|Home | About | Journals | Submit | Contact Us | Français|
THO/TREX, a conserved eukaryotic protein complex, is a key player at the interface between transcription and mRNP metabolism. The lack of a functional THO complex impairs transcription, leads to transcription-dependent hyperrecombination, causes mRNA export defects and fast mRNA decay, and retards replication fork progression in a transcription-dependent manner. To get more insight into the interconnection between mRNP biogenesis and genomic instability, we searched for HPR1 mutations that differentially affect gene expression and recombination. We isolated mutants that were barely affected in gene expression but exhibited a hyperrecombination phenotype. In addition, we isolated a mutant, hpr1-101, with a strong defect in transcription, as observed for lacZ, and a general defect in mRNA export that did not display a relevant hyperrecombination phenotype. In THO single-null mutants, but not in the hpr1 point mutants studied, THO and its subunits were unstable. Interestingly, in contrast to hyperrecombinant null mutants, hpr1-101 did not cause retardation of replication fork progression. Transcription and mRNP biogenesis can therefore be impaired by THO/TREX dysfunction without increasing recombination, suggesting that it is possible to separate the mechanism(s) responsible for mRNA biogenesis defects from the further step of triggering transcription-dependent recombination.
Gene expression requires the correct coupling of transcription to mRNA processing and export (7, 17, 30, 31). The final outcome of transcription is a ribonucleoprotein particle (mRNP) consisting of an mRNA molecule bound to proteins, which later function in export, mRNA surveillance, and translation. Transcription and assembly of the mRNP are tightly coordinated with the export of mRNA (42). A number of proteins have been identified that are important for the correct assembly of export-competent mRNP particles, including the THO complex, composed of Tho2, Hpr1, Mft1, and Thp2, as identified in Saccharomyces cerevisiae (9). THO has also been purified from humans and Drosophila, in which Hpr1 and Tho2 homologues are present (32, 40). In yeast and humans, it has been shown that THO is part of the TREX complex, which includes the Sub2/UAP56 and Yra1/Aly mRNA export factors (40). Functional connections between THO and the Sub2 and Yra1 export factors have been shown by synthetic lethal phenotypes of double mutants and by multicopy suppression (11, 19, 40).
Null mutants of the THO complex are impaired in transcription elongation (8, 23, 36). Additionally, these mutants exhibit a plethora of phenotypes, including transcription-dependent hyperrecombination (2), nuclear mRNA retention (40), and fast mRNA degradation (22, 46), as well as synthetic lethality in combination with mutations affecting transcription elongation such as spt4 (35) and with RNA export mutations such as mex67-5 and yra1-1 (19, 40). Altogether, these data suggest that THO plays a crucial role in mRNP metabolism at the interface between transcription and mRNP export. Mutants with changes affecting the Sub2-Yra1 component of TREX, the Mex67-Mtr2 mRNA export factor, the Nab2 hnRNP, or the Thp1-Sac3-Sus1 RNA export complex exhibit transcription and genetic instability phenotypes similar to those of THO mutants (12, 14, 19).
The THO complex is exclusively recruited to transcribed chromatin (20, 40, 46). Recruitment of the Sub2 and Yra1 components of the THO/TREX complex to transcribed DNA is partially dependent on THO (46). In contrast to THO, binding of Sub2 to transcribed chromatin is RNA dependent (1). The relevance of THO in the early steps of mRNP biogenesis is supported by the observation that transcription impairment in hpr1Δ mutants is suppressed if the nascent mRNA is self-cleaved by an artificially engineered hammerhead ribozyme. In these mutants, the nascent mRNA can form RNA-DNA hybrids (R loops) linked to both impairment of transcription and transcription-associated recombination (18). The observation that depletion of the vertebrate ASF/SF2 splicing factor in chicken DT40 cells and human HeLa cells also leads to genomic instability linked to R loop formation (21) indicates that a number of mRNA processing enzymes may have a role in preventing RNA-dependent structures that trigger genome instability. Our current view is that the THO complex participates in cotranscriptional assembly of export-competent mRNP during transcription elongation and contributes to the prevention of the RNA from forming structures that hamper transcription and could compromise genome stability (3).
Despite the link between impaired transcription and hyperrecombination in THO mutants, it is not clear whether impaired gene expression by itself is sufficient to induce hyperrecombination or whether hyperrecombination is possible in the absence of impaired gene expression. Thus, we searched for THO mutants compromised in gene expression that did not affect genetic integrity and vice versa. Hyperrecombination mutations that did not have a significant effect on transcription were not found. Instead, we isolated a new allele, hpr1-101 (L586P), which causes a strong decrease in transcript accumulation but does not show a relevant hyperrecombination phenotype. The in vivo stability of THO and its subunits was strongly reduced in THO-null mutants such as hpr1Δ but not in hpr1 point mutants. The absence of associated hyperrecombination seems to be due to the fact that, in contrast to hpr1Δ, transcription impairment in hpr1-101 does not hamper replication fork progression. Our results indicate that THO dysfunction can cause transcription defects that are not linked to impairment of replication fork progression, opening the possibility that THO links optimal mRNP biogenesis with transcription elongation, apart from its effect on genomic integrity.
The strains used in this study are listed in Table Table1.1. All of the strains are isogenic to W303-1A, except LAU3-10A and LAU3-10D, which are congenic segregants from a SChY58A × AYW3-1B cross. Strains WH101-1A and WH103-1A, isogenic to W303-1A but harboring the hpr1-101 and hpr1-103 alleles in the HPR1 chromosomal locus, were obtained by gene replacement in SChY58a (hpr1Δ::KAN). Plasmids pRS313, pRS316 (38), YCp70, YCpA13 (4), pRS314GLlacZ (15), YEp351-SUB2, pCM184-LAUR (19), pRS316-GAL1lacZ, pSCh202, pSCh204, pSCh206 (8), pPHO5-Rib+-lacZ (Rib+), pPHO5-ribm-lacZ (ribm), pGL-Rib+, pGL-ribm (18), and pRWY005 (43) have been described previously. Plasmid pRS313-GZ was constructed by cloning the GAL1pr::lacZ fusion from pRS316-GAL1lacZ into pRS313 in vivo as previously described (29). pRS316-HPR1 was obtained by cloning the 6.55-kb HindIII HPR1 fragment from YCpA13 into the HindIII site of pRS316. Plasmids YCphpr1-101 to YCphpr1-104 were obtained by in vivo cloning of linear mutagenized HPR1 PCR fragments into YCp70 in yeast strain LAU3-10A as described below.
In vitro mutagenesis of HPR1 was performed according to reference 27, with some modifications. PCR amplification of HPR1 was done under suboptimal conditions with 3.25 mM MgCl2, 0.25 mM MnCl2, 0.2 mM dTTP, 0.2 mM dGTP, 0.2 mM dCTP, 0.04 mM dATP, and 1.75 U of Expand High Fidelity Taq polymerase (Roche). YCpA13 (1 μg) was used as the template, and the oligonucleotides 5′ GCCACGTTTGTTTACTCG 3′ and 5′ CTATCATCCAACGCTTCC 3′ were used as primers. The LAU3-10A strain (hpr1Δ), previously transformed with pRS313-GZ, was cotransformed with the purified, PCR-mutagenized HPR1 mixture and the linear BamHI-XbaI fragment of YCpA13, which corresponds to YCp70 carrying 200 bp of HPR1 flanking sequences on each side. In vivo recombination between the PCR-induced hpr1 alleles and the linear vector led to formation of the YCphpr1 plasmids in the cells. Transformants thus obtained were plated on synthetic complete (SC) medium without histidine and leucine containing limited amounts of adenine (1.5 mg/liter) for visualization of red sectors resulting from recombination of the chromosomal leu2-k::ADE2-URA3::leu2-k repeat system. Colonies were then replicated onto SC without histidine and leucine containing 2% galactose to allow expression of the GAL1pr::lacZ fusion. On the following day, a solution containing 0.25 M phosphate buffer (pH 7.5), 0.1% sodium dodecyl sulfate (SDS), 400 μg/ml 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal), and 0.5% agarose was added and β-galactosidase activity was scored by blue color formation after 2 h at 30°C. Colonies transformed with either YCpA13 or YCp70 were used as positive (wild type) or negative (hpr1Δ) controls, respectively.
The THO complex was purified by using the C-terminally tandem affinity purification (TAP) epitope-tagged Tho2 protein and following standard procedures (33) and using specifications previously reported (19). Mft1 was detected by Western analysis of purified THO complexes previously separated by 10% SDS-polyacrylamide gel electrophoresis (PAGE) and transferred to nylon membranes. After treatment with phosphate-buffered saline containing 0.1% Tween 20 and 5% milk, proteins were detected with anti-Mft1 antibody and peroxidase-conjugated goat anti-rabbit immunoglobulin G (IgG). Blots were washed with phosphate-buffered saline-0.1% Tween 20 and developed by enhanced chemiluminescence reactions (Amersham). The Tho2-TAP epitope fusion was detected in whole cell extracts with peroxidase antiperoxidase (PAP; Sigma) antibody by following the manufacturer's instructions as described previously (19).
Cells were grown overnight to an optical density at 600 nm of 0.2 to 0.3 in SC medium containing 1% glycerol-lactate and 1% raffinose as a carbon source. After bringing the culture to 2% galactose, cells were grown for another 3.5 to 4 h prior to DNA isolation. DNA of 100 ml of cells was isolated with cetyltrimethylammonium bromide (44), dissolved in a final volume of 100 μl of Tris-EDTA, and stored at 4°C. Prior to analysis, 50 μl of DNA was double digested with SmaI-SacI. After digestion, DNA was precipitated and subjected to two-dimensional (2D) gel electrophoresis as described previously (6), with the following modifications. DNA was separated for 20 h in the first dimension on a 0.5% agarose gel at 1 V/cm and for 14 h in the second dimension on a 1.2% agarose gel at 4 V/cm and 4°C in the presence of 0.3 μg/ml ethidium bromide. After alkaline transfer of the DNA onto a Hybond-N+ membrane, the membrane was hybridized against a 32P-labeled 1.2-kb fragment obtained by PCR covering ARS1 and part of the 3′-lacZ gene. Membranes were exposed on a Fuji 3600 Phosphorimaging system and analyzed with the imaging software supplied.
Recruitment of Sub2, Tho2, and RNAPII to chromatin was determined by chromatin immunoprecipitation (ChIP) analyses on the endogenous PMA1 gene in yeast strains containing a 5′-end TAP epitope-tagged version of either the SUB2 or the THO2 gene inserted at its chromosomal locus, as previously described (16). Briefly, immunoprecipitations were performed with Ig-Sepharose for TAP epitope-tagged proteins or with anti-Rpb1-CTD monoclonal antibody 8WG16 (Berkeley Antibody Company) and protein A-Sepharose for RNA polymerase II (RNAPII). The GFX purification system (Amersham) was used for the last DNA purification step. We used 20- to 27-bp oligonucleotides for PCR amplification of PMA1 sequences. For Tho2 recruitment, we amplified two fragments, localized at positions 292 to 321 and 2273 to 2242, whereas for Sub2 and RNAPII recruitment, we only amplified the 2273-to-2242 fragment. In both cases, we used the PCR of the intergenic region at positions 9716 to 9863 of chromosome V as a negative control. The PCR products obtained were electrophoresed in a 15% acrylamide gel, stained with ethidium bromide, and quantified in a Fuji FLA-3000. The relative abundance of each DNA fragment was calculated as the ratio of each DNA fragment to the intergenic-region quantification results of the precipitated fractions normalized with respect to the corresponding ratios of the input fractions.
For analysis of RNAPII processivity, a strain harboring either YLR454 or lacZ under the control of the GAL1 promoter was used (23). RNAPII abundances at different positions of the genes were assayed as indicated, by using ChIP with anti-Rpb1-CTD monoclonal antibody 8WG16 and protein A-Sepharose as previously described (23). We used 20- to 30-bp oligonucleotides for PCR amplification of two fragments of YLR454, positions 3 to 43 and 7621 to 7674; of two fragments of lacZ, positions 715 to 1025 and 2247 to 2467; and of the 9716-to-9863 intergenic region of chromosome V that was used as a control. Real-time quantitative PCR was performed with SYBR green dye in the 7500 Real Time PCR system of Applied Biosystems by following the manufacturer's instructions.
Recombination frequencies were calculated as previously described (18). For each genotype, the recombination frequencies are given as the average and standard deviation of the median recombination value obtained from two or three different transformants with 6 to 12 independent colonies per transformant. β-Galactosidase and phosphatase assays and Northern analyses were performed according to previously published procedures (18). mRNA export assays were performed by in situ poly(A)+ RNA localization with a synthetic digoxigenin-labeled oligo(dT)20 as previously described (5, 14).
Mutagenic PCR-generated hpr1 alleles were cloned into the YCp70 plasmid by in vivo homologous recombination (see Materials and Methods). hpr1 mutations that differentially affected gene expression and recombination were screened by a color assay based on the chromosomal leu2-k::ADE2-URA3::leu2-k direct-repeat construct (4). With this construct, wild-type cells form white colonies (Ade+) whereas hpr1 cells form red-sectoring colonies (Ade+ Ade−). Defects in gene expression were determined with a GAL1pr::lacZ fusion construct which results in a dark blue color in expression-competent wild-type colonies and a light blue color in expression-deficient hpr1 mutant colonies. Screening of 5,346 transformant colonies permitted the identification of four hpr1 mutants (hpr1-101 to hpr1-104) which, upon subsequent analyses, were verified as being preferentially hyperrecombinant or defective in lacZ expression. The hpr1 alleles were sequenced, and the mutations responsible for the amino acid changes of each allele were determined (Fig. (Fig.1A).1A). hpr1-101 harbors two point mutations, but only one (L586P) leads to amino acid substitution. Interestingly, although the primary sequence of the Hpr1 protein is not particularly conserved, this mutation lies in a highly conserved region (Fig. (Fig.1B).1B). hpr1-102 has a −1 frameshift that leads to a premature stop codon and a carboxy-terminally truncated Hpr1 protein. hpr1-103 and hpr1-104 harbor multiple point mutations leading to 12 and 6 amino acid substitutions, respectively.
Next, recombination frequencies and β-galactosidase activities were determined in a LAU3-10A hpr1Δ strain carrying the leu2-k::ADE2-URA3::leu2-k and GAL1pr::lacZ systems, respectively, and transformed with the YCp70::hpr1 series of plasmids (hpr1-101 to hpr1-104), YCpA13 (HPR1), or the empty vector YCp70 (hpr1) (Fig. (Fig.1C).1C). hpr1Δ cells display a strong reduction in β-galactosidase activity (a 25-fold decrease) that correlates with a strong increase in recombination (773-fold above the wild-type level). Like hpr1Δ cells, hpr1-101 and hpr1-102 cells exhibit a significant reduction in β-galactosidase activity (4.3- and 6.7-fold decreases, respectively) which, in contrast to hpr1Δ, was accompanied by a weak increase in recombination (5- and 50-fold, respectively). Instead, β-galactosidase activity in hpr1-103 and hpr1-104 cells was poorly affected (1- to 2-fold decreases), while recombination values (239- and 567-fold above the wild-type level) were close to the values obtained in hpr1Δ. We therefore decided to further analyze hpr1-101 and hpr1-103 in order to understand the basis for the different phenotypes they produce.
The strength of the gene expression defect and hyperrecombination phenotypes of hpr1Δ has been shown to be dependent on the DNA sequence (10). Thus, transcription through the lacZ bacterial gene is severely compromised in hpr1Δ and leads to strong hyperrecombination, but the transcription defect and hyperrecombination phenotypes are weak or barely detectable in other DNA sequences, such as that of PHO5. To understand the connection between the gene expression and recombination phenotypes observed in the hpr1 point mutants, we further analyzed the most representative alleles, i.e., hpr1-101 (strong gene expression defect and weak hyperrecombination) and hpr1-103 (weak gene expression defect and strong hyperrecombination). Thus, we constructed isogenic strains harboring single chromosomal copies of hpr1-101 (WH101-1A) and hpr1-103 (WH103-1A) at the HPR1 locus by gene replacement.
Recombination analysis in the L-lacZ system, containing lacZ flanked by leu2 direct repeats, revealed that, compared to hpr1Δ (recombination 87-fold above the wild-type level), hyperrecombination was weak in hpr1-101 (8-fold) and moderate in hpr1-103 (20-fold) (Fig. (Fig.2A).2A). Next, we analyzed recombination in the L-PHO5 and GL-lacZ systems, the latter of which is the same as L-lacZ except that transcription is driven by the GAL1 promoter. With either actively transcribed L-PHO5 or GL-lacZ under repressive conditions (2% glucose), we obtained wild-type recombination levels in the L-PHO5 and GL-lacZ systems in both hpr1-101 and hpr1-103, respectively. These results indicate that hyperrecombination was transcription and sequence dependent in both mutants. In order to determine whether transcription impairment and hyperrecombination in these mutants are associated, we analyzed transcript levels in the same system used to measure recombination. Northern analysis showed similar L-PHO5 transcript levels for all of the strains (Fig. (Fig.2B).2B). Importantly, in the hpr1-103 mutant the transcript levels of lacZ mRNA expressed from the L-lacZ system only reached 75 to 80% of the wild-type levels, which clearly differs from the β-galactosidase results obtained in the GAL1pr::lacZ system (Fig. (Fig.1C1C and data not shown). Thus, when transcription was analyzed in the same system in which hyperrecombination was detected, the hpr1-103 mutant showed the same phenotypes as hpr1Δ, although to a lesser extent. As the L-lacZ and GAL1pr::lacZ systems differ in the length of the transcribed DNA sequence, it seems that the hpr1-103 effect on transcription was detectable in long DNA sequences. This is consistent with previous observations indicating that hyperrecombination is only observed in association with a detectable transcription defect (2).
Since hpr1-101 did not display a significant hyperrecombination phenotype compared with hpr1Δ, we wondered whether this mutation leads to a real impairment in RNAPII transcription, as is the case for hpr1Δ strains. First we determined gene expression from GAL1pr::lacZ and GAL1pr::PHO5 in hpr1-101 mutant cells compared to wild-type and hpr1Δ cells. As shown in Fig. Fig.3A,3A, whereas β-galactosidase activity was clearly diminished in hpr1-101, acid phosphatase was not affected. This pattern is similar, although to a lesser extent, to the results obtained for the null hpr1Δ strain.
Next, we determined whether the reduction of mRNA levels in hpr1-101, as in hpr1Δ, was different, depending on the gene tested. We analyzed the mRNA levels in various open reading frames (ORFs) driven by the strong GAL1 promoter. These ORFs included lacZ from Escherichia coli (3 kb and 53% GC content); PHO5 (1.5 kb and 40% GC), YAT1 (2 kb and 51% GC), and YRL454 (8 kb and 41% GC) from S. cerevisiae; and LAC4 (3 kb and 41% GC) from Kluyveromyces lactis. As shown in Fig. Fig.3B,3B, hpr1-101 causes a reduction as great as that produced by hpr1Δ in the mRNA levels of lacZ and YRL454 but has little effect on PHO5, consistent with the enzymatic analysis (Fig. (Fig.1).1). However, it has only a moderate effect on YAT1 and LAC4 mRNA levels. These results suggest that the strength of the deficiency in mRNA accumulation in hpr1-101 cells is also ORF dependent.
In hpr1Δ strains, RNAPII processivity, defined as the ability of RNAPII to travel the entire length of a gene, is impaired (23). Although mRNA levels were affected in hpr1-101, it was conceivable that RNAPII processivity was not impaired in this strain, therefore separating RNAPII processivity from hyperrecombination. To test this hypothesis, we determined the processivity of RNAPII in hpr1-101 along the lacZ and YRL454 genes by ChIP. This was performed with an anti-Rpb1 antibody and quantitative standard and real-time PCRs at the 5′ end and 3′ end of each coding sequence. As shown in Fig. Fig.4,4, RNAPII is significantly reduced toward the 3′ end of lacZ in hpr1Δ, as well as hpr1-101 (Fig. (Fig.4A).4A). In YRL454, RNAPII was significantly reduced at the 3′ end in the hpr1Δ strain and to a lesser extent in the hpr1-101 strain (Fig. (Fig.4B).4B). These results are consistent with the mRNA accumulation results observed for each gene and indicate that RNAPII processivity is diminished in hpr1-101. Thus, we conclude that impaired RNAPII processivity by itself is not sufficient to lead to hyperrecombination. Instead, in hpr1-103, RNAPII processivity along lacZ was not affected (Fig. (Fig.4B),4B), in agreement with the expression data (Fig. (Fig.1C),1C), but it was affected for the long (8-kb) YLR454 ORF to the same extent as in hpr1-101 (Fig. (Fig.4B).4B). This is consistent with the conclusion that hpr1-103 affects transcription to a minor degree but sufficiently to cause transcription-dependent hyperrecombination.
The transcriptional defect of THO-null mutants such as hpr1Δ has been shown to be associated with nuclear accumulation of poly(A)+ RNA caused by defects in mRNA export (19, 40). The functional relationship of THO with mRNA export is provided by the synthetic lethality of RNA export factor mutants in the absence of a functional THO complex, as demonstrated by the hpr1 mex67-5 ts double mutants. To explore whether the hpr1-101 and -103 point mutants were affected in RNA export, we first determined the capacity of the new mutations to rescue the synthetic lethality of the hpr1Δ mex67-5 mutant. For this, we constructed hpr1Δ mex67-5 strains carrying a URA3-based plasmid containing wild-type HPR1. These strains were transformed with a plasmid containing either mutant allele hpr1-101 or hpr1-103, and the capacity of each allele to rescue hpr1Δ mex67-5 lethality was determined by growth in SC medium plus 5-fluoroacetic acid (FOA). As shown in Fig. Fig.5A,5A, the ability of each hpr1 allele to rescue hpr1 mex67-5 lethality is linked to its transcriptional defect. hpr1-103, which had little effect on gene expression, almost completely rescued the synthetic lethality, whereas hpr1-101, which confers strong transcription defects, did not, suggesting a direct correlation between transcription and impaired mRNA export. Analysis of other hpr1 alleles isolated in this study, such as hpr1-102 and hpr1-104, confirmed this correlation between transcription defect and synthetic lethality with mex67-5 (data not shown).
We next analyzed the nuclear export of poly(A)+ mRNA by in situ hybridization with a 20-mer oligo(dT) fluorescent probe in the wild-type, hpr1-101, hpr1-103, and hpr1Δ strains (Fig. (Fig.5B).5B). Incubation of the cells at 37°C led to nuclear poly(A)+ mRNA accumulation in hpr1Δ and hpr1-101 cells. However, in hpr1-103 cells, mRNA accumulation was only observed after 6 h of incubation at 37°C. These results indicate that the gene expression defect in hpr1Δ and hpr1-101 cells directly correlates with their mRNA export defect.
The transcription impairment and hyperrecombination phenotypes of hpr1Δ cells have been shown to be dependent on structures mediated by the nascent mRNA (18). To assay the effect of the nascent mRNA on transcription, we used the previously defined Rib+ and ribm constructs (Fig. (Fig.6A),6A), in which a PHO5-Rib-lacZ transcriptional fusion containing either a wild-type (Rib+) or a mutated (ribm) Hammerhead ribozyme sequence was placed under the control of the GAL1 promoter (18). In both constructs, a 2.2-kb long mRNA is transcribed, but in the Rib+ construct the active hammerhead ribozyme cleaves the transcript, liberating a 1.8-kb mRNA. Meanwhile, RNAPII travels farther, producing a 0.6-kb mRNA. Northern analysis of the ribm construct revealed that less transcript was accumulated in hpr1Δ and hpr1-101 cells than in wild-type and hpr1-103 cells. However, ribozyme cleavage in the Rib+ construct led to an increase in the resulting 0.6-kb transcript so that similar mRNA levels were found in all strains.
Apparently, ribozyme cleavage was capable of increasing the amount of mRNA transcripts in hpr1Δ and hpr1-101 cells. In a previous work, we showed that ribozyme cleavage was capable of partially suppressing the hyperrecombination phenotype observed in hpr1Δ cells. Thus, we asked whether ribozyme cleavage might also be sufficient to reduce the hyperrecombination phenotype observed in hpr1-103 cells. Hyperrecombination was assayed with the GL-Rib+ and GL-ribm repeat systems containing PHO5, followed by active Rib+ or inactive ribm sequences between 0.6-kb-long leu2 direct repeats, respectively (Fig. (Fig.6B).6B). In the GL-ribm system, recombination was strongly increased in hpr1Δ (285-fold), moderately increased in hpr1-103 (17-fold), and poorly affected in hpr1-101 (3-fold). In all three strains, recombination levels were reduced two- to threefold by the ribozyme-mediated cleavage of the nascent mRNA. Therefore, both the transcription defect of hpr1-101 and the hyperrecombination of hpr1-103 are similarly dependent on the nascent mRNA, as shown for hpr1Δ mutants.
The THO complex is a highly stable complex which can be purified with salt concentrations as high as 1.2 M and remains stable in vivo in mutants of the TREX complex such as sub2Δ (9, 19). However, the integrity of the THO complex has not been studied in mutants of the THO subunits. To determine whether the different phenotypes of hpr1-101, hpr1-103, and the THO-null mutants were related to the stability of THO, the complex was purified from the different mutant strains. With a TAP epitope-tagged Tho2 protein, the complex was purified from whole cell extracts derived from wild-type, hpr1-101, hpr1-103, hpr1Δ, mft1Δ, and sub2Δ strains by two-step affinity purification (19, 33). Western blot analysis with a PAP antibody showed that the amount of TAP-Tho2 was similar in whole cell extracts of hpr1-101 and wild-type cells (Fig. (Fig.7A).7A). Although TAP-Tho2 seems slightly reduced in hpr1-103 and sub2Δ and the amount of Tho2 varies between experiments (Fig. (Fig.7A),7A), it was similar in all cases to that in wild-type cells. In contrast, in hpr1Δ and mft1Δ, loading of similar amounts of cell extract onto the gel did not permit detection of a signal corresponding to the full TAP-Tho2 protein (Fig. (Fig.7A).7A). Only when 10-fold more whole cell extract was loaded was a weak signal corresponding to the TAP-Tho2 protein detectable (Fig. (Fig.7A).7A). As THO2 transcript levels have been found to be similar in wild-type and hpr1Δ cells (13), the decrease in Tho2 must result from a posttranscriptional effect, suggesting that THO subunits may only be stable in the form of a complex.
Analysis of the TAP-Tho2 purified THO complex by gradient SDS-PAGE revealed that in the wild type, hpr1-101, hpr1-103, and sub2Δ, the four THO subunits (Tho2, Hpr1, Mft1, and Thp2) were present (Fig. (Fig.7B).7B). However, in hpr1Δ and mft1Δ, only the TAP-Tho2 protein was enriched by the TAP system. Thus, if one of the core proteins of THO is missing, the integrity of the THO complex is greatly reduced. In order to find out whether the remaining proteins form at least some kind of complex, we used an anti-Mft1 antibody to detect remaining traces of Mft1 in the TAP-Tho2 purified samples. Similar amounts of Mft1 were detected in wild-type, hpr1-101, hpr1-103, and sub2Δ cells, whereas no signal was detected in hpr1Δ and mft1Δ cells when similar amounts of TAP-purified Tho2 were loaded (Fig. (Fig.7C).7C). However, when 10-fold more Tho2 protein was loaded onto the gel, Mft1 could be detected in hpr1Δ TAP-Tho2 purified fractions (Fig. (Fig.7C).7C). These results indicate that the THO complex is unstable if one of its components is absent, such as Hpr1 or Mft1, but is stable in both the hpr1-101 and hpr1-103 mutants, as well as in mutants in which a TREX component, such as Sub2, is absent.
Since, contrary to hpr1Δ, THO is stable in hpr1-101 and hpr1-103 cells, we wondered if THO was also recruited into active chromatin in these mutants. With the same TAP-Tho2 strains as before, we found that hpr1-101 and hpr1-103 recruited the TAP-Tho2 subunit of THO to a transcriptionally active PMA1 gene (Fig. (Fig.8)8) at levels clearly above that of hpr1Δ mutants. Therefore, the two point mutations analyzed are not null alleles and form THO complexes that are competent for chromatin recruitment.
The facts that the presence of THO at the site of transcription elongation helps recruit Sub2 to transcribed chromatin (1, 46) and that multicopy SUB2 is able to suppress hpr1Δ (11, 19) are consistent with the hypothesis that Sub2 recruitment to the nascent mRNA plays a key role in gene expression. As THO is as stable in both hpr1-101 and hpr1-103 mutants as in wild-type cells, we wondered whether their respective phenotypes could be caused by different levels of Sub2 recruitment to transcribed chromatin. While testing this possibility, we first noticed that with a lacZ-URA3 translational fusion under the control of the tet promoter (LAUR system) (19), SUB2 overexpression suppressed the transcriptional defects of both the hpr1-101 and hpr1-103 mutants (Fig. (Fig.9A).9A). This was consistent with the possibility that Sub2 was not efficiently recruited to transcribed chromatin in both mutants. Again, hpr1-103 showed a transcription defect in extra-long DNA sequences (Fig. (Fig.9A)9A) that reinforces the conclusion that hpr1-103 is a leaky rather than a separation-of-function mutation.
Sub2 recruitment to active chromatin was directly assayed by ChIP in strains carrying a SUB2-TAP epitope fusion at the SUB2 chromosomal locus with Sepharose IgG and anti-RNAPII antibodies. As shown in Fig. Fig.9B,9B, there was no significant difference in the recruitment of RNAPII to the constitutively transcribed PMA1 gene between the wild-type and hpr1 mutant strains tested. Nevertheless, recruitment of Sub2-TAP was as inefficient in hpr1-101 and hpr1-103 cells as in hpr1Δ cells (55 to 60% of the wild-type level). Therefore, we can conclude that the specific gene expression phenotype of hpr1-101 mutants is not caused by different abilities to recruit Sub2 to the nascent mRNA and that inefficient Sub2 recruitment is not enough to trigger either the recombination or the transcription phenotype of THO mutants.
We have recently shown that in hpr1Δ mutants, hyperrecombination correlates with cell cycle-specific transcription (43). While S-phase-specific transcription was capable of inducing recombination, G2-specific transcription was not. At the molecular level, replication of the transcribed lacZ gene was accompanied with a slowdown of the replication fork during S phase. This slowdown was dependent on both transcription and RNA. We therefore reasoned that the absence of hyperrecombination in hpr1-101 mutants could be explained if reduced transcription was not associated with slowed-down replication fork progression. To test this possibility, we analyzed replication intermediates of plasmid pRWY005, which contains a replication origin facing the lacZ ORF driven by the GAL1 promoter (Fig. 10A). As shown in Fig. 10B, in contrast to the wild-type intermediates, hpr1Δ replication intermediates appear to accumulate around the inflection point (white arrow) and the bubble arc signal gets weaker, indicating a slowdown of fork progression within lacZ. However, such an accumulation of simple-Y molecules was not obvious in hpr1-101 cells, and the intensities of the bubble arc signals were similar in the hpr1-101 and wild-type strains. In contrast, replication intermediates from hpr1-103 cells accumulated at the inflection point and toward the 2n spike, although to a lesser extent than in hpr1Δ (Fig. (Fig.2).2). These results are consistent with the idea that transcription-associated hyperrecombination of THO mutants is linked to replication fork progression defects (43) and suggest that transcription impairment in hpr1-101 mutants does not lead to structures capable of slowing down or disrupting replication fork progression. Instead, the mild transcription defect of hpr1-103 seems to be sufficient to impair replication fork progression, consistent with its hyperrecombination phenotype.
Here we report the isolation and molecular characterization of novel THO complex mutations hpr1-103, which produces mutants similar to hpr1-null mutants but with weaker phenotypes, and hpr1-101, which impairs transcription and RNA export without causing hyperrecombination. As in hpr1Δ, the transcriptional defect of hpr1-101 is dependent on the nascent RNA from highly transcribed long and GC-rich DNA sequences, but it is not sufficient to impair replication fork progression. Therefore, this study indicates that hyperrecombination in all of the THO mutants so far studied is always linked to a transcription defect but that impaired RNAPII processivity alone is not sufficient to induce hyperrecombination as long as replication is not affected. These findings suggest that it is possible to separate the mechanism(s) responsible for mRNA biogenesis defects from the further step of triggering transcription-dependent recombination by THO/TREX dysfunction.
Impaired transcription and hyperrecombination are hallmark phenotypes of mutants of the THO complex (Fig. (Fig.2).2). Out of the screen for mutants that allow the dissection of these two phenotypes, hpr1-103, together with hpr1-102 and hpr-104, contains multiple point mutations and shows intermediate effects on gene expression and genetic instability, which can be explained by leakiness. Indeed, one of these mutations, hpr1-102, is a carboxy-terminal truncation similar to the first, leaky, hpr1-1 mutation identified (4). Of the three leaky mutants, hpr1-103 is particularly interesting because of its clear hyperrecombination phenotype and weak effect on transcription. A close look at its phenotypes validates our model in which hpr1-mediated hyperrecombination is a consequence of defective transcription. Thus, transcription impairment in hpr1-103 is detected in DNA sequences such as the lacZ-URA3 fusion, and RNAPII processivity impairment is detected in long ORFs such as YLR454. In the direct-repeat construct (L-lacZ) in which hyperrecombination was observed in hpr1-103, transcription is also clearly reduced, validating the general conclusion that hyperrecombination in THO mutants is linked to a transcription defect. Indeed, hyperrecombination is only observed in hpr1-103 when the recombination system is transcribed and depends on the nascent mRNA molecule. Consistently, in hpr1-103 mutants, recruitment of the Sub2 subunit of TREX to transcribed chromatin is impaired and poly(A)+ RNA is accumulated in the nuclei, although to a lesser extent, as in hpr1Δ-null mutants, indicating that hpr1-103 is also impaired in mRNP formation. The phenotypes of hpr1-103 are also in agreement with the idea that various kinds of aberrant intermediates could be formed in the presence of a mutated THO complex with different outcomes in transcription and genomic instability. Therefore, the levels of hyperrecombination may not necessarily be proportional to the strength of the transcriptional defect but may reflect the formation of threshold levels of aberrant transcriptional intermediates, such as the RNAPII-DNA-RNA ternary complex, stalled or paused RNAPII, etc.
In contrast to the hpr1-102 to -104 leaky alleles, mRNA accumulation of DNA sequences such as lacZ in hpr1-101 cells are strongly reduced, but this is not linked to hyperrecombination. Although a weak increase in recombination is observed in this mutant, the strong gene expression defect, which is similar to that of hpr1Δ in lacZ, strongly suggests that its major type of transcription defect is not linked to hyperrecombination. The hpr1-101 mutation probably affects the quality of the mRNP formed as well, and thus its transcription impairment is linked to a synthetic lethality defect in combination with mex67-5 and a weak mRNA export defect and is dependent on the nascent mRNA molecule. However, our results clearly show that hpr1-101 cells have a major impairment in transcription elongation, as determined by ChIP, which reflects a defect in RNAPII processivity, as previously shown for hpr1Δ (23, 36). Indeed, hpr1-101 is synthetic lethal with the spt4Δ mutant of transcription elongation (data not shown), as also observed for hpr1Δ (35). Furthermore, if the low levels of RNA were primarily due to an increased RNA decay in hpr1-101, one could predict suppression in cells lacking the Rrp6 3′-5′ exo-RNase subunit of the nuclear exosome. However, lacZ mRNA levels remained low in hpr1-101 rrp6Δ double mutants (data not shown).
The gene expression defects of hpr1-101 are similar in intensity to those observed for hpr1Δ, including a selective impairment of transcription and a nuclear accumulation of poly(A)+ RNA that are consistent with suboptimal formation of export-competent mRNPs. In favor of this conclusion is the fact that, similar to hpr1Δ and hpr1-103 cells, hpr1-101 cells are impaired in the ability to recruit Sub2 to transcribed chromatin. Sub2, like its human ortholog UAP56 (37), may be an RNA-dependent ATPase required for the cotranscriptional formation of an export-competent mRNP particle. It functions as a heterodimer together with the Yra1 RNA export factor as part of the TREX complex (39, 40) and is recruited to the transcribed chromatin in a THO-dependent manner (46). Our observation that in hpr1-101 cells, overexpression of Sub2 suppresses the mRNA accumulation defect as in hpr1-103 and hpr1Δ cells is consistent with the idea that the formation of export-competent mRNPs in hpr1-101 mutants is caused by their inability to recruit Sub2 properly to the nascent mRNA. This means that in hpr1-101 mutants, a dysfunctional THO complex incapable of promoting proper mRNP biogenesis and export may be formed. However, the fact that Sub2 recruitment is equally impaired in hpr1-101 and hpr1-103, as well as in hpr1Δ, mutants, despite the fact that the intensities of their transcriptional and mRNA export defects are different, strongly suggests that THO has an additional role in mRNP biogenesis, regardless of its capability to help recruit Sub2 into transcribed DNA.
An outcome of our work is that THO is a biochemical and functional unit different from Sub2, suggesting that THO may constitute a functional core complex apart from the Sub2-Yra1 heterodimer. Null mutations in presumably all THO genes, as shown for HPR1 and MFT1, but not in the Sub2 component of TREX, led to the dissociation of the whole THO complex and the degradation of presumably all THO subunits, as shown for Tho2 (Fig. (Fig.7)7) (19). This could explain the previously reported similarity of phenotypes among the different THO-null mutants (9, 40). As the THO complex present in hpr1-101 and hpr1-103 is loaded onto active chromatin at similar level as in the wild type, the different gene expression defects of these mutants might be based on distinct interactions between THO and the nascent mRNP or the transcriptional apparatus or on the possibility that THO has several biochemical activities, which would be differentially affected by each point mutation.
The hpr1-101 allele leads to an L586P amino acid substitution in a well-conserved domain among Hpr1 orthologues from different species from yeast to humans (Fig. (Fig.1B).1B). Despite the fact that the biological function of this protein domain in the context of the whole protein is unknown, it seems that this mutation could define a functional motif relevant to the mRNP biogenesis function of THO. Although the change from L to P could introduce a hinge in the protein structure and therefore completely destroy the ability of the protein to interact with other components of THO, our protein purification analysis revealed that this is not the case and that the L residue is essential for its role in mRNP biogenesis and its capacity to facilitate Sub2 recruitment to actively transcribed DNA. The similarity of the phenotypes of sub2Δ cells and THO-null mutants (9, 19, 40) suggests that the integrity of a functional THO complex is not sufficient to guarantee proper mRNP biogenesis and export. Our study reveals the need for a functional TREX complex (THO-Sub2, Yra1) at the transcription site for the formation of an export-competent mRNP particle and confirms that THO constitutes a core structural unit apart from Sub2-Yra1. This may be consistent with the facts that, as in S. cerevisiae (9), THO has been purified independently of Sub2 in Drosophila (32); that humans hHpr1 and hTho2 can be purified apart from UAP56-Aly, orthologues of Sub2-Yra1 (24); and that Sub2 recruitment to active chromatin, contrary to THO recruitment, depends on RNA in yeast (1).
The difference in THO integrity between hpr1Δ and hpr1-101 confirms that, despite the strong effect on transcription and mRNA export of hpr1-101, the latter is not a null mutation. This conclusion is also clear from the observation that hpr1-101, in contrast to all THO-null single mutants, does not lead to hyperrecombination. This nonhyperrecombinant phenotype suggests that in THO/TREX mutants, RNA-dependent transcriptional intermediates resulting from impaired transcription elongation and defective cotranscriptional mRNP formation may not necessarily be sufficient to trigger hyperrecombination. This seems to be the case for hpr1-101. Such a nonhyperrecombinogenic intermediate may involve the nascent mRNA, as self-cleavage of the nascent mRNA affected transcript accumulation and recombination rates in all mutants, regardless of whether or not they were hyperrecombinant (Fig. (Fig.6)6) (18). It may be that the nascent RNA is different or that if it is the same, it is insufficient by itself to trigger hyperrecombination. Instead, a further recombinogenic step would be required. Several reports suggest that R loops can be intermediates associated with hyperrecombination in different systems (18, 21, 45). It may be that R loops represent a previous or cooperative step in hyperrecombination and, in addition, an mRNA-RNAPII-DNA tertiary structure able to obstruct replication or the action of a modifying enzyme (such as hAID in class switching), etc., would be required to trigger recombination.
We have recently shown that transcription impairment of hpr1Δ along a highly transcribed lacZ gene causes a retardation of replication, observed as a slowdown of fork progression (43). This slowdown of fork progression could partially be suppressed by ribozyme-mediated cleavage of the mRNA. As mitotic recombination is a postreplicative DNA repair pathway of, primarily, replication-dependent DNA breaks (25, 26, 34), it seems that in THO-null mutants only RNA-dependent transcriptional intermediates that interfere with replication are able to induce recombination. Consistent with this idea, transcription-dependent hyperrecombination in hpr1Δ is S phase dependent (43). Indeed, replication is impaired in the hyperrecombinant hpr1-103 mutant but not in hpr1-101, consistent with the idea that hyperrecombination in THO mutants correlates with replication fork impairment.
Different possibilities could explain why, in the absence of a fully functional THO complex, replication fork progression is affected whereas this is not the case in hpr1-101 cells. One possibility is that elongating RNAPII or another RNA-protein intermediate could be released from the transcription site in wild-type and hpr1-101 cells via, directly or indirectly, a THO-dependent function still present in Hpr1-101 complexes. Consistently, THO remains stable and seems to be recruited into chromatin in both hpr1-101 and hpr1-103 mutants. Further research is required to understand this, and the hpr1-101 mutation may be a useful tool with which to dissect both the mechanism(s) leading to impaired mRNP biogenesis and the transcription-dependent structure responsible for replication impairment and genomic instability in both THO/TREX mutants and wild-type cells.
In summary, this study indicates that transcription elongation in THO mutants can be impaired without leading to a downstream intermediate at the site of transcription responsible for hyperrecombination, unless such an intermediate impairs replication fork progression. The RNAPII processivity defect in THO mutants may reflect the inability of RNAPII to progress further if mRNP is not properly formed. Whether or not there is a checkpoint mechanism mediated by THO that causes RNAPII to halt or reduce elongation is an attractive possibility to be tested. In this scenario, a further mRNA-dependent step (formation of aberrant mRNA-RNAPII-DNA tertiary structures, etc.) able to impair replication fork progression would be required to cause hyperrecombination. This could be a general feature conserved from bacteria to eukaryotes, as transcription may be an obstacle to replication (28, 41).
We thank H. Gaillard for reading the manuscript, S. Jimeno for providing the Tho2-TAP mft1Δ MTT1 strain, E. Hurt, T. Lithgow, and K. Struhl for kindly providing reagents, and D. Haun for style supervision.
Grants from the Ministry of Science of Education of Spain (BMC2000-0409 and SAF2003-00204) and Junta de Andalucía (CVI102) supported this work. P.H. was the recipient of a predoctoral training grant from the Spanish Ministry of Science and Education.
Published ahead of print on 14 August 2006.