-gtt agt att gtc atc gtt att ata tct caa tat ttc caa tta ttt acg gag aag ctt cgt acg ctg cag g-3′); to position 116,500, RERIGHTYCR1WI (5′-cat agt att cag aag agg aga gag gat gtg att tat caa tat ctg cat tac tcg aaa agt aaa taa aca a-3′) and RERIGHTYCR1WII (5′-tta tag ttt tct agt ggg aat atc ctc agt aga ggg caa agt tca tta ctg aag ctt cgt acg ctg cag g-3′). To move the centromere from its natural locus to position 117,000, a NheI–BamHI fragment from pMJ611 (kindly provided by Michael Lichten) bearing a deletion of the centromere marked by KanMX has been cotransformed with a long-flanking homology (LFH) PCR fragment bearing the centromere sequences linked to the HPH gene. This fragment was obtained as previously described (Amberg et al. 1995; Wach 1996) using pEC10 as a matrix and the primers LFcen31 (5′-tgt gcg gcc aga aaa atc tt-3′), LFcen32 (5′-cag atc cac tag tgg cct atg cta aat tat taa aat acg cca a-3′), LFcen33 (5′-gcg tac gaa gct tca gct gaa cgc tca ggg act ctt tct aat g-3′), and LFcen34 (5′-aat tgc ctt gaa aga aaa tg-3′).

Insertion of the URA3-RE cassette at position 74 kb on chromosome III in the FUS1-YCL026c intergenic region was performed by transformation of pRET1 digested with KpnI and XbaI.

Strains in which the genes involved in mating-type switching have been moved to chromosome V were constructed as follows. We started from a strain deleted for both donors and bearing the ade3∷GAL-HO allele (JKM179) (Lee et al. 1998). MAT has been deleted by transformation of a PCR fragment bearing the TRP1 gene flanked by proximal and distal sequence to MAT. This fragment was amplified from the plasmid CV13 with the primers matTRP1U (5′-gtt tcc att gga aaa gta aat cac tga ggt cag ttg cac cgc aca att cag cat tgg tga cta ttg agc acg-3′) and matTRP1L (5′-tac aaa tac ata gac ata aac aaa aga ggc aag tag ata agg gta tag ccc tca gta ata acc tat ttc tta gc-3′). Then, pSP1-HMR digested with NotI and pSP2-RE with BssHII were used to introduce, respectively, HMRα and RE on the left arm of chromosome V between the two ORFs YEL073C and RMD6 at position 9 kb. MATa was inserted at the ARG5,6 locus at position 296 kb by transformation with a PCR fragment containing MATa linked to HPH and flanked by the 50 first and last nucleotides of the ARG5,6 ORF. This PCR fragment was obtained using pEC2 as a matrix and the primers ARG5MATU (5′-atg cca tct gct agc tta ctc gtc tcg aca aag aga ctt aac gct tcc aaa atg ggc gat att ttg ata c-3′) and ARG5MATL (5′-tca gac acc aat aat ttt att ttc agg gat acc agc ata ctc tcc ata acg tga agc cga agg taa cta-3′). HMRα-BamHI was introduced between YER186C and YER187C at position 566 kb on the right arm of chromosome V by transformation with a PCR fragment consisting of HMRα-BamHI sequence linked to the KanMX4 gene and flanked by sequences corresponding to the 566-kb region. PCR was performed using pEC3 as a matrix and the primers khs1HMRU (5′-atc gtt aga ttg ttt gca ata att ttt gca ctg ctt ttg gta gcg tac tca agg gtc caa taa act tac t-3′) and khs1HMRL (5′-ccc tgg tac cgc tgc cac gat ttg atc gat gag cat gct cac att atg tat tcc tca ttc cgt tat atg t-3′). The leu2-K allele was introduced in place of the LEU2 cassette linked to the HMRα locus at position 9 kb by the integration and popping-out of a URA3-leu2-K plasmid as described previously (Lichten et al. 1987). Another LEU2 cassette was then introduced instead of the YER064C ORF. To do so, we created a PCR fragment containing the LEU2 cassette flanked by the 50 first and last base pairs of YER064C ORF, using a plasmid bearing the LEU2 cassette as a matrix and the primers yer064cLEU2U (5′-atg att gac gat act gag aac tcc aaa att cat ttg gaa ggt agc cat aag gaa tcc caa caa tta cat c-3′) and yer064cLEU2L (5′-tta ttt tga aat agt act gaa tgg ttt gga aaa ggt gga att ttc aat ctc ttg agg gaa ctt tca cca t-3′). The LEU2 cassette was mutated to leu2-R by integration and popping-out of a URA3-leu2-R plasmid as described previously (Lichten et al. 1987).

Insertion of RE near the URA3 locus on chromosome V in a strain carrying the leu2 ectopic recombination system on chromosome III and V (Wu and Haber 1995) was performed by transformation with an LFH–PCR fragment (Amberg et al. 1995; Wach 1996) containing RE linked to the NAT gene (Goldstein and McCusker 1999) and flanked by sequences surrounding position 99 kb, 17-kb distal from URA3. The primers used were LFYEL028W1 (5′-gta ata ttc cgt cac aag agg c-3′), LFYEL028W2 (5′-ttg ttt att tac ttt tcg agg gat ctg ata taa cta cga tgc-3′), LFYEL028W3 (5′-cct gca gcg tac gaa gct tcc gag ttt cag taa gta ctc agg-3′), and LFYEL028W4 (5′-gga gaa gtg acg ctt tta gag c-3′), and the matrix used was pDB25.

Insertion of RE near the ARG4 locus on chromosome VIII in a strain carrying the arg4 ectopic recombination system on chromosome VIII (arg4-RV) and V (arg4-Bg) (E. Coïc, A. Adjiri and F. Fabre, unpublished results) was performed by transformation with a PCR fragment containing RE linked to the NAT gene, flanked by 50-bp sequences of chromosome VIII sequences located 15 kb centromere-proximal to ARG4 at the YHR009C locus. The primers used were YHR009CRENATU (5′-gga ctt cca aga ttc taa aag att gtc caa ttt ctg gat tga gag ccc acc tcg aaa agt aaa taa aca a-3′) and YHR009CRENATL (5′-gaa aaa ttc acg gtc gtt aac ttt tag ttc agc aag aat agc gta agg tgg aag ctt cgt acg ctg cag g-3′).

Analysis of donor preference:

Quantification of donor preference on Southern blot was performed as previously described (Wu et al. 1998). Signals were quantified using ImageQuant V1.2 (Molecular Dynamics).

Measurement of spontaneous recombination rates:

Spontaneous formation of LEU⁺ by recombination was quantified by a fluctuation test based on a minimum of nine independent cultures of each strain, initiated from 200 cells and grown to saturation (Lea and Coulson 1949).

RESULTS

HMLα is still accessible to a MATa recipient located on the same chromosome arm when RE is deleted:

In MATα cells, exclusion of HMLα is observed at several positions on the left arm of chromosome III, suggesting that the entire left arm could constitute a closed domain inaccessible to MAT (Wu and Haber 1996). If so, the introduction of MAT into this domain should reestablish HML usage. However, it is also possible that a signal propagated along the left arm inhibits the opening of the chromatin at HML, preventing strand invasion at this donor. In that case, the relocation of MAT on the left arm should not change the status of donor preference. Therefore, we measured HMLα usage in strains where the MATa locus was moved from its original position to the left arm of chromosome III, at the HIS4 locus, ∼50 kb from HML (Figure 1). The relocation of MATa does not change the usage of HML (90%). However, deletion of RE shows only a weak decrease in HML usage (83%). This result is very different from the nearly complete exclusion of HML observed when RE is deleted in strains bearing MATa at its normal position. We conclude that HML can be used as a donor when RE is deleted if the recipient is located within the left arm. Therefore, RE seems to make the left arm accessible to other regions of the genome rather than control the opening of the chromatin of the HML donor itself.

Figure 1.—

HMLα is the preferred donor when MATa is located on the left arm in RE⁺ and RE⁻ strains. The coordinates of the loci involved in this experiment are indicated in kilobases. HML usage has been quantified at least three times on independent Southern blots. The means and standard errors are shown. (A) The MAT locus has been moved from its natural locus to the HIS4 locus (ECY359). (B) RE has been deleted (Δ) in a strain bearing MATa on the left arm at the HIS4 locus (ECY360). (C) RE has been reinserted near the right donor HMRα-BamHI (ECY362). (D) Representative Southern blots showing the products of MAT switching in strains bearing MAT on the left arm at the HIS4 locus. Genomic DNA digested with the BamHI and NspI restriction enzymes was probed with a MAT-distal DNA fragment. In addition to the parental MATa fragment, the products MATα (from HMLα) and MATα-BamHI (from HMRα-BamHI) are shown.

RE is portable on the left arm of chromosome III:

Wu and Haber (1996) showed that the activation of a donor decreased with its distance from RE (Figure 2B). They also show that RE was functional on the left arm but it was not clear whether this activity was still dependent on the natural copy of RE. We moved the 1.2-kb RE to position 74 kb on chromosome III, 45 kb to the right of its usual location (Figure 2E). The RE sequence was isolated from possible transcription occurring from one of the flanking genes, by inserting the strong CYC1 gene transcription terminator (Zaret and Sherman 1982) at both ends of RE, to prevent impinging transcripts from affecting RE activity. The new RE (RE₇₄) was introduced in a set of strains containing a donor sequence at different locations along chromosome III and deleted for the wild-type RE (RE₂₉). The new RE was also introduced in an isogenic set of strains containing the RE at the wild-type location.

Figure 2.—

Moving RE to another location on the left arm of chromosome III does not impair its function. (A) Physical map of chromosome III showing the loci involved in this experiment and their position in kilobases. This map is not to scale and the dotted lines represent interruptions in the sequence of the chromosome. (B) Usage of HMLα or of HMRα-LEU2 inserted at different positions on the left arm of chromosome III (41 kb, HIS4, or LEU2) in strains deleted for HMLα (Wu et al. 1996). The donors are represented by shaded horizontal bars. The vertical shaded bars represent the left donor usage as the percentage of a population of switched cells, determined on Southern blots. The solid square represents RE sequences. The solid circle represents the centromere. The horizontal solid rectangle represents MATa. (C) Same as B but the cells are deleted for RE (Δ) (Wu et al. 1996). (D) Same as B but an additional RE is inserted at position 74 kb. (E) Same as D but RE at 29 kb has been deleted. HML usage was measured at least three times on independent Southern blots except when HMRα-LEU2 was inserted at LEU2. Means and standard errors are shown. Strains used in these experiments: (B and C) Wu et al. (1996); (D) 12 kb: GFR18, 41 kb: GFR14, 66 kb: GFR15; (E) 12 kb: GFR22, 41 kb: GFR17, 66 kb: GFR24. (F) Representative Southern blot analysis of the product of mating-type switching in strains bearing either HMLα or (in strains deleted for HMLα) the HMRα donor inserted at different positions on the left arm of chromosome III (Wu et al. 1996). HML usage was quantified on Southern blots after digestion of the genomic DNA with the BamHI and StyI restriction enzymes. A MAT-distal probe detects the products MATα and MATα-BamHI, as well as the parental MATa sequence and a second MAT distal fragment. The extra band observed in the GFR17 corresponds to the unrepaired HO-cut fragment in a fraction of the population.

In strains deleted for the wild-type RE (RE₂₉) and containing the new RE (RE₇₄), activation of the different donors is distance dependent (Figure 2E). The donors closest to RE₇₄ show the greater preference (77 and 76%). There is a decrease in activation for the donor located at 41 kb (59%) and a further decrease for the donor located at 12 kb (23%). It should be noted that when RE is absent from both positions, HML usage is ∼10% (Figure 2C). In addition, strains containing an RE₂₉ and an RE₇₄ were constructed (Figure 2D). Compared to the strain bearing only RE₂₉, an increase in activation of the donor at HIS4 was observed, due to the nearby presence of RE₇₄. We conclude that RE is able to act in a different chromosomal context and that it can activate the use of a donor sequence located at different positions on the left arm of chromosome III, in a distance-dependent manner, spanning ∼50 kb. RE is therefore a context-independent activating element, at least when located on the left arm of chromosome III.

The centromere of chromosome III constitutes a barrier to RE function:

If the two arms of chromosome III constitute two separate domains whose accessibility depends on mating type, the centromere might be able to inhibit RE activity, for example, by preventing RE influence from propagating along the two arms of the chromosome. To test the effect of the centromere on RE activity, we used strains in which RE was relocated close to the centromere. HML was deleted and a copy of HMRα was introduced into the LEU2 locus as the left-arm donor. The LEU2 locus was chosen so that the donor maintains an ∼25-kb distance from RE. We moved RE between the centromere and the donor inserted at LEU2 or on the right arm (Figure 3). When RE is moved to position 112.5 kb, on the same arm as the donor, the left-arm donor usage is 50%. This is significantly higher than the 12% observed in absence of RE (Wu and Haber 1996) (Figure 2). However, when RE is moved to the right arm, 500 bp from the centromere, the left-arm donor usage drops to 26%. This effect is also observed when RE is moved 2 kb from the centromere (32.5%, Figure 3B). We conclude that the centromere partially inhibits RE, but only when RE and the donor are separated by the centromere. It is possible that placing RE near a centromere intrinsically reduces its effect, as the activation of leu2∷HMR is less than when RE is placed a similar distance distal to the donor, at 74 kb.

Figure 3.—

The centromere constitutes a barrier to RE activity. (A) Physical map of chromosome III showing the loci involved in this experiment and their positions in kilobases on the chromosome. (B) Effect of the centromere on RE function. The percentage of HMRα usage (vertical shaded bar) introduced at the LEU2 locus in strains deleted for HML has been measured with RE relocated to three different positions around the centromere: 112.5 kb (ECY139), 115 kb (ECY137), 116.5 kb (ECY138). ECY135 and ECY136 have HML at its normal location and with RE present or deleted, respectively. The solid circle represents the centromere. The solid squares represent RE. (C) To assess the negative effect of the centromere on RE function, the centromere has been moved to the right of RE when at position 116.5 kb (ECY169). Donor usage was quantified at least three times on independent Southern blots; means and standard errors are shown. (D) Representative Southern blot analysis of the product of mating-type switching in strains bearing the RE in the region of the centromere. Genomic DNA was digested with BamHI and HindIII restriction enzymes. Using a Yα fragment as a probe leads to the detection of the products MATα and MATα-BamHI.

The decrease in RE activity is probably not linked to the increased distance between RE and the left-arm copy of HMR, which goes only from 21.5 to 24 or 25.5 kb. However, the different sequence context surrounding RE could be responsible for the difference observed. To be sure that the decrease in RE activity is linked to the presence of the centromere between RE and the nearby donor, we simultaneously deleted CEN3 and inserted a cloned centromere to the right of RE in the strain bearing RE at position kb 116.5. This relocation places RE on the same arm as the nearby donor (Figure 3C). Usage of the left-arm donor in this strain is increased to 56%. We conclude that the centromere inhibits RE activity when it separates RE from the nearby donor. The centromere seems to be an important but not decisive factor to isolate the right arm from the action of RE and to ensure the exclusive activation of the left arm in MATa cells.

The exclusion of HMR observed when MAT is on the left arm can be suppressed only partially by RE:

When MAT is on the left arm, HMR is excluded or outcompeted as a donor. This may be the consequence of a mating type-independent tethering of HMR in the nucleus. It has been shown previously that the usage of HMR is only weakly regulated by the mating type (Wu et al. 1997) and that the stronger HMR usage in MATα cells occurs by default (Wu et al. 1996). We asked if the introduction of RE near HMRα-BamHI in a reΔ strain bearing MAT on the left arm would have the same strong effect on HMRα-BamHI usage that it does on the left arm. Quantification of donor preference in such a strain shows that HML usage is decreased from 83 to 47% (Figure 1C). Thus, RE adjacent to HMRα-BamHI improves its use when MAT is on the left arm, but this activation is not nearly as strong as the preferential use of HML when MAT is on the right arm. The difference does not reside in the sequence differences between HMR and HML, as it has been shown previously that RE increases the left arm donor usage to the same extent when HMR is inserted in place of HML (Wu and Haber 1996). It is possible that the reduced activation of HMR reflects an inherent constraint on the movement of the right arm of chromosome III or that RE does not have appropriate targets to fully activate recombination.

tRNA^Thr linked to HMR does not affect donor preference:

Several families of tRNA genes colocalize with 5S ribosomal DNA and U14 small nucleolar RNA at the nucleolus; these associations could have an important impact on chromosome architecture (Thompson et al. 2003). The tRNA gene tT(AGU)C is located 1163 bp distal to the HMR locus and could be responsible for the tethering of HMR, although the localization of the tRNA^Thr family of genes was not studied specifically in the work of Thompson et al. (2003). We measured donor preference in MATa and MATα strains bearing a deletion of the tRNA^Thr gene tT(AGU)C. We found that donor preference was not affected in either mating type (data not shown), indicating that the putative localization of sequences near HMR to the nucleolus does not affect donor preference.

RE is not very efficient in activating a trans-located donor to compete with a cis-located donor:

When HML is deleted from chromosome III and HMRα introduced 9 kb from the left telomere of chromosome V (Figure 4A) this donor is very poorly used to switch MATa cells. This result confirms that RE cannot act on a donor located in trans (Wu and Haber 1996). The introduction of a copy of RE near the donor at position 24 kb on chromosome V increases its usage to 30%, showing that RE can activate a donor located outside of chromosome III. However, this activation is less strong than on the left arm of chromosome III. This could mean that RE cannot fully overcome the constraint imposed by a donor located on another chromosome to compete with a donor located on the same chromosome that carries MAT. It is also possible that RE does not have appropriate targets on this chromosome to promote its function. We decided to move MAT and the right donor HMRα-BamHI on chromosome V to understand better the influence of the position of the different mating-type cassettes and the activation conferred by RE.

Figure 4.—

Effect of RE on donors located on chromosome V. (A) RE activates poorly a trans-located donor on chromosome V. Donor preference has been measured in strains derived from AN406 in which HMLα has been deleted from chromosome III and integrated near the left telomere of chromosome V at position 9 kb (AN409). The effect of RE on the donor located on chromosome V has been quantified in a strain in which RE has been introduced on chromosome V at position 24 kb (GF26). A representative Southern blot analysis, similar to that carried out in Figure 2, is shown. (B) RE is strongly active on the same chromosome V-located donor when the mating-type switching apparatus is moved entirely on chromosome V. In this experiment, HMLα, MAT, and HMRa have been deleted from chromosome III. HMRα has been reintroduced at position 9 kb on chromosome V, MATa at the ARG5,6 locus, and HMRα-BamHI at the KHS1 locus. HMRα usage has been measured in strains with (ECY429) or without (ECY427) RE at position 24 kb. A representative Southern blot analysis of the product of MAT switching is shown. Genomic DNA was analyzed as in Figure 3. In both experiments, donor usage was measured at least three times on independent Southern blots; means and standard errors are shown.

RE is functional when the mating type switching apparatus is moved on chromosome V:

To test if RE requires chromosome III-specific sequences to activate the right arm, we moved HML, MAT, and HMR from chromosome III to a similar position on chromosome V (Figure 4B), which is about twice the size of chromosome III. The original RE was not deleted because we have previously shown that it could not activate a donor located on another chromosome III (Wu and Haber 1996 and data cited above). HMRα was integrated near the left telomere around position 9 kb. We have already shown that the replacement of the HMLα sequences by those of HMRα does not change donor preference on chromosome III (Wu and Haber 1996). HMRα-BamHI was integrated 10 kb from the end of the right arm around position 566 kb. MATa was introduced on the right arm in replacement of the ARG5,6 gene around position 296 kb.

Quantification of donor preference in such a strain, which does not have RE on chromosome V, shows 40% use of the left-arm donor (Figure 4B). Here, the choice of the donor appears to be nearly random. The sequences responsible for the exclusion of the left donor in the absence of RE are apparently specific for chromosome III. When RE is inserted near the left donor at position 24 kb on chromosome V, usage of the nearby donor increases to 93%. This result means that RE can strongly activate an adjacent donor located on another chromosome. It is possible that both HMR and HML carry a tethering sequence that RE can counteract.

RE depends on chromosome III sequences for its function:

The activity of RE on chromosome V could be related to the presence of responsive sequences within HMR on the left arm of this chromosome. To test if RE does not depend on any chromosome III sequences for its function, we asked if RE could increase the spontaneous rate of recombination when it was inserted near a leu2-K allele inserted in chromosome V but without the proximity of HM sequences. Wu and Haber (1995) previously showed that in MATa but not MATα cells the wild-type RE on chromosome III could increase the rate of spontaneous recombination between two leu2 alleles located in ectopic positions, one replacing HML and the other one at the URA3 locus on chromosome V. We confirmed here that the presence of RE at its natural position increases the rate of recombination fivefold compared to the rate observed when RE is deleted (Figure 5A). However, a strain bearing RE on chromosome V, 17 kb centromere-distal to leu2-K at the URA3 locus, and deleted for the resident RE, does not show any induction of the rate of recombination compared to a strain without RE.

Figure 5.—

RE depends on chromosome III sequences for its function. (A) Effect of RE on the rates of spontaneous recombination between two leu2 heteroalleles, one located in place of HML on chromosome III and the other integrated at the URA3 locus on chromosome V. Spontaneous rates of recombination have been measured in strains bearing RE at its natural locus (XW485, top), deleted for RE (ECY382, middle), or bearing RE near the leu2 allele located on chromosome V (ECY389, bottom). The proportion of the use of leu2-K as a donor among LEU⁺ recombinants is also indicated. (B) RE does not induce spontaneous recombination rate outside of chromosome III even when located near a good donor. The spontaneous recombination rate between two arg4 alleles, located at the ARG4 locus on chromosome VIII and at the URA3 locus on chromosome V, is shown in a strain in which RE has been inserted near ARG4 (ECY272) and in the control without RE (ECY269). The proportion of the usage of arg4-Bg as the donor is shown for strain ECY269.

In our previous experiments, we showed that the leu2-K allele, located near MAT, was preferentially gene converted when recombining with leu2-R inserted in place of HML (Wu and Haber 1995); RE also stimulated spontaneous leu2 recombination when leu2-K was inserted on chromosome V. These results supported the argument that RE was not acting to enhance the initiation of recombination at a nearby sequence. Rather, RE would enable long-distance interactions between two homologous sequences that would result in Leu⁺ recombinants by possibly preventing invisible sister chromatid repair at the leu2-K recipient. We find that the same bias is seen in the experiments we report here. We measured the proportion of leu2-K used as a donor among selected LEU⁺ recombinants (Figure 5A). We determined which allele was converted by popping out the ura3-52-LEU2-URA3 integration on chromosome V, selecting for Ura⁻ colonies on 5-fluoro-orotic acid (5-FOA) medium (Boeke et al. 1987). If the ura3-52-LEU2-URA3 integration carried the converted Leu⁺ locus, then the Ura⁻ derivatives also would become Leu⁻. If hml∷leu2-R had been converted, the strain would remain Leu⁺. In the strain carrying RE at its natural locus, 6/102 (5%) remained Leu⁺ indicating that the leu2-K allele on chromosome V was most often converted. This proportion was not statistically different in strains deleted for RE or carrying RE on chromosome V: 10/97 (10%) and 11/59 (19%) remained Leu⁺, respectively, indicating that the leu2-K allele was usually converted. It seems that the spontaneous lesions that initiate spontaneous recombination are more frequent around the leu2-K allele surrounded by pBR322 sequences.

To pursue further whether RE would activate spontaneous ectopic recombination, we used a second system in which two arg4 alleles were both located outside of chromosome III and each was used frequently as donors in generating Arg⁺ recombinants. We inserted RE 15 kb centromere-proximal to the arg4-RV allele at its normal chromosome location. This strain also carries arg4-Bg at the URA3 locus on chromosome V (Figure 5B). These two alleles are converted with equal frequency among the Arg⁺ recombinants (E. Coïc, A. Adjiri and F. Fabre, unpublished results; Figure 5B). The frequency of recombinants was not affected by the presence of RE in a MATa cell. Thus, even when RE is inserted near a good donor, we were unable to detect any increase in the spontaneous formation of Arg⁺ recombinants. Therefore, we conclude that the RE activity detected in the strain bearing all the genetic elements involved in mating-type switching on chromosome V was dependent on some sequences within the inserted HMR locus. We can also conclude the left arm of chromosome III must harbor other sequences that are sufficient to support RE activity, since RE located on chromosome III can activate ectopic recombination even if HML is deleted.

DISCUSSION

Our genetic studies of donor preference lead us to several new conclusions about chromosome architecture, nuclear organization, and RE function. First, our experiments rule out hypotheses suggesting that donor preference is established through the mating type-regulated changes in the chromatin structure of the donors. When MAT is moved to the left arm of chromosome III in a strain deleted for RE, HML becomes the preferred donor. Thus HML can be used efficiently as a donor even when RE is deleted, which means that this locus is still accessible for strand invasion. We suggest that the structure of the left arm of chromosome III in the nucleus constitutes an isolated domain, making it inaccessible to sequences on the right arm during the process of homology search. The strong use of HML when MATα is on the left arm may be a combination of HML's proximity and that the left-arm domain now excludes HMR from interacting with MAT in its novel location. The idea that the left arm is tethered or folded into an isolated domain is illustrated by using chromosomally tethered fluorescent proteins and deconvolution microscopy to show that HML's motion is strongly constrained in both MATα and RE-deleted MATa strains, compared with MATa (Bressan et al. 2004). Additionally, the three-dimensional configuration of MAT, HML, and HMR is mating-type dependent, the distance between HML and the other cassettes being greater in MATα cells (Bressan et al. 2004). We interpret these results to say that there is a constitutive tethering or folding of the left arm, which is relieved in MATa cells through the action of RE.

RE is portable. It can act over ∼50 kb to increase donor usage even when it is placed close to the centromere of chromosome III. The presence of the centromere between RE and a donor reduces RE's effect, but the centromere is not a formidable barrier.

The placement of all the genetic elements involved in mating-type switching onto chromosome V also shows that the left arm of chromosome III possesses some specific exclusion capability. In strains where mating-type switching occurs on chromosome V, donor usage is nearly random. Therefore, the architecture of chromosome V does not inherently lead to the exclusion of one of the donors. On chromosome V, the distance between each donor and MAT (300 and 260 kb) are similar, while on chromosome III HML is significantly more distant from MAT (∼200 kb) than from HMR (∼90 kb). This difference could explain the absence of exclusion of the left donor on chromosome V. However, we know from previous experiments that on chromosome III when the donors are placed in a way that the distance to MAT is similar (e.g., when HML is moved near LEU2), the left donor is still excluded (Wu and Haber 1996); being farther from MAT is therefore unlikely to be the cause of the exclusion of a donor. We believe donor exclusion on chromosome III-L must be linked to yet undescribed sequences located on that arm.

In a chromosome III translocation, even when HML is adjacent to RE in the context of chromosome III sequences, this donor located in trans is not able to compete effectively with an HMR donor located in cis (Haber et al. 1981; Wu et al. 1997). Although RE facilitates more frequent encounters between allelic sites on homologous chromosomes in MATa diploids (Bressan et al. 2004), Houston et al. (2004) reported that RE did not stimulate interchromosomal recombination between heteroalleles. The difficulty of using a donor located in trans suggests that chromosomes could define territories in the yeast nucleus without being totally isolated from each other, as it has been described in other organisms (reviewed by Spector 2003; Taddei et al. 2004). However, it is known that recombination rates are similar between heteroalleles placed in allelic or in ectopic positions (Lichten and Haber 1989) and that RE when located on chromosome III can activate recombination fivefold between two ectopic alleles located on different chromosomes (Wu and Haber 1995) (Figure 5). We believe that competition assays between two donors for a single recipient can detect more subtle differences in the ability of different regions of the genome to interact than a recombination system involving only two partners. For example, because the DNA checkpoint can restrain cells from dividing until damage is repaired, even an inherently poor donor may eventually succeed much more often in isolation than when this donor is in competition with an efficient donor.

RE function is independent of its position on the left arm of chromosome III (Figure 2) and can activate a cis-located copy of HMR on another chromosome (Figure 4). However, a copy of RE located on chromosome V cannot increase LEU2 or ARG4 ectopic recombination in the absence of HMR sequences (Figure 5). One interpretation of this result is that the cloned segment of HMR carries a tethering sequence that RE can counteract. On chromosome III there must be constraining sequences other than simply those in or immediately around HML or HMR, as RE can activate a leu2-k allele replacing HML and surrounding sequences on chromosome III (Figure 5). These additional constraining sequences, as well as the precise identity of those within HMR, still have to be identified. We have shown that gene silencing imposed by Sir2/Sir3/Sir4 is not responsible for reducing the left-arm usage in the leu2 heteroallele assay (Wu and Haber 1995; E. Coïc and J. E. Haber, unpublished results), ruling out the involvement of the silencer sequences E and I.

In summary, the architecture of chromosome III is the foundation of donor preference regulation. Our study shows that the two arms constitute two independent domains, inaccessible from each other in the absence of RE activity. This isolation is consistent with the idea that there is specific tethering of the left arm. This organization of the two arms in domains is different from the organization of the chromosome in territories, since REs increase intrachromosomal interaction more strongly. Finally, we show that RE works through the entire left arm of chromosome III, a mechanism that is dependent on other sequences of the chromosome. The determination of these sequences should give us new insights about RE function.