Diverse Roles of RNA Polymerase II-associated Factor 1 Complex in Different Subpathways of Nucleotide Excision Repair*

Transcription-coupled repair (TCR) and global genomic repair (GGR) are two pathways of nucleotide excision repair (NER). In Saccharomyces cerevisiae, Rad26 is important but not absolutely required for TCR. Rpb4, a nonessential RNA polymerase II (Pol II) subunit that forms a subcomplex with Rpb7, and the Spt4-Spt5 complex, a transcription elongation factor, have been shown to suppress Rad26-independent TCR. The Pol II-associated factor 1 complex (Paf1C) has been shown to function in transcription elongation, 3′-processing of mRNAs, and posttranslational modification of histones. Here we show that Paf1C plays a marginal role in facilitating Rad26-dependent TCR but significantly suppresses Rad26-independent TCR. The suppression of Rad26-independent TCR is achieved by cooperating with Spt4-Spt5. We propose a model that, in the absence of Rad26, a lesion is “locked” in the active center of a Pol II elongation complex, which is stabilized by the coordinated interactions of Rpb4-Rpb7, Spt4-Spt5, and Paf1C with each other and with the core Pol II. We also found that Paf1C facilitates GGR, especially in internucleosomal linker regions. The facilitation of GGR is achieved through enabling monoubiquitination of histone H2B lysine 123 by Bre1, which in turn permits di- and trimethylation of histone H3 lysine 79 by Dot1. To our best knowledge, among the NER-modulating factors documented so far, Paf1C appears to have the most diverse functions in different NER pathways or subpathways.

Transcription-coupled repair (TCR) and global genomic repair (GGR) are two pathways of nucleotide excision repair (NER). In Saccharomyces cerevisiae, Rad26 is important but not absolutely required for TCR. Rpb4, a nonessential RNA polymerase II (Pol II) subunit that forms a subcomplex with Rpb7, and the Spt4-Spt5 complex, a transcription elongation factor, have been shown to suppress Rad26-independent TCR. The Pol II-associated factor 1 complex (Paf1C) has been shown to function in transcription elongation, 3-processing of mRNAs, and posttranslational modification of histones. Here we show that Paf1C plays a marginal role in facilitating Rad26-dependent TCR but significantly suppresses Rad26-independent TCR. The suppression of Rad26-independent TCR is achieved by cooperating with Spt4-Spt5. We propose a model that, in the absence of Rad26, a lesion is "locked" in the active center of a Pol II elongation complex, which is stabilized by the coordinated interactions of Rpb4-Rpb7, Spt4-Spt5, and Paf1C with each other and with the core Pol II. We also found that Paf1C facilitates GGR, especially in internucleosomal linker regions. The facilitation of GGR is achieved through enabling monoubiquitination of histone H2B lysine 123 by Bre1, which in turn permits di-and trimethylation of histone H3 lysine 79 by Dot1. To our best knowledge, among the NER-modulating factors documented so far, Paf1C appears to have the most diverse functions in different NER pathways or subpathways.
Almost all cellular organisms are equipped with multiple DNA repair pathways to contend with constantly occurring DNA damage caused by endogenous and exogenous DNAdamaging agents (1). Nucleotide excision repair (NER) 2 is a DNA repair pathway that removes a wide variety of bulky, helix-distorting lesions that generally obstruct transcription and normal replication, such as UV-induced cyclobutane pyrimidine dimers (CPDs). NER is a multistep reaction that requires the coordinated action of over 30 proteins implicated in damage recognition, helix opening, lesion verification, dual incision of the damaged strand bracketing the lesion, removal of an oligonucleotide containing the lesion, gap-filling DNA synthesis, and ligation. Transcription-coupled repair (TCR) is an NER pathway dedicated to rapid repair in the transcribed strand (TS) of actively transcribed genes. Global genomic repair (GGR) is the other NER pathway that removes lesions throughout the genome, including the nontranscribed strand (NTS) of actively transcribed genes. The two NER pathways share most of the common NER factors but differ in the damage recognition step.
TCR is believed to be initiated by an RNA polymerase stalled at a lesion in the TS of active genes (2). However, TCR in eukaryotic cells appears to be extremely complicated, and the biochemical mechanism of the process is still largely unknown. In the budding yeast Saccharomyces cerevisiae, Rad26, a DNAstimulated ATPase that is homologous to the human CSB protein (3), plays an important role in TCR (4). However, TCR is not solely dependent on Rad26 because a substantial extent of TCR still occurs in rad26⌬ cells (5)(6)(7). Rpb9, a nonessential subunit of the 12-subunit (Rpb1-12) RNA polymerase II (Pol II), has been shown to be required for Rad26-independnet TCR (5). Interestingly, the Rad26-independent TCR has been shown to be suppressed by at least three proteins, namely Rpb4 (5), Spt4 (8), and Spt5 (9). Rpb4 is another nonessential Pol II subunit that forms a subcomplex with Rpb7. The Rpb4-Rpb7 subcomplex is associated with the 10-subunit core Pol II through a "wedge" structure on Rpb7, "pushing" the clamp of the core Pol II to a closed position (10,11). Spt4 forms a complex with Spt5, which physically interacts with Pol II (12). It was found recently that the role of Spt4 in suppressing Rad26-independent TCR is indirectly achieved by protecting Spt5 from degradation and by stabilizing the interaction of Spt5 with Pol II (9). Furthermore, the C-terminal repeat (CTR) domain of Spt5, which contains 15 copies of a six-amino acid sequence that can be phosphorylated by the Bur kinase, is responsible for suppressing Rad26-independent TCR (9).
Rad7, Rad16 (13), and Elc1 (14) are specifically required for GGR in yeast. The exact roles of these proteins in GGR are not yet clear. It has been proposed that the Rad7-Rad16 complex may act as an ATP-dependent motor that translocates along the DNA in search of damage, and upon encountering a lesion, the complex is stalled, which may remodel and open damaged chromatin, thereby facilitating recruitment of other NER factors (15). Elc1 has been shown to be a component of an E3 ubiquitin ligase that contains Rad7 and Rad16 (16). This E3 ubiquitin ligase has been shown to ubiquitinate Rad4, an essential NER factor required for both GGR and TCR. Optimal NER correlates with the ubiquitination of Rad4 but not its subse-quent degradation (16). Elc1 has also been suggested to be a component of another ubiquitin ligase complex, which contains Ela1, Cul3, and Roc1 but not Rad7 and Rad16, and is required for ubiquitination and degradation of Rpb1, the largest subunit of Pol II (17).
The basic repeating component of chromatin in eukaryotic cells is the nucleosome, which is composed of 146 base pairs of DNA wrapped around a protein octamer containing two molecules of each of the four core histones H2A, H2B, H3, and H4 (18). Although the packaging of DNA in chromatin can restrict the NER machinery, especially the GGR machinery, from accessing sites of DNA damage, limited pieces of evidence have emerged recently that chromatin metabolism may also play an active role in the repair process (19). For example, acetylation of histone H3 on lysine 9 and/or 14 by the acetyltransferase Gcn5 facilitates GGR (20,21). Also, SWI/SNF, an ATP-dependent chromatin-remodeling complex, has been shown to be recruited to chromatin upon induction of DNA damage by UV (22). A critical piece of evidence indicating the active engagement of chromatin in GGR is the recent discovery that methylation of histone H3 lysine 79 (H3K79), catalyzed by the histone methyltransferase Dot1, is required for GGR but plays no role in TCR in yeast (23). The Lys-79 residues of the two histone H3 molecules contained in a nucleosome are located at the top and bottom surfaces of the nucleosome disk and most likely regulate interactions with exogenous proteins (24). It was proposed that the methylated H3K79 may serve as a docking site for the GGR machinery on the chromatin (23).
The highly conserved Pol II-associated factor 1 complex (Paf1C), which is abundant in simple and complex eukaryotic cells, directly interacts with Pol II and chromatin at both promoter regions and throughout the coding regions of genes (for a recent review, see Ref. 25). In yeast, Paf1C is composed of Paf1, Rtf1, Cdc73, Leo1, and Ctr9, whereas in human cells, Paf1C also contains Ski8. Paf1C is involved in a variety of cellular processes, including transcription elongation, 3Ј-processing of mRNAs, and modification of chromatin. Genome-wide gene expression analyses have shown that Paf1C affects transcrip-tion of a small number of yeast genes, among which are many cell wall biosynthetic genes and a subset of cell cycle-regulated genes but no NER genes (26,27).
Although Paf1C binds to Pol II, the major functions of Paf1C may be independent of Pol II. Indeed, loss of Rtf1 or Cdc73, which results in loss of Paf1 factors from chromatin and from the Pol II complex, has little phenotypic consequence (28). Also, loss of Paf1, which results in severe phenotypes and reduced amounts of other Paf1C components, has little effect on the abundance or chromatin distribution of Pol II (28). Furthermore, Paf1C has been shown to be required for Bre1catalyzed monoubiquitination of histone H2B lysine 123 (H2BK123) (29 -31), which is in turn partially required for dimethylation and absolutely required for trimethylation of H3K79 by Dot1 (32)(33)(34). However, these histone modifications are not specifically limited to the transcribed regions of the genome (35)(36)(37)(38), supporting the idea that a fraction of Paf1C that is not associated with Pol II is able to promote these histone modifications.
In this paper, we identified Paf1C as a new suppressor of Rad26-independent TCR. We also discovered that Paf1C plays a marginal role in facilitating Rad26-dependent TCR and enhances GGR, especially in internucleosomal linker regions. Furthermore, we present evidence that the different roles of Paf1C in the different NER pathways or subpathways are manifested through different mechanisms.

EXPERIMENTAL PROCEDURES
Yeast Strains and Plasmids-Genes that were deleted individually or combinatorially in yeast cells are shown in Table 1. The bre1⌬ strain (MATa his3⌬1 leu2⌬0 met15⌬0 ura3⌬0 bre1::Kan) and its isogenic wild type BY4741 strain were purchased from Open Biosystems. Additional bre1⌬ strains were created in other strain backgrounds. Cells expressing mutant H3K79A and H2BK123A histones and their isogenic wild type strains YBL574 and FY406 (39) were kindly provided by Dr. Ali Shilatifard (Stowers Institute for Medical Research). Gene deletions in cells expressing the mutant histones and in the wild type strain Y452 (MAT␣ ura3-52 his3-1 leu2-3 leu2-112) were created using procedures described previously (5). Cells with their genomic PAF1 gene tagged with three consecutive FLAG (3ϫFLAG) sequences were created using PCR products amplified from plasmid p3FLAG-KanMX, as described previously (40). A multicopy plasmid (pGAL-SPT5) overexpressing 3ϫFLAG-tagged Spt5 under the control of the GAL10 promoter and single-copy centromeric plasmids encoding the fulllength or the CTR-deleted Spt5 were created as described previously (9). UV Sensitivity Assay-All strains used for the UV sensitivity test were created in the Y452 background. Yeast cells were grown at 30°C in YPD medium (2% peptone, 1% yeast extract, 2% glucose), and sequential 10-fold dilutions were made. The diluted samples were spotted onto YPD plates. When the spots had dried, the plates were irradiated with different doses of 254-nm UV light. The plates were incubated at 30°C for 3-4 days in the dark prior to being photographed.
UV Irradiation, Repair Incubation, and Genomic DNA Isolation-Yeast cells were grown at 30°C in minimal media containing 2% glucose or 2% galactose (to induce a gene under the control of the GAL1 promoter) to late log phase (A 600 Ϸ 1.0), washed twice with ice-cold water, resuspended in ice-cold 2% glucose (for glucose cultures) or 2% galactose (for galactose cultures), and irradiated with 80 J/m 2 of 254-nm UV light. Onetenth volume of a stock solution containing 10% yeast extract and 20% peptone was added to the irradiated cell suspension. The cells were incubated at 30°C in the dark to allow them to repair their DNA, and aliquots were collected at different time points. Genomic DNA was isolated from the cells as described previously (5).
NER Analysis of UV-induced CPDs at Nucleotide Resolution-The induction and repair of CPDs at individual sites in each strand of the RPB2 gene were measured using a method we developed previously (41)(42)(43). Briefly, ϳ1 g of genomic DNA was digested with DraI to release the RPB2 fragment and incised at CPDs with an excess amount of purified T4 endonuclease V (Epicenter). Two biotinylated oligonucleotides were then used to specifically "fish out" and label the TS and NTS of the RPB2 gene fragment, respectively. One pmol of one of the oligonucleotides was mixed with each of the samples. The mixtures were heated at 95°C for 5 min to denature the DNA and then cooled to an annealing temperature of around 50°C. One hundred g of streptavidin magnetic beads (Dynabeads M-280 Streptavidin, Invitrogen) was added to each of the mixtures to capture the strand of the RPB2 fragment hybridized to the biotinylated oligonucleotide. The other unwanted genomic DNA fragments were washed away. The fragments captured on the magnetic beads were 3Ј-end-labeled with [␣-32 P]dATP and Sequenase Version 2 (U.S. Biochemical Corp.). The labeled fragments were resolved on sequencing gels, which were then dried and exposed to a phosphorimaging screen (Bio-Rad). The signal intensities at gel bands corresponding to CPD sites were quantified by using Quantity One software (Bio-Rad). The percentages of CPDs remaining at individual sites after different times of repair incubation were calculated, and the times required for repairing 50% of CPDs (t1 ⁄ 2 ) were obtained by either linear or second order polynomial regression.
Immunoprecipitation of Pol II Complex-Yeast cells were cultured at 30°C in minimal medium to late log phase and harvested. The cells harvested from a 25-ml culture were washed with and resuspended in 0.5 ml of immunoprecipitation buffer (50 mM Tris-Cl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 0.4 mM Na 4 VO 3 , 10 mM Na 4 P 2 O 7 , 10 mM sodium fluoride, 0.5% Nonidet P-40, 1% Triton X-100, 0.1% SDS, 0.2 mM phenylmethylsulfonyl fluoride, and protease inhibitors) (44). The cells were broken by vortexing with acid-washed glass beads, and cell debris was removed by centrifugation at 20,000 ϫ g for 10 min at 4°C. Fifty l of the lysate was saved as "input." The remaining lysate was added with 15 g of 8WG16 (Neoclone), which recognizes the C-terminal heptapeptide repeats of Rpb1, the largest subunit of Pol II (45). The mixture was incubated at 4°C overnight with gentle rotation. Protein A-coated agarose beads (Sigma) were added to the mixture and incubated at 4°C for 3 h with gentle rotation. The beads were washed twice with immunoprecipitation buffer containing 0.5 M NaCl and twice with immunoprecipitation buffer containing 0.15 M NaCl. Bound proteins were eluted by boiling the beads in 50 l of 2 ϫ SDS-polyacrylamide gel loading buffer.
Chromatin Immunoprecipitation (ChIP) Assay-ChIP was carried out as described previously with slight modifications (46). Briefly, log phase cells with or without UV irradiation were fixed with 1% formaldehyde for 20 min at room temperature and lysed in buffer A (50 mM HEPES, pH 7.5, 140 mM NaCl, 1% Triton X-100, 0.1% sodium deoxycholate, 1 mM EDTA, and phosphatase and protease inhibitors) by vortexing with glass beads. The cell lysates were sonicated to shear the DNA to an average size of ϳ300 bp by using a Bioruptor (UCD-200, Diagenode). Chromatin fragments were immunoprecipitated with antibodies recognizing mono-, di-, and trimethylated H3K79 and total histone H3 (Abcam). Cross-links were reversed by incubation at 65°C overnight. DNA in the immunoprecipitated chromatin samples was purified by using the DNA Clean & Concentrator kit (Zymo Research). A 150-bp fragment in the RPB2 gene coding region (corresponding to nucleotides 177-327 downstream of the translation start codon ATG) was amplified by PCR using the purified DNA samples as templates.
Western Blot-Protein samples were resolved on an SDSpolyacrylamide gel and transferred onto a polyvinylidene difluoride membrane (Immobilon-P, Millipore). Protein bands of interest were detected with primary and secondary antibodies and SuperSignal West Femto Maximum Sensitivity Substrate (Pierce). The blots were scanned with the ChemiDoc TM XRS ϩ system (Bio-Rad). Band intensities were quantified using Quantity One software (Bio-Rad).

Paf1C Plays a Marginal Role in Facilitating
Rad26-dependent TCR-To determine the role of Paf1C in TCR, we deleted genes encoding Paf1C components in rad16⌬ cells, which are deficient in GGR (13), so that TCR can be unambiguously analyzed. Yeast cells were cultured to late log phase, UV-irradiated, and incubated in a repair medium for various lengths of time. Total genomic DNA was isolated, digested with a restriction enzyme to release a fragment of the constitutively expressed RPB2 gene, and incised at the CPDs with an excess amount of T4 endonuclease V (47). The TS of the RPB2 gene fragment was "fished out" with a biotinylated oligonucleotide and streptavidin magnetic beads, radioactively labeled at the 3Ј-end, and resolved on a DNA sequencing gel. As can be seen in Figs. 1A and 2, rapid TCR, which initiates ϳ40 nucleotides upstream of the transcription start site of the RPB2 gene, occurred in rad16⌬ cells. The TCR rates were marginally but reproducibly slower in rad16⌬ cells lacking a Paf1C component (Figs. 1 (B and C) and 2) (data not shown), indicating that Paf1C plays a marginal role in facilitating TCR.
In yeast, TCR is dependent on Rad26 and Rpb9 (5). To determine whether the marginal role of Paf1C in facilitating TCR is dependent on Rad26, Rpb9, or both, we analyzed TCR in rad16⌬ rpb9⌬ and rad16⌬ rpb9⌬ rtf1⌬ cells, where only Rad26-dependent TCR is operative. The TCR rate was also marginally but reproducibly slower in the rad16⌬ rpb9⌬ rtf1⌬ cells than in the rad16⌬ rpb9⌬ cells (Figs. 1 (D and E) and 2), indicating that the marginal role of Paf1C in facilitating TCR is dependent on Rad26.
Additional deletion of RTF1 in rad16⌬ and rad16⌬ rpb9⌬ cells caused increased UV sensitivity (Fig. 3), indicating that Paf1C is not epistatic to Rad16 or Rpb9 and supporting the notion that Paf1C facilitates Rad26-dependent TCR. However, the effect of the RTF1 deletion on UV sensitivity appeared to be greater than would be expected from the marginal deficiency in Rad26-dependent TCR caused by the deletion. It is therefore likely that, besides the marginal role in facilitating Rad26-dependent TCR, Paf1C may function in other repair and/or DNA damage tolerance pathway(s).
Paf1C Suppresses Rad26-independent TCR-We then asked what role Paf1C may play in Rad26-independent TCR. Surprisingly, elimination of a Paf1C component in rad16⌬ rad26⌬ cells, where only Rad26-independent TCR is operative, resulted in enhanced repair (Figs. 4 (compare B-E with A) and 5), indicating that Paf1C suppresses Rad26-independent TCR.
In agreement with our previous studies (5,6), no repair can be seen in rad16⌬ rad26⌬ rpb9⌬ cells (Figs. 4F and 5). Additional elimination of a Paf1C component did not result in restoration of TCR in rad16⌬ rad26⌬ rpb9⌬ cells (Figs. 4G and 5). These results indicate that Paf1C suppresses Rad26-independent TCR that is absolutely dependent on Rpb9.
Having found the role of Paf1C in suppressing Rad26-independent TCR, we wanted to elucidate how this might occur. Several factors, such as Rpb4 (5), Spt4, and Spt5 (through its CTR domain) (8,9) have been shown to also suppress Rad26independent TCR. It is possible that Paf1C acts cooperatively with these factors in the suppression. We therefore attempted to investigate the functional interactions between Paf1C and these factors. We found that deletion of RPB4 and a Paf1C gene is synthetically lethal for the cell (not shown), which prevented us from further exploring the functional interaction between Paf1C and Rpb4. We therefore turned to Spt4 and Spt5.

Paf1C Does Not Suppress Rad26-independent TCR in Cells
Lacking Spt4, and Vice Versa-In agreement with a previous report (8), TCR was more rapid in rad16⌬ rad26⌬ spt4⌬ cells than in rad16⌬ rad26⌬ cells (Figs. 4 (compare A and H) and 5). We wondered if Paf1C and Spt4 suppress Rad26-independent TCR through a common pathway. If they do, Paf1C will not suppress Rad26-independent TCR in cells lacking Spt4, and vice versa. Indeed, the TCR rate in rad16⌬ rad26⌬ cells lacking both a Paf1C component and Spt4 was slightly slower, rather than faster, than those lacking either a Paf1C component or Spt4 (Figs. 4 (compare I with B-E and H) and 5). In other words, elimination of a Paf1C component did not further release Rad26-independent TCR in cells lacking Spt4, and vice versa. This indicates that the roles of Paf1C and Spt4 in suppressing Rad26-independent TCR are likely to be through a common pathway. The slight slowdown of TCR in rad16⌬ rad26⌬ cells lacking both a Paf1C component and Spt4 may be attributed to the fact that both Paf1C and Spt4 function in transcription elongation, and elimination of both may synergistically cause a transcriptional elongation defect.
The Role of Paf1C in Suppressing Rad26-independent TCR Is Not Subsidiary to That of Spt5-Spt4 forms a complex with Spt5 (12). Unlike Spt4, which is dispensable, Spt5 is essential for cell viability (48). It has been proposed that Spt5 may directly interact with Pol II and play a more fundamental role in transcription, whereas Spt4 may be associated with Pol II through interaction with Spt5 (49). We recently found that Spt4 actually suppresses Rad26-independent TCR indirectly by protecting Spt5 from degradation and by stabilizing the interaction of Spt5 with Pol II (9). Indeed, overexpression of Spt5 suppresses Rad26-independent TCR in cells lacking Spt4 (9). To test if the role of Paf1C in suppressing Rad26-independent TCR is also subsidiary to that of Spt5, we overexpressed Spt5 in rad16⌬ rad26⌬ rtf1⌬ cells. The overexpression did not affect the TCR rate in these cells (Figs. 4 (compare J with B) and 5), indicating that Paf1C is indispensable for suppressing Rad26-independent TCR even when an excess amount of Spt5 is present. In other words, both Paf1C and Spt5 are required for suppressing  Rad26-independent TCR, and the role of Paf1C in the suppression is not subsidiary to that of Spt5.
The Association of Paf1C with Pol II Is Facilitated by the CTR Domain of Spt5-We have found that the CTR domain of Spt5, which is dispensable for cell viability and is not required for interactions with Spt4 and Pol II, is responsible for suppressing Rad26-independent TCR (9). Previous studies by others have indicated that Spt5 and Paf1C have extensive genetic and physical interactions (50). We wondered if the Spt5 CTR is responsible for recruiting Paf1C to Pol II, thereby forming a larger complex that suppresses Rad26-independent TCR. We created yeast strains whose genomic SPT5 gene is deleted and complemented with a single-copy centromeric plasmid encoding the full-length or CTR-deleted Spt5. Three consecutive FLAG sequences (3ϫFLAG) were tagged to the coding sequence of the genomic PAF1 gene in these yeast cells. The 3ϫFLAG tag did not cause any noticeable deficiency for the cells (not shown). We immunoprecipitated the Pol II complex with antibody 8WG16, which recognizes the C-terminal heptapeptide repeats of Rpb1, the largest subunit of Pol II (45). The presence of Pol II and 3ϫFLAG-tagged Paf1 in the immunoprecipitates was examined on a Western blot by using the anti-Rpb1 (8WG16) and an anti-FLAG antibody, respectively. The level of 3ϫFLAG-tagged Paf1 in cells expressing the CTR-deleted Spt5 is ϳ1.6-fold that in cells expressing the full-length Spt5 (Fig. 6, Input), presumably due to an unknown compensation mechanism in the absence of the Spt5 CTR. The 3ϫFLAG-tagged Paf1 coimmunoprecipitated with Pol II in cells expressing the CTRdeleted Spt5 is ϳ30% of that in cells expressing the full-length Spt5 (Fig. 6, IP), indicating that the Spt5 CTR facilitates the association of Paf1C with Pol II. In view of the observations that Paf1C is not subsidiary to Spt5 in suppressing Rad26-independent TCR and that the association of Paf1C with Pol II is facil-itated by the Spt5 CTR, it is likely that a complex that includes at least Spt4-Spt5 and Paf1C is responsible for the suppression.
Paf1C Is Not Epistatic to Either Rad26 or Spt4/Spt5-In line with a role for Spt4 and the Spt5 CTR in suppressing Rad26independent TCR, elimination of either Spt4 (8,9) or the Spt5 CTR (9) in rad16⌬ rad26⌬ (or rad7⌬ rad26⌬) cells restores the UV resistance of these cells to the level of rad16⌬ (or rad7⌬) cells. However, elimination of Spt4 or the Spt5 CTR in RAD26 ϩ cells does not affect UV sensitivity, which agrees with the observations that Spt4 and the Spt5 CTR do not affect GGR and Rad26-dependent TCR (8,9). However, instead of restoring UV resistance, additional elimination of any Paf1C component in rad16⌬ rad26⌬ cells enhanced UV sensitivity ( Fig. 7; rad16⌬ rad26⌬ cells with other Paf1C components eliminated not shown). Also, elimination of Paf1C components enhanced the UV sensitivity of rad16⌬ rad26⌬ spt4⌬ cells (Fig. 7). These results indicate that Paf1C is not epistatic to either Rad26 or Spt4/Spt5, and besides having a role in suppressing Rad26-independent TCR together with Spt4/Spt5, Paf1C functions in other repair and/or DNA damage tolerance pathway(s).
Paf1C Facilitates GGR, Especially in Internucleosomal Linker Regions-To determine if Paf1C plays a role in GGR, we measured repair of UV-induced CPDs in the NTS of the RPB2 gene. In principle, NER in either strand of a repressed gene may also reflect GGR. However, "noise" transcription, which commonly occurs in both strands of repressed genes in eukaryotic cells (51), may be able to initiate a certain level of TCR, which can be confused with GGR (52). Indeed, apparent NER that is dependent on Rad26 still occurs in all repressed genes (e.g. GAL1-10, PHO5, and ADH2) tested in GGR-deficient rad16⌬ cells (52). The noise transcripts may not be detected by traditional ways because they can be rapidly degraded after being produced in the cell (51). Active transcription from the TS of a gene may prevent noise transcription from the NTS. Therefore, NER in the NTS of an actively transcribed gene may reflect GGR better than that in either strand of a repressed gene. We have found that NER in the NTS of the RPB2 gene is absolutely dependent on the GGR-specific factors Rad7, Rad16, and Elc1 and thus appears to exclusively reflect GGR (14,52).
In wild type cells, CPDs were repaired at different rates at different sites in the NTS of the RPB2 gene (Figs. 8A and 9). The repair rates generally correlated with nucleosome positioning, being slowest in the central regions of nucleosomal core DNA and fastest in the internucleosomal linker regions (Figs. 8A (marked with brackets on the right) and 9). This indicates that nucleosome structure inhibits GGR, in agreement with previous reports (e.g. see Refs. 6 and 53). In cells lacking a Paf1C component, GGR was still apparent but significantly compro- Yeast cells whose genomic PAF1 gene was tagged with 3ϫFLAG and whose genomic SPT5 gene was deleted and complemented with a single-copy centromeric plasmid encoding either the full-length or CTR-deleted Spt5 were cultured to late log phase. Pol II complexes were immunoprecipitated from these cells with antibody 8WG16 (anti-Rpb1). The immunoprecipitation input and immunoprecipitates (IP) were subjected to Western blot, and the presence of Rpb1 and 3ϫFLAG-tagged Paf1 (Paf1-3F) on the blot was detected with 8WG16 and anti-FLAG antibody, respectively. Numbers below the blot indicate relative levels of 3ϫFLAG-tagged Paf1 (normalized to the level of Rpb1) (the level in cells expressing the full-length Paf1 is set as 1). FIGURE 7. Deletion of rtf1 in rad16⌬ rad26⌬ and rad16⌬ rad26⌬ Spt4⌬ cells enhances UV sensitivity. Saturated cultures of yeast strains were sequentially 10-fold diluted and spotted onto YPD plates. When the spots had dried, the plates were irradiated with the indicated doses of 254-nm UV light. The plates were incubated at 30°C for 3-4 days in the dark prior to being photographed. mised (Figs. 8 (compare A with B and C) and 9). Indeed, in the internucleosomal linker regions, the times required for repairing half of the CPDs in Paf1C-eliminated cells were about twice as long as those in wild type cells (Figs. 8 (A-C; compare bands marked with brackets on the right of A) and 9). We noticed, however, that the difference of the repair speeds between Paf1C-eliminated and wild type cells in the nucleosomal core DNA (especially in the central regions of the nucleosomal core) was not as dramatic as in internucleosomal linker regions, which could be due to the fact that GGR was quite slow in nucleosomal core DNA even in wild type cells (Figs. 8A and 9). It appears that Paf1C cannot effectively overcome the inhibitory effect of positioned nucleosomes on GGR, especially in the central regions of nucleosomal core DNA. Taken together, our results indicate that Paf1C facilitates GGR, especially in internucleosomal linker regions.
Paf1C Facilitates GGR through Enabling Monoubiquitination of H2BK123 by Bre, Which in Turn Permits Di-and Trimethylation of H3K79 by Dot1-We recently found that methylation of H3K79 by Dot1 is required for GGR (23). Monoubiquitination of H2BK123, which is catalyzed by the E3 ubiquitin ligase Bre1, has been shown to be partially required for dimethylation and absolutely required for trimethylation but is dispensable for monomethylation of H3K79 (33,34). We also found that Bre1 and monoubiquitination of H2BK123 facilitates GGR, especially in internucleosomal linker regions (23).
Paf1C has been shown to be required for bringing Bre1 to the chromatin to monoubiquitinate H2BK123. This function of Paf1C appears to be achieved by direct and selective interaction with Bre1 (54). We wondered if the role of Paf1C in facilitating GGR is through Bre1-mediated monoubiquitination of H2BK123, which in turn permits di-and trimethylation of H3K79. Deletion of a Paf1C gene in bre1⌬, dot1⌬, H2BK123A, and H3K79A mutant cells did not significantly affect their GGR patterns (Figs. 8 (D-J), 9, and 10). Also, the UV sensitivity of rft1⌬ cells is similar to that of bre1⌬ or dot1⌬ cells, being about 10 times more sensitive than wild type cells (Fig. 11). Although Dot1 is required and Bre1 is partially required for GGR, dot1⌬ cells are not significantly more UV-sensitive than bre1⌬ cells (Fig. 11), reflecting the fact that the UV sensitivity may not always faithfully reflect repair capacity. Indeed, previous studies have shown that elimination of the GGR-specific factor Elc1 (14) or the TCR-specific factor Rad26 (4,5) in otherwise wild type cells does not cause any detectable UV sensitivity. Combined deletions of RTF1  (65). The arrow on the right indicates the transcription start site. Brackets on the right of A indicate bands of CPDs located in the internucleosomal linker regions that were repaired rapidly in WT cells but more slowly in the mutants. with either BRE1 or DOT1 did not result in additional UV sensitivity relative to the single mutants (Fig. 11), indicating that RTF1 is epistatic to BRE1 and DOT1.
We then examined if Paf1C affects H3K79 methylation. Indeed, like bre1⌬ and H2BK123R mutant cells (33), rtf1⌬ cells showed undetectable trimethylation, dramatically reduced dimethylation, and increased monomethylation of H3K79 (Fig.  12A). To measure the states of H3K79 methylation in the RPB2 gene, which we used to map GGR and TCR, we performed a ChIP assay, by using antibodies recognizing mono-, di-, and trimethylated H3K79, respectively. The states of H3K79 methylation in the RPB2 gene appeared to reflect cellular levels of the different states of H3K79 methylation (Fig. 12B). UV irradiation did not seem to significantly affect H3K79 methylation overall or in the RPB2 gene (Fig. 12).
Taken together, our results indicate that Paf1C facilitates GGR through enabling monoubiquitination of H2BK123 by Bre1, which in turn permits di-and trimethylation of H3K79 by Dot1.

DISCUSSION
We discovered that Paf1C plays a marginal role in facilitating Rad26-dependent TCR but significantly enhances GGR, especially in internucleosomal linker regions. We also identified Paf1C as a new suppressor of Rad26-independent TCR. Because different NER pathways or subpathways can be superimposed in the same gene, it can be challenging to dissect the multiple functions of a factor in the different NER pathways or subpathways, especially if the factor acts positively in one NER pathway or subpathway but negatively in another. We have successfully dissected the different functions of Paf1C in the different NER pathways or subpathways. This successful dissection can be attributed to two technological aspects. First, genes that are specifically required for different NER pathways or subpathways have been identified, and it is possible to create desirable combinations of deletions of these genes in haploid yeast cells. Second, nucleotide-level NER analysis methods are available, which allow for unambiguous comparison of the repair rates among different sites/regions in the same DNA fragment. To the best of our knowledge, among the NER-modulating factors documented so far, Paf1C has the most diverse functions in the different NER pathways or subpathways.
How Is the Marginal Role of Paf1C in Facilitating Rad26-dependent TCR Achieved?-We found that Paf1C plays a marginal role in facilitating Rad26-dependent TCR. It is unlikely that the marginal role is manifested by promoting expression of a common NER factor(s) that is shared by the TCR and GGR machineries. Indeed, Paf1C also facilitates GGR, especially in internucleosomal linker regions. However, the facilitation of GGR by Paf1C is achieved by enabling ubiquitination of H2BK123 by Bre1 and di-and trimethylation of H3K79 by Dot1, and these histone modifications have been shown to play no role in TCR (including Rad26-dependent TCR) (23). Also, genome-wide gene expression analyses have shown that Paf1C affects transcription of a small subset of yeast genes, among which are many cell wall biosynthetic genes and a subset of cell cycle-regulated genes but no NER genes (26,27).
It has been shown that, in the absence of DNA damage, Rad26 associates with the coding sequence of a gene in a transcription-dependent manner (55). Induction of DNA damage does not appear to cause more recruitment of Rad26 to an actively transcribed gene. It is therefore likely that Rad26 is intrinsically associated with Pol II, thereby "priming" the transcription machinery competent for TCR. The marginal role of Paf1C in facilitating Rad26-dependent TCR may be achieved by enhancing the priming. We found that a small amount of Rad26 indeed coimmunoprecipitates with Pol II. However, the coimmunoprecipitation is not affected by the presence of Paf1C, and the association of Paf1C with Pol II is not affected by the presence of Rad26 (not shown). Therefore, if and how Paf1C enhances the priming remains to be determined. Rad26 plays a role in transcription elongation (56). It is also quite possible that the marginal effect of Paf1C in Rad26-dependent TCR is due to  . RTF1 is epistatic to DOT1 and BRE1. Saturated cultures of yeast strains were sequentially 10-fold diluted and spotted onto YPD plates. When the spots had dried, the plates were irradiated with the indicated doses of 254-nm UV light. The plates were incubated at 30°C for 3-4 days in the dark prior to being photographed. FIGURE 12. Paf1C is partially required for dimethylation and absolutely required for trimethylation of H3K79. A, Western blots showing cellular H3K79 methylation states. The asterisk on the right indicates the band of an unknown protein that cross-reacts with the trimethylated H3K79 (H3K79me3) antibody. B, ChIP assay measuring H3K79 methylation sates in the RPB2 gene. Agarose gels show PCR products of a 150-bp RPB2 fragment (corresponding to nucleotides 177-327 downstream of the translation start codon ATG) immunoprecipitated by using the indicated antibodies. The arrows on the right indicate the RPB2 fragment amplified. Asterisks on the right indicate PCR primer dimmers. ϩUV samples were from cells irradiated with 80 J/m 2 of 254-nm UV and then incubated in a repair medium for 30 min.
its pleiotrophic effect, which is likely to affect transcription rather than a specific effect on TCR.
How Does Paf1C Suppress Rad26-independent TCR?-Pol II is a globular enzyme with a deep central cleft (10,11). The DNA template enters and travels along the base of this cleft to the active site. On one side of the cleft is a flexible clamp structure. Binding of the Rpb4-Rpb7 subcomplex to the 10-subunit core Pol II "pushes" the clamp to the closed position (10,11). RNA polymerases (57) and Spt4/Spt5 (58) from all three kingdoms of life (bacteria, archaea, and eukaryotes) are conserved. The Spt4-Spt5 complex has not been co-crystallized with an intact RNA polymerase (49). However, an archaeal Spt4-Spt5 has been co-crystallized with the clamp domain of an archaeal RNA polymerase (49). Based on the archaeal co-crystal structure, a model of the complete yeast Pol II-Spt4-Spt5 elongation complex has been proposed. This model suggests that the NGN domain of Spt5 binds to the Pol II clamp and closes the central cleft to lock nucleic acids and render the elongation complex stable and processive. Spt4 binds to the other side of the NGN domain of Spt5 and points away from the Pol II surface. The KOW1 domain of Spt5 may contact DNA and/or exiting RNA, and such contacts could contribute to Pol II elongation complex stability and may also involve the Rpb4-Rpb7 subcomplex. The locations of other domains (KOW2 to -4 and the CTR) of Spt5 are currently unpredictable (49). How Paf1C interacts with Pol II is presently unknown. However, the interaction may be at least partially through Sp4-Spt5. Paf1C and Spt5 have been shown to have extensive genetic and physical interactions (50). Furthermore, optimal association of Paf1C with Pol II is dependent on Spt4 (59) and the Spt5 CTR (Fig. 6).
Structure-function analyses of Pol II elongation complexes containing a T-T CPD in the TS showed that the CPD slowly passes a translocation barrier and enters the Pol II active site. The 5Ј-T of the CPD then directs uridine misincorporation into the elongating mRNA, which stalls the translocation of Pol II (60). Our results indicate that both Paf1C and the Spt4-Spt5 complex are required for suppressing Rad26-independent TCR, and the two complexes may exert the suppression through a common pathway. Our previous results showed that Rpb4 is also required for suppression of Rad26-independent TCR (5). It is therefore likely that, in the absence of Rad26, a lesion is "locked" in the active center of a Pol II elongation complex, which is stabilized by the coordinated interactions of Spt4-Spt5, Rpb4-Rpb7, and Paf1C with each other and with the core Pol II. Elimination of any of these factors may destabilize the Pol II elongation complex, making it possible for TCR to take place. The role of Rad26 in TCR may be achieved indirectly by destabilizing the Pol II elongation complex because in the absence of any of these TCR suppressors, Rad26 is dispensable for TCR. This model may explain why Spt4-Spt5, Rpb4-Rpb7, and Paf1C suppress TCR only in the absence of Rad26. However, it remains to be understood how Rad26 destabilizes the Pol II elongation complex and how TCR takes place in the absence of both Rad26 and the TCR suppressors.
Although Paf1C and Spt4-Spt5 may suppress Rad26-independent TCR through a common pathway, Paf1C appears to be more "peripheral" for the suppression. First, TCR is somewhat slower in rad16⌬ rad26⌬ cells lacking a Paf1C component than in rad16⌬ rad26⌬ spt4⌬ cells (Figs. 4 and 5). Second, elimination of Spt4 restores TCR not only in cells lacking Rad26 but also, to a certain extent, in cells lacking both Rad26 and Rpb9 (61) (data not shown). Therefore, both Rad26 and Rpb9 appear to facilitate TCR indirectly rather than by directly recruiting NER factors. However, elimination of a Paf1C component restores TCR only in cells lacking Rad26 and not in cells lacking both Rad26 and Rpb9 (Figs. 4 and 5). The Spt4-Spt5 complex appears to interact with Pol II more directly and more tightly than Paf1C because the association of Paf1C with Pol II is dependent on Spt4 and the Spt5 CTR, whereas the association of Spt4-Spt5 with Pol II is not affected by Paf1C (59) (this study and data not shown). Therefore, elimination of Spt4 may cause more destabilization of the Pol II elongation complex than Paf1C, thereby allowing a higher extent of restoration of TCR in rad26⌬ and rad26⌬ rpb9⌬ cells.
How does Paf1C Facilitate GGR?-Our results indicate that Paf1C facilitates GGR through enabling monoubiquitination of H2BK123 by Bre1, which in turn permits di-and trimethylation of H3K79 by Dot1. Paf1C is a transcription elongation factor and travels along with Pol II. How can Paf1C facilitate GGR in inactive genes or intergenic regions? In fact, although Paf1C is essential for H2BK123 ubiquitination, which is in turn the prerequisite for di-and trimethylation of H3K79 by Dot1, these histone modifications are not specifically limited to the transcribed regions of the genome. Indeed, ϳ90% of all histone H3 is methylated on Lys-79, and the relative levels of H3K79 methylation have little correlation with the transcriptional activity of a gene (35,36,38). Also, the levels of H2BK123 ubiquitination do not seem to be correlated with the transcriptional activity of a gene (37). The widespread features of these histone modifications make them ideal for mediating GGR. It is possible that a fraction of Paf1C that is not associated with Pol II is able to promote the histone modifications. There is evidence that Paf1C has functions independent of Pol II (28). As discussed above, Spt4 and the Spt5 CTR are important for the association of Paf1C with Pol II. However, unlike elimination of a Paf1C component, elimination of Spt4 or the Spt5 CTR does not cause any defect in GGR. This supports the notion that Paf1C may enable the histone modifications and facilitate GGR independently of Pol II. However, it remains to be elucidated how methylation of H3K79 is engaged in GGR.
Epistatic Interactions of Paf1C with NER-modulating Factors-Although it plays a marginal role in facilitating Rad26-dependent TCR, suppresses Rad26-independent (Rpb9-dependent) TCR along with Spt4-Spt5, and facilitates GGR, Paf1C is not epistatic to Rad26, Rpb9, Spt4, or Rad16 in terms of UV resistance (Figs. 3 and 7). Therefore, besides modulating different NER pathways or subpathways, Paf1C must have an additional function that confers the cells with additional UV resistance. The additional function is probably the activation of DNA damage checkpoints, which enhances the resistance of cells to DNA damage (62). Cells lacking Paf1C, Dot1, and Bre1 have been shown to have a similar defect in activation of DNA damage checkpoints (63,64), implying that the roles of these factors in checkpoint activation are mostly manifested through their common ability to directly or indirectly permit di-and trimethylation of H3K79. Therefore, the epistatic relationship among Paf1C, Dot1, and Bre1 can be explained by their com-mon effects on di-and trimethylation of H3K79, which play roles not only in GGR but also in checkpoint activation.