Depleting Components of the THO Complex Causes Increased Telomere Length by Reducing the Expression of the Telomere-Associated Protein Rif1p

Telomere length is regulated mostly by proteins directly associated with telomeres. However, genome-wide analysis of Saccharomyces cerevisiae mutants has revealed that deletion of Hpr1p, a component of the THO complex, also affects telomere length. The THO complex comprises four protein subunits, namely, Tho2p, Hpr1p, Mft1p, and Thp2p. These subunits interplay between transcription elongation and co-transcriptional assembly of export-competent mRNPs. Here we found that the deletion of tho2 or hpr1 caused telomere lengthening by ∼50–100 bps, whereas that of mft1 or thp2 did not affect telomere length. Since the THO complex functions in transcription elongation, we analyzed the expression of telomere-associated proteins in mutants depleted of complex components. We found that both the mRNA and protein levels of RIF1 were decreased in tho2 and hpr1 cells. RIF1 encodes a 1917-amino acid polypeptide that is involved in regulating telomere length and the formation of telomeric heterochromatin. Hpr1p and Tho2p appeared to affect telomeres through Rif1p, as increased Rif1p levels suppressed the telomere lengthening in tho2 and hpr1 cells. Moreover, yeast cells carrying rif1 tho2 or rif1 hpr1 double mutations showed telomere lengths and telomere silencing effects similar to those observed in the rif1 mutant. Thus, we conclude that mutations of components of the THO complex affect telomere functions by reducing the expression of a telomere-associated protein, Rif1p.


Introduction
Telomeres are the structure at the ends of eukaryotic linear chromosomes [1,2]. They are essential for the maintenance of chromosome integrity, and protect natural DNA ends from being recognized as double-strand breaks. In most organisms, the telomeric DNA is composed of short, tandemly repeated sequences with a strand rich in guanine residues (G-strand) running 59 to 39 toward the end of the telomere. For example, the telomeric sequences in the brewer's yeast Saccharomyces cerevisiae are ,250-300 base pair-long TG 1-3 /C 1-3 A repeats. Telomeres are maintained by telomerase in most eukaryotes [3]. Telomerase is a ribonucleoprotein containing a catalytic protein component TERT (telomerase reverse transcriptase) and an associated RNA moiety, TER, which serves as the template to extend telomeric DNA sequences. In S. cerevisiae, these two components are encoded by EST2 [4,5] and TLC1 [6], respectively.
Telomere length homeostasis involves the coordination of telomere lengthening and shortening processes. Although the complete mechanism of telomere length regulation remains to be elucidated, controlling telomerase activity appears to be a major part of the process [7]. For example, the binding of telomerase Tlc1 RNA by yeast Ku proteins contributes to telomere length regulation [8]. The mutation of YKU70 or YKU80 causes shortening of telomere length [9][10][11][12][13][14]. It has also been reported that Pif1p helicase negatively regulates telomerase by removing telomerase from the telomeric DNA [15,16]. As a consequence, mutation of PIF1 has been shown to lengthen telomeres [17]. These telomere-associated proteins affect telomere length by directly regulating the telomerase activity [7]. Regulation of telomerase could also be achieved through posttranslational modification of telomere-associated proteins. For example, phosphorylation of the telomere binding protein Cdc13p by DNA damage response kinases Tel1p/Mec1p is required for recruiting telomerase onto telomeres for telomere replication [18]. Short telomeres have been observed in cells missing either TEL1 or MEC1 [19][20][21].
Telomere length homeostasis can also be regulated by the repressor/activator protein 1, Rap1p. in S. cerevisiae [22]. It has been reported that several rap1 temperature-sensitive mutants display phenotypes consistent with the role of Rap1p as a negative regulator of telomere length [23][24][25]. Rap1p binds to a loosely defined recognition site within the double-stranded TG [1][2][3] telomeric DNA tracts to affect telomere functions [23][24][25]. Regulation of telomere length by Rap1p appears to be mediated by Rif1p and Rif2p, which bind to the protein-interaction domain at the C-terminus of Rap1p. Deletion of the nonessential RIF1 and RIF2 genes results in extensive telomere elongation [26,27].
In addition to affecting telomerase and telomere-associated proteins, genes affecting telomere length have also been identified as being involved in diverse cellular functions [31][32][33]. Among them, deletion of HPR1 has been shown to increase telomere length by ,50-150 bp. HPR1 is a component of the THO complex (a suppressor of the Transcriptional defects of Hpr1 mutants by Overexpression) that is involved in transcriptionelongation and the export of transcribed mRNAs [34,35]. The THO complex is a conserved nuclear complex that is formed by four protein subunits: Hpr1p, Tho2p (Rlr1p), Mft1p, and Thp2p in S. cerevisiae. Mutations in THO components cause defects in transcription-dependent hyper-recombination and mRNA export [36][37][38][39]. Moreover, there is evidence that DNA:RNA hybrids are formed co-transcriptionally, resulting in impaired transcription elongation of GC-rich and long DNA sequences and defective transcription-associated recombination [40]. However, it is unclear how mutations of THO components cause changes in telomere length. Here we found that mutations of THO components affect telomeres by reducing the expression of Rif1p. Our results reveal a novel mechanism by which telomere functions could be affected indirectly through perturbing the expression of a telomere-associated component.

Mutations of THO complex components HPR1 and THO2 affect telomere length
In a genome-wide screen, mutation of HPR1 was found to cause a ,50-150 bp increase in telomere length [31]. Since Hpr1p is a component of the THO complex, we first tested whether the other three THO components also affect telomere length. In S. cerevisiae, the middle repetitive sequences known as the Y9 elements are found in the subtelomeric regions of most chromosomes [41]. In wild-type cells, XhoI digestion produces a ,1.3 kbp fragment from the ends of Y9-bearing chromosomes that contains ,870 bp of Y9 and the terminal tract of ,350-430 bp of TG 1-3 /C 1-3 A DNA. However, the lengths of Y9-bearing telomeres in both hpr1 and tho2 cells appeared ,75 bp longer than those in wild-type cells (Fig. 1A). This telomere length phenotype was not progressive, as increased cell divisions did not further affect its length (data not shown). Interestingly, mutations of the other two THO complex components, MFT1 and THP2, did not affect telomere length (Fig. 1A), although it has been reported that mutations of these two genes cause slightly shorter telomeres [31]. The reason for the apparent discrepancy is unclear. Telomere lengths of yeast cells harboring both hpr1 and tho2 mutations were also analyzed. As shown in Figure 1A (right panel), hpr1 tho2 double mutations did not further increase the telomere length. Thus, HPR1 and THO2 might affect telomere length through the same pathway. As a control, the temperature sensitivity of the resulting strains were analyzed since mutations of THO components also exhibit severe growth defects at 37uC [42,43]. Indeed, all of the THO mutant cells showed the expected temperature-sensitive phenotype (Fig. 1B).
To determine if the telomere lengthening in hpr1 and tho2 cells required the telomerase activity, deletion of telomerase RNA component TLC1 was introduced into hpr1 and tho2 cells, respectively. Yeast cells losing telomerase RNA undergo progressive telomere shortening which eventually leads to cell death [6]. As shown in Figure 1C, telomere lengths in hpr1 tlc1 and tho2 tlc1 cells were similar to those in tlc1 cells. The results indicate that the long telomeres observed in hpr1 and tho2 cells required functional telomerase. The effect of recombination on telomere lengthening of hpr1 and tho2 cells was also analyzed. A rad52 mutation was introduced into tho2 cells and telomere length was examined. As shown in Figure 1D, rad52 did not affect telomere length in either wild-type or tho2 cells. The results suggest that telomere lengthening observed in THO complex mutants is not due to telomere recombination.
RIF1 expression is reduced in hpr1 and tho2 cells Next, we explored the mechanism by which HPR1 and THO2 affect telomere length. The THO complex is involved in transcription elongation. Hybrids of template DNA and transcribed RNA are formed during transcription in mutants of the THO complex components [40]. As a result, the transcription of long genes is preferentially repressed [39,44]. We hypothesized that mutations of HPR1 and THO2 might affect telomere length by influencing the expression of genes that encode telomereassociated proteins. We first surveyed the length of telomereassociated proteins ( Table 1). Among those we examined, RIF1 is the longest gene with a size of 5,751 bps that encodes a protein with 1,916 amino acid residues. To test if hpr1 and tho2 affect the expression of the long genes, we analyzed the Rif1p protein level in mutant cells. As shown in Figure 2A, Rif1p levels were decreased by ,4 folds in both hpr1 and tho2 cells. In comparison, SIR4 (4,077 bps), CDC13 (2,775 bps), RAP1 (2,484 bps), and RIF2 (1,188 bps) were also analyzed. The results shown in Figure 2A indicate that hpr1 and tho2 did not affect the expression of SIR4, CDC13, or RAP1. The Rif2p level was only slightly decreased in these two mutants. Since a yeast strain with 9 repeats of Myc attached to the 39 end of the chromosomal RIF1 gene was used in our analysis, the effect of Myc-tagging on Rif1p was also analyzed. The RIF1-myc9 tagged allele caused similar, if not identical, effects on telomere length and telomere silencing to those observed in cells with the wild-type allele, implying that the tagged Rif1p is expressed and functional (Fig. 2B).
It has been reported that the expression levels of yeast genes increase with the G+C contents in wild-type and both hpr1 and tho2 cells [44,45]. The G+C contents of the genes encoding telomere-associated proteins were also surveyed ( Table 1). Among a total of 16 genes analyzed, all of them showed a G+C content within the range of 34-44%, similar to the ,40% average G+C content of all yeast genes. As a comparison, the E. coli lacZ gene has a G+C content of ,56%. According to our findings, the expression of telomere-associated genes in hpr1 and tho2 cells did not appear to be affected by their G+C content ( Fig. 2A).
To test if the reduction of Rif1p protein is specific to hpr1 and tho2 mutations, Rif1p protein expression was also determined in mft1 and thp2 cells. We discovered that the Rif1p protein level remained the same in mft1 or thp2 cells (Fig. 2C), indicating that the reduction in Rip1p levels is specific to hpr1 and tho2 mutations. This result is also consistent with our observation that only hpr1 and tho2 but not mft1 or thp2 cells showed long telomeres (Fig. 1A).
To determine whether the reduced Rif1p level was due to transcription defects caused by hpr1 and tho2 mutations, the RNA level of RIF1 was analyzed. Total RNAs were prepared from yeast cells and hybridized with specific probes against GCN1 (8,019 bps), RIF1 (5,751 bps), RIF2 (1,188 bps), and TUB1 (1,460 bps) genes. As shown in Figure 2D, both hpr1 and tho2 mutations showed defects in transcribing the two long genes GCN1 and RIF1. The effect is specific, as the transcript levels of short genes RIF2 and TUB1 were not affected by the mutations. Thus, our results are in agreement with the previous findings which suggest that hpr1 and tho2 mutations affect the transcription of long genes. Consequently, both the protein and RNA levels of RIF1 were significantly reduced in these cells.
THO2 and HPR1 share the same pathway of RIF1 to regulate telomere length Given the role of RIF1 as a negative regulator of telomere length and the finding that the deletion of RIF1 in yeast causes elongation of telomere length [27], it is likely that the long telomeres observed in hpr1 and tho2 cells were due to the reduction of RIF1 expression. This simple scenario predicts that the telomere lengths of rif1 hpr1 or rif1 tho2 cells should not be longer than that in the rif1 cells. Indeed, the telomere lengths of cells carrying rif1 hpr1 and rif1 tho2 mutations were similar to those in the rif1 cells (Fig. 3A). In comparison, we found that telomere lengths were further elongated when rif2 was introduced into hpr1 or tho2 cells (Fig. 3A). Similarly, combinatorial effects were also observed when pif1-m2 mutation was introduced into tho2 cells (Fig. 3A). These results show that Hpr1p and Tho2p might affect telomere length through the same pathway as Rif1p.
It is also interesting to note that TEL1 is a long gene (8,364 bps) and its expression level is decreased in hpr1 cells [44]. Using real time RT-PCR, we found that the amount of TEL1 RNA in hpr1 or tho2 cells was decreased to ,10-15% of the wild-type level (Fig. 3B, left panel). In comparison, the RNA level of GCN1 was also decreased in hpr1 or tho2 cells, consistent with our previous observation ( Fig. 2 and 3B). Although Tel1p is not directly associated with telomeres, it participates in telomerase recruiting and its mutation has been shown to shorten telomeres [19]. To evaluate the contribution of tel1 on telomere length maintenance in THO mutants, telomere length of tel1 tho2 cells was determined. As shown in Figure 3B (right panel), telomere length in both wildtype and tho2 mutant cells was decreased by the same extent after the introduction of a tel1 mutation. The results suggest that although TEL1 RNA level is also decreased in THO mutants, the observed telomere lengthening in hpr1 or tho2 cells is not due to the decreased Tel1p level.
Our model also predicts that increasing the Rif1p level should suppress the long telomere phenotype associated with hpr1 and tho2 Yeast DNA was isolated from the YPH499 yeast strain with indicated mutations, digested with XhoI, separated on a 1% agarose gel and analyzed by Southern blotting. The blots were hybridized with probes prepared using the 600-bp sequence from the Y9 element 39 end. The position of the Y9 telomere is indicated. (B) Temperature-sensitive growth of THO complex mutants. Aliquots of ten-fold serial dilutions of yeast cells were spotted on YEPD plates and incubated at 30 or 37uC for three days. (C) Telomerase is required for the telomere lengthening effects in hpr1 and tho2 cells. Yeast cells carrying the indicated mutations were freshly sporulated and grown on YEPD plates at 30uC after 1 or 2 restreaks. Total yeast DNA was then isolated and analyzed by Southern blotting. (D) Recombination is not required for telomere lengthening in tho2 cells. Total yeast DNA from the indicated strains was isolated and analyzed by Southern blotting. doi:10.1371/journal.pone.0033498.g001 mutations. Initial attempts to overexpress RIF1 using a cen plasmid did not yield a detectable elevation of Rif1p levels in hpr1 or tho2 cells. Accordingly, there was no change in telomere length in these cells (data not shown). We then increased Rif1p levels by introducing a high-copy number 2 m plasmid carrying RIF1 into yeast cells. As shown in Figure 3C, increasing Rif1p reduced telomere length in wild-type cells. Long telomeres observed in hpr1 and tho2 cells were suppressed when Rif1p levels were brought back to the wild-type level. Significantly, there appeared to be a negative correlation between telomere lengths and the expression levels of Rif1p in the cells we examined. The results provide strong evidence supporting that the long telomeres observed in hpr1 and tho2 cells are caused by the reduction in the expression level of Rif1p.

Enhancement of telomere silencing in hpr1 and tho2 cells
The function of RIF1 on telomeres is not limited to telomere length regulation. It has been reported that rif1 also enhances the telomere silencing effect [26,46]. To test if mutations of THO components affect telomere silencing, expression of URA3 at varying distances from telomeres was analyzed in strains carrying THO mutations. As shown in Figure 4A, stronger repression of URA3 was observed in cells harboring a tho2 mutation. The enhancement of URA3 repression was observed in all three strains tested and was most obvious in UCC509. The effect of hpr1 and rif1 on telomere silencing was then evaluated using UCC509. We anticipated that hpr1 would affect telomere silencing in a way similar to that of tho2 mutation and that the addition of the hpr1 or tho2 mutation to the rif1 mutant should not further enhance telomere silencing. The results shown in Figure 4B, tho2 and hpr1 showed similar enhancement of telomere silencing, ant the enhancement in both THO mutants was less than that conferred by the rif1 mutation. The addition of the tho2 or hpr1 mutation had no further effect on telomere silencing in the rif1 mutant. Thus, mutations on the two THO components also affect telomere silencing through a pathway mediated by RIF1.
Overexpressing SUB2 cannot restore the Rif1p level or telomere length in tho2 and hpr1 cells A high-copy suppressor screen using lacZ fused to a strong promoter has found that SUB2 suppresses the transcriptional defect of hpr1 [47][48][49]. SUB2 encodes a DEAD-box RNA helicase that is involved in splicesosome assembly and mRNA exporting [50][51][52]. Sub2p binds to Yra1p and the THO complex to form a TREX (TRanscription/EXport) complex that couples elongation and exporting of mRNA [35,52,53]. To test whether overexpressing SUB2 also recovers RIF1 expression and restores telomere length in THO mutants, tho2 and hpr1 cells with a plasmid carrying SUB2 or sub2-5 were analyzed for Rif1p levels and telomere length. The sub2-5 allele expresses a mutant Sub2p with a Q308R mutation near the helicase motif IV, which does not cause any detectable defects in splicing activity or spliceosome assembly [54]. However, the expression level of sub2-5 is higher than that of wild-type SUB2 in cells for unknown reasons [55]. As shown, overexpressing SUB2 or sub2-5 could rescue the growth defect of tho2 and hpr1 at 37uC (Fig. 5A), but did not restore the Rif1p level (Fig. 5B). Our results of real time RT-PCR experiments also showed that the RIF1 RNA in both tho2 and hpr1 cells was not restored to the wild-type level by overexpressing SUB2 (Fig. 5C). Consequently, telomere length in the tho2 and hpr1 cells remained long (Fig. 5D). Thus, while it is unclear why the Rif1p level cannot be restored through overexpressing SUB2 or sub2-5, the results support our conclusion that the lowered Rif1p level is the cause for long telomeres in tho2 and hpr1 cells.

Discussion
Telomere length homeostasis is mainly achieved through regulation of telomerase and telomere-associated proteins. Here we discovered that THO complex components Hpr1p and Tho2p also affect telomere length and telomere silencing. Long telomeres and enhanced telomere silencing were observed in hpr1 and tho2 cells. We also discovered that RIF1 expression was greatly reduced in hpr1 and tho2 cells, which is in agreement with a previous observation that showed selective reduction of long transcripts by THO mutations [36]. Moreover, our genetic analyses using double mutations and overexpression demonstrated that telomere phenotypes observed in hpr1 and tho2 cells were caused by reduction of Rif1p levels. Thus, we have provided evidence that THO mutations exert their effects on telomeres through affecting the expression of Rif1p, the largest telomere-associated protein.
It has been shown that the hpr1 and tho2 mutations have the highest impact on gene expression and recombination among THO mutations [56]. Similarly, here we found that mutations on HPR1 and THO2 affected both telomere length and telomere silencing, whereas mutations on the other two THO components, Mft1p and Thp2p, did not affect Rif1p level and telomere lengths.
Thus, it appears that the roles of Hpr1p and Tho2p are functionally distinct from those of Mft1p and Thp2p. Although the function of each component within the THO complex and the detailed mechanism underlying the differential impacts of individual THO mutations are not clearly defined, there appears to be a hierarchical difference within these components in terms of their in vivo biological relevance. Among these four components, the Tho2p and Hpr1p subunits are more important for telomere homeostasis.
Mutations of components of the THO complex cause a reduction in the transcription of yeast genes [34,36-38,45,57]. Using the E. coli lacZ gene as a reporter, it has been shown that the level of gene transcription decreases with increasing gene length [39]. Further analysis has revealed that high G+C content might contribute to the reduced expression of lacZ [39]. Here we demonstrate a preferential reduction of RIF1 expression in hpr1 or tho2 cells. The result is consistent with the prediction of preferential reduction of long genes in THO mutants. We did not find noticeable reduction in the expression of high G+C content genes in hpr1 or tho2 cells. It is possible that because the overall G+C contents of the genes encoding telomere-associated proteins are low, the effect of G+C content becomes less apparent. In any case, the G+C content did not appear to have a role in modulating the expression of genes encoding telomere-associated proteins in cells depleted of the THO complex. A genome-wide expression analysis using microarray has showed that the TEL1 level was significantly decreased in hpr1 cells [44]. Our analysis also shows that the TEL1 RNA levels are decreased in both tho2 and hpr1 cells. Because cells harboring a tel1 mutation have short telomeres [19], a combinatory effect of telomere shortening and lengthening was observed in the tho2 tel1 double mutant cells (Fig. 3B). This suggests that the telomere lengthening observed in tho2 or hpr1 cells is not due to the reduced TEL1 RNA levels. It is possible that although TEL1 expression is reduced in THO mutants, the amount of Tel1p is still sufficient to maintain its function on telomeres. In contrast, a ,40-50% reduction of RIF1 RNA level causes a ,75-80% reduction of Rif1p level, which is sufficient to contribute to an increase of telomere length. The results implicate that the Rif1p level has to be tightly regulated to stably maintain the telomere length. Of note, although we found significant reduction of RIF1 expression in both tho2 and hpr1 cells, RIF1 was not identified among the repressed genes in THO mutants in the genome-wide expression analyses. The microarray analyses considered the genes with expression levels at least 1.5-fold above or below the wild-type values as significantly different [44]. It is possible that the ,40-50% reduction of the RIF1 RNA level in tho2 and hpr1 cells evaded the detection by microarray analyses.
Through a high-copy suppressor screen, it has been discovered that overexpressing SUB2 suppressed both the growth and transcription defects of tho2 and hpr1 mutants [47][48][49]. It is surprising to us that the Rif1p levels were not restored by overexpressing SUB2 or sub2-5 in tho2 and hpr1 cells, although the growth defects were abrogated. The cause of the temperaturesensitive growth in tho2 and hpr1 cells is unclear. Given roles of the THO complex in transcription and mRNA export, it is possible that the growth defect is due to the impairment of bulk transcription-, mRNA export, or the generation of transcriptional byproducts upon the collapse of these gene expression processes. Elucidation of the involving molecular mechanism awaits the identification of target yeast genes affected by the THO complex mutations. Alternatively, it is also possible that due to a separation of function conferred by the sub2 alleles, SUB2 overexpression leads to suppression of temperature sensitivity but not telomere phenotypes. Indeed, it has been reported that SUB2 overexpression in tho2 cells only recovers ,20-30% of the lacZ activity driven by a GAL1 promoter [49], further supporting that SUB2 might function differently in suppressing growth defect and telomeres. No matter what mechanism is used by Tho2p and Hpr1p in   suppressed the growth defect of tho2 or hpr1 cells. Strains of the indicated genotypes derived from THO2/tho2, or HPR1/hpr1 diploids carrying SUB2 or sub2-5 overexpressing plasmids were grown on YC plates at 30u or 37uC. (B) SUB2 or sub2-5 overexpression did not restore the Rif1p level in tho2 or hpr1 cells. Immunoblotting assays were carried out as described. (C) SUB2 overexpression did not restore the RIF1 level in tho2 or hpr1 cells. Total mRNA from the indicated strains was isolated and analyzed for the RIF1 RNA level using real time RT-PCR. Results were presented as relative levels normalized to the wild-type expression level. The bars were standard deviations calculated using data from three independent experiments. (D) SUB2 or sub2-5 overexpression did not restore the telomere length in tho2 or hpr1 cells. Telomere length analyses were performed as previously described. doi:10.1371/journal.pone.0033498.g005 controlling gene expression, our results still clearly indicate that the lowered Rif1p level is the cause of long telomeres in tho2 and hpr1 cells.

Telomere length determination
To determine telomere lengths, yeast DNA was prepared, digested with XhoI, and separated on 1% agarose gels. Next, DNA fragments were transferred to a Hybond N + paper (Amersham) for hybridization using a random primed Y9 DNA probe [61].

Northern Blotting analysis
Yeast cells were grown to an OD 600 of 0.6 and then RNA was extracted using the hot phenol extraction protocol. Briefly, yeast cells were harvested by centrifugation and quickly resuspended in cold TES (10 mM Tris pH 7.5, 10 mM EDTA, 0.5% SDS). Equal volumes of acid phenol (Phenol : Chloroform = 5 : 1, pH 4.5, Sigma) were added and the samples were incubated at 65uC for 60 min. Acid phenol extraction was repeated several times until the interface was clear. The extracted RNA was extracted with equal volumes of CHCl 3 and precipitated by ethanol. Ten mg of the extracted RNAs were separated by 1.2% formaldehyde agarose gels, transferred onto Hybond N + membranes (Amersham), and hybridized with DNA probes prepared by random priming. The primers used to prepare PCR amplified DNA probes were: Rif1 forward (

Assay for telomere position effect
Strains UCC507, UCC509, and UCC511 were used to evaluate the effect of telomere silencing. These strains were crossed with YPH500 carrying tho2, hpr1, or rif1 mutations. The resulting diploid strains were induced into meiosis and the resulting tetrads were dissected. Haploid strains with URA3 placed near the telomere and tho2, hpr1, or rif1 mutations were then selected. These cells were streaked on YC plates and grown for 3 days at 30uC. Colonies were resuspended in water, and aliquots of different dilutions were spotted on YC plates, YC plates lacking uracil or plates containing 5-FOA. Plates were incubated at 30uC until colonies formed. Usually, it takes 3 days for colonies to form on YC plates, and it takes 6-7 days on plates containing 5-FOA.