Protosilencers in Saccharomyces cerevisiae Subtelomeric Regions

Corresponding author: Eric Gilson, Laboratoire de Biologie Moléculaire et Cellulaire, UMR5665 CNRS/ENSL, Ecole Normale Supérieure de Lyon, 46 Allée d'Italie, 69364 Lyon Cedex 07, France., Eric.Gilson{at}ens-lyon.fr (E-mail)

Communicating editor: S. HENIKOFF

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Saccharomyces cerevisiae subtelomeric repeats contain silencing elements such as the core X sequence, which is present at all chromosome ends. When transplaced at HML, core X can enhance the action of a distant silencer without acting as a silencer on its own, thus fulfilling the functional definition of a protosilencer. Here we show that an ACS motif and an Abf1p-binding site participate in the silencing capacity of core X and that their effects are additive. In addition, in a variety of settings, core X was found to bring about substantial gene repression only when a low level of silencing was already detectable in its absence. Adjoining an X-STAR sequence, which naturally abuts core X in subtelomeric regions, did not improve the silencing capacity of core X. We propose that protosilencers play a major role in a variety of silencing phenomena, as is the case for core X, which acts as a silencing relay, prolonging silencing propagation away from telomeres.

SILENCING is a form of repression that is not promoter specific and can extend over large chromosomal regions. A classical example is the position effect variegation (PEV) of gene expression due to proximity to heterochromatin (reviewed in HENIKOFF 2000 ). In Saccharomyces cerevisiae, silencing is found at silent mating-type loci (HM) within rDNA repeats and in subtelomeric regions [telomere position effect (TPE)]. Key components of silent chromatin in S. cerevisiae include the complex of silent information regulators Sir2p, Sir3p, and Sir4p that is formed through homo- and heterotypic interactions, as well as histones H3 and H4 (for a review, see GARTENBERG 2000 ). Both establishment and maintenance of silencing depend on cis-acting elements known as silencers that nucleate assembly of a SIR complex. HM loci are flanked by the E and I silencers, composed of specific combinations of binding sites for Abf1p, the origin replication complex (ORC), and Rap1p. These act as a "surface" for Sir interaction, either directly or through Sir1p, as in the case of Orc (TRIOLO and STERNGLANZ 1996 ). At telomeres, the SIR complex interacts with tandemly reiterated Rap1p molecules bound to telomere-repeat sequences. Silencing is thought to result from the subsequent "spreading" of the SIR complex across the silenced locus, involving interactions between H3/H4 and Sir3p/Sir4p. Isolated binding sites for silencer binding proteins, although unable to initiate silencing at a silencer-free HM locus, have been shown to cooperatively interact with intact, distant silencers to strengthen silencing (BOSCHERON et al. 1996 ). They have been called protosilencers by analogy with protoenhancers. Recently, protosilencers were shown to autonomously maintain silencing in G1-arrested cells but not in cycling cells (CHENG and GARTENBERG 2000 ). This contrasts with bona fide silencers that stably maintain silencing during the cell cycle.

A new type of silencing element was recently described in S. cerevisiae subtelomeric X and Y' repeats. It corresponds to the invariant part of X elements, called core X, and to an internal segment of Y' (LOUIS et al. 1994 ; LOUIS 1995 ; FOUREL et al. 1999 ; PRYDE and LOUIS 1999 ). These elements act as relays prolonging the inward propagation of telomeric silencing. Albeit unable to induce silencing on their own when inserted at HML, they can reinforce the effect of a weak silencer, thus behaving like protosilencers (FOUREL et al. 1999 ). Interestingly, these silencing elements may be involved in the SIR-dependent inhibition of replication initiation from subtelomeric ARS consensus sequences (ACSs; STEVENSON and GOTTSCHLING 1999 ) and in the silencing of Ty5-1 retrotransposon at telomere III-L (VEGA-PALAS et al. 1997 ). Different sequence motifs that may play a role in silencing have been previously recognized in core X (LOUIS 1995 ). These are: (i) an ORC-binding site (ACS), also found in Y'; (ii) an Abf1p-binding site; (iii) a GC-rich sequence that is nearly identical to the meiosis specific regulatory sequence URS1, which allows tethering of the Rpd3p histone deacetylase through binding of the Ume6p factor (KADOSH and STRUHL 1997 ). Subtelomeric silencing elements are separated from telomere repeats by insulator elements named STARs (for subtelomeric antisilencing region) that can protect a reporter gene from silencing when interposed between it and a silencer (FOUREL et al. 1999 ). Within X elements, the STAR sequence (named X-STAR) coincides with the more variable STR region (LOUIS 1995 ; FOUREL et al. 1999 ). Tbf1p and Reb1p proteins bind to multiple sites within STARS and likely contribute an essential role in their activity (FOUREL et al. 1999 , FOUREL et al. 2001 ; KOERING et al. 2000 ; Fig 1A).

View larger version (50K): In this window In a new window Download PPT slide   Figure 1. Core X silencing activity involves redundant elements. (A) Schematic representation of S. cerevisiae chromosome II right end and chromosome XI left end, with emphasis on the X subtelomeric element. Tandem arrowheads, TG1-3n telomeric repeats. Core X was arbitrarily subdivided into two 300-bp segments, named X1 and X2, and represented as hatched boxes. The STR sequence is shown as a stippled box. The positions of an ACS and of binding sites for Abf1p, Reb1p, and Tbf1p are indicated by symbols. (B) Left: Schematic representation of the left end of chromosome VII after recombination at the ADH4 locus, telomere seeding, and chromosome fragmentation using pURTEL-based constructs (FOUREL et al. 1999 ), including fragments from the telomere II-R X element (see MATERIALS AND METHODS). Arrows indicate the 5' to 3' direction of URA3 transcription, and the X element reversed orientation in strain GF51. In strain GF77, the STR and the X1 fragments are artificially placed adjacently, which is illustrated by a space separating the corresponding boxes. Shown is a representative experiment. At least three independent transformants were analyzed for each strain, and for each of them cultures started from at least two independent colonies were serially diluted and grown on synthetic complete (SC), on SC + 5-FOA, and on SC - uracil (SC-U) medium. SC + 5-FOA allows only growth of cells that do not express URA3. Right: Each diamond indicates the ratio of colonies growing on either SC + 5-FOA or SC-U vs. SC for a single culture, considered as measurements of URA3 repression and expression, respectively. The histogram bar represents the average of the values obtained for a given strain and medium, as indicated above the panel.

The efficiency of silencers upon ectopic chromosomal insertion decreases as the distance from telomeres increases (MAILLET et al. 1996 ). However, their silencing capacity can be restored at a long distance away from telomeres either by overexpressing Sir proteins or by decreasing the ability of Sir proteins to bind telomeres. This substantiates the view that telomeres play the role of reservoirs of silent chromatin components (MAILLET et al. 1996 ; MARCAND et al. 1996 ; MARTIN et al. 1999 ). Core X seemingly functions as a silencer in the vicinity of a telomere but not at HML, i.e., 13 kb away from telomere III-L. Core X was therefore speculated to be a weak silencer that is potentiated in its natural telomeric environment by the specific enrichment in Sir proteins. Here we addressed this hypothesis by dissecting core X silencing activity and by asking whether core X might behave as a bona fide silencer under other circumstances.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plasmid constructions:
Molecular biology techniques were performed as described in SAMBROOK et al. 1989 . The Expand HiFi system (Boehringer Mannheim) was employed to carry out PCRs according to the manufacturer's recommendations.

Plasmids used in end replacements at telomere VII-L or to modify the HML locus were obtained by inserting various fragments at the BamHI site of pURTEL or at the BclI site or pE-i, respectively, as described in FOUREL et al. 1999 . GF1, GF3, GF6, GF19, GF21, GF25, GF31, GF48, GF53, GF62, GF63, GF64, GF65, and GF72 were previously obtained and analyzed in FOUREL et al. 1999 . PCR amplification of natural subtelomeric regions was carried out using plasmids p19X10 and p89H9 [containing, respectively, chromosome II-R and XI-L end (LOUIS and BORTS 1995 ), kindly provided by E. Louis] as templates and oligonucleotides listed in Table 1. The PCR amplification reactions used to generate the various constructs were performed with the following pairs of primers: STR(IIR), 19C1 + 19C2; X1(IIR), 19A1 + 19A2; X2(IIR), 19B1 + 19B2; coreX(IIR), 19A1 + 19B2; X(IIR), 19A1 + 19C2; X2-STR(IIR), 19B1 + 19C2; STR(XI-L), STRD-Ba + STRA-BgBis; Rap1p (RAP)/ STR(XI-L), STRD-RAP + STRA-BgBis. Note that the limits of the core X fragment were defined as in LOUIS et al. 1994 , i.e., as corresponding to the highly conserved portion of 450–475 bp (PRYDE et al. 1995 ) plus a telomere distal sequence up to 600 bp. The construct used to generate GF76 strain was linearized with BamHI and the STR(IIR) fragment was inserted to produce GF77 strain. An ACS motif (ACS) and a binding site for Abf1p (ABF) were obtained through annealing of ACS- and ACS+ and ABF- and ABF+ oligonucleotides, respectively, and inserted at the BamHI site of pURTEL, and of the construct used to generate GF3. The Rap1p-binding site in RAP corresponds to that found in UAS (GIESMAN et al. 1991 ). The oligonucleotide ACS contains a perfect ARS consensus sequence and the indicated flanking region is identical to the flanking sequence of the ACS in HMR-E. The oligonucleotide ABF contains a perfect Abf1p-binding site consensus and the flanking sequence is identical to that used in the synthetic silencer (MCNALLY and RINE 1991 ).

Replicative plasmids harboring a modified HML locus were obtained through PCR amplification of this locus from the genomic DNA of the corresponding strains using the HML-E and HML-I primers and insertion at the NotI site of pRS315 (SIKORSKI and HIETER 1989 ). These primers contain a terminal BsaI site and adjacent sequence engineered to create NotI compatible ends after BsaI digestion of the PCR products.

p2µASir3His and the corresponding control p2µAHis were obtained from p2µASir3 and pAAH5 (MAILLET et al. 1996 ) by inserting a HIS3 fragment at the unique BglII and PvuII sites, respectively. These plasmids allow both leucine and histidine prototrophy.

Yeast strains, media, and methods:
Genetic manipulations and growth of yeast are as described in FOUREL et al. 1999 . Yeast strains described in this article harboring a modified VII-L telomere are derivatives of W303-1a (MATa ade2-1 ura3-1 his3-11,15 leu2-3,112 trp1-1 can1-100) and those with a modified HML locus have a S150-2B background (MATa leu2-3,112 ura3-52 trp1-289 his3D gal2 gal4:LEU2). Deletion in the HDF1 gene was obtained by replacement of the full-length open reading frames (ORFs) by the KanMX4 cassette as described in PRYDE and LOUIS (1999). A large number of control strains was previously described in FOUREL et al. 1999 . All telomere fragmentation and loci replacement were confirmed by Southern blot analysis.

For overexpressing Sir3p and/or Sir4p, the appropriate strains were transformed with the high-copy plasmids p2µASir3His (see above) and pFP320 (MAILLET et al. 1996 ) or the corresponding control plasmids p2µAHis and pRS424 and grown on medium lacking histidine and tryptophane. Yeast strains harboring HML-related replicative plasmids were obtained through transformation of W303-1A with the corresponding plasmids and selection on SC- leucine medium.

Analysis of URA3 expression:
The variegated expression of URA3 was monitored essentially as described in FOUREL et al. 1999 by spotting 10 µl of serial dilutions of overnight culture onto SC, SC - uracil (SC-U), and SC + 5-FOA (1 g/liter) and allowing growth for 3 days at 30°. In situations in which particularly low silencing levels were considered (Fig 4 and Fig 5), 1-ml cultures were plated on 10-cm 5-fluoroorotic acid (5-FOA)-containing dishes of appropriate selective medium and assayed classically by spotting 10-µl dilutions on corresponding 5-FOA-lacking medium to estimate the total cell number. After growth for 4 days at 30°, three representative colonies were picked from dishes harboring colonies, which were sometimes of varying size, and serial 10-µl dilutions were spotted on SC and SC-U to address possible mutation of the reporter URA3 gene. A total of 12–24 cultures seeded with independent transformants were assayed for each strain, and the average value indicated by the histogram bar was calculated as the cumulation of 5-FOA-resistant colonies harboring a functional URA3 reporter among the estimated total number of colonies for all 12–24 cultures.

View larger version (59K): In this window In a new window Download PPT slide   Figure 2. Core X silencing activity is recapitulated by single binding sites for either Rap1p or Abf1p proteins or the ORC complex. The ACS and Abf1p- and Rap1-binding sites inserted in the subtelomeric position at a fragmented VII-L telomere included consensus sequences and were of distinct origins (see MATERIALS AND METHODS). (A) The isolated STR segment in strains GF3 and GF78–82 is derived from chromosome XI left telomere. (B) Deletion of the HDF1 ORF was carried out in strains GF1, GF53, GF79, and GF81 to obtain strains GF66, GF67, GF83, and GF84, respectively, as described in MATERIALS AND METHODS. Details are otherwise the same as in the legend for Fig 1.

View larger version (77K): In this window In a new window Download PPT slide   Figure 3. Adjoining X-STAR sequences do not improve core X silencing capacity at HML. The HML-E and I silencers are drawn as boxes, solid when intact and open when carrying the e1 and i (I-242) deletions (MAHONEY and BROACH 1989 ; BOSCHERON et al. 1996 ). Most of the HML locus was replaced by URA3. Fragments derived from the telomere II-R X element were inserted between HML-E and URA3. When no colony grew on SC + 5-FOA from 10 µl of undiluted culture, the fraction of 5-FOA-resistant cells was estimated to be <0.0005%, as indicated. Details are otherwise the same as in the legend for Fig 1.

View larger version (55K): In this window In a new window Download PPT slide   Figure 4. Overexpression of Sir3p and Sir4p enables core X silencing activity at HML. Sir3p and Sir4p were overexpressed from high-copy plasmids, and control strains carried the corresponding parental vectors. Silencing was assayed by a particularly sensitive method and integrity of the URA3 reporter gene was verified prior to taking into account a 5-FOA-resistant colony, as described in MATERIALS AND METHODS. The diamonds indicate the ratio of 5-FOA-resistant colonies for each of the 12–24 individual transformants assayed. The details are otherwise the same as in the legend for Fig 3.

View larger version (64K): In this window In a new window Download PPT slide   Figure 5. Plasmid-borne core X behaves as a protosilencer. The HML locus from appropriate yeast strains described in Fig 3 was amplified and inserted in the pRS315 vector, represented by an oval, giving rise to the plasmids represented on the left. Details are otherwise the same as in the legends for Fig 3 and Fig 4.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Core X silencing activity involves redundant elements:
Insertion of an X-STAR sequence (Fig 1A) between telomere repeats and a URA3 reporter gene strikingly impedes its silencing, as indicated both by a decrease in the frequency of 5-FOA-resistant colonies—a compound toxic for cells expressing a functional URA3 gene product (BOEKE et al. 1984 )—and an increase of the ratio of colonies able to grow to a normal size on a medium lacking uracil (SC-U; Fig 1B, compare lines 1 and 3, 2 and 4). This decrease in silencing is due to the insulator activity of X-STAR, also called STAR activity, which is itself counteracted by the silencing activity of core X upon insertion of a full-length X element, thus restoring predominant URA3 silencing (Fig 1B, lines 7–8) (FOUREL et al. 1999 ).

In an attempt to identify elements responsible for core X silencing capacity, the core X sequence of II-R telomere was arbitrarily dissected into two 0.3-kb fragments (X1 and X2, Fig 1A). Strikingly, the combination of either X1 or X2 with the X-STAR region restored substantial silencing as compared to the insertion of the X-STAR region by itself, thus qualitatively recapitulating the activity of the whole core X element (Fig 1B, compare lines 3, 7, 11, and 15 with lines 2, 4, 8, 12, and 16). Silencing levels were nevertheless not quite as high as when full-length core X sequences were provided, suggesting that each of X1 and X2 do not recapitulate all the silencing capacities of full-length core X elements. This conclusion is strengthened by the fact that the insertion of full-length core X displaces URA3 0.3 kb farther away from the telomere and would thus theoretically decrease the influence of TPE on URA3 expression as compared with the insertion of either X1 or X2. Of note, the X2 fragment clearly strengthens telomeric silencing upon insertion between URA3 and telomere repeats. This does not hold true for X1 and core X fragments, as better visualized on SC-U (Fig 1B, compare lines 2, 6, 10, and 14). Similar results were obtained using subtelomeric sequences derived from chromosome XI-L end (data not shown). We conclude that core X consists of at least two independent functional elements, which cooperate to yield full core X silencing activity.

Core X silencing activity is recapitulated by elementary protosilencers—single binding sites for Rap1p, Abf1p, or the ORC complex:
The presence of putative ORC- and Abf1p-binding sites was previously noted through sequence inspection of the X1 and X2 fragments, respectively (LOUIS 1995 ; Fig 1A). These elements, together with Rap1-binding sites, constitute the building blocks of silencers at the silent mating-type cassettes. In addition, silencing at HM loci and at telomeres harboring an X element were found to share many properties, such as a partial requirement for Sir1p and only a weak dependence on the HDF1 gene product (FOUREL et al. 1999 ). The ORC- and Abf1p-binding sites therefore emerged as strong candidates in mediating core X silencing activity.

Oligonucleotides, including either an ACS or an Abf1p-binding site (named ABF), were found to substitute for core X in restoring significant silencing levels (Fig 2A, compare GF78 and GF80 to GF1, GF3, and GF6). However, the silencing effect was not as strong as that of core X and was actually slightly weaker than that of the corresponding X2 and X1 fragments. Significantly, whereas insertion of the X2 fragment by itself resulted in enhanced TPE (see Fig 1B, GF74), this was not the case with the ABF oligonucleotide (Fig 2A, GF81). Conversely, abutting an ACS oligonucleotide to telomere repeats improved TPE, as better visualized on SC-U (Fig 2A, compare lines 2 and 8). Core X silencing activity was also recapitulated by targeting Gal4 chimera containing either Orc1p N terminus or Abf1p transcription activation domain, to UASg binding sites placed in subtelomeric position (data not shown). These findings suggest that full core X activity involves the additive cooperation of several protosilencers among which are the Abf1p- and ORC-binding sites. This conclusion is also supported by the impairment of subtelomeric silencing upon mutation of the ACS or the Abf1p-binding site (PRYDE and LOUIS 1999 ). We previously showed that the effect of deleting the HDF1 gene on TPE is not as profound for a telomere harboring a core X sequence as for strains carrying only reporter genes next to telomere repeats (FOUREL et al. 1999 ). A similar partial loss of HDF1 dependency is also observed upon the insertion of either an ACS or an ABF oligonucleotide (Fig 2B, compare GF66, GF67, GF83, and GF84), indicating that individual ACS and ABF sequences can contribute to the overall TPE.

Inspection of core X sequences did not reveal the presence of any potential Rap1p-binding site complying with the previously defined consensus. However, short tracts of TG1-3 repeats are frequently found on the centromere proximal side of Y' elements, embedded between X and Y' sequences. These provide Rap1p-binding sites that might potentially cooperate with core X. Therefore, a Rap1p oligonucleotide (named RAP) was also assayed. It was found to counteract X-STAR-mediated antisilencing when inserted as a single copy, much like ACS and ABF oligonucleotides (Fig 2A, GF82).

Thus, core X subtelomeric silencing activity is partially recapitulated by individual binding sites for the ORC, Abf1p, and Rap1p proteins, and elementary protosilencers seemingly cooperate in an additive manner to yield full core X silencing activity.

Adjoining X-STAR sequences does not improve core X silencing capacity at HML:
Yeast subtelomeric insulator and silencing elements seemingly operate in an independent manner (FOUREL et al. 1999 ). However, core X is almost always found adjacent to an X-STAR sequence (LOUIS 1995 ). Analogous silencer/insulator combinations have been previously described in other systems and organisms (MIHALY et al. 1997 ; AYOUB et al. 1999 ; DONZE et al. 1999 ; SRIVASTAVA et al. 2000 ), which might point to common functional determinants. Therefore, although X-STAR sequences are not required for core X silencing activity under conditions of optimal operation in telomere vicinity, we thought it was possible that they may contribute to core X silencing activity in less favorable settings such as at the HM loci. We thus compared the silencing capacity of a full-length X element to that of core X at the HML locus.

A series of isogenic yeast strains was generated in which most HML was replaced by URA3, flanked by various combinations of HML-E, HML-I, deletion derivatives, and telomere II-R X-derived sequences. The presence of either HML-E or HML-I in control strains bearing no insert resulted in significant silencing levels of URA3, although not as high as with both silencers intact (FOUREL et al. 1999 ; see also Fig 3 and compare GF25, GF19, and GF48). In contrast, not a single colony grew out of 300,000 cells on 5-FOA-containing medium in the context of both HML-E and -I silencers deleted, even in the presence of core X (FOUREL et al. 1999 ; see also Fig 3, GF62 and GF63) or of a full-length X sequence (Fig 3, EL1, EL2). The full-length X element is therefore as incapable as core X of autonomously eliciting silencing at the HML locus.

Adjoining an X-STAR sequence to core X did not enhance its protosilencer activity as revealed in cooperation with a distant HML-I silencer (Fig 3, compare EL3 and EL4 with GF64 and GF65). We previously reported that, at telomeres, the synergy between core X and telomeric repeats is not disrupted by interposed STAR sequences (FOUREL et al. 1999 ). This holds true also at HML when core X is separated from URA3 and HML-I by an intervening X-STAR sequence (Fig 3, compare GF65 and EL4). Moreover, the antisilencing effect of X-STAR is overcome by core X when X-STAR is bracketed between core X and a silencer (Fig 3, compare GF19, GF21, and EL5), similar to previous observations at telomere (FOUREL et al. 1999 ).

Overexpression of Sir3p and Sir4p enables core X silencing activity at HML:
We next wondered whether overexpression of Sir3p and Sir4p might confer silencing capacity to core X at HML in the absence of any silencer. As shown above (Fig 3) and confirmed here using a more sensitive assay (see MATERIALS AND METHODS), URA3 did not exhibit any silencing at HML in the absence of a silencer and under wild-type expression levels of Sir proteins, regardless of the nearby insertion of X-derived sequences (Fig 4, GF31, GF62, and EL1). An average of 30 colonies grew from 4.5 x 10⁷ cells spread on 5-FOA-containing medium, all of which had acquired a mutation in the URA3 gene (data not shown). However, overexpression of Sir3p and Sir4p in the absence of any known silencing element yielded a ratio of 4.4 x 10^-5% 5-FOA-resistant colonies that had an intact URA3 gene, suggesting a weak but statistically significant silencing of URA3 (Fig 4, EL15 and data not shown). Three hypotheses may be proposed to account for this unanticipated observation. First, this silencing might depend upon unknown silencing elements within the HML locus that would remain unaffected by the deletions of the silencers. Alternatively, overexpression of the Sir proteins might allow propagation of silenced chromatin along chromosomal DNA all the way from the III-L telomeric area to the HML locus, i.e., >13 kb. In support of this hypothesis, silencing has been reported to extend up to 22 kb away from telomere V-R upon Sir3p overexpression (RENAULD et al. 1993 ). Finally, it is also conceivable that promiscuous nucleation of silent chromatin complexes might occur under high intranuclear Sir concentration, as previously proposed by HOLMES et al. 1997 .

Silencing levels of URA3 increased to 0.01% upon Sir3p and Sir4p overexpression when full-length X or core X sequences were inserted (Fig 4, EL16 and EL17). Importantly, Sir3p and Sir4p overexpression did not improve HML-E and -I silencing capacity at HML but rather had a slight negative effect (Fig 4, EL18 and EL19). This modest trans-dominant effect might be accounted for by the perturbation of the optimal stoichiometry between Sir proteins, as previously suggested (BUCK and SHORE 1995 ; MAILLET et al. 1996 ). Thus, overexpression of Sir3p and Sir4p enables a weak core X-mediated silencing at HML and, again in this situation, X-STAR sequences seem not to contribute to X activity.

Plasmid-borne core X behaves as a protosilencer:
One may hypothesize that the absence of detectable, autonomous silencing activity of core X inserted at HML results from topological constraints due to the position of this locus within a chromosome, whereas X elements naturally stand in an immediate subtelomeric location. In light of the capacity of silencer elements to induce silencing in cis when carried on plasmids (ABRAHAM et al. 1984 ; FELDMAN et al. 1984 ), we decided to monitor the silencing capacity of a plasmid-borne core X sequence to address this hypothesis. A series of strains was generated through transformation of W303-1A cells with pRS315-based plasmids carrying copies of HML constructs with URA3 as a reporter gene as described above. These plasmids were selected on leucine dropout medium both following transformation and during the silencing assays. Strain ELC, which harbors the parental pRS315 vector, was found to be resistant to 5-FOA; as expected, as it does not carry a functional URA3 gene (Fig 5). Some rare 5-FOA-resistant colonies (1 in 7 x 10⁶) grew from cultures containing an URA3 derivative of pRS315 devoid of any silencer (EL7). Those harbored an intact URA3 gene as verified by quantitative restreaking on SC-U (data not shown). We believe they arose by spontaneous genetic mutations that confer either silencing-independent 5-FOA-resistant or Sir-independent silencing of URA3 (LIN et al. 1990 ; CHI and SHORE 1996 ; DULA and HOLMES 2000 ). Alternatively, one may envisage the existence of a background level of silencing for plasmid-borne reporter genes, which might be accounted for by the preferential positioning of plasmids at the nuclear periphery, for example. This level of 5-FOA resistance observed for an HML locus deprived of silencers sets up the detection limit of our "on-plasmid" silencing assay.

Inserting core X by itself did not significantly raise silencing above this limit (Fig 5, compare EL7 and EL8). As expected, significant silencing levels were observed in the presence of HML-E and/or HML-I silencers (Fig 5, EL9, EL11, and EL13). It is worth noting that, for identical silencing reporter cassettes, silencing is lower in the plasmid context than in the native setting at HML (compare Fig 5 with GF25, GF19, and GF48 in Fig 3). These results are consistent with previous reports (ABRAHAM et al. 1984 ; FELDMAN et al. 1984 ; BRAND et al. 1985 ; MAILLET et al. 1996 ).

Finally, core X potently synergized with HML-I in establishing a stronger than HML-E-driven silencing level (Fig 5, compare EL9, EL10, and EL11). Thus, core X does not appear to act as a silencer when carried on a plasmid, but rather as a protosilencer in that it requires cooperation with a silencer, as previously observed at HML.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

What distinguishes a protosilencer from a silencer?
The detailed characterization of core X silencing activity presented here confirms and extends the idea that core X qualitatively behaves like an isolated protosilencer. It is, however, more efficient, owing to the cooperation of several elementary protosilencers, including an ACS and an Abf1p-binding site. It is noteworthy that, although a functional URS1 sequence can strengthen telomeric silencing, we were unable to ascribe any silencing effect to the URS1-like sequence found in core X (G. FOUREL and E. GILSON, unpublished results). Although not characterized in detail, silencing elements previously described in Y' subtelomeric repeats (FOUREL et al. 1999 ), as well as short internal tracts of TG1-3n telomere-like sequences (LOUIS 1995 ), may be anticipated to display comparable properties.

Interestingly, core X activity could be demonstrated only upon cooperation with a bona fide silencer, with a telomere, or upon Sir protein overexpression. In these environments, core X can amplify preexisting silencing to levels approaching those conferred by classical silencers. In contrast, authentic silencers have been shown to be functional in various chromosomal contexts thought to be exempt of any basal silencing. One may envision that protosilencers and core X qualitatively behave as authentic silencers, i.e., nucleating the de novo assembly of a silencing-competent complex, but only when an extremely high concentration of silencing factors is provided (BOSCHERON et al. 1996 ; LUSTIG 1998 ). In a silencing-free environment, conversely, binding of silent chromatin components by protosilencers would be counter-balanced by their dissociation rate, never reaching the threshold that would allow nucleation of a silencing-competent complex. Thus, in spite of a continuum of organization between core X and authentic silencers, both containing protosilencers as active silencing elements but arranged differently, a quantitative difference in their capacity to recruit silent chromatin components would translate into a qualitative one owing to threshold effects. A similar model has been proposed for Drosophila to explain why a single copy of the white transgene can recruit the heterochromatin Protein-1 (HP1) but does not display silencing features, whereas multimer arrays of white behave similarly to heterochromatin blocks (DORER and HENIKOFF 1994 , DORER and HENIKOFF 1997 ; FANTI et al. 1998 ). The silencing activity of core X as well as other protosilencers may alternatively depend upon either a particular state of silencing-conducive chromatin or on the presence of classical but unstable silent chromatin, which would besides allow quasi-normal expression of an inserted reporter gene. Both of these hypotheses are in agreement with the recent demonstration that protosilencers stabilize preestablished silent chromatin (CHENG and GARTENBERG 2000 ).

We therefore propose that authentic silencers (master silencer in Fig 6) can autonomously elicit silencing (Fig 6A). In contrast, protosilencers, i.e., elementary blocks of silencing elements or more elaborate sequences such as core X, allow silencing to affect neighboring portions of the genome only in a silencing-conducive environment (Fig 6, b–d). These features predict that an array of protosilencers can function as silencing relays and propagate silencing through "spreading" of silent chromatin at an extensive distance away from silencers. In addition, this propagation can be discontinuous, possibly owing to the involvement of insulator elements (FOUREL et al. 1999 ; Fig 6D). Alternatively, the cooperation of multiple protosilencers may alleviate the need for a master silencer and create a fully functional silencing structure (Fig 6E). A similar model involving cooperating relays was previously proposed to explain the spread of silencing by PcG proteins (PIRROTTA 1997 ; STRUTT et al. 1997 ).

View larger version (21K): In this window In a new window Download PPT slide   Figure 6. Mechanisms of silencing by silencers and protosilencers. This model emanates from this study as well as other studies quoted in the text. A master silencer can autonomously establish and maintain silencing (a), at least in certain chromosome contexts, whereas a protosilencer cannot do so in the vast majority of settings (b). A protosilencer amplifies, stabilizes, or prolongs silent chromatin propagation only in silencing-conducive environments (c and d). These were found to correspond to chromosomal regions in which silencing preexists, sometimes at very low levels. This is the case, for instance, in the vicinity of telomeres and silencers (c), but also at a larger distance (d) and even in the absence of a residing silencing element at HML upon overexpression of Sir3p and Sir4p. Cooperation between the master silencer and the protosilencer must then involve discontinuous silent chromatin propagation. Alternatively, multiple protosilencers may be envisioned to create an autonomous silencing structure (e). Functional cooperation as illustrated by arrows likely involves direct physical interaction and coalescence of the silent domains, as discussed in FOUREL et al. 1999 .

General involvement of protosilencers in silencing phenomena:
In S. cerevisiae, an isolated Rap1p-binding site in the promoter of 1/2 genes, centrally located in between the HML-E and -I silencers, acts as a protosilencer and contributes to the maintenance of silencing at this locus (BOSCHERON et al. 1996 ; CHENG and GARTENBERG 2000 ). In Schizosaccharomyces pombe, elements flanking the mat2-P and mat3-M mating cassettes facilitate the bidirectional spreading of silencing from a central K-region silencer throughout the locus and enhance silencing stability (THON et al. 1999 ; AYOUB et al. 2000 ). They are seemingly devoid of silencing activity on their own and thus act as protosilencers (AYOUB et al. 2000 ). The REII element may further include a bordering insulator, an arrangement similar to that of X subtelomeric repeats of S. cerevisiae (AYOUB et al. 1999 , AYOUB et al. 2000 ; FOUREL et al. 1999 ).

Vertebrate Hox genes are repressed early on in development through a silencing-type mechanism and are then activated in a spatio-temporal sequence colinear with the position of the genes within their cluster. Dissection of the HoxD complex suggested the existence of a master silencer at the 5' end and of protosilencers, relay elements in between genes (KONDO et al. 1998 ; KONDO and DUBOULE 1999 ). Amazingly, the relay element between mouse Hoxd-13 and Hoxd-12 genes can be functionally replaced by the orthologous zebrafish DNA that displays no apparent nucleotide homology (KONDO et al. 1998 ). A decrease in available components of silent chromatin, following a dilution effect in the course of embryo development, would progressively render relays located the farthest from the silencer inoperative. This would account for the orientated, progressive opening of chromatin toward the 5' end (VAN DER HOEVEN et al. 1996 ; KONDO et al. 1998 ).

Recent studies revealed the possibility of a local and transient break in colinearity upon insertion of a copy of the most anterior Hoxb gene at the most posterior position in the HoxD complex (KMITA et al. 2000 ). Thus, this master silencer/protosilencer relay system allows silencing discontinuity, as previously shown in S. cerevisiae subtelomeric regions (FOUREL et al. 1999 ; PRYDE and LOUIS 1999 ; Fig 6D). In an analogous manner, in female mammals, spreading of heterochromatic inactivation throughout one X chromosome from the inactivation center involves silencing relays that may consist of LINE-1 elements (L1s) and some DNA segments containing significantly fewer L1s can escape inactivation (BAILEY et al. 2000 ; LYON 2000 ). Cis discontinuity in silencing was also recently observed in a classical system of PEV in Drosophila (TALBERT and HENIKOFF 2000 ).

Tandem repeat-induced silencing in Drosophila (DORER and HENIKOFF 1997 ), the X inactivation center in mammals (BAILEY et al. 2000 ), as well as rRNA silencing in S. cerevisiae (BRYK et al. 1997 ; SMITH and BOEKE 1997 ) may rather be envisioned to operate according to the model in Fig 6E.

Overall, it appears that protosilencers critically assist with silencers and may cooperate to generate autonomously silencing structures in a variety of systems and organisms.

	FOOTNOTES

¹ Present address: Laboratoire de Biologie Cellulaire et Moléculaire du Développement des Plantes, Université de Bordeaux, Talence 33405, France.
² Present address: Laboratoire du cytosquelette, CEA/DBMS, Grenoble 38054, France.

	ACKNOWLEDGMENTS

We thank Ed Louis and Stéphane Marcand for plasmids and for invaluable discussions, and Pierre-Antoine Desfossez for reading the manuscript. This work was supported by La Ligue Nationale contre le Cancer.

Manuscript received November 29, 2000; Accepted for publication February 12, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ABRAHAM, J., K. A. NASMYTH, J. N. STRATHERN, A. J. S. KLAR, and J. B. HICKS, 1984  Regulation of mating-type information in yeast: negative control requiring sequences both 5' and 3' to the regulated region. J. Mol. Biol. 176:307-331[Medline].

AYOUB, N., I. GOLDSHMIDT, and A. COHEN, 1999  Position effect variegation at the mating-type locus of fission yeast: a cis-acting element inhibits covariegated expression of genes in the silent and expressed domains. Genetics 152:495-508[Abstract/Free Full Text].

AYOUB, N., I. GOLDSHMIDT, R. LYAKHOVETSKY, and A. COHEN, 2000  A fission yeast repression element cooperates with centromere-like sequences and defines a mat silent domain boundary. Genetics 156:983-994[Abstract/Free Full Text].

BAILEY, J. A., L. CARREL, A. CHAKRAVARTI, and E. E. EICHLER, 2000  Molecular evidence for a relationship between LINE-1 elements and X chromosome inactivation: the Lyon repeat hypothesis. Proc. Natl. Acad. Sci. USA 97:6634-6639[Abstract/Free Full Text].

BOEKE, J. D., F. LACROUTE, and G. R. FINK, 1984  A positive selection for mutants lacking orotidine-5'-phosphate decarboxylase activity in yeast: 5-fluoro-orotic acid resistance. Mol. Gen. Genet. 197:345-346[Medline].

BOSCHERON, C., L. MAILLET, S. MARCAND, M. TSAI-PFUGFELDER, and S. M. GASSER et al., 1996  Cooperation at a distance between silencers and proto-silencers at the yeast HML locus. EMBO J. 15:2184-2195[Medline].

BRAND, A. H., L. BREEDEN, J. ABRAHAM, R. STERNGLANZ, and K. NASMYTH, 1985  Characterization of a silencer in yeast: a DNA sequence with properties opposite to those of a transcriptional enhancer. Cell 41:41-48[Medline].

BRYK, M., M. BANERJEE, M. MURPHY, K. E. KNUDSEN, and D. J. GARFINKEL et al., 1997  Transcriptional silencing of Ty1 elements in the RDN1 locus in yeast. Genes Dev. 11:255-269[Abstract/Free Full Text].

BUCK, S. W. and D. SHORE, 1995  Action of a RAP1 carboxy-terminal silencing domain reveals an underlying competition between HMR and telomeres in yeast. Genes Dev. 9:370-384[Abstract/Free Full Text].

CHENG, T.-H. and M. R. GARTENBERG, 2000  Yeast heterochromatin is a dynamic structure that requires silencers continuously. Genes Dev. 14:452-463[Abstract/Free Full Text].

CHI, M.-H. and D. SHORE, 1996  SUM1–1, a dominant suppressor of SIR mutations in Saccharomyces cerevisiae, increases transcriptional silencing at telomeres and HM mating-type loci and decreases chromosome stability. Mol. Cell. Biol. 16:4281-4294[Abstract].

DONZE, D., C. R. ADAMS, J. RINE, and R. T. KAMAKAKA, 1999  The boundaries of the silenced HMR domain in Saccharomyces cerevisiae. Genes Dev. 13:698-708[Abstract/Free Full Text].

DORER, D. R. and S. HENIKOFF, 1994  Expansion of transgene repeats causes heterochromatin formation and gene silencing in Drosophila. Cell 77:993-1002[Medline].

DORER, D. R. and S. HENIKOFF, 1997  Transgene repeat arrays interact with distant heterochromatin and cause silencing in cis and in trans. Genetics 147:1181-1190[Abstract].

DULA, M. L. and S. G. HOLMES, 2000  MGA2 and SPT23 are modifiers of transcriptional silencing in yeast. Genetics 156:933-941[Abstract/Free Full Text].

FANTI, L., D. R. DORER, M. BERLOCO, S. HENIKOFF, and S. PIMPINELLI, 1998  Heterochromatin protein 1 binds transgene arrays. Chromosoma 107:286-292[Medline].

FELDMAN, J. B., J. B. HICKS, and J. R. BROACH, 1984  Identification of the sites required for repression of a silent mating-type locus in yeast. J. Mol. Biol. 178:815-834[Medline].

FOUREL, G., E. REVARDEL, C. E. KOERING, and E. GILSON, 1999  Cohabitation of insulators and silencing elements in yeast subtelomeric regions. EMBO J. 18:2522-2537[Medline].

FOUREL, G., C. BOSCHERON, E. REVARDEL, E. LEBRUN, and Y.-F. HU et al., 2001  An activation-independent role of transcription factors in insulator function. EMBO J. 2:124-132.

GARTENBERG, M. R., 2000  The Sir proteins of Saccharomyces cerevisiae: mediators of transcriptional silencing and much more. Curr. Opin. Microbiol. 3:132-137[Medline].

GIESMAN, D., L. BEST, and K. TATCHELL, 1991  The role of RAP1 in the regulation of the MAT locus. Mol. Cell. Biol. 11:1069-1079[Abstract/Free Full Text].

HENIKOFF, S., 2000  Heterochromatin function in complex genomes. Biochim. Biophys. Acta 1470:1-8.

HOLMES, S. G., A. B. ROSE, K. STEUERLE, E. SAEZ, and S. SAYEGH et al., 1997  Hyperactivation of the silencing proteins, Sir2p and Sir3p, causes chromosome loss. Genetics 145:605-614[Abstract].

KADOSH, D. and K. STRUHL, 1997  Repression by UME6 involves recruitment of a complex containing Sin3 corepressor and Rpd3 histone deacetylase to target promoters. Cell 89:365-371[Medline].

KMITA, M., F. VAN DER HOEVEN, J. ZAKANY, R. KRUMLAUF, and D. DUBOULE, 2000  Mechanisms of Hox gene colinearity: transposition of the anterior Hoxb1 gene into the posterior HoxD complex. Genes Dev. 14:198-211[Abstract/Free Full Text].

KOERING, C. E., G. FOUREL, F. KLEIN, E. BINET-BRASSELET, and T. LAROCHE et al., 2000  Identification of high affinity Tbf1p binding sites within the budding yeast genome. Nucleic Acids Res. 28:2519-2526[Abstract/Free Full Text].

KONDO, T. and D. DUBOULE, 1999  Breaking colinearity in the mouse HoxD complex. Cell 97:407-417[Medline].

KONDO, T., J. ZÀKÀNY, and D. DUBOULE, 1998  Control of colinearity in AbdB genes of the mouse HoxD complex. Mol. Cell 1:289-300[Medline].

LIN, C. I., G. P. LIVI, J. M. IVY, and A. J. KLAR, 1990  Extragenic suppressors of mar2(sir3) mutations in Saccharomyces cerevisiae. Genetics 125:321-331[Abstract].

LOUIS, E. J., 1995  The chromosome ends of Saccharomyces cerevisiae. Yeast 11:1553-1573[Medline].

LOUIS, E. J. and R. H. BORTS, 1995  A complete set of marked telomeres in Saccharomyces cerevisiae for physical mapping and cloning. Genetics 139:125-136[Abstract].

LOUIS, E. J., E. S. NAUMOVA, A. LEE, G. NAUMOV, and J. E. HABER, 1994  The chromosome end in yeast: its mosaic nature and influence on recombinational dynamics. Genetics 136:789-802[Abstract].

LUSTIG, A. J., 1998  Mechanisms of silencing in Saccharomyces cerevisiae. Curr. Opin. Gen. Dev. 8:233-239[Medline].

LYON, M. F., 2000  LINE-1 elements and X chromosome inactivation: a function for "junk" DNA? Proc. Natl. Acad. Sci. USA 97:6248-6249[Free Full Text].

MAHONEY, D. J. and J. BROACH, 1989  The HML mating-type cassette of S. cerevisiae is regulated by two separate but functionally equivalent silencers. Mol. Cell. Biol. 9:4621-4630[Abstract/Free Full Text].

MAILLET, L., C. BOSCHERON, M. GOTTA, S. MARCAND, and E. GILSON et al., 1996  Evidence for silencing compartments in the yeast nucleus: a role for telomere proximity and Sir protein concentration in silencer-mediated repression. Genes Dev. 10:1796-1811[Abstract/Free Full Text].

MARCAND, S., S. W. BUCK, P. MORETTI, E. GILSON, and D. SHORE, 1996  Silencing of genes at nontelomeric sites in yeast is controlled by sequestration of silencing factors at telomeres by Rap1 protein. Genes Dev. 10:1297-1309[Abstract/Free Full Text].

MARTIN, S. G., T. LAROCHE, N. SUKA, M. GRUNSTEIN, and S. M. GASSER, 1999  Relocalization of telomeric Ku and SIR proteins in response to DNA breaks in yeast. Cell 97:621-633[Medline].

MCNALLY, F. J. and J. RINE, 1991  A synthetic silencer mediates SIR-dependent functions in Saccharomyces cerevisiae. Mol. Cell. Biol. 11:5648-5659[Abstract/Free Full Text].

MIHALY, J., I. HOGGA, J. GAUSZ, H. GYURKOVICS, and F. KARCH, 1997  In situ detection of the Fab-7 region of the bithorax complex into a chromatin domain boundary and a Polycomb-response element. Development 124:1809-1820[Abstract].

PIRROTTA, V., 1997  Pc-G complexes and chromatin silencing. Curr. Opin. Genet. Dev. 7:249-258[Medline].

PRYDE, F. E. and E. J. LOUIS, 1999  Limitations on natural TPE in yeast. EMBO J. 18:2538-2550[Medline].

PRYDE, F. E., T. C. HUCKLE, and E. J. LOUIS, 1995  Sequence analysis of the right end of chromosome XV in Saccharomyces cerevisiae: an insight into the structural and functional significance of subtelomeric repeat sequences. Yeast 11:371-382[Medline].

RENAULD, H., O. M. APARICIO, P. D. ZIERATH, B. L. BILLINGTON, and S. K. CHHABLANI et al., 1993  Silent domains are assembled continuously from the telomere and are defined by promoter distance and strength, and by SIR3 dosage. Genes Dev. 7:1133-1145[Abstract/Free Full Text].

SAMBROOK, J., E. F. FRITSCH and T. MANIATIS, 1989 Molecular Cloning: A Laboratory Manual, Ed. 2. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

SIKORSKI, R. S. and P. HIETER, 1989  A system of shuttle vectors and yeast host strains designed for efficient manipulation of DNA in Saccharomyces cerevisiae. Genetics 122:19-27[Abstract/Free Full Text].

SMITH, J. S. and J. D. BOEKE, 1997  An unusual form of transcriptional silencing in yeast ribosomal DNA. Genes Dev. 11:241-254[Abstract/Free Full Text].

SRIVASTAVA, M., S. HSIEH, A. GRINBERG, L. WILLIAMS-SIMONS, and S.-P. HUANG et al., 2000  H19 and Igf2 monoallelic expression is regulated in two distinct ways by a shared cis acting regulatory element upstream of H19. Genes Dev. 14:1186-1195[Abstract/Free Full Text].

STEVENSON, J. B. and D. E. GOTTSCHLING, 1999  Telomeric chromatin modulates replication timing near chromosome ends. Genes Dev. 13:146-151[Abstract/Free Full Text].

STRUTT, H., G. CAVALLI, and R. PARO, 1997  Co-localization of Polycomb protein and GAGA factor on regulatory elements responsible for the maintenance of homeotic gene expression. EMBO J. 16:3621-3632[Medline].

TALBERT, P. B. and S. HENIKOFF, 2000  A reexamination of spreading of position-effect variegation in the white-roughest region of Drosophila melanogaster. Genetics 154:259-272[Abstract/Free Full Text].

THON, G., K. P. BJERLING, and I. S. NIELSEN, 1999  Localization and properties of a silencing element near the mat3-M mating-type cassette of Schizosaccharomyces pombe. Genetics 151:945-963[Abstract/Free Full Text].

TRIOLO, T. and R. STERNGLANZ, 1996  Role of interactions between the origin recognition complex and SIR1 in transcriptional silencing. Nature 381:251-253[Medline].

VAN DER HOEVEN, F., J. ZÀKÀNY, and D. DUBOULE, 1996  Gene transpositions in the HoxD complex reveal a hierarchy of regulatory controls. Cell 85:1025-1035[Medline].

VEGA-PALAS, M. A., S. VENDITTI, and E. DI MAURO, 1997  Telomeric transcriptional silencing in a natural context. Nat. Genet. 15:232-233[Medline].

This article has been cited by other articles:

				L. Valenzuela, N. Dhillon, R. N. Dubey, M. R. Gartenberg, and R. T. Kamakaka Long-Range Communication between the Silencers of HMR Mol. Cell. Biol., March 15, 2008; 28(6): 1924 - 1935. [Abstract] [Full Text] [PDF]

				M. A. Mondoux and V. A. Zakian Subtelomeric Elements Influence But Do Not Determine Silencing Levels at Saccharomyces cerevisiae Telomeres Genetics, December 1, 2007; 177(4): 2541 - 2546. [Abstract] [Full Text] [PDF]

				M. A. Rehman, G. Fourel, A. Mathews, D. Ramdin, M. Espinosa, E. Gilson, and K. Yankulov Differential Requirement of DNA Replication Factors for Subtelomeric ARS Consensus Sequence Protosilencers in Saccharomyces cerevisiae Genetics, December 1, 2006; 174(4): 1801 - 1810. [Abstract] [Full Text] [PDF]

				Y. Zou, Q. Yu, Y.-H. Chiu, and X. Bi Position Effect on the Directionality of Silencer Function in Saccharomyces cerevisiae Genetics, September 1, 2006; 174(1): 203 - 213. [Abstract] [Full Text] [PDF]

				S. Ferrari, K. C. Simmen, Y. Dusserre, K. Muller, G. Fourel, E. Gilson, and N. Mermod Chromatin Domain Boundaries Delimited by a Histone-binding Protein in Yeast J. Biol. Chem., December 31, 2004; 279(53): 55520 - 55530. [Abstract] [Full Text] [PDF]

				E. Lebrun, G. Fourel, P.-A. Defossez, and E. Gilson A Methyltransferase Targeting Assay Reveals Silencer-Telomere Interactions in Budding Yeast Mol. Cell. Biol., March 1, 2003; 23(5): 1498 - 1508. [Abstract] [Full Text] [PDF]