Human Mitochondrial DNA Polymerase Metal Dependent UV Lesion Bypassing Ability

Human mitochondrial DNA contains more UV-induced lesions than the nuclear DNA due to lack of mechanism to remove bulky photoproducts. Human DNA polymerase gamma (Pol γ) is the sole DNA replicase in mitochondria, which contains a polymerase (pol) and an exonuclease (exo) active site. Previous studies showed that Pol γ only displays UV lesion bypassing when its exonuclease activity is obliterated. To investigate the reaction environment on Pol γ translesion activity, we tested Pol γ DNA activity in the presence of different metal ions. While Pol γ is unable to replicate through UV lesions on DNA templates in the presence of Mg2+, it exhibits robust translesion DNA synthesis (TLS) on cyclobutane pyrimidine dimer (CPD)-containing template when Mg2+ was mixed with or completely replaced by Mn2+. Under these conditions, the efficiency of Pol γ′s TLS opposite CPD is near to that on a non-damaged template and is 800-fold higher than that of exonuclease-deficient Pol γ. Interestingly, Pol γ exhibits higher exonuclease activity in the presence of Mn2+ than with Mg2+, suggesting Mn2+-stimulated Pol γ TLS is not via suppressing its exonuclease activity. We suggest that Mn2+ ion expands Pol γ′s pol active site relative to Mg2+ so that a UV lesion can be accommodated and blocks the communication between pol and exo active sites to execute translesion DNA synthesis.

Human mitochondria DNA is replicated by DNA polymerase gamma (Pol γ), which consists of a catalytic subunit Pol γA and a dimeric accessory subunit Pol γB. Pol γ possesses activities of 5′-3′ polymerization (pol) for DNA synthesis, 3′-5′ exonuclease (exo) for proofreading, and 5′deoxyribose phosphate lyase for DNA repair. Pol γB has no intrinsic enzymatic activity, but it accelerates polymerization rate, increases affinity to DNA, and enhances processivity of the holoenzyme (Johnson et al., 2000;Johnson and Johnson, 2001;Lee et al., 2010). Pol γA belongs to the A-family polymerases that are single polypeptide enzymes structurally resembling a right-hand shape and containing fingers, palm, and thumb domains (Steitz, 1999;Garcia-Diaz and Bebenek, 2007;Bębenek and Ziuzia-Graczyk, 2018). Other replicative A-family DNA polymerase members include bacterial DNA polymerase I and phage T7 DNA polymerase.
All known DNA polymerases require divalent metal ions to catalyze DNA synthesis. A general mechanism of two-metalion catalysis established a foundation for DNA and RNA polymerases' enzymatic reactions (Steitz and Steitz, 1993). An exonuclease-deficient Pol γ variant was shown to have limited translesion DNA synthesis (TLS) activity on UV lesion-containing DNA in the presence of Mg 2+ , suggesting that Pol γ possesses intrinsic TLS ability (Kasiviswanathan et al., 2012). Mitochondria are the main storage site of intracellular Mn 2+ ion (Maynard and Cotzias, 1955;Liccione and Maines, 1988;Gavin et al., 1990;Gavin et al., 1992), which raises an important question on impact of Mn 2+ or Mg 2+ /Mn 2+ mixture on Pol γ mtDNA replication as well as TLS.
We report here studies of the A-family member Pol γ lesion bypassing across UV lesions in the presence of different metal ions. We show that Pol γ is unable to replicate past the CPD-or (6-4)PP-containing DNA template in the presence of Mg 2+ , but displays efficient TLS ability comparable to that on the non-damaged template in the presence of Mn 2+ or Mg 2+ /Mn 2+ mixture. Mn 2+ promotes TLS activity without involvement of the polymerase's exonucleolytic activity. The Mn 2+mediated trans-UV lesion DNA synthesis appears uniquely to mitochondrial DNA polymerase and is absent in other A-family DNA polymerases tested. These results also give a novel insight into the metal-regulated error recognition and communication between the pol and exo active sites in Pol γ.

RESULTS
Mn 2+ Ion Stimulates Robust Translesion DNA Synthesis Across CPD but Not (6-4)PP DNA templates (49 nt) containing either a CPD (T-CPD) or (6-4)PP (T-(6-4)PP) ( Table 1) at the 28th position from the 3′end was synthesized as described in Methods. Pol γ translesion replication was first tested on a 27 nt primer (P N ) annealed to the T-CPD where the 3′-thymine (3′-T) in the CPD serves as the coding base ( Figure 1A, lanes 9-16, Table 1). Timedependent (0-20 min) primer extension assay was conducted in the presence of Mg 2+ . As a control, a parallel assay was carried out on a non-damaged template (T-ND) with identical sequence except that the CPD is replaced by 2 dT ( Figure 1A, lanes 1-8, Table 1). In anticipation of higher polymerase activity on T-ND, the Pol γ concentration was reduced (see Methods). For convenience, only the full-length product formation rate at normalized enzyme concentration was used for rate calculation. Full-length formation rate of Pol γ on T-ND was 4.5 nM/s ( Figure 1C). However, on T-CPD, the primer was not extended, but instead was degraded to P N-1 to P N-3 over time ( Figure 1A lanes 9-16).
To evaluate the effect of high Mn 2+ content in mitochondria on Pol γ activity, we substitute Mg 2+ with Mn 2+ and repeated the above experiments. In stark contrast to results with Mg 2+ , Pol γ exhibited strong TLS across CPD in the presence of Mn 2+ ( Figure 1B, lanes 9-16). The rates of full-length product formation by Pol γ on T-CPD were calculated to be 0.58 nM/ s, which is only 1.7-fold lower than that on T-ND at 1 nM/s ( Figure 1C). Interestingly, Pol γ is 4.5-fold less efficient on the non-damaged template in the presence of Mn 2+ than in the presence of Mg 2+ .
Previous studies showed that exonuclease-deficient (exo-) Pol γ displayed limited intrinsic CPD-bypassing ability (Kasiviswanathan et al., 2012). To assess whether Mn 2+ plays a redundant role with exonuclease activity in Pol γ TLS activity, Pol γ exo-(D198A/E200A) (Spelbrink et al., 2000) was analyzed on T-ND and T-CPD. In the presence of Mg 2+ , Pol γ exo-extended a small amount of the primer P N over the CPD site to full-length product, consistent with the previously reported data (Kasiviswanathan et al., 2012). However, a predominant amount of primer was extended by only a single nucleotide and formed product P N+1 (Figure 2A, lanes 1-8). Accumulation of the P N+1 indicates that silencing exonuclease enables the polymerase to synthesize across the 3′-T but not 5′-T of the CPD, suggesting that synthesis using the 5′-T as a coding base is the rate-limiting step of Pol γ TLS.
In the presence of Mn 2+ , Pol γ exo-exhibited significantly elevated TLS activity ( Figure 2B, lanes 9-16). The rate of Pol γ exo-full-length product formation on T-CPD was calculated to be 0.81 nM/s with Mn 2+ , which is more than 800-fold increase than that with Mg 2+ (0.001 nM/s) and exceeds that of wild-type (0.58 nM/s) ( Figure 2C), confirming that suppressing exo activity indeed increases TLS in the presence of either metal ion. P N+1 product was drastically diminished, indicating Mn 2+ facilitates Pol γ to overcome the energy barrier at the rate-limiting   step. Similar to wild-type enzyme, Mn 2+ exhibits inferior activity (eightfold lower) on Pol γ exo-synthesis on T-ND than Mg 2+ . When translesion synthesis was tested on (6-4)PP lesion containing template, wild-type Pol γ showed no activity in the presence of either Mg 2+ or Mn 2+ (Figure 3, lanes 4 and 5). However, Pol γ exo-extended the primer to P N+1 with Mg 2+ , similarly to that on the T-CPD (Figure 3, lanes 7 and 8). Similarly, no full-length product was formed in the presence of Mn 2+ , and only P N+1 and a small amount of P N+2 was formed ( Figure 3, lanes 9-10). Taken together, these results show that exonuclease activity contributes to the Pol γ′s lack of TLS activity, and Mn 2+ further promotes Pol γ′s TLS ability independent of exonuclease activity.

Mn 2+ Stimulation of Pol γ TLS Is Independent of Exo Activity
The results thus far showed Mn 2+ promotes TLS ability to both Pol γ wild-type and exo-variant, suggesting Mn 2+ functions independent of exonuclease activity. To test the conclusion, we tested the exo activity of Pol γ with Mn 2+ and compared it to that with Mg 2+ . We defined the rate of exonucleolysis as disappearance of the substrate and exonuclease events following the initial reaction were discounted, thus it is an underestimate of the actual rate. The rate of exonucleolysis of Pol γ on single-stranded DNA P N is 12.6 min −1 with Mn 2+ , sevenfold faster than that of Mg 2+ at 1.8 min −1 (Figures 4A,C). Pol γ exo activity was also assayed on duplex P N /T-ND ( Figure 4B). The excision rate of the duplex P N /T-ND in the presence of Mn 2+ is at 3.9 min −1 , which is about eightfold higher than Mg 2+ at 0.5 min −1 ( Figure 4C). Our results are consistent with a previous report where Mn 2+ stimulates porcine liver Pol γ excision of terminal mismatch (Longley and Mosbaugh, 1991). As Mn 2+ increased exo activity of Pol γ on P N /N-ND, which is free of error, it suggests that primer is sent more frequently from the pol to the exo site. As suppressing exo activity as well as addition of Mn 2+ that increases exo activity both promote Pol γ′s TLS, we predict that Mn 2+ must reduce error recognition ability of reading correct Watson-Crick base pairing geometry, which in turn promotes erroneous shuttling of the primer strand to the exo site.

Pol γ CPD Bypassing Activity in the Presence of Metal Ions Mixture
We show that Mg 2+ and Mn 2+ play opposite roles on Pol γ TLS. Since both ions are present in mitochondria, we tested the effects of metal ion mixture. Pol γ′s translesion synthesis assays were carried out at a fixed, near physiological concentration of one metal ion and varying the concentrations of the other. Specifically, Mg 2+ was kept constant at 1.0 mM, and Mn 2+ varied from 0.1 to 10 mM ( Figure 6A); Mn 2+ was kept constant at 0.4 mM, and Mg 2+ varied from 0.1 to 10 mM ( Figure 6C). Titrations of a single metal were also performed ( Figure 5). As each dNTP is capable of binding to one of either metal ion and with higher affinity to Mn 2+ , the precise concentrations of free Mg 2+ and Mn 2+ ions for catalysis cannot be accurately calculated, but, in the presence of a total 0.8 mM dNTPs, it should be significantly lower than the initial concentrations.
The TLS activity with mixed metal ions differs from that with a single metal ion alone. For example, assays with constant 1.0 mM Mg 2+ and increasing Mn 2+ , Pol γ began to show TLS activity at 0.1 mM Mn 2+ ( Figure 6A, lane 10-18, 6B). Nonetheless, different from Pol γ producing 97% FL product with 0.1 mM Mn 2+ on T-ND ( Figure 5B), it only produced 5% FL TLS product with 0.1 mM Mn 2+ / 1.0 mM Mg 2+ on T-CPD ( Figure 6B); the polymerase produced 95% FL product with 0.2 mM Mn 2+ alone on T-ND ( Figure 5B) but no activity with 1.0 mM Mg 2+ on T-CPD ( Figure 5C, lanes1-9), yet producing 12% FL TLS product with metal mixture ( Figure 6B). These results suggest that Pol γ′s metal binding sites are unlikely occupied by the same metal ion. We thus hypothesize that both metals are bound to the polymerase simultaneously. We should note that under the reaction condition Pol γ TLS activity was observed (0.2 mM Mn 2+ and 1.0 mM Mg 2+ , Figure 6A, lane 13), and assume dNTPs exhibit equal affinity to Mg 2+ and Mn 2+ ions and ignore metal binding to DNA, the free Mn 2+ ions available for Pol γ catalysis is~67 µM {0.2 − 0.8p[ 0.2 (1 + 0.2) ]} (mM), below the reported physiological concentration of Mn 2+ in mitochondria (Corkey et al., 1986;Romani, 2011).

Mn 2+ Does Not Stimulate Translesion Synthesis in Other Family A Polymerases
Previous studies show that Mn 2+ does not stimulate TLS activity in other A-family DNA polymerase members, such as E. coli Pol I and Klenow Fragment (Moore et al., 1981;Rabkin et al., 1983). To further evaluate the uniqueness of Mn 2+ activity on Pol γ, we analyzed Mn 2+ function on another A-family member, T7 DNA polymerase (T7 DNAP), a structural homolog of Pol γ. Neither Mg 2+ nor Mn 2+ enables wild-type T7 DNAP to bypass CPD ( Figures 7A,B, lanes 8-14), while both metals support activity on T-ND ( Figures 7A,B lanes 1-7). Similar to Pol γ, T7 DNAP exonuclease activity appeared to be higher in the presence of Mn 2+ than Mg 2+ . The results suggest Mn 2+ exclusively stimulates Pol γ′s TLS activity among replicative A-family polymerases.

DISCUSSION
Perhaps due to lack of UV lesion repair mechanism, mtDNA bares more pyrimidine dimers than chromosomal DNA (Birch-Machin et al., 2013). Nuclear DNA polymerases assume distinct functions in replication or lesion bypassing, but mitochondrial DNA polymerase, Pol γ, is the sole replicase responsible for synthesis on normal as well as lesion-containing mtDNA. Thus, the mitochondrial replicase will unavoidably encounter these bulky photoproducts during mtDNA replication. We present here studies of Pol γ replicating on CPD-and (6-4)PPcontaining DNA templates. We showed that Pol γ exerts

Pol γ May Bind to Both Mg 2+ and Mn 2+ in Mitochondria
Manganese is an essential element for biological functions and many ubiquitous enzymatic reactions in mitochondria. Mn 2+ is rapidly transported into mitochondria via the uniporter with activated Ca 2+ but slowly exported (Chance, 1965;Gavin et al., 1992), thus the organelle concentration is higher than that in cytosol. Mitochondrial concentration of Mn 2+ is reported at 0.1-0.35 mM with an upper limit of~1 mM in brain, heart, and liver (Konji et al., 1985;Gunter et al., 2004), and a concentration of Mg 2+ was 0.37-0.95 mM with upper limit of~1.5 mM in the heart (Corkey et al., 1986;Romani, 2011). We showed that, under the sub-stoichiometric concentrations of Mg 2+ /Mn 2+ , Pol γ displayed significant trans-CPD synthesis ability and presented features of binding to both metal ions. Based on these results, we hypothesize that Pol γ could bind to both metals in vivo and is therefore capable of replicating on non-damaged and CPD-containing templates. At very low Mg 2+ or Mn 2+ concentration, Pol γ is still able to replicate on both non-damaged and CPD-containing templates. Although the exact affinities of Mg 2+ and Mn 2+ to Pol γ are not known, it must be higher than the affinities toward dNTPs. Mg 2+ or Mn 2+ will first bind to metal A site in Pol γ, then Pol γ will selectively recruit incoming nucleotide pre-bound to metal ion, or metal B site. This feature of Pol γ partially compensates for lack of UV-lesion repair mechanism and TLS polymerase in mitochondria.
It is interesting to speculate Pol γ replication fidelity in the presence of Mg 2+ /Mn 2+ mixture. At a high concentration of Mn 2+ (above mM), many DNA polymerases become mutagenic and exhibit elevated replication errors with Mg 2+ (Sirover and Loeb, 1976). A simplistic hypothesis would be that Mn 2+ suppresses the proofreading exonuclease activity in DNA polymerases. However, others (Longley and Mosbaugh, 1991) and we show here that Pol γ displayed increased exo activity in the presence of Mn 2+ relative to that of Mg 2+ . Thus, high concentrations in Mn 2+ -induced replication errors are not by direct inhibition of exo activity, rather by hindrance of error recognition. Nevertheless, at low concentrations of Mn 2+ , DNA polymerases can replicate accurately. For example, at 2 µM free Mn 2+ concentration, E. coli Pol I was shown to synthesize DNA with similar error frequency as with Mg 2+ (Beckman et al., 1985). We predict Pol γ would also replicate with high fidelity at a physiological concentration of Mn 2+ and Mg 2+ . Needless to say, the hypothesis should be rigorously tested experimentally in future investigations. Lack of TLS activity on (6-4)PPcontaining template by Pol γ suggests two possible outcomes: there may be another TLS polymerase that can overcome this UV lesion, or the damaged mtDNA is eliminated by a mitochondrial degradation mechanism as observed in C. elegans and primary human fibroblasts upon UV radiation damage (Bess et al., 2012;Bess et al., 2013).

Selective Stimulation of CPD Bypassing Among Pol I Family Members
Pol γ is the only A-family polymerase found to have an intrinsic TLS activity across UV lesions as a wild-type enzyme. Studies from the Copeland and Meyer groups showed that silencing Pol γ′s exonuclease activity led to limited CPD bypassing activity, despite the diminished activity for wild-type enzymes (Miller and Grollman, 1997;Kasiviswanathan et al., 2012). We show that the TLS activities of both wild-type and exonuclease-deficient Pol γ are greatly stimulated by the presence of Mn 2+ . Additionally, plant and yeast mitochondrial DNA polymerases also exhibited TLS activity similar to human Pol γ (unpublished results). However, such activities are not found in other replicative A-family DNA polymerases. E. coli Pol I and Klenow Fragment lack intrinsic TLS activity and only insert one nucleotide against 3′-T in the CPD in presence of Mn 2+ (Moore et al., 1981;Rabkin et al., 1983;Smith et al., 1998). We showed here that wild-type T7 DNAP does not display Mn 2+ -stimulated TLS activity. The Mn 2+stimulated TLS activity appears to be reserved only to mitochondrial DNA polymerases, perhaps to take advantage of higher Mn 2+ concentration in the mitochondria.

Manganese Ion Likely Alters Pol γ9s Pol Active Site Structure
A distinct structural difference between high fidelity DNA polymerase such as Pol γ and bona fide TLS polymerases such as Pol η is the pol active site conformation and polymerase-enforced template DNA conformation. Crystal structures of Pol γ showed that Pol γ belongs to the A-family polymerases with characteristic fingers, thumb, and palm domains (Lee et al., 2009). Upon binding to a correct incoming nucleotide, the fingers domain undergoes large open-closed conformational changes to align the γ-phosphate of dNTP, 3′-OH of the primer, and catalytic metal ions for optimal catalysis for phosphodiester bond formation (Lee et al., 2009;Szymanski et al., 2015;Sohl et al., 2015). Significantly, the template bends 90°at the coding base (n) and downstream neighboring residue (n+1) (Figure 8, right). DNA template bending in Pol γ is accomplished by the helices O and O1 in the finger domain. Such template strand bending is thought to prevent template 'slippage' and ensuring replication fidelity. A non-bendable dimer is deterred from entering the pol site and stalls replication (Li et al., 2004). Indeed, the structure of T7 DNA polymerase, a Pol γ homolog, complexed with CPD lesion (Li et al., 2004) shows that after incorporating a nucleotide opposing the 3′-T of CPD, base pairing with the 5′-T of CPD is severely crooked, preventing replication at the n+1 position.
Contrary to the closed pol site in high-fidelity polymerases ternary conformation, the pol active site is more open in Y-family TLS polymerases Pol η. Structure of DNA polymerase η (Pol η) complexed with a CPD-containing template and primer duplex reveals that the bulky CPD can be easily accommodated in the pol site due to the preformed, spacious pol active site and highly mobile little-fingers and thumb domains, shared features in Y-family DNA polymerases (Figure 8, left) (Yang and Woodgate, 2007;Biertümpfel et al., 2010). Unlike replicative DNA polymerases, the fingers subdomain in Y-family DNA polymerases does not go through large conformational changes. Rather, the enzyme is already poised for DNA synthesis even in the absence of incoming nucleotides. Littlefinger and thumb domains go through the largest conformational changes and seem to confer specific TLS activities (Boudsocq et al., 2004;Yang and Woodgate, 2007).
To provide a structural interpretation for Pol γ metaldependent TLS activity, we composited a ternary complex of Pol γ, primer/CPD-template, and an incoming nucleotide by docking CPD-containing substrate from T7 DNA polymerase complex (PDB: 1SL2) onto Pol γ ternary complex (PDB: 4ZTZ) after superposition on the two pol active sites (Figure 8, middle). The resulting complex showed that the CPD would sterically clash with the O1 helix fingers domain severely if no conformational change occurs. For Pol γ to accommodate the bulky CPD, the fingers domain would have to adopt a more open configuration resembling the Pol γ-DNA binary complex. In addition, Mn 2+ may also enlarge the pol site in a way that no longer bends the template strand, allowing the bulky thymine dimers to enter effortlessly. We thus hypothesize that Mn 2+ may alter Pol γ structure from high-fidelity configuration to that of TLS polymerases. Under this idea, when Pol γ replicates on a non-damaged DNA, the catalytic site is not challenged for which a catalytic site reshaping is not needed ( Figure 9A). But on a CPD-containing substrate, due to its volume, Pol γ′s catalytic site is reshaped in the presence of Mn 2+ or a mixture with Mg 2+ ( Figure 9B). More studies should be addressed to better understand this mechanism.
Pol ι, another Y-family DNA polymerase, has better TLS activity across UV lesions in the presence of Mn 2+ compared to Mg 2+ (Frank and Woodgate, 2007). The mechanism behind Mn 2+ -mediated TLS in Pol ι is not by large conformational changes of protein. Crystal structures of Pol ι do not display noticeable overall structural changes in the presence of Mn 2+ compared to Mg 2+ except for small changes near divalent cations in the active site (Choi et al., 2016). Mn 2+ supports more optimal octahedral coordination compared to Mg 2+ , explaining the higher activity and fidelity in presence of Mn 2+ . In addition, Pol ι does not go through large conformational change upon dNTP binding like replicative DNA polymerases do, which gives molecular explanation behind its ability to tolerate bulkier lesions in its wide and rigid active site (Nair et al., 2006). For replicative polymerases like Pol γ, finger domain movement upon dNTP binding is essential. It is possible that Mn 2+ binding induces sufficient structural changes to accommodate CPD.
Mn 2+ -induced conformational changes are seen in other polymerases. The pol active site of Pol η widens when bound to Mn 2+ relative to being bound to Mg 2+ and Ca 2+ (Yang and    Ling et al., 2003;Rechkoblit et al., 2006;Weng et al., 2018). The proteolysis pattern for a catalytic subunit of herpes simplex virus DNA polymerase (UL30) bound to platinated DNA differs in the presence of Mn 2+ than in the presence of Mg 2+ , implying that UL30 undergoes conformational changes between Mn 2+ to Mg 2+ bound states (Villani et al., 2002). If Mn 2+ actually alters Pol γ pol site and the fingers domain resemble that in Pol η, Pol γ would be able to accommodate the rigid thymine dimer in the enlarged pol site and to incorporate incoming nucleotide opposite to the dimer without stringent geometry check for W-C base pairing complementarity. Crystal structures of RB69 DNA polymerase, replicative B-family DNA polymerase, in the presence of Mg 2+ or Mn 2+ do not display significant conformational changes (Xia et al., 2011). It is likely the same for Pol γ when it is replicating on T-ND. However, when Pol γ is challenged with bulky lesion like CPD, the Mn 2+mediated flexibility in the pol site may allow enlargement to accommodate CPD (Figure 8). Currently, structural studies revealed only two catalytic metal ions in the pol site; however, EPR study of E. coli Pol I showed at least 25 Mn 2+ binding sites were present (Slater et al., 1972). Whether the catalytic metal ions or additional metal ions are involved in reshaping the pol site warrants further investigation.
Wild-type T7 DNA polymerase bound to thioredoxin was purchased from New England Biolab (Ipswich, MA). Concentrations of the proteins were determined on the NanoDrop ™ spectrophotometer (Thermo Scientific, Waltham, MA) based on the absorbance at 280 nm. All experiments reported in this study were carried out with the same batch purified proteins. The protein concentrations are determined by A280 absorption and specific extinction coefficients, not active site concentration. The active site concentrations of purified Pol γ are routinely measured to be 80-84%.

DNA Substrates
All oligonucleotides are synthesized by Integrated DNA Technologies (Morrisville, NC) except for templates containing CPD and (6-4)PP ( Table 1). The oligonucleotides containing the CPD and the (6-4)PP were synthesized as described previously (Murata et al., 1990;Iwai et al., 1996). Complementary oligonucleotides were mixed in solution containing 50 mM Tris (pH 7.8), 50 mM NaCl, and 1 mM EDTA (pH 8), then were annealed by heating at 95°C for 5 min followed by slowly cooling to room temperature overnight. For all activity assays, 5′-end of the primer strands were labeled using T4 polynucleotide kinase (New England Biolab, Ipswich, MA) using γ-32 P-ATP (PerkinElmer, Waltham, MA).

Polymerase Activity Assays
Three primer/template duplexes were formed by primer P N (27 nt) annealed to a 49 nt non-damaged template (T-ND), a CPD-containing template (T-CPD), or a (6-4)PP-containing template (T-(6-4)PP) ( Table 1). For UV product-containing DNA, 200 nM wild-type or exonuclease-deficient Pol γ was pre-incubated with 1000 nM P N /T-CPD or P N /T-(6-4)PP in Buffer N (25 mM HEPES pH 7.5, 140 mM KCl, 1 mM EDTA, 5% glycerol, and 100 μg/ml BSA) at 37°C for 5 min. For T-ND, 100 nM wild-type or exonuclease-deficient Polγ was preincubated with 1000 nM P N /T-ND in Buffer N. The reaction was started by the addition of equal volume of Buffer C (Buffer N supplemented with 400 µM dNTP mix and 40 mM MgCl 2 or MnCl 2 ) to a final concentration of 200 µM dNTP mix and 20 mM MgCl 2 or MnCl 2 . Reaction was quenched at the indicated times by adding ninefold excess of Buffer Q (80% Formamide, 50 mM EDTA pH 8, 0.1% SDS, 5% glycerol, and 0.02% bromophenol blue). Quenched samples were heated at 95°C for 5 min and resolved on a 23% polyacrylamide gel containing 7 M urea. Gels were soaked in solution containing 50% methanol and 20% glycerol prior to drying overnight under vacuum at 50°C. Gels were exposed on a phosphor screen, which was imaged using Amersham Typhoon RGB scanner (Cytiva, Marlborough, MA). The intensities of bands were quantified using ImageQuant TL 8.2 (Cytiva, Marlborough, MA). Graphs were plotted using the mean and standard deviation in SigmaPlot 14 (Systat Software, San Jose, CA). To obtain the rate of fulllength product formation, we used the following equation: rate IFL I0 , where I FL is the intensity of the full-length product band at 120 s, and I 0 is the intensity of the original substrate band at 0 s. We normalized the enzyme concentration as polymerase activity assays on T-CPD contained 100 nM Pol γ whereas polymerase activity assays on T-ND contained 50 nM Pol γ.
For polymerase activity assays using wild-type T7 DNA polymerase, a total of 400 nM T7 DNA polymerase was preincubated with 200 nM of P N /T-ND or P N /T-CPD in Buffer N at 37°C for 5 min. The reaction was started by addition of equal volume of Buffer C to a final concentration of 200 µM dNTP mix and 20 mM MgCl 2 or MnCl 2 . Quench step, electrophoresis, and autoradiography were carried out identically to polymerase activity assays using P N .

Exonuclease Assays
A total of 400 nM Polγ was pre-incubated with 1000 nM single stranded primer P N or duplex P N /T-ND in Buffer N, and the reaction was started by addition of an equal volume of Buffer C without dNTPs to a final concentration of 20 mM MgCl 2 or MnCl 2 . Quench step, electrophoresis, autoradiography, and quantification were carried out identically to polymerase activity assays. The experimental data were fitted to an equation y Ae −kt , where A is the amplitude and k is the rate of excision. Graphs were plotted using the mean and standard deviation in SigmaPlot 14 (Systat Software, San Jose, CA).

Polymerase Assays With Mixed Metal Ions
For polymerase assay using mixed metal ions, 1000 nM wild-type Polγ was pre-incubated with 1000 nM P N /T-ND or P N /T-CPD in Buffer N at 37°C for 5 min. The reaction was started by addition of an equal volume of Buffer C to final concentrations of either constant 1 mM MgCl 2 and varied MnCl 2 (0.1, 0.2, 0.3 0.4, 0.5, 1, and 10 mM) concentrations, or constant 0.4 mM MnCl 2 and varied MgCl 2 (0.1, 0.2, 0.4, 0.4, 0.5, 1, and 10 mM) concentrations. Reaction was quenched 5 min after adding ninefold excess of Buffer Q. Electrophoresis, autoradiography, and quantification were carried out identically to polymerase activity assays.

Polymerase Assays With Single Metal Ions
For polymerase assay using single metal ions, 200 nM wild-type Polγ was pre-incubated with 200 nM P N /T-ND or P N /T-CPD in Buffer N at 37°C for 5 min. Reaction was started by addition of an equal volume of Buffer C containing MgCl 2 or MnCl 2 (0.025, 0.05, 0.1 0.2, 0.4, 0.8, 1.0, and 10 mM). Reaction was quenched 5 min after adding ninefold excess of Buffer Q. Electrophoresis, autoradiography, and quantification were carried out identically to polymerase activity assays.

Experimental Replicas
Experiments with corresponding graphs containing error bars were repeated three times. The rest of the experiments with corresponding graphs not containing error bars were repeated twice.

DATA AVAILABILITY STATEMENT
The data supporting the conclusions of this article will be made available by the corresponding authors, without undue reservation.

AUTHOR CONTRIBUTIONS
JP, NB-T, LB, and YWY designed experiments, JP, NB-T, SI, and GH executed the experiments, JP and NB-T wrote the first draft, and all authors contributed to the final manuscript.

FUNDING
The work is supported in parts by grants from NIH to YWY and SJT (R01AI134611 and R01NS095747), the James W.
McLaughlin Fellowship Fund to JP, and an endowment from the Sealy and Smith Foundation to the Sealy Center for Structural Biology and Molecular Biophysics at UTMB.