Molecular Genetics and Genomics

, Volume 279, Issue 2, pp 123–132

Different physiological relevance of yeast THO/TREX subunits in gene expression and genome integrity


  • María García-Rubio
    • Departamento de Biología Molecular, CABIMERCSIC, Universidad de Sevilla
    • Departamento de Genética, Facultad de BiologíaUniversidad de Sevilla
  • Sebastián Chávez
    • Departamento de Genética, Facultad de BiologíaUniversidad de Sevilla
  • Pablo Huertas
    • Departamento de Genética, Facultad de BiologíaUniversidad de Sevilla
    • The Wellcome Trust and Cancer Research UK Gurdon InstituteUniversity of Cambridge
  • Cristina Tous
    • Departamento de Biología Molecular, CABIMERCSIC, Universidad de Sevilla
    • Departamento de Genética, Facultad de BiologíaUniversidad de Sevilla
  • Sonia Jimeno
    • Departamento de Biología Molecular, CABIMERCSIC, Universidad de Sevilla
    • Departamento de Genética, Facultad de BiologíaUniversidad de Sevilla
    • Departamento de Biología Molecular, CABIMERCSIC, Universidad de Sevilla
    • Departamento de Genética, Facultad de BiologíaUniversidad de Sevilla
  • Andrés Aguilera
    • Departamento de Biología Molecular, CABIMERCSIC, Universidad de Sevilla
    • Departamento de Genética, Facultad de BiologíaUniversidad de Sevilla
Original Paper

DOI: 10.1007/s00438-007-0301-6

Cite this article as:
García-Rubio, M., Chávez, S., Huertas, P. et al. Mol Genet Genomics (2008) 279: 123. doi:10.1007/s00438-007-0301-6


THO/TREX is a conserved nuclear complex that functions in mRNP biogenesis and plays a role in preventing the transcription-associated genetic instability. THO is composed of Tho2, Hpr1, Mft1 and Thp2 subunits, which associate with the Sub2-Yra1 export factors and Tex1 to form the TREX complex. To compare the functional relevance of the different THO/TREX subunits, we determined the effect of their null mutations on mRNA accumulation and recombination. Unexpectedly, we noticed that a full deletion of HPR1, hpr1ΔK, conferred stronger hyper-recombination phenotype and gene expression defects than did hpr1ΔH, the allele encoding a C-terminal truncated protein which was used in most previous studies. We show that tho2Δ and, to a lesser extent, hpr1ΔK are the THO mutations with the highest impact on all phenotypes, and that sub2Δ shows a similar transcription-dependent hyper-recombination phenotype and in vivo transcription impairment as hpr1ΔK and tho2Δ. Recombination and transcription analyses indicate that THO/TREX mutants share a moderate but significant effect on gene conversion and ectopic recombination, as well as transcription impairment of even short and low GC-content genes. Our data provide new information on the relevance of these proteins in mRNP biogenesis and in the maintenance of genomic integrity.


THO complexSub2Genetic instabilitymRNP biogenesisTranscriptionTranscription-associated recombination


Transcription leads to the production of an mRNA molecule associated with a number of proteins that constitute an export-competent ribonucleoprotein particle, the mRNP. Many mRNA binding proteins have a later function in mRNA export, mRNA surveillance or translation. THO is a conserved eukaryotic complex, first identified in the yeast S. cerevisiae as composed of stoichiometric amounts of the proteins Tho2, Hpr1, Mft1 and Thp2 (Chávez et al. 2000). Later, homologous complexes were purified from humans and Drosophila (Strasser et al. 2002; Rehwinkel et al. 2004). A number of data indicate that the THO complex plays a functional role at the interface between transcription and mRNA export (Vinciguerra and Stutz 2004; Aguilera 2005). Thus, THO is recruited to transcribed chromatin while it is absent from non-transcribed chromatin (Strasser et al. 2002; Zenklusen et al. 2002; Kim et al. 2004), THO null mutants are viable but show impaired transcription elongation (Chávez et al. 2000; Rondón et al. 2003b; Mason and Struhl 2005), transcription-dependent hyper-recombination (Prado et al. 1997; Aguilera 2002), nuclear accumulation of poly(A)+, (Strasser et al. 2002) enhanced mRNA decay (Libri et al. 2002; Zenklusen et al. 2002), and impaired transcription-coupled repair (Gaillard et al. 2007). In addition, THO mutations are synthetic lethal with mutations affected in post-initiation steps of transcription such as spt4 (Rondón et al. 2003a), spt6 (Burckin et al. 2005), paf1 (Chang et al. 1999) or rna14 (Luna et al. 2005), as well as with RNA export mutations such as mex67-5 (Jimeno et al. 2002), and with sub2 and yra1 mutations (Strasser et al. 2002; Zenklusen et al. 2002). In yeast and humans, THO has been shown to be part of a larger complex termed TREX, together with the Sub2/UAP56 and Yra1/Aly mRNA export factors (Strasser et al. 2002). Functional connections between THO and the Sub2 and Yra1 export factors have been inferred from the synthetic lethal phenotypes of double mutants, the capacity of Sub2 overexpression to suppress THO mutations, and the phenotypic similarities of THO and sub2 mutations (Fan et al. 2001; Jimeno et al. 2002; Strasser et al. 2002).

Genetic and molecular analysis of the THO complex revealed a connection between mRNP biogenesis and genetic stability, which has also been inferred from the analysis of mutations in the Sub2-Yra1 component of TREX, Mex67-Mtr2 mRNA export factor, Nab2 hnRNP or Thp1-Sac3-Sus1 mRNA export complex (Fischer et al. 2002; Jimeno et al. 2002; Gallardo et al. 2003). The implication of THO in mRNP biogenesis is further supported by the observation that transcription impairment in hpr1ΔH mutants is suppressed if the nascent mRNA is self-cleaved by an artificially engineered hammerhead ribozyme (Huertas and Aguilera 2003). Furthermore, we have shown that the nascent mRNA can form RNA–DNA hybrids which are linked to the transcription-associated recombination. Currently, the THO complex is thought to participate in co-transcriptional assembly of export-competent mRNP during transcription elongation in a yet undefined manner.

Hpr1 and Tho2 are present in the THO complex of the three organisms from which it has been purified so far and have homologues in all eukaryotes with available genome sequences. This is consistent with the functional relevance of these two proteins in a basic and conserved nuclear process such as mRNP biogenesis. The deletion mutations of the four structural genes of THO show similar phenotypes of transcription impairment, hyper-recombination and nuclear mRNA accumulation, consistent with the idea that THO is a structural and functional unit. However, some differences in phenotype strength have been observed (Piruat and Aguilera 1998; Chávez et al. 2000). In order to define this putative difference and understand its biological meaning, we have undertaken a comparative analysis of the different phenotypes of representative mutants of THO/TREX. Our results indicate that the subunits with the major impact on gene expression and recombination are Tho2, Hpr1 and Sub2. Unexpectedly, the original hpr1ΔH mutation used in many previous studies, which encodes the first 112 N-terminal-amino acids of Hpr1, is slightly leaky with respect to the null hpr1ΔK mutation. tho2Δ and, to a lesser extent hpr1ΔK, confer a moderate but significant increase in inverted-repeat, allelic, and plasmid–chromosome recombination. Null mutations of HPR1, THO2 and SUB2 impair transcription of genes that are shorter and with a lower G-C content than previously shown for hpr1ΔH. This analysis adds further understanding and opens new questions about the in vivo physiological relevance of the different THO/TREX subunits.

Materials and methods

Strains and plasmids

All strains are isogenic to W303-1A, except the BY4741 (wild type) and its isogenic thp2Δ (BY-HR167). No growth differences were found for the mutants used in the two different backgrounds. The complete null allele of HPR1 is hpr1ΔK (hpr1Δ::KAN) (SChY58a strain), whereas hpr1ΔH (hpr1Δ::HIS3) (U678-4C) refers to the 112 N-terminal amino acid deletion allele. Yeast strains previously described are WMK-1A (mft1Δ::KAN) (Chávez et al. 2000), RK2-6C (tho2Δ::KAN) (Piruat and Aguilera 1998) and a Ura segregant of DLY23 (sub2Δ::HIS3) (Libri et al. 2001). For LOH assays the 8D8AC57 wild-type diploid strain (Aguilera and Klein 1988), and the congenic diploids carrying the hpr1/hpr1 and tho2/tho2 mutations were used. Strains used as reporters of mating efficiency were F4 (MATa thr4) and F15 (MATα thr1 arg4). The chromosomal recombination system his3P::INV and the plasmid-recombination systems have been previously described [Santos-Rosa and Aguilera (1994) references therein]. Plasmids used to determine recombination frequencies (pRS316, pSCh206, pSCh204, p414-GL-lacZ), and mRNA expression levels (pSCh255, pSCh247, pSCh227, p416GAL1-lacZ and pSCh202) have been described previously (Prado et al. 1997; Piruat and Aguilera 1998). The new plasmid-borne gene expression system tet::LYS2 is based on a 4.2-kb long LYS2 fragment that was placed under the control of the tet promoter in centromeric plasmid pCM184 (Gari et al. 1997).

Analysis of recombination

Recombination frequencies were calculated as previously described (Santos-Rosa and Aguilera 1994). For each genotype, the recombination frequencies are given as the average and standard deviation of the median recombination value obtained from two to three different transformants using 6 to 12 independent colonies per transformant. The frequency of MATa/MATα diploid cells that become able to mate with F4 and F15 reporter strains was taken as the frequency of loss-of-heterozygosity (LOH). Such LOH events can occur either by heteroallelic gene conversion between the two MAT alleles or by loss of chromosome III (Aguilera and Klein 1988).

mRNA analysis

Northern analyses were performed according to standard procedures. Kinetics of mRNA accumulation in the GAL1 promoter-based constructs was carried as described (Chávez et al. 2000). Mid-log phase plasmid-transformed cells were diluted in 3% glycerol–2% lactate synthetic complete medium SC and diluted into identical fresh media to an OD600 of 0.5 and incubated for 16 h. Galactose was then added and samples were taken for Northern blot analyses at different times. For Northern analyses of tet-driven expression, cells carrying the tet:LYS2 construct were grown to exponential phase in the absence of doxycyclin (transcription activation conditions).


Different growth phenotypes of mutant alleles of the THO/TREX subunits

Since the best characterized THO mutation hpr1Δ::HIS3 is an allele expressing a truncated Hpr1 protein of 112 amino acids (to be called hpr1ΔH from now on), we constructed a complete deletion of HPR1 by replacing the whole HPR1 gene with the KanMX4 cassette, which confers G418 resistance (Wach et al. 1997). The new null strain (hpr1ΔK), constructed in the W303 genetic background in which most previous studies have been performed, is viable and shows the thermosensitive (ts) phenotype previously observed for hpr1ΔH (data not shown). We used this strain to make a comparative analysis of growth efficiency at 30°C with other null mutants of THO/TREX. As can be seen in Fig. 1, sub2Δ was the strain that grew the worst of all strains tested. This is consistent with the observation that Sub2 is essential in most genetic backgrounds or the mutant grows poorly (Libri et al. 2001). tho2Δ and hpr1ΔK were the most affected among the THO mutants tested, whereas mft1Δ and thp2Δ grew similar to wild-type cells. Interestingly, hpr1ΔH grew slightly slower as compared to wild type cells but significantly better than hpr1ΔK. These results opened the possibility that the original hpr1ΔH mutation used in most previous studies could be leaky and that there was a hierarchical role of the different subunits of THO in terms of their in vivo biological relevance, the conserved Tho2 and Hpr1 subunits being the most important. Consequently, we asked next whether this difference also applied to two of the hallmark phenotypes of THO mutants, hyper-recombination and gene-expression defects.
Fig. 1

Comparative phenotypic analysis of THO/TREX mutants. Growth efficiency of wild type WT1 (W303–1A), wild type WT2 (BY4741), mft1Δ (WMK-1A), thp2Δ (BY-HR167), hpr1ΔH (U678-4C), hpr1ΔK (SChY58a), tho2Δ (RK2-6C), sub2Δ::HIS3 (Urasegregant of DLY23), on YPED medium at 30°C. Photographs were taken after 4 days at 30°C

Null tho2Δ and hpr1ΔK confer stronger hyper-recombination than hpr1ΔH, which is not restricted to direct repeat recombination

THO complex mutants confer strong stimulation of recombination that is dependent on RNAPII-driven transcription elongation and increases in other forms of genetic instability such as plasmid and chromosome loss (Santos-Rosa and Aguilera 1994; Chávez et al. 2000). We have extended our analyses on recombination to the new hpr1ΔK mutant in order to get further insight into the phenotype strength of the different mutants. The frequency of recombination of the hpr1ΔK and tho2Δ null strains was characterized in four DNA-repeat recombination systems and compared to that of hpr1ΔH. Three of the systems are located in the plasmids and are based on a 0.6 kb internal repeats of the LEU2 gene in direct (L and L-PHO5) or inverted (SU) orientation and were used for the analysis of deletions and inversions, which are scored as Leu+ recombinants. The fourth system (his3P::INV) is chromosomal and is based on a 3.0-kb inverted repeat containing the his3Δ5′ and his3-k alleles. It was used for the study of gene conversions, which are scored as His+ recombinants.

As can be seen in Fig. 2, whereas hpr1ΔH caused no significant increase of deletions in the L system, which contains no intervening sequence between the repeats, hpr1ΔK led to a fourfold increase in deletions. This difference between the two strains was more dramatic in L-PHO5, which contains the PHO5 coding sequence between the leu2 repeats. Whereas hpr1ΔH caused an eightfold increase above wild-type levels, recombination was increased 122-fold in hpr1ΔK. Even though the hyper-recombination phenotype of hpr1ΔK was evident in both the systems, it was not as high as that conferred by tho2Δ (58- to 805-fold). These results indicate that hpr1ΔH does not behave as a complete null allele.
Fig. 2

Effect of hpr1ΔK and tho2Δ mutations on deletion, inversion, gene conversion, hetero-allelic recombination and LOH. Recombination frequencies of wild-type (W303-1A), hpr1ΔH (U678-4C), hpr1ΔK (SChY58a) and tho2Δ (RK2-6C) strains transformed with centromeric plasmids, pRS316-L, pSCh206, and pRS314-SU containing the corresponding direct and inverted repeat systems. The mRNA synthesized by each system is drawn as an arrow. Gene conversion was determined in wild type and isogenic mutant strains carrying the chromosomal inverted repeat system his3P::INV. Ectopic recombination was determined as the recombination frequency between two leu2 heteroalleles (leu2-r) and (leu2-k) located on heterologous chromosomes or, with the leu2-allele in the replicative plasmid pAA547, and the leu2-k allele in the natural chromosomal locus (plasmid–chromosome system). LOH was determined in the wild-type diploid strain 8D8AC57 and the congenic hpr1/hpr1 and tho2/tho2 mutants (see "Materials and methods"). A vertical line inside mutant allele boxes indicates the approximate site of each mutation. The recombination frequencies are average of three median values obtained from three independent fluctuations tests each based on six dependent colonies

Given the fact that hpr1ΔK and tho2ΔK exhibited stronger recombination phenotypes than hpr1ΔH, we wondered whether these mutations also affected other recombination events in which the hpr1ΔH allele had previously been shown to have little effect compared to the hyperrecombination observed between direct-repeat sequences (Santos-Rosa and Aguilera 1994; Freedman and Jinks-Robertson 2004). Figure 2 shows that inversions increased poorly in the hpr1ΔH mutants (eightfold), whereas a strong increase was observed in hpr1ΔK (109-fold) and tho2Δ (809-fold). We extended our analysis to heteroallelic gene conversion, for which we used the inverted repeat system his3P::INV carrying the his3Δ5′ and his3-k alleles in inverted orientation. His+ gene conversions in this chromosomal system were poorly affected in hpr1ΔH mutants but strongly increased in hpr1ΔK (87-fold) and tho2Δ (300-fold). We determined the effects of hpr1ΔK and tho2Δ on recombination between the two leu2 heteroalleles located on heterologous chromosomes (III and XV), as well as on plasmid–chromosome recombination between a plasmid carrying the leu2-r allele and the leu2-k allele at its natural chromosomal locus. As can be seen in Fig. 2, ectopic recombination between the heterologous chromosomes was increased threefold in hpr1ΔK and fivefold in tho2Δ, whereas plasmid–chromosome recombination was increased 30-fold in tho2Δ cells. Additionally, we determined the effect of hpr1ΔK and tho2Δ on LOH frequency, as determined by mating efficiency occurring by heteroallelic recombination at the chromosomal MAT locus or by loss of chromosome III (see "Materials and methods"). The frequency of acquisition of mating competence of heteroallelic MATα/MATa diploid strains (LOH) was increased fourfold in hpr1ΔK mutants and 100-fold in tho2Δ as compared to the wild type. Thus, our results indicate that tho2Δ and hpr1ΔK have a strong effect on inversions and gene conversion and a moderate but significant effect on ectopic recombination, tho2Δ being the mutant with the highest impact on recombination.

Strong gene expression defect of highly transcribed short and low GC-content DNA sequences in null hpr1ΔK and tho2Δ mutants

DNA recombination and gene expression analyses of the hpr1ΔH mutants revealed that their transcription elongation impairment was preferentially observed in GC-rich and long DNA sequences under the control of the GAL1 promoter (Chávez et al. 2001). Since both hpr1ΔK and tho2Δ strains are strongly affected in recombination between direct repeats when a short and low GC-content DNA sequence such as PHO5 is transcribed from the LEU2 promoter in the L-PHO5 construct (Fig. 2), it seemed likely that transcription impairment in these mutants would be stronger in short or low GC-content DNA sequences than in hpr1ΔH strains. As can be seen in Fig. 3, mutants hpr1ΔH, hpr1ΔK and tho2Δ show a strong reduction in the kinetics of transcript accumulation of YAT1 (53% GC; 2 kb) and a LYS2 fragment (42% GC; 4.2 kb) under the control of the GAL1 promoter. However, transcription of LAC4 gene (41% GC; 3kb), the low GC-content lacZ ortholog of Kluyveromyces lactis, was strongly reduced to 5–12% of the wild type mRNA levels in both tho2Δ and hpr1ΔK strains, respectively but only to 50% of wild type levels in hpr1ΔH [Fig. 3; Chávez et al. (2001)]. A similar strong defect for GAL1::lacZ and a weaker but significant transcript reduction for GAL1::PHO5 have been observed in hpr1ΔK (data not shown) as previously shown for tho2Δ (Piruat and Aguilera 1998). Since all results shown so far refer to GAL1 fusion constructs, it is conceivable that the differences in transcription impairment observed between the mutations might rely on some defects in GAL1 promoter activation. In order to test this possibility, we used a new construct, tet::LYS2, in which a 4.2 kb fragment of the LYS2 gene is placed under the control of the doxycyclin-regulated tet promoter. As can be seen in Fig. 4, mRNA levels in the tet::LYS2 system are clearly reduced in tho2Δ and hpr1ΔK and to a lesser extent in hpr1ΔH and mft1Δ. These results confirm that the null hpr1ΔK allele causes stronger transcription impairment than hpr1ΔH, and that both tho2Δ and hpr1ΔK are the THO mutations with the strongest impact on transcription. Therefore, the hpr1ΔH mutant used in most previous studies may have some residual activity.
Fig. 3

Transcription analysis of hpr1ΔK and tho2Δ versus the wild type as a function of GC-content and length of the transcribed DNA sequence. Northern analysis of mRNA levels in different plasmid constructs containing LAC4 (pSCh255), YAT1 (pSCH247) and LYS2 (pSCh227) under the control of the GAL1 promoter in strains W303-1A (wild type), hpr1ΔK (SChY58a) and tho2Δ (RK2-6C) are shown. Mid-log phase plasmid-transformed cells were diluted in 3% glycerol–2% lactate synthetic complete medium SC and diluted into identical fresh media to an OD600 of 0.5 and incubated for 16h. Galactose was then added and samples were taken for Northern blot analyses at different times. RNA levels in arbitrary units (A.U.) were obtained by quantification of signal intensities in a FUJI FLA 3000 and normalized with respect to rRNA levels of each sample. Wild-type mRNA levels were taken as 100%. The mRNA curves corresponding to hpr1ΔH (dashed lines) are taken from previously published results (Chávez et al. 2001) and are included for comparison. The averages of two experiments are plotted
Fig. 4

Comparative effect of mutant alleles of different subunits of the THO complex on transcription. Northern analysis of mRNA levels in tet:LYS2 construct in wild type hpr1ΔK (SChY58a), hpr1ΔH (U678-4C), tho2Δ (RK2-6C), and mft1Δ (WMK-1A) under conditions of tet activation. Cells were grown to exponential phase in the absence of doxycyclin (transcription activation conditions). Other details as in Fig. 3

A major role for the Sub2 component of TREX in mRNP biogenesis and transcription-associated genetic stability

The observation that THO physically interacts with Sub2, Sub2 overexpression suppresses hpr1Δ, and sub2Δ strains show similar transcript accumulation defects, hyper-recombination, and transcription-coupled repair impairment such as THO mutants (Fan et al. 2001; Jimeno et al. 2002; Gaillard et al. 2007) suggests that Sub2 and THO are functionally related. However, the transcription dependency of the hyper-recombination phenotype of sub2Δ cells and the relevance of their in vivo transcription impairment in comparison with other mutants of the THO complex has not been established. We addressed these questions directly and showed that mRNA accumulation kinetics are strongly affected in sub2Δ cells, both in GAL1pr::lacZ (56% GC, 3 kb) and GAL1pr::PHO5 (41% GC, 1.5 kb) constructs (Fig. 5). Even though the effect is stronger for lacZ transcripts, as shown for hpr1Δ and tho2Δ (Piruat and Aguilera 1998; Chávez et al. 2001), the significant reduction in PHO5 transcript accumulation reveals that Sub2 is a key player in mRNP biogenesis, since it is also required for transcription of short and low GC-content genes.
Fig. 5

Transcription analysis of sub2Δ mutants. Expression analysis of GAL1pr::lacZ and GAL1pr::PHO5 constructs. Northern analysis of lacZ and PHO5 mRNAs derived from the GAL1 promoter in the wild-type (W303-1A) and sub2Δ strains transformed with plasmids p416-GAL1-lacZ or pSCh202. Other details are as described for Fig. 3

We analyzed the effect of sub2Δ on recombination in previously described systems based on 0.6 kb leu2 direct-repeats. These are the GL-lacZ, in which transcription is driven from the GAL1 promoter and is inactive in glucose (low or no transcription), the L-lacZ system in which transcription is driven from the LEU2 promoter and traverses the leu2 repeats and the lacZ intervening region, and the L-PHO5 system, identical to L-lacZ but containing PHO5 as intervening sequence. As can be seen in Fig. 6, sub2Δ showed a weak increase in recombination (fivefold) under conditions of low or no transcription, a 37-fold increase when transcription traverses the PHO5 sequence and a 521-fold increase when it traverses the lacZ sequence. Therefore, our results indicate that hyper-recombination is transcription-dependent in sub2Δ, and that it is primarily observed when transcription traverses long and GC-rich DNA sequences like lacZ, even if some hyper-recombination is also observed in L-PHO5, consistent with the reduction of PHO5 transcript accumulation. These results are similar to those of tho2Δ and hpr1ΔK, indicating that Sub2 is as important as the most relevant components of THO in preventing transcription-associated hyper-recombination.
Fig. 6

Recombination analysis of sub2Δ mutants. Recombination frequencies were determined in wild-type (W301-A) and sub2Δ strains transformed with plasmids pSCh204 and pSCh206 carrying the direct-repeat systems L-lacZ and L-PHO5, respectively, in which transcription is under the LEU2 promoter. Shown is a recombination frequency under repressed conditions OFF (glucose) in cells transformed with plasmid pRS314GLlacZ containing the GL-lacZ recombination assay, in which transcription is under the control of the GAL1 promoter. Other details are as described for Fig. 2


In this study, we show that the gene expression defect and hyper-recombination phenotypes conferred by THO/TREX mutations depend on the mutated subunit. Whereas Hpr1, Tho2 and Sub2 are the subunits with the strongest phenotypes, the effects of Thp2 and Mft1 are weak, as determined by growth, thermosensitivity, mRNA accumulation, and recombination (Chávez et al. 2000). The hpr1Δ::KAN complete deletion confers stronger transcription and recombination phenotypes than the original hpr1Δ::HIS3 mutation used in most previous studies. Furthermore, we show that sub2Δ shows transcription-dependent hyper-recombination and transcription impairment in sequences with different GC content similar to THO mutants. Two major physiological findings are derived from this study. First, tho2Δ and hpr1ΔK, the mutations with the strongest hyper-recombination phenotype, confer a moderate but significant increase in inverted-repeat, allelic, and plasmid–chromosome recombination. Second, even though the longer and the higher the GC-content of a gene, the stronger the effect of THO mutations on transcription, an effect that is clearly seen in moderate alleles such as hpr1ΔH or mft1Δ, transcription of short and low GC-content genes is also affected in the most potent alleles (hpr1ΔK, tho2Δ and sub2Δ). Altogether, these results open a new view of the physiological relevance of the different subunits of THO/TREX in transcription and genome instability.

An unexpected observation of this study is the leakiness of hpr1Δ::HIS3 (hpr1ΔH) (Aguilera and Klein 1990), the original and most common deletion mutation of HPR1. This allele encodes a C-terminal-truncated protein containing the first 112 amino acids. These mutants are affected in the transcription and are strongly hyper-recombinant. The observation that the complete hpr1ΔK deletion exacerbates all these phenotypes indicates that hpr1ΔH must retain some residual THO/TREX activity. The difference between the two alleles is particularly evident in recombination and transcription of systems based on short and low GC-content ORFs. The difference in phenotype strength between the two strains could explain differences in genetic interaction with other mutations; for example whereas null hpr1ΔK is synthetic lethal with spt4Δ (Rondón et al. 2003a), double hpr1ΔH spt4Δ mutants are sick, but viable. Similarly, it could explain the intriguing observation that THO2 was identified as a multicopy suppressor of hpr1ΔH (Piruat and Aguilera 1998), while it is not able to suppress the null hpr1ΔK allele (data not shown).

Interestingly, the two largest and most conserved subunits of THO are those with the major physiological relevance. tho2Δ and hpr1ΔK are the mutants with the most severe growth phenotype and the strongest transcription-defects and hyper-recombination phenotypes. Analysis of other phenotypes, such as the synthetic lethality of hpr1ΔK and tho2Δ with spt4 and spt5 transcription mutants and with mex67-5 export mutant confirms this conclusion, since synthetic lethality is not observed with mft1Δ and thp2Δ (Jimeno et al. 2002; Rondón et al. 2003a). Furthermore, RNA export defects have been shown to be stronger in tho2Δ than in leaky hpr1ΔH, which itself shows a more severe phenotype than mft1Δ and thp2Δ [Strasser et al. (2002); data not shown]. Similar differences were found when the effects of these mutations on transcription of repeat-containing genes were analyzed (Voynov et al. 2006). Indeed, a number of studies have been performed with mft1Δ strains as representative of THO mutants, presumably due to the fact that these strains do not show growth problems, in contrast to hpr1Δ and tho2Δ (Libri et al. 2002; Zenklusen et al. 2002). The phenotypes of mft1Δ are indeed weaker than those of null hpr1ΔK and tho2Δ. It is worth noting that, in spite of these differences, THO behaves as a stable structural and functional unit. In fact the gene length-dependent defects in mRNA biogenesis caused in mft1Δ and thp2Δ are stronger than those produced by the absence of other proteins involved in transcription elongation (Morillo-Huesca et al. 2006). We have recently observed that the stability of THO is seriously compromised when one subunit of Hpr1, Tho2 or Mft1 is depleted, (Huertas et al. 2006). A more detailed structural and functional analysis of THO will be required to investigate whether the leaky phenotype found in some of the mutants might be due to the formation of residually active THO complexes below detection levels in Westerns, or a hypothetical capability of unstable and incomplete complexes to partially complement the function of THO in mRNP biogenesis.

One hallmark phenotype of THO mutants is their hyper-recombination phenotype, which is clearly observed for deletions occurring between direct repeats, as first characterized in the original hpr1 point mutants (Aguilera and Klein 1990). A detailed recombination analysis revealed that the main events stimulated in leaky hpr1ΔH strains were deletions occurring by single-strand annealing (SSA), whereas allelic recombination was less affected, its effect varying from no increase to a low but significant increase depending on the recombination assay used (Santos-Rosa and Aguilera 1994; Freedman and Jinks-Robertson 2004). Recently, we have shown that the hyper-recombination in hpr1ΔH is linked to the replication fork progression impairment occurring at the region where transcription defects are detected (Wellinger et al. 2006). It is likely that the transcription-dependent recombinogenic intermediate, presumably mediated by replication fork stalling, gets efficiently repaired by recombination using a close partner, such as an intrachromosomal repeat, but not with allelic sequences located either in homologous or heterologous chromosomes. The latter events may not be efficiently stimulated by replication impairment, explaining why direct-repeat recombination is the primary event observed in these mutants.

Our analyses of null tho2ΔK and hpr1ΔK mutants confirm that their hyper-recombination phenotype consists primarily of SSA events leading to deletions, but they also show increases in allelic recombination between plasmid and chromosome, gene conversion, recombination between inverted repeats, and LOH (Fig. 2). Nevertheless, there are clear differences in recombination frequencies between tho2Δ and hpr1ΔK, which may indicate a more relevant role of Tho2 in mRNP biogenesis and the THO complex, consistent with the fact that Tho2 is the largest subunit. This needs further investigation to be completely understood. In summary, it appears that the recombinogenic transcription-dependent intermediates formed in THO mutants are not only competent for Rad51-independent SSA, but are capable of initiating and completing a Rad51-dependent strand-exchange mechanism of gene conversion, although to a lesser extent.

Another hallmark phenotype of THO mutants is transcription elongation impairment. The defects in gene expression of the hpr1ΔH leaky mutation are stronger for long and GC-rich transcribed DNA sequences (Chávez et al. 2001), suggesting that the relevance of THO is higher with increasing DNA length and GC-content, but lower or undetectable for short and GC-poor DNA sequences. Recently, it has been reported that THO is required for transcription of genes with internal repeats (Voynov et al 2006). Importantly, this study reveals that THO is also important for other DNA sequences, despite weaker effects. Both null hpr1ΔK and tho2Δ strains show a significant reduction in transcription efficiency of all sequences tested, even though the effect is still more severe for long and GC-rich DNA sequences. This general transcription defect might be related to poor growth and strong ts phenotypes of these mutants, and is consistent with a high relevance of THO in nuclear metabolism. Furthermore, it would explain the lethality of THO mutants in Drosophila cell lines and mouse embryos (Rehwinkel et al. 2004; Wang et al. 2006).

Finally, an important component of TREX is the Sub2 RNA-dependent ATPase, which was re-isolated as multicopy suppressor of hpr1ΔH strains and was subsequently shown to lead to strong hyper-recombination and transcription impairment when deleted (Fan et al. 2001; Jimeno et al. 2002; Rondón et al. 2003b). However, whether the hyper-recombination phenotype of sub2 mutants is transcription-dependent and the in vivo effect on transcription is like that of THO mutants have never been shown. This is an important issue since a number of mRNA metabolism and processing mutants have been shown to cause hyper-recombination independently of transcription, and other TREX mutants, like tex1, do not show transcription or hyper-recombination phenotypes (Luna et al. 2005). Here, we show that sub2Δ hyper-recombination depends on transcription, and that the in vivo effect on transcription is as strong as in tho2Δ mutants. These results, together with the facts that multicopy SUB2 suppresses THO mutants (Jimeno et al. 2002), suggest that Sub2 might be a key player in the connection of mRNP biogenesis and transcription-dependent genome instability, the modulation of Sub2 recruitment to active chromatin being the ultimate determinant for TAR in mRNP biogenesis mutants. Indeed, the fact that the THO complex can be purified in similar amounts in wild-type and sub2Δ cells, despite the strong phenotypes of sub2Δ (Jimeno et al. 2002), is consistent with this key role of Sub2.

In summary, our work reveals a hierarchy of THO/TREX subunits in gene expression and genetic integrity as determined by the effect of their mutations on different phenotypes. All these results should help to clarify the variability of phenotypes observed in different studies using distinct THO/TREX mutants, and provide a better comprehension of the functional relevance of THO/TREX in living cells.


We would like to thank H. Gaillard for critical reading of the manuscript, and D. Haun for style supervision. This work was supported by the Ministry of Science and Education of Spain (grants SAF2003–00204 and BMC2006–05260) and Junta de Andalucía (CVI-102).

Copyright information

© Springer-Verlag 2007