Genome-Wide Analysis of mRNA Accumulation and 4-NQO- and MPA-Sensitivity
We were interested in analyzing and comparing the genes conferring resistance to 4-NQO and MPA. To explore the functional consequences of the treatments with 4-NQO and MPA, we first determined the expression levels of the whole genome after treating wild-type cells with either 75 ng/ml 4-NQO or 50 µg/ml MPA during 30 min each. Expression of a total of 2374 genes was evaluated by microarray analysis (data available at www.ncbi.nlm.nih.gov/geo/
under the access number GSE11561). Among these genes, mRNA levels that were at least 2-fold above or below mock treated cells were found for 376 genes in cells treated with 4-NQO and for 295 genes in cells treated with MPA (Table S1
). Of the genes affected by either treatment, very few were coincident, which is consistent with the fact that 4-NQO and MPA affect different cellular processes (). Ontology analysis of the 644 genes showing significant variations in expression levels indicated that there is not a relevant class of genes specifically affected by any of the compounds used (data not shown).
Analysis of genes similarly affected by 4-NQO and MPA.
For the analysis of genes required for resistance to 4-NQO and MPA, a collection of 4894 yeast haploid knock-out mutants, covering 85% of all yeast genes and virtually all (99,4%) of yeast non-essential genes, were grown in SC medium supplemented either with 150 ng/ml 4-NQO or 25 µg/ml MPA. Cells were incubated at 30°C and growth monitored after 48, 72, and 120 for 4-NQO and 72 hours for MPA. Analysis of the strains showing at least 80% growth inhibition by 4-NQO at the indicated time led to a classification of sensitive strains into three groups (Table S2
): group A contains 189 strains whose sensitivity to 4-NQO is observed from early on (48 h) and is maintained over time, B contains 308 strains whose sensitivity was observed only at early time points (48 h), and C contains 100 slow-growing strains whose sensitivity is observed later on (from 72 h). Direct comparison of cell sensitivity to different DNA-damaging agents and stress conditions (4-NQO, methyl-methanesulfonate [MMS], menadione [Mnd], UV, and 37°C) in a selection of 4-NQO-sensitive mutants validated our high-throughput results and confirmed that 4-NQO in addition to being a ‘UV-mimetic’ agent, causes oxidative damage (Table S3
). In the same line, comparison of the mutants found sensitive to 4-NQO with the mutants found sensitive to at least one of five oxidants 
revealed that 35% of the 4-NQO sensitive strains are sensitive to oxidative damage as well. As compared to a genome-wide study in which deletion strains were pooled and grown competitively in the presence of 4-NQO 
, our set of 4-NQO sensitive strains contains the 10 strains found as top sensitive and 31 out of the 37 most sensitive strains.
Correspondingly, 85 MPA-sensitive deletions showing at least 50% growth inhibition were found (Table S4
), of which 40 had been reported in a previous analysis for MPA sensitivity of the yeast disruptome 
and 45 had not been described as MPA-sensitive to date. We focused on 289 4-NQO-sensitive (groups A and C) and 68 MPA-sensitive deletions showing at least 60% growth inhibition. Ontology analysis of the genes identified for each drug is consistent with the fact that 4-NQO affects DNA repair whereas MPA affects transcription (). Only 25 mutations led to sensitivity to both compounds (). From these mutations, 13 identified known genes involved in aspects of transcription and mRNP biogenesis, and 12 identified genes involved in protein modification, intracellular trafficking, and primary and secondary metabolism. Direct comparison of the genes showing significant variations in their patterns of expression, as determined by microarray analyses, with those showing sensitivity to either MPA or 4-NQO reveals no obvious correlation (data not shown). Therefore, a higher expression of a gene in the presence of 4-NQO or MPA does not imply requirement for resistance, but rather it is the result of an adaptation of the cell to the new conditions of growth.
Effect of SAGA, CTK, Mediator, Ccr4-Not, Bre1-Rad6, and Fun12 in Transcription Elongation
We have previously shown that mutants impaired in transcription elongation display lower efficiency in the transcription of long vs.
short transcription units 
. To gain insight into the putative defects in RNAPII transcription of 45 MPA-sensitive mutants, the ratios of acid phosphatase activity for a long (PHO5-lacZ
a short transcription unit (PHO5
), which is taken as an approximate measurement of G
ccumulation of m
RNA (GLAM) were determined (Table S4
). GLAM-ratios were previously used as an indirect estimation of RNAPII elongation 
. Similar GLAM-ratios were obtained with the two long transcription units (Figure S1
). Out of the assayed mutants, 24 showed values below 0.5, which were taken as criteria for candidates with defects in RNAPII elongation. 6 mutants carried deletions of genes encoding subunits of protein complexes that were previously related to RNAPII elongation, including RNAPII (rpb4Δ
), THO (tho2Δ
), Spt4-Spt5 (spt4Δ
), and PAF (rtf1Δ
). 7 mutants carried deletions of genes encoding proteins affecting transcription, for which an implication in transcription elongation has been proposed in the past. These included subunits of SAGA (spt3Δ
, CDK (bur2Δ
, and Ccr4-Not (ccr4Δ
complexes as well as proteins involved in H2B ubiquitylation (bre1Δ
, and lge1Δ
. Another 4 mutants carried deletions of genes encoding proteins affecting transcription, but that had not been related to RNAPII elongation. These included subunits of Mediator (med2Δ
) and Ccr4-Not (not5Δ
) complexes as well as the Uvs1 putative transcription factor (YPL230wΔ
). The remaining 7 mutants affected proteins whose function has not been previously linked to transcription (tpd3Δ
, and ubp15Δ
) or is unknown (YJR018wΔ
). To evaluate the significance of these putative new links with RNAPII elongation, and because some mutants of the SAGA (spt7Δ
), CDK (ctk1Δ
), and Mediator (med12Δ
) complexes did not exhibit expression deficiencies, we extended our analysis to deletions of other functionally related genes.
Gene expression was analyzed in all viable SAGA deletions (). We found that, in addition to spt20Δ and spt3Δ, the absence of three other subunits (Hfi1, Sgf73, and Sgf29) showed a clear gene expression defect (GLAM<0.5) while the remaining 5 viable deletions (gcn5Δ, ada2Δ, ubp8Δ, ngg1Δ, and sgf11Δ) were only poorly or not affected, as observed for spt7Δ and spt8Δ.
Gene expression analyses of selected MPA-sensitive and functionally related mutants.
Since the H2B-ubiquitylases Rad6, Bre1, and Lge1, two PAF subunits (Cdc73 and Rtf1) as well as Bur2 belonged to the MPA-sensitive mutants exhibiting low GLAM-ratios, we extended the analysis to mutants of COMPASS, the complex responsible for H3-K4 methylation. Low GLAM values were observed for some of the COMPASS and H2B ubiquitylation mutants (), suggesting that both H2B ubiquitylation and H3-K4 methylation may be important for transcription elongation. Interestingly, analysis of Ctk1-Ctk2-Ctk3 (CTDK-I)—the other cyclin-dependent protein kinase involved in transcription elongation 
—suggests that, despite ctk1Δ
being MPA-sensitive, CTDK-I might be dispensable for the expression of long genes.
Most viable deletions lacking Mediator subunits and the viable deletions lacking other subunits of the Ccr4-Not complex were also assayed (). The results indicated that in addition to Med2 and Med15, the Med18, Med19 and Med20 subunits might affect transcription elongation, whereas the remaining viable subunits may be dispensable. For the Ccr4-Not complex, not4Δ showed low expression of long genes, as observed for not5Δ and ccr4Δ, whereas this was not the case for caf1Δ and not3Δ.
Finally, since the only known function of Fun12 is related to translation initiation, we assayed other viable deletions of translational machinery elements playing a role during initiation (Gcn2, Gcn3, Hcr1 and Tif3). To understand the low GLAM-ratios of tpd3Δ, all other deletions lacking subunits of the PP2A complex were assayed. shows that neither the translation initiation machinery nor the PP2A complex influence gene expression in a gene length-dependent way.
Since the analysis of GLAM-ratios relies on measurement of enzymatic activities, we decided to assess directly the efficiency of transcription of representative mutants using an in vitro
elongation assay. This assay is based on a plasmid (pGCYG1-402) in which a hybrid GAL4-CYC1
promoter containing a Gal4 binding site is fused to a 1.88-kb DNA fragment coding for two G-less cassettes. The first cassette is right downstream of the promoter and is 84-nt-long. The second is located 1.48-kb from the promoter and is 376-nt-long. The efficiency of elongation is determined in whole cell extracts (WCEs) by the values of the ratio of accumulation of the 376- versus the 84-nt-long G-less RNA fragments after RNase T1 digestion 
WCEs from representative mutants of Mediator (med15Δ), CTK (bur2Δ), Bre1-Rad6 (bre1Δ), SAGA (spt20Δ), Ccr4-Not (not5Δ), and PP2A (tpd3Δ) complexes as well as the translation initiation mutant fun12Δ were analyzed. As can be seen in , bre1Δ, spt20Δ, and not5Δ WCEs transcribed the 376-nt G-less cassette with efficiencies around or below 60% of the wild-type levels. These results indicate severe defects in transcription elongation in those subunits of the Bre1-Rad6, SAGA, and Ccr4-Not complexes. Strikingly, fun12Δ cell extracts also led to a clear transcription elongation phenotype in our assay (62%). WCEs extracts of bur2 and med15 mutants were moderately affected in transcription elongation, with efficiencies ranging from 68 to 77% of wild-type levels. Transcription elongation efficiencies of tpd3Δ WCEs reached wild-type levels, indicating that this mutant was fully transcription elongation-proficient in this assay.
In vitro transcription elongation.
In addition, we aimed at testing whether the candidate mutations changed the distribution of RNAPII along a transcribed unit, as an alternative method to measure elongation 
. Therefore, RNAPII occupancy was analyzed by ChIP for the representative mutants selected for the in vitro
assay in the LAUR expression system 
, which contains a 4.15 kb lacZ
translational fusion under the control of the Tet
promoter. The presence of RNAPII was determined at a 5′-end and a 3′-end region of the lacZ
sequence, as well as within the fused URA3
gene (). The spt20Δ
mutants and, to a lesser extent, bur2Δ
were impaired in elongation in this assay, as less RNAPII was found toward the 3′-end than at the 5′-end of the transcription unit. Strikingly, RNAPII appeared to accumulate toward the 3′-end of the gene in med15Δ
cells. No significant changes in RNAPII distribution were observed in the bre1Δ
, and tpd3Δ
mutants, the latter of which is consistent with the in vitro
Therefore, our results indicate that subunits of SAGA, Ccr4-Not, Mediator, and, to a lesser extent, CDK affect transcription elongation, as seen with the three different assays tested while the effect of Bre1-Rad6 and the translation factor Fun12 on transcription elongation depends on the assay used.
Genetic Analysis of UV Sensitivity in the Absence of Global Genome Repair
Given the strong dependency of TC-NER on RNAPII transcription and the fact that the few proteins known to be involved in TC-NER are related to transcription, we made use of the results of our MPA-sensitivity screen to select 18 mutants encoding for transcription factors, protein de-ubiquitylase, H2B-ubiquitylase, subunits of the CDK, SWI/SNF, SAGA, Mediator, PAF, Ccr4-Not complexes, and RNAPII and look for those possibly involved in TC-NER. For this purpose, we abolished GG-NER by deleting the RAD7
gene in the chosen mutants. In the absence of GG-NER, deficiencies in TC-NER lead to increased UV-sensitivity, a phenotype that we screened for by drop assay (). Growth of each double mutant was compared to the growth of rad7Δ
, giving rise to the classification of 5 mutants as not more sensitive than rad7Δ
, and spt7Δ
) and 4 mutants as slightly more sensitive to UV than rad7Δ
, and med12Δ
), this effect being more obvious when higher UV doses were used (data not shown). The remaining 9 strains were much more sensitive to UV than rad7Δ
, and not5Δ
). Rad6 and Rpb9 are known to be involved in post-replication repair of UV-damaged DNA and TC-NER, respectively 
. However, the 7 other mutants have no known connection to any UV-damaged DNA repair pathway. These mutants include subunits of the mediator (med2Δ
), SAGA (spt3Δ
), PAF (rtf1Δ
), and Ccr4-Not (not5Δ
) complexes. Given the fact that other subunits of the Mediator and SAGA complexes were represented in the moderately UV-sensitive strains (med12Δ
, and spt8Δ
), we focused on the PAF and Ccr4-Not complexes for a more detailed analysis.
UV sensitivity in the absence of global genome repair in selected MPA-sensitive mutants.
TC-NER Is Impaired in Cells Defective in the PAF and Ccr4-Not Complexes
To refine the UV sensitivity analysis of PAF and Ccr4-Not mutants in the absence of GG-NER, UV survival curves were performed for all viable PAF and Ccr4-Not mutants (rtf1Δ, cdc73Δ, paf1Δ, leo1Δ, not5Δ, not4Δ, not3Δ, caf1Δ, ccr4Δ) alone or in combination with the rad7Δ mutation (). The rtf1Δ, cdc73Δ, leo1Δ, not3Δ, caf1Δ and ccr4Δ single mutants show no increased UV sensitivity as compared with wild-type cells. However, upon UV irradiation viability of the corresponding double mutants dropped below the levels of the rad7Δ single mutant. The paf1Δ, not5Δ and not4Δ single mutants showed a moderate UV sensitivity, reaching levels very close to that of rad7Δ in the case of paf1Δ and not4Δ. Nevertheless, the viability of the paf1Δ rad7Δ, not5Δ rad7Δ, and not4Δ rad7Δ double mutants dropped far below the levels of the corresponding single mutants upon UV irradiation.
PAF and Ccr4-Not mutants are sensitive to UV in the absence of global genome repair.
Both the PAF and the Ccr4-Not complexes have been previously linked to the DNA damage checkpoint pathway 
. Therefore, we wondered whether the observed UV sensitivity might rely on checkpoint activation defects. Firstly, we checked the GLAM-ratios of cells lacking the DNA damage checkpoint protein Rad9 (). No transcription defects were observed in this assay. Secondly, we performed UV survival curves of the DNA damage checkpoint rad9Δ
and the bre1Δ
mutants alone or in combination with the rad7Δ
mutation (). The rad9Δ
single mutant was sensitive to UV irradiation, as previously shown 
. Deletion of the GG-NER factor Rad7 increased the UV sensitivity of rad9Δ
cells. Together, these data indicate that a functional DNA damage checkpoint response is important for viability upon UV irradiation both in repair proficient and in GG-NER deficient cells. Surprisingly, the bre1Δ
mutant behaved differently, as the single mutant was not UV-sensitive while the bre1Δ rad7Δ
double mutant was not more sensitive to UV irradiation than the rad7Δ
single mutant. Finally, we analyzed the impact of the rad9Δ
mutation on the UV survival of the rtf1Δ
, rtf1Δ rad7Δ
, and not5Δ rad7Δ
strains. A similar set of UV survival curves were performed with the TC-NER mutant rpb9Δ
as a control. As shown in , a synergistic effect was observed in the absence of GG-NER in mutants of the PAF and Ccr4-Not complexes, since the rft1Δ rad9Δ
and not5Δ rad9Δ
mutants were as sensitive to UV irradiation as the rad9Δ
mutant alone, while both the rtf1Δ rad7Δ rad9Δ
and the not5Δ rad7Δ rad9Δ
strains were significantly more sensitive to UV than the corresponding double mutants. In the TC-NER deficient rpb9Δ
strains, a synergistic effect with rad9Δ
was observed independently of the rad7Δ
mutation. Consequently, the enhanced UV sensitivity of mutants of the PAF and Ccr4-Not complexes in the absence of GG-NER is not due to Rad9-dependent checkpoint activation failure.
The increased UV sensitivity of PAF and Ccr4-Not mutants in the absence of global genome repair is not due to checkpoint activation failure.
Thus, because UV sensitivity in the absence of GG-NER is a phenotype mostly associated with TC-NER deficiencies, we tested whether PAF and Ccr4-Not are required for proficient TC-NER by monitoring the repair rates on the transcribed (TS) and non-transcribed (NTS) strands of the constitutively expressed RPB2
gene. Molecular analysis of strand-specific removal of UV photoproducts was performed in rtf1Δ
cells. Wild-type and TC-NER-deficient tho2Δ 
strains were used as controls. Repair at various time points after UV irradiation was determined in a 4.4-kb RPB2
restriction fragment by T4 endonuclease V (T4 endoV) digestion -resulting in ssDNA cleavage at CPD sites- followed by alkaline electrophoresis and indirect end-labeling with strand-specific probes (). Non-irradiated and DNA not treated with T4 endoV show the intact restriction fragment. Repair of CPDs is visualized by a time-dependent increase of the T4 endoV-resistant fraction of restriction fragments. In rtf1Δ
cells, repair of the TS was significantly reduced compared to wild-type level. As observed by UV sensitivity assays in the absence of GG-NER (), not5Δ
cells were more strongly affected in TS repair than rtf1Δ
cells. The repair deficiencies of these PAF and Ccr4-Not mutants were comparable to those of the TC-NER-deficient tho2Δ
strains. In the NTS, in contrast to the GG-NER-deficient rad7Δ
strain, the repair levels of rtf1Δ
were similar to wild-type and tho2Δ
cells, indicating that GG-NER is not significantly affected in rtf1Δ
Transcription coupled repair is impaired in PAF and Ccr4-Not deficient cells.
A number of factors have been implicated in the repair of DNA lesions encountered by the RNAPII in eukaryotes, but our knowledge on the mechanisms of TC-NER is scarce. Since proteasome-mediated degradation of UV damage-stalled RNAPII complexes is believed to be alternatively required for DNA repair, we tested whether the effect of PAF and Ccr4-Not effect on TC-NER was dependent on RNAPII degradation. For this, we performed an epistatic analysis of the PAF mutant rft1Δ
with both the def1Δ
and the rpb9Δ
mutants, which are deficient in RNAPII degradation in response to UV in yeast 
. As can be seen in , a synergistic enhancement of the UV sensitivity was observed in both cases, the rft1Δ def1Δ
and the rft1Δ rpb9Δ
mutants being more sensitive to UV than the corresponding single mutants. Similarly, the rft1Δ def1Δ rad7Δ
and the rft1Δ rpb9Δ rad7Δ
triple mutants were more sensitive to UV irradiation than the corresponding double mutants. These results suggest that the TC-NER phenotype of PAF mutants is not due to an alteration of the Def1- or Rpb9-mediated degradation of UV damage-stalled RNAPII.
Synergistic increase of UV sensitivity phenotypes in rft1 def1 and rft1 rpb9 double mutants.
Taken together, our results place the PAF and the Ccr4-Not5 complexes as new factors needed for efficient TC-NER.