key: cord-0305485-y9tjvn4n
authors: Bezalel-Buch, Rachel; Cheun, Young K.; Roy, Upasana; Schärer, Orlando D.; Burgers, Peter M.
title: Bypass of DNA Interstrand crosslinks by a Rev1-DNA polymerase ζ complex
date: 2020-05-10
journal: bioRxiv
DOI: 10.1101/2020.05.09.085902
sha: 79d10dca01349190044f68c7f2a26e00545bf788
doc_id: 305485
cord_uid: y9tjvn4n

DNA polymerase ζ (Pol ζ) and Rev1 are essential for the repair of DNA interstrand crosslink (ICL) damage. We have used yeast DNA polymerases η, ζ, and Rev1 to study translesion synthesis (TLS) past a nitrogen mustard-based ICL with an 8-atom linker between the crosslinked bases. The Rev1-Pol ζ complex was most efficient in complete bypass synthesis, by 2-3 fold, compared to Pol ζ alone or Pol η. Rev1 protein, but not its catalytic activity, was required for efficient TLS. A dCMP residue was faithfully inserted across the ICL-G by Pol η, Pol ζ, and Rev1-Pol ζ. Rev1-Pol ζ, and particularly Pol ζ alone showed a tendency to stall before the ICL, whereas Pol η stalled just after insertion across the ICL. The stalling of Pol η directly past the ICL is attributed to its autoinhibitory activity, caused by elongation of the short ICL-unhooked oligonucleotide (a six-mer in our study) by Pol η providing a barrier to further elongation of the correct primer. No stalling by Rev1-Pol ζ directly past the ICL was observed, suggesting that the proposed function of Pol ζ as an extender DNA polymerase is also required for ICL repair.

DNA Polymerase z (Pol z) is a multi-subunit B-family DNA polymerase enzyme that is involved in translesion DNA synthesis (TLS) (1) . Unlike the other B family members (Pol a, Pol d, Pol e), which show a high replication fidelity, Pol z has a low fidelity (2, 3) . Pol z is a four-subunit DNA polymerase (Rev3, Rev7, Pol31, Pol32), sharing the Pol31 and Pol32 subunits with Pol d (4, 5) .

Its interaction with the replication clamp PCNA increases its activity, presumably by increasing the processivity of the enzyme (5, 6) . Rev1 DNA polymerase is a Y family DNA polymerase with a unique dCMP transferase activity (7) . However, while Saccharomyces cerevisiae REV1 is absolutely required for damage-induced mutagenesis (8) , its catalytic activity is not (9) , suggesting that the non-catalytic functions of Rev1 are important. Therefore, Rev1 is considered mainly as a scaffold protein onto which the mutasome is assembled. It promotes mutagenesis by essential interactions with other factors, including mono-ubiquitinated PCNA and Pol z (10-13).

Cells deficient for Pol z or Rev1 are sensitive to DNA damage and are among the most hypersensitive to treatment with interstrand cross-linking (ICL) agents, suggesting that they have a key role in ICL repair (14) (15) (16) (17) (18) (19) . DNA ICLs covalently link two strands of the double helix, thereby providing a complete block to both the DNA replication and the transcription machineries. ICLs can be formed by endogenous sources as well by antitumor agents such as nitrogen mustard and cisplatin (20, 21) . The removal of ICLs from genomes necessarily requires a complex repair process and several pathways for ICL repair have been described. Although ICL repair also occurs in the G1 phase of the cell cycle, the removal of ICLs is most critical during S-phase, where they provide an absolute block to replication (22) . ICLs can be encountered by the collision of two replication forks (23) , or a unidirectional fork can traverse the ICL in a FancM-dependent step followed by priming and fork resumption downstream of the ICL (24) . Regardless, all ICL repair pathways require an unhooking step to separate the two crosslinked strands and a DNA synthesis step to restore the duplex. Studies of the repair of sitespecific plasmid-based ICLs in Xenopus egg extracts have provided a biochemical framework for understanding ICL repair and illustrate what roles DNA polymerases play in ICL repair (23) .

Activation of the Fanconi anemia (FA) pathway at the ICL leads to the recruitment of the ERCC1-XPF and possibly other endonucleases to unhook the crosslink from one of the two strands (25, 26) . The unhooked ICL may be trimmed further by an exonuclease such SNM1A/Pso2 (27) . The unhooked and processed ICL is then bypassed by translesion synthesis (TLS) DNA polymerases to generate one continuous duplex that can be used as a template to fix the second strand (23) . While this FA-dependent pathway can act on most ICLs, several variants of this general pathway exist for ICLs formed by psoralen, abasic sites or acetaldehydes (28, 29) .

Studies in the xenopus system showed that depletion of Rev1 from the extract did not affect the efficiency and kinetics of insertion at the ICL site, however, it did have an inhibitory effect on the extension reaction (13) . Similarly, depletion of the Rev7 inhibited the extension of the repair product past the insertion opposite a cisplatin ICL, in agreement with the known role of Pol z as an extender polymerase (23) . Biochemical studies have shown that various TLS polymerases can bypass ICLs (21, (30) (31) (32) , but conclusive studies of the bypass activity of purified Pol z and Rev1, are missing. In this paper, we provide critical new information regarding the functions of Pol z and Rev1 on an ICL substrate that mimics an unhooked nitrogen mustard lesion (30, 33) .

Our studies show that Rev1 stimulates ICL bypass by Pol z, but its enzymatic activity is largely dispensable. Both Pol h and Rev1-Pol z, but not Pol d can mediate ICL bypass. However, Pol h shows very strong stalling directly past insertion at the ICL. In contrast, Rev1-Pol z proceeds more efficiently past the ICL, suggesting that the enzyme is suitable for complete TLS.

We and others have previously shown that the structure of the ICL that links two bases as well as the length of the duplex around an ICL affects bypass by DNA polymerases (30, 31, 33) . We used our previously published method to generate a stable nitrogen mustard ICL mimic with an 8 atom crosslink through a double reductive amination reaction of an aldehyde ICL precursor with dimethylethylenediamine embedded in a 6mer duplex (8a-6bp ICL, Figure 1A ) (34, 35) . To the initially obtained ICL on a 39mer template, we ligated 5' and 3' biotinylated oligos to generate a 93mer substrates with biotinylated ends. Our design was based on the consideration that i) when replication forks stall, about 20 nucleotides remain unreplicated between the leading strand 3'-end and the ICL, and ii) that nucleolytic processing of the non-template strand leaves the unhooked ICL in a duplex of only a few nucleotides (23) . The biotin moieties at either template terminus were introduced to bind streptavidin to form blocks to prevent PCNA from sliding off the DNA (6). The control substrate (undamaged DNA) lacks the ICL and the 6 nucleotides of dsDNA, which would not form a stable duplex with the template strand in the absence of the ICL.

We have previously shown that Pol z can be purified as a stable 5-subunit complex together with Rev1 (36) . A complex with the identical activity can also be obtained by simply mixing Pol z together with Rev1, thereby reconstituting the 5-subunit complex. In this paper, we have reconstituted the various complexes of Rev1-Pol z by mixing Rev1, or Rev1 mutants, together with wild-type four-subunit Pol z.

The DNA substrates were pre-incubated with the single-stranded DNA binding protein RPA (replication protein A), and PCNA (proliferating cell nuclear antigen) was loaded by RFC (replication factor C) and ATP, as shown in Figure 1B . The reaction was started by addition of the individual DNA polymerases. dNTPs were present at concentrations that prevail during the DNA damage response in yeast (36) . Replication of the undamaged control DNA was carried out very efficiently, with the exception of replication by Rev1-Pol z (lane 5, discussed below).

The presence of the streptavidin moiety at the template 5'-end sterically limits full extension by Pol d (lanes 1,2) and Pol z (lanes 4,5), compared to Pol h (lane 3), and this pattern was also displayed with the ICL-containing substrates.

Pol d stalled mainly at the nick position, four nucleotides prior to encountering the crosslinked G residue ( Figure 1C , lanes 6-9), consistent with its known idling function at DNA nicks (37) . However, about 2-3% full-length product was observed at every time point, which could be due to incomplete crosslinking of the substrate or a low level of bypass by Pol d (see also Supplementary Figure S2A ). This non-specific background observed by wild-type Pol d was subtracted from the percentage of extension products observed with the TLS enzymes. First, we eliminated the proofreading function of Pol d (Pol d-exo -, D520V). This reduced idling at the nick (37) , but still allowed only minimal bypass synthesis over the 30 min time-course of the assay ( Figure 1C , lanes 10-13; Figure 1E ).

Based on previous studies of TLS by Pol h, we have divided the replication products into three classes for the purpose of quantification (33) . Products up to the ICL are designated as "approach", those that have inserted a nucleotide opposite the crosslinked G, plus an additional three nucleotides past the ICL are designated as "bypass", and those longer than that up to fulllength are designated as "extension". Quantification is shown in Figure 1E . The bypass of this particular ICL has been studied previously, however these model studies were carried out with Pol h alone, without the PCNA clamp and at low salt concentrations (33) . TLS by Pol h alone was most efficient at low salt (50 mM NaCl, Supplementary Figure S2B ). For sake of consistency, all studies in this paper were carried out at 100 mM NaCl. Under these conditions, the limited bypass of the ICL by Pol h was stimulated substantially by PCNA ( Figure 1C , compare lanes 14-16 with 17-19 & Figure S2B , compare lanes 7-9 with 10-12, and 13-15 with [16] [17] [18] . However, as observed previously (33) , the majority of products are still stalled at the +1 to +3 positions. In contrast, Pol z stalled predominantly prior to the ICL, similar to proofreadingdefective Pol d (lanes [20] [21] [22] . However, significant bypass and extension was observed.

Notably, the lack of products at the +1 to +3 positions, together with the presence of fully replicated DNA indicates that, once a nucleotide has been inserted opposite the ICL-G position, this product is favorably accommodated in the Pol z binding site to allow continued extension.

This remarkable difference between Pol h and Pol z is shown graphically in Figure 1D .

Three dramatic changes were observed with Rev1-Pol z as TLS enzyme compared to Pol z alone. First, strong pause sites were observed during replication of normal (undamaged) DNA ( Figure 1C , lanes 4 and 5). The inhibition of Pol z activity by Rev1 on undamaged DNA is the subject of a different study (Bezalel-Buch, R. and Burgers, P. unpublished data).

Replication of the ICL-DNA also shows these ICL-independent distant pause sites, in addition to ICL-dependent pause sites directly ahead of the crosslink. Second, synthesis past the ICL is enhanced by the inclusion of Rev1 ( Figure 1C , lanes [23] [24] [25] . Third, once bypass has been achieved, further replication of the downstream template is once more inhibited as is evident from the presence of multiple, distant stall sites. After 30 min, the total extension products by Rev1-Pol z are about 30% compared to 10% with Pol z alone ( Figure 1C -E).

The catalytically inactive form of Rev1 stimulates Pol z-mediated ICL bypass.

The data in Figure 1 show clearly that Rev1 has an important role in Pol z-mediated ICL bypass and extension. To assess whether this is due to the catalytic function of Rev1 or to its proposed scaffolding function, we repeated the TLS assay with Rev1 cd (Rev1-DE467,468AA), a catalyticinactive mutant of Rev1 ( Figure 2A , (9)). Remarkably, the efficiency of ICL bypass and the extension by Rev1 cd -Pol z was higher than by Pol z alone, and it is diminished only slightly from that observed with wild-type Rev1-Pol z ( Figure 2B ). The major difference is the stronger accumulation of stalled products directly prior to the ICL, suggesting the participation of the catalytic activity of Rev1 at the ICL. Interestingly, the major lesion-independent stall site at position 46 is still present albeit reduced by 30-55% ( (across a dC template) is associated with misincorporation of dCMP with a 15-20% efficiency (3 determinations) . Misincorporation is carried out by Rev1's dCMP transferase activity, since it is not present with Pol z alone, and eliminated with Rev1 cd -Pol z ( Figure 2C ). Therefore, the inhibitory mechanisms of Rev1 on non-damaged DNA involve both its catalytic and scaffolding functions. Within the sensitivity of detection (>95%), all three enzymes (Pol z, Rev1-Pol z, Rev1 cd -Pol z) faithfully introduced a dCMP across the ICL-G position ( Figure 2C ).

It is not entirely clear where and when during DNA replication and translesion synthesis, a TLS polymerase takes over DNA synthesis. Is it when replication stalls ~20 nt before the ICL (on the leading strand), or only when the ICL position is reached? A recent study in Xenopus egg extract suggests that the approach is carried out by a replicative DNA polymerase, up to one nucleotide before the ICL position, and bypass and extension is performed by a complex of Rev1 and Pol z (13). We determined the efficiency of ICL bypass as a function of the distance between the primer terminus and the ICL ( 

While it is well recognized that Pol z and Rev1 are indispensable for the TLS of ICLs in eukaryotic cells, Pol h can contribute to the efficiency of this process (38) . Switching between Pol h and other DNA polymerases has been observed with pyrimidine dimer lesions (39) (40) (41) .

Does switching also occur during the TLS of ICLs, and with what efficiency? This is of particular interest since Pol h stalls at positions past the ICL, which in theory should provide good substrates for extension by Pol z or Rev1-Pol z. Therefore, we investigated a potential collaboration between DNA polymerases in ICL bypass. In the assay we used all combinations of Pol d, Pol h, Pol z and Rev1 ( Figure 4A ). We were surprised by the observation that the combination of Pol h and Rev1-Pol z did not give the anticipated synergy in bypass. We had predicted that this combination would give excellent TLS because Pol h alone would produce mostly products that stalled past the ICL, while Pol z or Rev1-Pol z showed no stalling at these positions. Therefore, we expected that Pol z or Rev1-Pol z would readily extend the Pol h intermediates to full-length products ( Figure 1D ).

However, this was not observed. We carried out a two-step bypass assay, in which the ICL product was replicated by Pol h in the first stage, and then further extended with other DNA polymerases in the second stage ( Figure 4B ). While Pol d was inhibitory in the second stage, We next examined the possibility that the Pol h was bound up in a stalled, nonexchangeable (frozen) complex with the +1 to +3 bypass products, blocking access by Rev1-Pol z. This hypothesis needed testing, in spite of the observation that the 3'-exonuclease activity of Pol d did have access to these stalled products ( Figure 4B, lane 3) . After the first stage, the DNA was reisolated by ethanol precipitation and subjected to the second stage assay ( Figure   4C ). Analogous results to those in Figure 4B were obtained. While Rev1-Pol z promoted significant propagation of the Pol h bypass products, it was still limited, yielding very few fulllength products during the 30 minute incubation.

Translesion synthesis in the cell requires mono-ubiquitination of PCNA at Lys-164 (42) .

Mono-ubiquitinated PCNA shows increased binding to Rev1 and it stimulates Rev1 activity in certain sequence contexts (10, 12, 43) . However, mono-ubiquitinated PCNA does not stimulate Pol z activity (10) . The main proposed function of monubiquitination is to recruit Rev1, and thereby Pol z to damaged chromatin for TLS. In a control experiment, in which PCNA and ubiquitinated PCNA were compared as accessory factors for TLS of the ICL, we detected no significant difference in bypass efficiency (Supplementary Figure S5) .

We next investigated whether the bypass products made by Pol h were mere elongation products of the 40-mer primer, or whether other changes to the ICL substrate had occurred that were responsible to led to an inhibition of the formation of full-length products. If primer elongation, for instance to the +2 position, were the only result of replicating the ICL substrate with Pol h, then that +2 product should be equivalent in activity to a substrate made by hybridizing the +2 primer to the ICL template. In the experiment in Figure 5 , different length primers, with the same 5'-end but with their 3'-termini at the -6 position (the standard primer), at 0 (across the ICL), or at +2 or +6 position, were hybridized to the ICL template. Their reactivities were assessed with each of the four enzymes. For the purpose of this experiment, we defined replication product from ten nucleotides past the ICL to full-length as extension products. When the series of substrates was tested with Pol h, we saw some improvement in full-length product with the +2 primer compared to the standard -6 primer ( Figure 5A ), but stalling was still a major problem. Only the +6 primer escaped the problem of stalling. Pol d extended only the +6 primer efficiently, suggesting that the double-stranded DNA binding cleft of Pol d is very sensitive to the ICL-induced structural alterations from Watson-Crick base-pairs ( Figure 5B ). Nevertheless, even the +6 substrate was compromised for extension by Pol d. About 20 % was subject to degradation by Pol d's proofreading activity, as follows from the generation of products at positions around the nick. This was observed only with the ICL substrate and not with the control substrate (compare lanes 14 and 15) .

In sharp contrast, Pol z extended even the (0) substrate with very high efficiency, while the -6 substrate shows the usual stalling just prior to the ICL ( Figure 5C ). Rev1-Pol z also fully extended the primer opposite the ICL as well as the +2 and +6 primer ( Figure 5D ). The continued stalling further downstream of the ICL is a result of the negative regulatory activity of Rev1 on Pol z, and was also observed with the control template ( Figure 5D , e.g. compare lane 10 with 11).

One likely cause for the autoinhibitory activity of Pol h during ICL bypass could be that the 6-mer, which is crosslinked to the template is extended by Pol h. Consequently, after ICL bypass, further elongation would require strand displacement synthesis through the extended 6mer strand, which would be inhibitory for these TLS polymerases (30) . The assay described in Figure 6 lends support to this hypothesis. We incubated the unprimed ICL substrate, or the ss93-mer as control, with Pol h, followed by hybridization of the standard primer (-6), and then a second incubation with Pol h. The FAM-primer and its extension products were detected by FAM fluorescence (Figure 6A ), while all DNAs were detected with SYBR-gold staining ( Figure   6B ). The appearance of a Pol h-dependent slow-migrating species in (B) with the unprimed 93-ICL+6bp, but not with the un-crosslinked 93-mer template, is evidence that indeed, Pol h used the covalent 6-mer as a primer for extension, yielding a 93-ICL+37bp product upon replication to the end of the available template ( Figure 6B, lanes 7,8) . We propose that this extension product is inhibitory for full extension of the FAM-primer not only by Pol h, and also by Pol z or Rev1-Pol z.

For over a decade, a consensus model has emerged for TLS in eukaryotic cells (44) (45) (46) . This process is usually carried out by the tandem action of two DNA polymerases. The first DNA polymerase, the inserter, incorporates one or more nucleotides opposite the lesion, whereas the second DNA polymerase, the extender, continues replication. Depending on the type lesion, the inserter DNA polymerase is usually a Y-family DNA polymerase, such as Pol h, Pol k, or Pol i in mammalian cells, whereas Pol z is generally considered to be an extender DNA polymerase (1).

In yeast, Pol h is the only classical "inserter" Y-family enzyme. The second Y-family enzyme, Rev1 has a specialized obligatory function with Pol z, and it does so without generally requiring its unique deoxycytidyl transferase activity, including during TLS of cis-platin induced DNA adducts (47, 48) . One caveat to these genetic observations with Rev1 is that they do not distinguish between intra-and interstrand DNA lesions caused by cis-platin.

In our study of a highly replication-blocking ICL, the role of Pol z as an extender is readily apparent. Once the insertion of a nucleotide across the ICL has been made, extension by Pol z was facile, regardless of the presence of Rev1 ( Figure 5C, D) . But the question as to which DNA polymerase is involved in the insertion process remains. Based on Rev1 and Pol z depletion studies in Xenopus egg extracts, it has been suggested that Rev1-Pol z is not responsible for insertion across the ICL, but it is for the extension step (13, 23) . Therefore, it follows that another DNA polymerase should be responsible for the insertion step and Rev1-Pol z for extension.

The purpose of the current study was two-fold. First, to provide a biochemical framework for the activity of Rev1-Pol z at ICL lesions. While previous studies, including those from our own laboratories, have investigated the biochemical activities of Pol z and Rev1 at various lesions (9, 30) , these studies lacked the proper biochemical context, i. e. that of the 5-subunit Rev1-Pol z complex, together with its essential PCNA clamp, which may additionally by modified by ubiquitination at Lys-164. Secondly, we wanted to investigate which other DNA polymerase could participate in this process as a possible inserter enzyme, using both yeast Pol d and Pol h as candidates. One firm conclusion from our studies is that, under all conditions tested, Pol d was inhibitory to TLS, because of its tendency to degrade TLS insertion and extension intermediates with its proofreading exonuclease activity (Figure 4, 5B) . In fact, as far as six nucleotides past the ICL, the lesion still affects the decision-making process by Pol d, that of extension versus degradation ( Figure 5B, lanes 14-16) . We explored the possibility that ubiquitination of PCNA would reveal a unique step in TLS when multiple DNA polymerases were present. One of the proposed roles of ubiquitination is that it localizes Rev1, and by association Pol z to the lesion, perhaps favoring the competition between this TLS enzyme and the multitude of other PCNA client proteins for occupancy at the lesion. However, ubiquitinated PCNA did neither stimulate nor inhibit TLS by Rev1-Pol z in our studies, nor did it disfavor the inhibitory participation by Pol d (Supplementary Figure S5) . Studies of the role of PCNA ubiquitination in ICL-TLS in xenopus extracts were similarly uninformative (13) .

From an enzymatic perspective, Pol h would be an ideal enzyme to serve as an inserter for ICL. In previous studies, Pol h has been shown to replicate past several model ICL lesions (33, 49, 50) , and this is confirmed in the current study. Interestingly, in both the previous and our current studies, Pol h has the propensity to stall either after insertion at ICL, or just past the lesion (Fig. 1D) in ICL repair in yeast is not strong. A yeast rad30D mutant showed no sensitivity to a variety of ICLs (14, 53, 54) . Furthermore, loss of RAD30 showed no effect on the hypermutability to cisplatin (55) . By contrast, POLH deficient DT40 cells or XPV patients cells are mildly sensitive to cisplatin, consistent with a possible role of vertebrate Pol h in ICL repair, perhaps redundant with another polymerase (17, 56) .

Our studies with the five-subunit enzyme indicate that Rev1-Pol z may function as both an inserter and extender. Rev1 stimulated the synthesis of substantial percentage of extension products by Pol z, in a process which is only in part dependent on the catalytic activity of Rev 1 ( Figure 2 ). Our data on the dispensability of Rev1's catalytic activity are consistent with a genetic study, which shows that the catalytic inactive mutant of Rev1 does not sensitize yeast to treatment with cis-platin (48) . Of importance for successful TLS is that the primer terminus is in close proximity to the ICL upon the recruitment of Rev1-Pol z (Figure 3) . The function of Rev1 on non-damaged DNA is to inhibit DNA synthesis by Pol z, and this can be observed in all of our experiments. Therefore, engaging Rev1-Pol z too far from the ICL may result in premature stalling of the complex. Conversely, Rev1 also ensures that TLS is terminated soon after lesion bypass has been accomplished. While this type of regulation by Rev1 makes sense from a physiological point of view, our biochemical data cannot be easily be reconciled with a recent study, which suggests that Pol z can carry out extensive TLS in yeast, for stretches longer than 200 nucleotides, and introduce mutations during this process (57) . Currently, it is not clear whether these extended mutagenic stretches are the result of processive DNA synthesis by Pol z, perhaps after dissociation of Rev1, or the result of iterative binding of Rev1-Pol z on the gapped, damaged DNA, resulting in close, but individual mutational patches.

Cells with mutations in Pol z are among the most hypersensitive to ICL-forming agents, and here we show that Rev1-Pol z can function as both an inserter and extender polymerase in bypassing ICLs. Our results further suggest that it is likely that in a physiological context, the insertion step across ICL is aided by another DNA polymerase such as Pol h. The limited sensitivity of Pol h deficient cells suggests that this function may be redundant with an additional TLS polymerase. The definitive answer to this question will have to await better biochemical systems to study ICL repair in yeast or human cells or conversely, improved genetics in Xenopus laevis, allowing for the generation of multiple simultaneous mutations in ICL repair genes in egg extracts that provide the only trackable biochemical system to study ICL repair to date. In the meantime, our studies provide a framework for the understanding of the role of Rev1-Pol z in ICL repair.

Chemical reagents were purchased from Sigma-Aldrich (St.Louis, MO, USA). USER™ Mix, T4 polynucleotide kinase, and T4 DNA ligase were purchased from New England Biolabs (Ipswich, MA, USA).

The sequences of the oligonucleotides were listed in Table S1 . HPLC-purified fluorescent labeled primers, biotinylated extension oligonucleotides, and splint oligonucleotides were purchased from Integrated DNA Technologies (Coralville, IA, USA). Single-stranded oligonucleotides with ICL precursors, 39mer 8 atom 20 bp ICL that contains two franking uracil bases around the crosslinked base (39+20(6) DMEDA ICL), and 39mer 8 atom 6 bp ICL (39+6 DMEDA ICL) were synthesized and purified as previously reported (33) . The identity of the ICLs was confirmed by 

A 93 nt-long single-stranded DNA that contains an ICL precursor (T93-C2) was prepared by ligating T39-C2 to 5'ext and 3'ext. The reaction and the purification methods were identical to those of 8a-6bp-ICL.The use of T39-G instead of T39-C2 gave the control undamaged 93mer

substrate. An ICL was formed between T93-C2 and C20-C2 as previously (35) . In brief, the two complementary oligonucleotides were annealed together, and then the ICL precursors were unmasked by sodium periodate to expose the aldehydes. The subsequent crosslink reaction by N,N'-dimethylethylenediamine (DMEDA) in the presence of sodium cyanoborohydride at pH 5.4 efforted the formation of the 8 atom ICL between T93-C2 and C20-C2. The purification method was identical to that 8a-6bp-ICL. The purified product was analyzed by 10% denaturing polyacrylamide gel containing 7 M urea and 1X TBE. (Figure S1B) Proteins.

All proteins are the Saccharomyces cerevisiae species and were purified as previously described. Pol z, Rev1-Pol z, Rev1 and Rev1 mutants, Pol h, Pol d and Pol d-DV (D520V) were purified from yeast overexpression systems (5, 6, 36, 58) . RPA, PCNA and RFC were purified from E. coli overexpression systems (59) (60) (61) . 

The polymerase exchange assay was as the TLS assay, with modifications. The assay mixture contained 100µM AMP-CPP instead of ATP for PCNA loading by RFC (the a-bmethylene group allows proficient loading of PCNA but not replication by DNA polymerases when dNTPs are omitted) (62) . Subsequently, DNA polymerase(s) were added to the assay, and, after 30 sec, replication was initiated by addition of dNTPs. Reactions were terminated after 15 min and processed as described above.

Reactions were increased to 50 µl and allowed to proceed for 60min, stopped with final concentration of 50% formamide, 10 mM EDTA and 0.1% SDS, and loaded on a 12% polyacrylamide-7M urea preparative gel. Full-length products were extracted from the gel and purified by the ZR small RNA PAGE recovery kit (Zymo Research -R1070). Sequencing primer 5'-AAGCTGGAGCTCCACCGCGG-3' was annealed and sequencing was performed with the USB Sequenase Version 2.0 DNA sequencing kit (Thermo Fisher Scientific -70770). Bottom, sequencing analysis. The red arrow indicates the ICL-dG template position, which is replicated by insertion of dCMP, which is sequenced as a dG. The purple arrow, also shown in (A) at position 60, indicates a position where Rev1, but not Rev1 cd misincorporates dCMP across from template dC with a frequency of 15-20%. Reactions were carried out for 15 min with either the individual DNA polymerases, or with a combination, which was added as a pre-formed mixture, as shown. Note that all lanes containing Pol d show degradation products shorter than the 40 mer primer. These products were less than 10% of total. (B) Two-stage reaction with Pol h in the first 10 min incubation, followed by the indicated DNA polymerase in the second 10 min incubation. Lane 1, no second incubation; lane 2, incubation with Pol h continued. (C) Two stage ICL bypass assay. After the first incubation, the DNA was isolated (materials and methods), followed by second stage assay with the indicated DNA polymerase. Lane 1, product after first stage; lane 2, second stage incubation without added polymerase. 

Standard assays were increased to 20 µl with DNA concentrations increased to 20 nM, and the proteins accordingly. In the first incubation, assays were carried out with the indicated unprimed template. After 30 min with Pol h, reactions in lanes 3 and 8 were stopped for 1 min at 70 ˚C, and the (-6) 5'-FAM-40-mer primer was hybridized (a 10-fold excess of primer was used). Assays were reinitiated with RPA, PCNA, RFC and Pol h for 30 min, and analyzed on a 12% PAGE-urea gel, containing 25% formamide. The gel was first scanned in the Typhoon for FAM fluorescence (A), and subsequently stained with SYBR GOLD, after the gel was cut in two pieces. It was then scanned in the Typhoon, using the EtBr setting (B). In the inset at right, the bottom of the SYBR GOLD-stained gel was contrast enhanced to show weak FL synthesis in lane 8. 93-ICL+37 is the expected size if the covalently attached 6-mer performs as a primer for extension by Pol h. 93-ICL+20bp (lane 4) is used as a size marker ( Figure 1A ). *These replication products are a consequence of the high primer/template ratio of 10 in this experiment. They were not observed when primer and template were equimolar.

DNA polymerase zeta in DNA replication and repair

Eukaryotic polymerases iota and zeta act sequentially to bypass DNA lesions

The fidelity of DNA synthesis by yeast DNA polymerase zeta alone and with accessory proteins

Pol31 and Pol32 subunits of yeast DNA polymerase delta are also essential subunits of DNA polymerase zeta

A four-subunit DNA polymerase zeta complex containing Pol delta accessory subunits is essential for PCNA-mediated mutagenesis

Proliferating cell nuclear antigen promotes translesion synthesis by DNA polymerase zeta

Deoxycytidyl transferase activity of yeast rev1 protein

Mutants of yeast defective in mutation induced by ultraviolet light

Roles of yeast DNA polymerases delta and zeta and of Rev1 in the bypass of abasic sites

Ubiquitinated proliferating cell nuclear antigen activates translesion DNA polymerases eta and REV1

A novel function of DNA polymerase zeta regulated by PCNA

A ubiquitin-binding motif in the translesion DNA polymerase Rev1 mediates its essential functional interaction with ubiquitinated proliferating cell nuclear antigen in response to DNA damage

Regulation of the Rev1-pol zeta complex during bypass of a DNA interstrand cross-link

DNA interstrand crosslink repair during G1 involves nucleotide excision repair and DNA polymerase zeta

Translesion DNA synthesis polymerases in DNA interstrand crosslink repair

Regulation of Rev1 by the Fanconi anemia core complex

Multiple repair pathways mediate tolerance to chemotherapeutic cross-linking agents in vertebrate cells

REV3 and REV1 play major roles in recombination-independent repair of DNA interstrand cross-links mediated by monoubiquitinated proliferating cell nuclear antigen (PCNA)

Multiple roles of Rev3, the catalytic subunit of polzeta in maintaining genome stability in vertebrates

Advances in understanding the complex mechanisms of DNA interstrand cross-link repair

Involvement of translesion synthesis DNA polymerases in DNA interstrand crosslink repair

DNA interstrand crosslink repair and cancer

Mechanism of replication-coupled DNA interstrand crosslink repair

The DNA translocase FANCM/MHF promotes replication traverse of DNA interstrand crosslinks

XPF-ERCC1 acts in Unhooking DNA interstrand crosslinks in cooperation with FANCD2 and FANCP/SLX4

The Fanconi anemia pathway promotes replication-dependent DNA interstrand cross-link repair

Human SNM1A and XPF-ERCC1 collaborate to initiate DNA interstrand cross-link repair

Replication-Dependent Unhooking of DNA Interstrand Cross-Links by the NEIL3 Glycosylase

Alcohol-derived DNA crosslinks are repaired by two distinct mechanisms

Structure-dependent bypass of DNA interstrand crosslinks by translesion synthesis polymerases

Role for DNA polymerase kappa in the processing of N2-N2-guanine interstrand cross-links

Drosophila DNA polymerase theta utilizes both helicase-like and polymerase domains during microhomology-mediated end joining and interstrand crosslink repair

The structure and duplex context of DNA interstrand crosslinks affects the activity of DNA polymerase eta

Generation of DNA interstrand cross-links by post-synthetic reductive amination

Synthesis of structurally diverse major groove DNA interstrand crosslinks using three different aldehyde precursors

Yeast DNA polymerase zeta maintains consistent activity and mutagenicity across a wide range of physiological dNTP concentrations

Idling by DNA polymerase delta maintains a ligatable nick during lagging-strand DNA replication

DNA polymerase eta reduces the gamma-H2AX response to psoralen interstrand crosslinks in human cells

Enzymatic switching for efficient and accurate translesion DNA replication

DNA binding properties of human DNA polymerase eta: implications for fidelity and polymerase switching of translesion synthesis

Two-polymerase mechanisms dictate error-free and error-prone translesion DNA synthesis in mammals

RAD6-dependent DNA repair is linked to modification of PCNA by ubiquitin and SUMO

Ubiquitin-binding domains in Y-family polymerases regulate translesion synthesis

Multiple two-polymerase mechanisms in mammalian translesion DNA synthesis

Translesion DNA polymerases in eukaryotes: what makes them tick?

Translesion Synthesis: Insights into the Selection and Switching of DNA Polymerases

Vertebrate DNA damage tolerance requires the Cterminus but not BRCT or transferase domains of REV1

The DNA polymerase activity of Saccharomyces cerevisiae Rev1 is biologically significant

Replication bypass of N2-deoxyguanosine interstrand cross-links by human DNA polymerases eta and iota

Mutagenic Bypass of an Oxidized Abasic Lesion-Induced DNA Interstrand Cross-Link Analogue by Human Translesion Synthesis DNA Polymerases

Orchestrating the nucleases involved in DNA interstrand cross-link (ICL) repair

Mechanism and regulation of incisions during DNA interstrand cross-link repair

) S. cerevisiae has three pathways for DNA interstrand crosslink repair

Editor's Highlight: High-Throughput Functional Genomics Identifies Modulators of TCE Metabolite Genotoxicity and Candidate Susceptibility Genes

Hypermutation signature reveals a slippage and realignment model of translesion synthesis by Rev3 polymerase in cisplatin-treated yeast

A role for polymerase eta in the cellular tolerance to cisplatin-induced damage

DNA polymerase zetadependent lesion bypass in Saccharomyces cerevisiae is accompanied by error-prone copying of long stretches of adjacent DNA

RPA and PCNA suppress formation of large deletion errors by yeast DNA polymerase delta

Recombinant replication protein A: expression, complex formation, and functional characterization

Overproduction and affinity purification of Saccharomyces cerevisiae replication factor C

A Mutational Analysis of the Yeast Proliferating Cell Nuclear Antigen Indicates Distinct Roles in DNA Replication and DNA Repair

Resolving individual steps of Okazaki-fragment maturation at a millisecond timescale

We thank Arnold Groehler IV (IBS Center for Genomic Integrity) for analyzing ICL-containing oligonucleotides by LC-MS. This work was funded in part by grants from the National Institutes of Health (GM118129 to PMB, CA165911 to ODS) and by the Korean Institute for Basic Science (IBS-R022-A1 to ODS).

RBB, YKC, UR, ODS, and PMB planned this study. YKC and UR synthesized the substrates and RBB carried out the enzymatic studies. RBB, YKC, ODS, and PMB were involved in the interpretation of the results and the writing of the paper. All authors approved of the final version.

The authors declare no conflict of interest with this work.