Structure of the error-prone DNA ligase of African swine fever virus identifies critical active site residues

African swine fever virus (ASFV) is contagious and can cause highly lethal disease in pigs. ASFV DNA ligase (AsfvLIG) is one of the most error-prone ligases identified to date; it catalyzes DNA joining reaction during DNA repair process of ASFV and plays important roles in mutagenesis of the viral genome. Here, we report four AsfvLIG:DNA complex structures and demonstrate that AsfvLIG has a unique N-terminal domain (NTD) that plays critical roles in substrate binding and catalytic complex assembly. In combination with mutagenesis, in vitro binding and catalytic assays, our study reveals that four unique active site residues (Asn153 and Leu211 of the AD domain; Leu402 and Gln403 of the OB domain) are crucial for the catalytic efficiency of AsfvLIG. These unique structural features can serve as potential targets for small molecule design, which could impair genome repair in ASFV and help combat this virus in the future.


Response to the reviewers: We sincerely thank both reviewers for reading our manuscript with great care, and we also thank the reviewers for their helpful comments, encouragement, and criticisms. Based on these comments and suggestions, we have carefully revised our manuscript with major changes highlighted in red in the revised manuscript. We believe that the quality of our manuscript has been significantly improved. The following are our point-to-point responses to the reviewers' comments.
Reviewers' comments: Reviewer #1 (Remarks to the Author): The structural and biochemical work on African Swine Fever Virus DNA ligase by Chen and coworkers shows the presence of a novel domain involved in DNA binding which replaces the DBD domain, present for example in the human DNA ligase. Biochemical results on mutants also presented confirm unsurprisingly its role. The second particularity of this enzyme is the tolerance towards base mismatches at the repair site which has also been addressed by biochemical data. The biochemical data confirm essentially the much more exhaustive study by Lamarche et al, 2005. Despite the presence of 4 potentially exciting different high-resolution X-ray structures representing different states of ASFV DNA ligase (ATP bound open form in complex with nicked DNA, sealed and unsealed structure with a CT mismatch, unsealed structure with a CG base pair), the results are presented extremely inefficiently and it is difficult to extract any new information from the work presented. Kinetic data and the effect of some mutants are supposed to support the story but finally do not contribute much to the understanding. Some information about the dynamics of the process of DNA ligation could theoretically be extracted from the structural information on the different states (nick recognition, nick site adenylation on the 5' phosphate and sealing of the nick site by the nucleophilic attack of the 3' hydroxyl group) but the paper does not treat this very interesting dynamic aspect. The central question about the structural base of the tolerance towards base mismatches at the nick site remains unanswered. The relevant structural superpositions and comparisons with ligases from other organisms are missing. The bibliography is incomplete, in particular the work on the related human DNA ligase is neither sufficiently exploited nor cited such as the work by, Pascal et al, 2004, Ochi et al, 2010and Cotner-Gohara et al., 2010, which also deal with dynamic aspects of ligase action. In particular, the illustrations of the paper are extremely inefficient in supporting the two lines of argumentation. In summary, the work of the authors made highly valuable structural information available, which has been exploited very poorly. In this form the work is not suitable for publication in Nature Communications.

Response: We sincerely thank the reviewer for all the helpful comments, suggestions, and criticisms as well. We also thank the reviewer for confirming the importance and novelty of our structures. Based on these comments, we have carefully revised our manuscript. In addition to the structural comparison with the homologous protein, we also included many new catalytic assay results in the revised manuscript. in the three catalytic form structures. In the non-catalytic form structure, the nick DNA was bound by the NTD domains of two AsfvLIG molecules. Due to the potential strand switching and/or base shifting, the DNA electron density is averaged and no nick can be identified in the structure.
The Rcryst of the data on the open conformation is unacceptably high and needs commenting. Response: Thanks for the comment. We examined our data collection and refinement statistic table (Table S5)

, and the high Rcrystal (24.7%) is for the catalytic form AsfvLIG:CT2 structure. Unlike the other three AsfvLIG structures, the AsfvLIG:CT2 structure has much higher solvent content (63%). In the crystal lattice, no symmetry-related molecule interacts with the NTD domain, though the residues interacting with DNA are clearly ordered (please see the left panel of the figure below), several loop regions that do not bind DNA are disordered in some content (supported by the weak 2Fo-Fc electron density shown below; the map was contoured at the 1.0 sigma level). Disordering of these NTD residues may contribute to the high Rcrystal value of the structure.
The AsfvLIG:CT2 crystals are very fragile; to obtain the diffraction data reported in the manuscript, hundreds of crystals were tested previously. Recently, we screened many more AsfvLIG:CT2 crystals, and most of them diffracted weakly (worse than 3.5Å). Though a couple crystals diffracted to 2.7-2. 8

Å, refinement resulted in a similar Rcrystal value as the current structure. We then checked the Protein Database Bank (PDB) and compared the current structure to many structures with similar resolution (2.7Å); the Rcrystal of our structure is comparable. Therefore, we believe that our structure is acceptable and it can provide highly useful information, especially the conformation of the sealed C:T pair, which has been shown in Figs. 5B and S7C.
The Materials and Methods section is much too detailed.

Response: Thanks for the helpful comment. Some of the very detailed information has been deleted from the Materials and Methods section.
Reviewer #2 (Remarks to the Author): This paper contains structural and biochemical analysis of the DNA ligase from a porcine virus. The structure has strong agricultural implications, as a target against this virus. The authors have done an extensive analysis. Although the work reads well overall, it is somewhat colloquial and has some grammar errors, and could be improved with a native English editor. Scientifically, the most interesting aspect of this work is this DNA ligase can work on certain mismatches, and this work provides the structural rationale for how the mismatches are accommodated. Although the authors compare their protein qualitatively with the human protein, a quantitative analysis may reveal additional mechanisms. The second most interesting aspect is that the porcine virus has replaced the main DNA binding domain with its own. It is intriguing to consider what advantage this would have for the virus. In general the work appears interesting and worthwhile.

Response: We sincerely thank the reviewer for all the useful comments and encouragement. As suggested by the reviewer, we have redone many of the in vitro ligation assays using a time-course and calculated the reaction rates for all the WT and mutant proteins. AsfvLIG has several unique residues at the nick site, including Asn153, Leu221, Leu402, and Gln403; compared to Asp570 and Arg871 (which form a salt bridge), and Phe536 and Phe872 of HsLIG1 and the corresponding residues in other homologous proteins, the nick side conformation of AsfvLIG is much more flexible. Besides mismatched substrates, our new ligation assays (please see the updated Figs. 5E-F, 6D-E, S8, and S9 in the revised manuscript) revealed that mutations of these nick site residues also significantly lowered the ligation rates of the Watson-Crick paired substrates. These observations suggested that, instead of discrimination, these flexible nick site residues of AsfvLIG are evolved to accommodate both mismatched and matched substrates, resulting in the apparent low fidelity of the protein.
Also, as suggested by the reviewer, our manuscript has been edited by a native English speaker from a language services company.
1. Figure 1A is difficult to read. I cannot see where CG and CT are on the left and can only see 7 of 10 lines on the right. On the right side, scaling to 60% will allow more separation. How many times was the time course done independently (e.g. different days)? (This should be given for all biochemistry figures). Was there any bias observed when fitting the curves for the ligation catalytic reaction?

Response: Thanks for the helpful comments and suggestions. The right panel of Fig. 1A has been scaled to 60% and it was renamed as Fig. 1B in the revised manuscript. All the experiments were repeated three times on different days. In addition to the mean values, the standard deviation (SD) values are also given for all the biochemistry figures.
2. What is the minimum font on the figures? Some seem smaller than 10 pt font. Response: The minimum font is 10 pt on some figures. To make them more readable, we have increased the font size in the updated figures.
3. What is the sequence conservation in the AD and OB domains between human and Asfv? Is there domain similarity to archaeal or human ligase (Mol Cell. 2006 Oct 20;24(2):279-91;Biochemistry. 2010 Jul 27;49(29):6165-76)? In Figure 1B, it would be helpful to include the correspondence of each domain to the residue number. Response: We sincerely thank the reviewer for all the suggestions and wonderful references, which have been cited in the manuscript. AsfvLIG and HsLIG1 have some sequence conservation in their AD and OB domains; however, as depicted in the updated supplementary figure S1, the sequence conservation is very low. The sequence identity and similarity are about 10% and 20%, respectively, between the AD domains of the two proteins, and they are even lower for the OB domains, at only about 5% and 10%, respectively. Instead of AsfvLIG, the domain architectures of human ligases (please see the figure below) and archaeal ligases are more like each other; besides the AD and OB domains, the DBD domains are also highly conserved in the human and archaeal ligases. The residue numbers of each individual domains have been included in Figure 1B, which was renamed as Figure 1A in the revised manuscript. Figure 1D should be shown with a simulated annealing omit map. Response: Done as suggested. Figure 2B, it would be good to have the ATP position somehow demarked. Response: Thanks for the suggestion. For the HsLIG1 structure, the AMP that pyrophosphate linked to the 5'-P of the downstream DNA has been shown as red spheres in Figure 2B. Though one ATP was captured in the non-catalytic form AsfvLIG structure, it was not captured in the catalytic form structures. Therefore, we did not demark the ATP position in the AsfvLIG structure in Figure 2B. Figure 2 a model or a crystal structure? My interpretation of the text is that the CT1 crystal domains were overlaid onto another crystal form. It would be better if refinement used the CT1 domains as a restraint and the figure showed the refined model corresponding to the crystal (not an overlay). Response: Thanks for the helpful comments. Figure 2A is the real structure of the AsfvLIG:CT1 complex. To make it clearer, we have re-written the legend of Figure  2A in the revised manuscript. Table 1, can you provide the overall Wilson B and the average B factor for protein, DNA, water, and ATP? Response: Done as suggested.

In
8. A depiction of the ligase domains' interactions (highlighting those conserved) with the DNA would be helpful. Programs such as DNAPRODB can create these automatically. How is the binding of each domain related to the major and minor grooves? It is interesting that the NTD of the AsfLigase is binding deep into the major groove, possibly similar to the DBD of human ligase. Is the number or relative proportion of bases that it's interacting with similar to the human protein? Do the interactions lead to any distortion in the DNA? Response: We sincerely thank the reviewer for the great suggestion of this program. Fig. S5C). As revealed by the non-catalytic form AsfvLIG structure (Fig. 1C), interaction with NTD alone will not cause any distortion in the DNA; however, when the DNA is bound by all the three domains, it was bent in the nick site (Fig. S5B). 9. In line 191, the authors write "Compared to the OB domain, the AD domain forms many more interactions with the AsfvLIG NTD domain." What is the relative surface area, as calculated by a program such as PISA? From the figures, the area looks similar. Response: As calculated by the PISA program, the surface areas of the NTD, AD, and OB domains are 7239 Å 2 , 9368 Å 2 , and 5636 Å 2 , respectively. The relative surface area between the NTD and OB domains is 157 Å 2 , whereas it is 331 Å 2 between the NTD and AD domains. Figure 4 shows that the NTD is essential for DNA binding. How does that compare to human ligase DBD? What is LD? It is not defined in text or figure legend.

Response: As reported by Pascal and coworkers (Nature, 2004, 432:473-478), the DBD is important for DNA binding and catalysis of HsLIG1; deletion of DBD lowered the substrate binding affinity by >75-fold and reduced the catalytic efficiency of HsLIG1 by >4x10 5 -fold. As depicted in Figs. 4E and 4F, though the DNA binding ability of NTD is weaker than WT AsfvLIG, deletion of the NTD (for AsfvLIG N) completely abolished the protein's DNA binding and ligation activities. These observations suggested that, compared to HsLIG DBD, AsfvLIG NTD makes a similar (or even greater) contribution to the ligases' DNA binding and ligation activity. LD stands for AsfvLIG mutant with the NTD amino acids 85-92 replaced by two Gly residues. The corresponding figure has been moved to the supplementary section and renamed as Fig. S6E; LD was defined in both the main text and the figure legend.
11. The most interesting biological aspect of this structure is the ability to ligate mismatches. A faded gray outline underlay with the WC base pair would be helpful, if possible. Is there any selection for, neutral or against non-Watson Crick base pairing (e.g. Hoogsteen). Are there any compensating distortions in the neighboring base pairs? Fig. S7. The DNAs were bent in the catalytic form AsfvLIG complexes (Fig. S5B), however, the different nick site base pairs (C:G vs C:T) did not cause any additional compensating distortions in the neighboring base pairs (Fig. S7E).

Response: As shown in Fig. S2 and summarized in Table S2, AsfvLIG can catalyze the ligation reaction of various substrates. Among the non-Watson-Crick paired substrates, AsfvLIG prefers the substrates with a pyrimidine nucleotide at the template strands, such as DNA-TC, DNA-CT, DNA-TG, and DNA-CA. It can catalyze the ligation reaction of DNA-GG, but it dislikes the other three Hoogsteen-paired substrates (DNA-GA, DNA-AG, and DNA-AA). The Watson-Crick C:G pair was not highlighted with a faded gray outline in the manuscript, but the detailed conformation of the C:G pair and its comparison with the C:T pair have been included in supplementary
12. The 402 and 403 mutants are interesting, as they seem to be stabilizing the mispair ligation. How do the mutant work on Watson Crick base pairs? Would you predict that they would have less effect on those than on mismatches?

Response: We sincerely thank the reviewer for the wonderful suggestion. Besides DNA-CT and DNA-TC, we also performed in vitro ligation assays for the four Watson-Crick paired substrates; the results have been included in the revised manuscript (please see Figs. 5E-F, 6D, S8, and S9). Unexpectedly, the mutants with single (L402R or Q403F), double (L402R/Q403F), triple (N153D/L402R/Q403F), or quadruple (N153D/L211F/L402R/Q403F) mutations all showed much weaker ligation activity towards all tested substrates (both matched and mismatched)
. These observations strongly suggested that, instead of discrimination, the nick site residues have evolved to form a flexible conformation, which can accommodate various substrates, resulting in the apparent low fidelity of the protein.
13. It would be helpful to have the double, triple and quadruple mutants defined in the figure legend (and not just in the main text). Response: Done as suggested.
14. A figure showing the catalytic mechanism for ligation would provide more clarity on the impact of the changes in the porcine virus active site.

Response: Thanks for the helpful suggestion. The catalytic mechanism of ATPdependent ligase has been included in supplementary figure S3B in the revised manuscript.
15. One interesting aspect of the porcine viral ligase is the replacement of the DBD. What advantage for the virus would the new DBD provide? The human protein interacts with PCNA. Does the virus encode PCNA? Does the virus contain a PIP to interact with the host PCNA?

Response: In combination with the unique nick site residues, the DBD of AsfvLIG provides the protein with good tolerance in various base pairs; AsfvLIG, AsfvAP nuclease, and AsfvPolX (which is a low fidelity DNA polymerase) form one complete repair system, which can efficiently repair the DNA damage caused by the oxidative environment. In addition to the genome stability maintenance, this low fidelity repair system also plays an important role in quick genotype formation, which may provide the virus with a better ability of surviving under different conditions. Though it has not been experimentally verified, the virus encodes one protein (E301R) that shares weak sequence similarity with PCNA. However, the PIPcontaining protein that can interact with E301R or host PCNA has not been reported.
16. Please have a native English speaker look through the text. These are the sentences that I noticed, but there may be more. a. "Very surprise, no ligation activity was observed for the triple mutant even at high concentration" b. …. using Gel filtration buffer c. For catalysis assays, A 10-μL reaction system d. Suggestion. Remove "very". It occurs multiple times in the text and its presence is more distracting than informative.

Reviewers' Comments:
Reviewer #1: Remarks to the Author: The answers of the authors address only partially the issues I raised. I think, I was not explicit enough in my first review: a) The mutational work done on Asfv ligase, in particular the choice of the mutations, is irrelevant for the explanation of the error tolerance of the enzyme. The mutagenesis does not tell anything about the importance of the residues for fidelity or tolerance to mismatches. The choice of the corresponding residues in the human structure does not really make sense. The human residues F872, D570, R871 and F635 form through a salt bridge and stacking interactions an interaction network as shown correctly in Fig 6C. The change of isolated residues is not expected to lead to any acquired function. In order to be conclusive, the mutational analysis needed to be done on the human enzyme showing that the disruption of the interactions lead to an increased tolerance for mismatches at the ligation site. The mutagenesis only shows that indeed these residues contacting the minor groove are important for the catalytic efficiency. b) As AMP is absent it cannot be excluded that the "catalytic structures" are rather artifacts of the crystal packing which mimic pretty well the catalytic state. The results presented in Fig. 6E show clearly that a 2:1 complex incompatible with the crystal structure is possible in solution, what means that the presented structure only represents one of several possible structures where the OB domain may not contact the DNA. The manuscript states that the mutants affect dDNA binding, but in reality the mutants do not affect the affinity for DNA, they have only appear to favor dimerization of asfv ligase when it is bound to the substrate. The contribution of the AD domain to affinity has not been studied, only the one of the N-terminal domain alone and the full length protein have been studies. The construct N-terminal domain + AD domain should also be analyzed. It is very well possible, that this construct has already the full affinity which is not increased by the presence of the OB domain. c) An open form is also observed but hardly analyzed. By consequence the discussion is centered too much onto the "closed" form. d) The positioning of the 5' phosphate and the 3'-OH in presence and in absence of a mismatch should be shown, compared and analyzed. For these 4 points, the revised manuscript does not show any improvements. The only sound evidence presented concerns the use by asfv ligase of a new type of N-terminal domain instead of the canonical N-terminal domains of DNA ligases and the contribution of a loop of the OB domain binding in the minor groove to catalytic efficiency which represents the exploitable part of the mutagenesis experiments. In the first review, I have been hoping for a global strengthening of the message of the article and I did not go much in the details of the manuscript what I now do below: The bibliography is still incomplete: 1. A separate reference should be given for the localization of the DNA replication, which is Rojo et al., 1999, but the contribution of the nucleus and the nature of the replication intermediates seems still to be up to debate. It should also be mentioned that the DNA ligase may also play a role in the ligation of Okazaki fragments occurring during replication. 2. The sentence "Owing to the weak diffraction, no …" should be more explicit. What is "weak diffraction?" 3. How have the rmsd values been calculated? Based only on the superposition of Calpha atoms or of all backbone atoms. How many residues have been excluded from the calculation of the rmsd? 4. "One ATP was captured…" A bound ATP molecule is present in the non-catalytical structure. 5. "nick site composition" For example: The ligase residues in proximity of the nick in the substrate DNA. 6. Correct repairs "Unique features of ASFV repair enzymes", 7. One of the proposed modifications is problematic. In the section about the binding to nicked versus dsDNA: "To clarify the function of AsfvLIG NTD, we carried out an in vitro DNA binding assay using nick and duplex DNA-CG. WT AsfvLIG can bind both nick and duplex DNA-CG (Figs. 4E and S6C); within the concentration range of 0.2-0.8 M, the nick DNA binding affinity of WT AsfvLIG is about 2-fold higher than that of duplex DNA. Compared to WT AsfvLIG, the DNA binding affinity of NTD is much weaker: at a 1.6 M concentration, NTD only binds about 20% nick DNA-CG and 9% DNA-CG duplex. Similar nick DNA preference was also observed for HsLIG1 and its DBD domain previously22." Looking at Fig. S6C the affinity difference for both substrate does not appear to be obvious and a statement that both substrates are bound with similar affinity would be sufficient. 8. "In fact, within the AD and OB domains, Leu402 and Gln403 are the only two residues projecting into the minor groove of the DNA (Fig. S4C)." This sentence is in contradiction with Fig. S4A, which shows also Y363 for a minor groove interaction. 9. "To further support this hypothesis, we carried out in vitro DNA binding assays (Figs. 6E-F)." These results are misinterpreted. The DNA binding affinity is not changed but some mutants favor dimerization of asfv ligase in a mode incompatible with the "catalytic" crystal structure. 10. The newly introduced diagram S4B needs explanation. What does H6, H9, S8 etc. mean? 11. It should be stated that the analysis using kobs is less complete than previous work by Lamarche who did a full enzymatic characterization leading to kcat's and Kd's. 12. The curves for the fraction of ligated substrate seem to level off and reach a plateau value different from 100 % ligation for weakly active mutants or substrates (Fig. 1B, Fig. 5EF, Fig. S9). Could you think of an explanation? 13. Fig. 1B, Fig. 6D, Table S2 the given kobs values are not plausible, from the curves you would expect maximal values in the order of 10e-1 min-1, which is a factor 1000 bigger than the given numerical values. Most likely, the factor is 10e-3 not 10e-6 min-1. 14. Fig. 5. The order of the panel is non-standard. 15. Fig. S1: Giving the sequence of AsfarLig N-terminal domain in a panel A would be useful. It is a pity that its sequence only starts with the AD domain. The domains corresponding to the 2 different parts of the alignment should be indicated by bigger labels on top of panel A and B (which rather become panels B and C). The residue number for HsLIG1_AD on the left side of the top block is wrong. 16. Fig. S3B The catalytic residues such as the lysine should also be marked on the sequence of Fig. S1. There is an error in the scheme of the reaction. At the level of step 3 the link HO-P is missing. It would be better to write DNA and NH2-LysLIG on one line for step 2. 17. Fig. S4A. The contacts of the AD domain should also be shown for the non-catalytic complex. S4B: Besides the non-explained symbols H9, S8 etc. suddenly some residues appear with a 3-letter code. 18. Fig. S6: Panel E should be split at is combines currently two different techniques, EMSA and ligation assay. 19. Fig. S7: The interacting water of Fig. 5A should be shown in panel E and F. A homogenous coloring scheme for the central base between panels A-C and panels D-F would have been nice. Also the open CG and CT complexes should have been superposed around the 5'-phosphate and the 3'-OH group. A very similar position of the two structures around the nick would explain why the ligation reaction is evenly efficient for base pairs or CG mismatches. 20. Table S1: I guess that the FAM label is linked to the 5'phosphate as in the commercial 6FAM modification and not to the base, as drawn. Lines indicating base pairs are slightly offset. Do not draw continuous but dotted lines at the positions of the mismatches what applies also to Table S2. 21. Table S2: Uncertainties shall only be given with two significant decimals, which also determine the last digit of the result which is presented.
Reviewer #2: Remarks to the Author: Re: Crystal structures and implications of the error-prone DNA ligase of African swine fever virus.
The authors have improved the writing and added in more kinetic and structural analyses. Unfortunately, it is still not possible to assess the quality of the data, given the current figures (see below). Additionally, the greatest insight for how the viral ligase can ligate with mismatches, in comparison to the human ligases, remains obscure as currently reported.
1) Based on the figures provided, it is impossible to assess the quality of the time courses. For a proper determination of the activity, the rate must be linear over time. Many proteins display a burst of activity followed by a slower rate, indicative of product release being rate limiting. To measure kcat, the initial linear rate should be measured. In the current figures, the data points are crowded at the bottom of the graph, making it impossible to view the linearity of the enzymatic activity or to distinguish between different mutants. It is not clear why the authors chose their Y axes as so high, given that the highest point in the data is ~25% of the Y axis maximum. In the one figure where the highest point goes to 50% of the y axis (Fig. S9), the authors have drawn their line over the entire course of the enzymatic reaction, drawing through both burst and the slower rate. Recommend that the authors match their y-axis to the maximal data and to only measure the initial linear region of the graph. At least three points should be used to draw the lines. Data points should be randomly on either side of the line. The data point positions should not be biased, with initial and final points below the line and middle points above the line, as observed in Fig. S9.
2) The major insight coming from this work is the allowance of the mismatch being ligated. The authors propose that the viral active site is more flexible. When one considers flexibility in a structure, it is standardly considered that there are protein regions that adopt multiple conformations. Is this what the authors intended? Based on Figure S7, it appears that the DNA is flexible, with a lack of base pairing. A CT base should have two H-bonds, but it appears to have only one, suggesting strain. The position of the phosphodiester backbone appears to be maintained. Does the phosphodiester backbone overlay, when overlaying the protein chain for the GC and the CT complexes? If that is the case, are there more phosphodiester stabilizing residues in the AsfvLig than in Lig1, 3, or 4? What is missing in the AsfvLig from Lig 1, 3, and/or 4 that ensures W-C fidelity? Could AsfvLig work on single stranded RNA or DNA?
3) The authors included a catalytic reaction schematic, as requested. However, it is a simple one that could be used in a biochemistry paper without the structure. It would be more helpful and inclusive of the results being reported if that schematic includes the active site residues and geometry, as this is a structure paper where the active site geometry is known. 4) The conservation between AD and OBD domains in porcine and human DNA ligases should be provided in text. It is interesting that despite the lower conservation, the structure is conserved. 5) The reference to the Ligase IV structure was not added. Structural comparison to Ligase III and Ligase IV were not done, which could provide information on the what makes the porcine virus enzyme unique. As above in #2, what is different from AsfvLig and the other Lig structures that allows for non-WC specificity? A part of this question is knowing what in the other lig structures enforces WC specificity. 6) Please include PDB ID for structures reported in this paper. It is helpful to the readers, if they wanted to look at them in a graphics program.

We sincerely thank both reviewers for reading our manuscript with great care, and for their helpful comments, encouragement, and criticisms. Based on these comments and suggestions, we have carefully revised our manuscript with major changes highlighted in red in the revised manuscript. The following are our point-to-point responses to the reviewers' comments.
Reviewers' comments: Reviewer #1 (Remarks to the Author): The answers of the authors address only partially the issues I raised. I think, I was not explicit enough in my first review: a) The mutational work done on Asfv ligase, in particular the choice of the mutations, is irrelevant for the explanation of the error tolerance of the enzyme. The mutagenesis does not tell anything about the importance of the residues for fidelity or tolerance to mismatches. The choice of the corresponding residues in the human structure does not really make sense. The human residues F872, D570, R871 and F635 form through a salt bridge and stacking interactions an interaction network as shown correctly in Fig 6C. The change of isolated residues is not expected to lead to any acquired function. In order to be conclusive, the mutational analysis needed to be done on the human enzyme showing that the disruption of the interactions lead to an increased tolerance for mismatches at the ligation site. The mutagenesis only shows that indeed these residues contacting the minor groove are important for the catalytic efficiency. In consistent with previous study, our in vitro catalytic assays showed that WT HsLIG1 (Fig. S11) has very high ligation activity and high fidelity toward DNA substrates. Interestingly, besides the non-Watson-Crick paired DNA substrates, the Watson-Crick paired DNA ligation activity of HsLIG1-d was also significantly lowered (Fig. S12). And no clear ligation activity was observed for any DNA substrate for the HsLIG1-q mutant (Fig. S13). These observations clearly indicated that, in addition to fidelity, the four ligase residues (D570, F635, R871, and F872) also play critical role in the catalytic efficiency of HsLIG1. The related results are summarized in the figure below and in Fig.7 of the revised manuscript. In combination with the mutagenesis and in vitro catalytic assays (Figs. 6D, S2, S9, and S10) of AsfvLIG, our results clearly indicated that the ligase residues in proximity of the nick in DNA substrate could not be switched between AsfvLIG and HsLIG1.
Like all the characterized ATP-dependent DNA ligases, AsfvLIG catalyzes phosphodiester bond formation between adjacent 3'-OH and 5'-phosphate in DNA duplex through a similar three-step mechanism (Please see the Figure below and Fig. 8A in the revised manuscript). Conceptually, differentiation between matched and mismatched base pairs could take place in step 2 and/or step 3 of the reaction. Indicated by the concomitant accumulation of the adenylated DNA intermediates for the mismatches only, HsLig3 and Vaccinia virus DNA ligase attain part fidelity during step 3. However, no adenylated intermediates for either mismatched or matched DNAs were detected during the in vitro DNA ligation assays catalyzed by AsfvLIG (Fig. S2) and HsLig1 (Fig. S11). These observations indicate that once adenylation of the nick occurs, the subsequent nick closure step will quickly finish. In other words, these observations also indicate that HsLig1 and AsfvLIG do not actively discriminate between match and mismatch during step 3; the discrimination or tolerance of mismatched DNAs solely occurs in step 2.
In addition to in vitro catalysis, we also compared the in vitro DNA binding behaviors of AsfvLIG and HsLig1 (

demonstrated that AsfvLIG has very low adenylation activity towards DNAs with 3'-dideoxy-or 3'-amino-terminated nicks, compared to regular nick DNAs; these observations indicated that 3'-OH of the nick is a critical component of the active site architecture during 5'-P adenylation. We believe that the unique NTD domain and the unique residues at the nick site all contribute to the low fidelity of AsfvLIG. The extensive interactions between DNA and the NTD domain offer
AsfvLIG with strong binding affinity to all matched and mismatched DNAs. In addition to matched DNAs, the strong binding also gives AsfvLIG enough opportunity to catalyze the ligation of certain mismatched DNAs. In incorporation with the NTD domain, the four nick site residues will help the reorientation of DNA 3'-OH and facilitate the ligation reaction. Compared to other nick site residues, Gln403 appears to be more important for the catalytic efficiency of AsfvLIG. As exampled by the AsfvLIG:CT2 structure (Fig. 5B)

, Gln403 can form H-bond interaction with the mismatched C:T pair at nick 3'-end; in addition to catalytic efficiency, such interaction definitely also contributes to the low fidelity of AsfvLIG.
Different from AsfvLIG, HsLig1 is a high fidelity DNA ligase. Though HsLig1 also discriminates between match and mismatch during step 2 only, our in vitro binding assays showed (Fig. 8D) that HsLig1 has much weaker binding affinities towards the matched nick DNAs, compared to the mismatched nick DNAs. These observations clearly indicated that the active site architecture will undergo certain conformational change during the adenylation process of DNA 5'-P. To ensure the high fidelity of HsLig1, the matched DNAs must be much more efficient during this reposition process, compared to the mismatched nick DNAs. Based on the only available HsLig1:DNA complex structure, it has been previously proposed that HsLig1 can impose some local distortion on the duplex, resulting in the 3'-OH and the adenylated 5'-P in positions appropriate for nick sealing. However, more structures, especially the structures of HsLig1 complexed with mismatched DNAs, are required to fully understand the basis for discrimination between matched and mismatched DNAs during the 3'-OH and 5'-P reposition process. b) As AMP is absent it cannot be excluded that the "catalytic structures" are rather artifacts of the crystal packing which mimic pretty well the catalytic state. The results presented in Fig. 6E show clearly that a 2:1 complex incompatible with the crystal structure is possible in solution, what means that the presented structure only represents one of several possible structures where the OB domain may not contact the DNA. The manuscript states that the mutants affect dsDNA binding, but in reality the mutants do not affect the affinity for DNA, they have only appear to favor dimerization of asfv ligase when it is bound to the substrate. The contribution of the AD domain to affinity has not been studied, only the one of the N-terminal domain alone and the full length protein have been studies. The construct N-terminal domain + AD domain should also be analyzed. It is very well possible, that this construct has already the full affinity which is not increased by the presence of the OB domain.
Response: We totally agree with the reviewer that no AMP was captured in our "catalytic structure". The three catalytic structures belong to two different space groups: P2 1 2 1 2 1 for AsfvLIG:CT2 and P2 1 for both AsfvLIG:CG and AsfvLIG:CT1. HsLig1:DNA complex (Figs. 2C and S4B), we also compared our AsfvLIG:CT1 structure with the catalytic form HsLIG3:DNA complex (Fig. S9A), showing that these two structures have similar overall folds. Based on these observations, we believe that our structures represent the real catalytic form structure and they can provide useful information, such as DNA substrate binding, catalytic complex assembly, and interaction between the NTD domain and the other two domains of AsfvLIG.

The molecular packing of AsfvLIG:CT2 is very different from the other two structures in the crystal lattice. However, as indicated by the low root mean square deviations (rmsd, about 0.75 Å, based on the superposition of 404 pairs of Cα atoms) among them, the three complex structures are very similar to each other. Besides
We also thank the reviewer for the helpful comments regarding the binding affinity and binging state between AsfvLIG and DNA substrates. We have carefully rewritten the statements in the revised manuscript. Based on the reviewer's comments, we constructed and purified one AsfvLIG protein with the OB domain deleted (referred as ΔOB). ΔOB can bind nick DNA and duplex DNA. As depicted in the figure below, the DNA-binding affinity of ΔOB is stronger than that of NTD domain of AsfvLIG. However, compared to the full-length WT AsfvLIG, the DNA-binding affinity of ΔOB is much weaker. As mentioned by the reviewer, the L402R or Q403F mutation within the OB domain will favor the dimerization of AsfvLIG when it is bound to the substrate. In addition to the 1:1 complex, we also noticed a significant amount of bands corresponding to the complex with ΔOB:DNA molar ration of 2:1 during the in vitro DNA binding assay. To test whether these complexes are compatible with the catalysis, we also carried out in vitro ligation assay. As depicted in the figure above, c) An open form is also observed but hardly analyzed. By consequence the discussion is centered too much onto the "closed" form.
Response: We thank the reviewer for the helpful comments. The "open" form structure provides detailed information on ATP binding and it captured the complex with AsfvLIG:DNA molar ratio of 2:1, which is incompatible with the catalysis. In combination with the "closed" form structures, the "open" form structure also revealed some domain arrangement during the substrate binding. We have reflected all these key points in the revised manuscript. In terms of assembly of the catalytic complex and the detailed interaction between AsfvLIG and DNA, the "close" form structures provide much more valuable information. Compared to the "open" form structure (2.55 Å), the resolution (2.35 Å) of the "closed" form AsfvLIG:CT1 structure is also slightly higher, therefore, we mainly focus on the "closed" form structure during structural analysis.
d) The positioning of the 5' phosphate and the 3'-OH in presence and in absence of a mismatch should be shown, compared and analyzed.

Response: The positioning of the 5' phosphate and 3'-OH is similar in presence and in absence of mismatch. We have included one new panel (Fig. 5C) and discussed about the positioning of the 5' phosphate and 3'-OH in the revised manuscript.
For these 4 points, the revised manuscript does not show any improvements. The only sound evidence presented concerns the use by asfv ligase of a new type of N-terminal domain instead of the canonical N-terminal domains of DNA ligases and the contribution of a loop of the OB domain binding in the minor groove to catalytic efficiency which represents the exploitable part of the mutagenesis experiments. In the first review, I have been hoping for a global strengthening of the message of the article and I did not go much in the details of the manuscript what I now do below: 1. The bibliography is still incomplete. A separate reference should be given for the localization of the DNA replication, which is Rojo et al., 1999, but the contribution of the nucleus and the nature of the replication intermediates seems still to be up to debate. It should also be mentioned that the DNA ligase may also play a role in the ligation of Okazaki fragments occurring during replication.
Response: We sincerely thank the reviewer for the helpful comments and wonderful reference, which (reference #8) has been included in the revised manuscript. We also mentioned the involvement of DNA ligase in the ligation of Okazaki fragments in the introduction section.
2. The sentence "Owing to the weak diffraction, no …" should be more explicit. What is "weak diffraction?" Response: Thanks for the helpful comment. In addition to AsfvLIG-DNA complexes, we also screened the sample of apo-AsfvLIG in this study. Though we got some apo-AsfvLIG crystals, they did not diffract well; the typical resolutions are around 8-10 Å. The term "weak diffraction" was used to indicate the low resolution of the diffraction data in previous manuscript. To avoid the possible confusion, we have deleted this term in the revised manuscript.

How have the rmsd values been calculated? Based only on the superposition of Calpha atoms or of all backbone atoms. How many residues have been excluded from the calculation of the rmsd?
Response: The rmsd values are calculated based on the superposition of C-alpha atoms. In the revised manuscript, we compared AsfvLIG structure with all the three HsLIG structures, including HsLIG1, HsLIG3, and HsLIG4. In each case, there are different numbers of residues excluded from the AsfvLIG and HsLIG structure during rmsd calculation. Instead of the number of the excluded residues, we prefer to give the number of residues used for structural superposition, to keep the manuscript more concise. The detailed number has been given in the revised manuscript.
4. "One ATP was captured…" A bound ATP molecule is present in the non-catalytical structure.
Response: Thanks for the wonderful suggestion. We have replaced the sentence with "A bound ATP molecule is present in the non-catalytical structure" in the revised manuscript.
5. "nick site composition" For example: The ligase residues in proximity of the nick in the substrate DNA.
Response: Thanks for the wonderful suggestion. We have replaced the term with "The ligase residues in proximity of the nick in the substrate DNA" in the revised manuscript.
Response: Thanks for the correction. We have fixed the mistake in the revised manuscript.
7. One of the proposed modifications is problematic. In the section about the binding to nicked versus dsDNA: "To clarify the function of AsfvLIG NTD, we carried out an in vitro DNA binding assay using nick and duplex DNA-CG. WT AsfvLIG can bind both nick and duplex DNA-CG (Figs. 4E and S6C); within the concentration range of 0.2-0.8 μM, the nick DNA binding affinity of WT AsfvLIG is about 2-fold higher than that of duplex DNA. Compared to WT AsfvLIG, the DNA binding affinity of NTD is much weaker: at a 1.6 μM concentration, NTD only binds about 20% nick DNA-CG and 9% DNA-CG duplex. Similar nick DNA preference was also observed for HsLIG1 and its DBD domain previously22." Looking at Fig. S6C the affinity difference for both substrate does not appear to be obvious and a statement that both substrates are bound with similar affinity would be sufficient.

Response:
We have repeated the DNA binding assay for several more times. In consistent with our previous assays, the new results also suggest that WT AsfvLIG and NTD have some weak preference for the nick DNA. We thank the reviewer for the helpful comments and nice suggestion. In the revised manuscript, we have toned down the statements. Fig. S6C has been renumbered as Fig. S5C in the revised manuscript.
8. "In fact, within the AD and OB domains, Leu402 and Gln403 are the only two residues projecting into the minor groove of the DNA (Fig. S4C)." This sentence is in contradiction with Fig. S4A, which shows also Y363 form a minor groove interaction.
Response: We sincerely thank the reviewer for pointing our mistakes. Fig. S4A has been renumbered as Fig. S3A in the revised manuscript. By carefully examining the structures and Fig. S3A, we found that Leu211, Ala215, and Asn219 of AD domain and Tyr363, Leu402, and Gln403 of OB domain all interact with the DNA in the minor groove. Among them, Leu402 and 403 are the only two reside next to the nick. We have modified the sentence in the revised manuscript.
9. "To further support this hypothesis, we carried out in vitro DNA binding assays (Fig. 6E)." These results are misinterpreted. The DNA binding affinity is not changed but some mutants favor dimerization of asfv ligase in a mode incompatible with the "catalytic" crystal structure.
Response: Thanks for the correction. We have fixed the mistake in the revised manuscript.
11. It should be stated that the analysis using kobs is less complete than previous work by Lamarche who did a full enzymatic characterization leading to kcat's and Kd's.

Response: Thanks for the helpful comment. We have reflected this point and cited the reference in the first paragraph of the results section of the revised manuscript.
12. The curves for the fraction of ligated substrate seem to level off and reach a plateau value different from 100 % ligation for weakly active mutants or substrates (Fig. 1B, Fig.  5EF, Fig. S9). Could you think of an explanation?
Response: We are not sure about the exact basis underlying these phenomena, but we think that slow product releasing could be one potential cause of these phenomena.
13. Fig. 1B, Fig. 6D, Table S2 the given kobs values are not plausible, from the curves you would expect maximal values in the order of 10e -1 min -1 , which is a factor 1000 bigger than the given numerical values. Most likely, the factor is 10e -3 not 10e -6 min -1 .
Response: We thank the reviewer for the helpful comments. We have carefully reanalyzed our data, which supported our previous results. Of course, we agree with the reviewer, some K obs values can be written using the factor of 10e -3 min -1 , especially for the double, triple, and quadruple mutants of AsfvLIG. However, to keep the factor consistent, we prefer to use the factor of 10e -6 min -1 for all WT and mutants of AsfvLIG.
14. Fig. 5. The order of the panel is non-standard. Fig. 5 have been rearranged according to the standard mode.

Response: The panels in
15. Fig. S1: Giving the sequence of AsfvLig N-terminal domain in a panel A would be useful. It is a pity that its sequence only starts with the AD domain. The domains corresponding to the 2 different parts of the alignment should be indicated by bigger labels on top of panel A and B (which rather become panels B and C). The residue number for HsLIG1_AD on the left side of the top block is wrong.
Response: We thank the reviewer for the very helpful comments. As suggested by the reviewer, a new panel (Fig. S1A) has been included in the figure, showing the sequence of AsfvLIG NTD. The NTD, AD domain, and OB domain of the proteins are labelled on the right side of the panels A, B, and C, respectively. We also thank the reviewer for pointing out our mistake on the residue number of HsLIG1_AD, which has been fixed.
16. Fig. S3B: The catalytic residues such as the lysine should also be marked on the sequence of Fig. S1. There is an error in the scheme of the reaction. At the level of step 3 the link HO-P is missing. It would be better to write DNA and NH2-LysLIG on one line for step 2.
Response: We thank the reviewer for the very helpful comments. Fig. S3B has been moved to Fig. 8A in the revised manuscript. In addition to the catalytic lysine, we also marked the other catalytic residues, such as Glu203 and Glu291, with red asterisks in the figure. The ligase residues in proximity of the nick in the substrate DNA are marked with green asterisks.
Thanks for pointing out our mistake in the scheme of the reaction, which has been fixed. As suggested by the reviewer, DNA and NH 2 -LysLIG are written on one line for step 2.
17. Fig. S4A. The contacts of the AD domain should also be shown for the non-catalytic complex. S4B: Besides the non-explained symbols H9, S8 etc. suddenly some residues appear with a 3-letter code.
Response: We thank the reviewer for the very helpful comments. Fig. S4A has been renumbered as Fig. S3A in the revised manuscript. Unlike the catalytic form complex, in which DNA interacts with all the NTD, AD, and OB domains of AsfvLIG, the AD and OB domains do not interact with DNA in the non-catalytic form complex. Therefore, we did not include the non-catalytic complex in Fig. S3A.
H6, H9, S6, and S8 represent the helices 6 and 9, and strands 6 and 8 of AsfvLIG. The labels and legend of Fig. S4B have been updated in the revised manuscript. The 3-letter codes are also replaced by 1-letter codes in the updated Fig. S3B.
18. Fig. S6: Panel E should be split at is combines currently two different techniques, EMSA and ligation assay. 20. Table S1: I guess that the FAM label is linked to the 5'phosphate as in the commercial 6FAM modification and not to the base, as drawn. Lines indicating base pairs are slightly offset. Do not draw continuous but dotted lines at the positions of the mismatches what applies also to Table S4.
Response: We thank the reviewer for the helpful comments and suggestions. We have fixed the mistakes in both Table S1 and S5, corresponding to the original  Table S4. Table S2: Uncertainties shall only be given with two significant decimals, which also determine the last digit of the result which is presented. Table S2 and the text throughout the manuscript.

Reviewer #2 (Remarks to the Author):
Re: Crystal structures and implications of the error-prone DNA ligase of African swine fever virus.
The authors have improved the writing and added in more kinetic and structural analyses. Unfortunately, it is still not possible to assess the quality of the data, given the current figures (see below). Additionally, the greatest insight for how the viral ligase can ligate with mismatches, in comparison to the human ligases, remains obscure as currently reported.
Response: We sincerely thank the reviewer for the helpful comments. As suggested by both reviewers, we have compared AsfvLIG and all the three HsLIG structures in the revised manuscript (Figs. S1, S4, and S9). Though the AD and OB domains are conserved in all the four proteins, AsfvLIG possesses four very unique ligase residues (Asn153, Leu211, L402, and Q403) in the proximity of the nick in the substrate DNA. Previous, we showed that replacing these residues by the corresponding residues of HsLIGs significantly lowered AsfvLIG's DNA ligation activity. As suggested by reviewer 1, we have constructed and purified wild type and two mutants of HsLIG1, in which the nick site residues are replaced by the corresponding residues of AsfvLIG. Similar to the AsfvLIG mutation, our in vitro ligation assays showed that mutation of the nick residues of HsLIG1 also significantly impaired the protein's DNA ligating activity (Figs. 7 and S11-S13). These results clearly indicated that the nick site residues could not be switched between AsfvLIG and HsLIG1.
Like all the characterized ATP-dependent DNA ligases, AsfvLIG catalyzes phosphodiester bond formation between adjacent 3'-OH and 5'-phosphate in DNA duplex through a similar three-step mechanism. Conceptually, differentiation between matched and mismatched base pairs could take place in step 2 and/or step 3 of the reaction. Indicated by the concomitant accumulation of the adenylated DNA intermediates for the mismatches only, HsLig3 and Vaccinia virus DNA ligase attain part fidelity during step 3. However, no adenylated intermediates for either mismatched or matched DNAs were detected during the in vitro DNA ligation assays catalyzed by AsfvLIG (Fig. S2) and HsLig1 (Fig. S11). These observations indicate that once adenylation of the nick occurs, the subsequent nick closure step will quickly finish. In other words, these observations also indicate that HsLig1 and AsfvLIG do not actively discriminate between match and mismatch during step 3; the discrimination or tolerance of mismatched DNAs solely occurs in step 2.
In addition to in vitro catalysis, we also compared the in vitro DNA binding behaviors of AsfvLIG and HsLig1 (Figs. 8C-D). Similar to the four Watson-Crick paired DNAs, DNAs with mismatched base pairs at the 3'-end of the nick can all be efficiently bound by AsfvLIG; the binding affinities between AsfvLIG and all DNAs are very similar. Compared to AsfvLIG, HsLig1 has similar binding affinity to mismatched DNAs with pyrimidines (C or T) on the template strand; however, when the protein concentrations are within the range of 0.2-0.8 μM, the binding affinities between HsLig1 and mismatched DNAs with purines (A or G) on the template strands are significantly weaker than those of AsfvLIG. Very surprising, compared to the mismatched DNAs, the binding affinities between HsLig1 and the four matched DNAs are much weaker.
In consistent with previous study (Showalter et al., 2006, Chemical Review, 106: 340-360), the obvious different DNA binding behavior, unique fold of the NTD domain, and unique residues at the nick site all confirm that AsfvLIG is one atypical DNA ligase. Before any AsfvLIG structure was reported, Showalter and coworkers already demonstrated that AsfvLIG has very low adenylation activity towards DNAs with 3'-dideoxy-or 3'-amino-terminated nicks, compared to regular nick DNAs; these observations indicated that 3'-OH of the nick is a critical component of the active site architecture during 5'-P adenylation. We believe that the unique NTD domain and the unique residues at the nick site all contribute to the low fidelity of AsfvLIG. The extensive interactions between DNA and the NTD domain offer AsfvLIG with strong binding affinity to all matched and mismatched DNAs. In addition to matched DNAs, the strong binding also gives AsfvLIG enough opportunity to catalyze the ligation of certain mismatched DNAs. In incorporation with the NTD domain, the four nick site residues will help the reorientation of DNA 3'-OH and facilitate the ligation reaction. Compared to other nick site residues, Gln403 appears to be more important for the catalytic efficiency of AsfvLIG. As exampled by the AsfvLIG:CT2 structure (Fig. 5B), Gln403 can form H-bond interaction with the mismatched C:T pair at nick 3'-end; in addition to catalytic efficiency, such interaction definitely also contributes to the low fidelity of AsfvLIG.
Different from AsfvLIG, HsLig1 is a typical, high fidelity DNA ligase. Though HsLig1 also discriminates between match and mismatch during step 2 only, our in vitro binding assays showed (Fig. 8D) that HsLig1 has much weaker binding affinities towards the matched nick DNAs, compared to the mismatched nick DNAs. These observations clearly indicated that the active site architecture will undergo certain conformational change during the adenylation process of DNA 5'-P. To ensure the high fidelity of HsLig1, the matched DNAs must be much more efficient during this reposition process, compared to the mismatched nick DNAs. Based on the only available HsLig1:DNA complex structure, it has been previously proposed that HsLig1 can impose some local distortion on the duplex, resulting in the 3'-OH and the adenylated 5'-P in positions appropriate for nick sealing. However, more structures, especially the structures of HsLig1 complexed with mismatched DNAs, are required to fully understand the basis for discrimination between matched and mismatched DNAs during the 3'-OH and 5'-P reposition process.
1) Based on the figures provided, it is impossible to assess the quality of the time courses. For a proper determination of the activity, the rate must be linear over time. Many proteins display a burst of activity followed by a slower rate, indicative of product release being rate limiting. To measure kcat, the initial linear rate should be measured. In the current figures, the data points are crowded at the bottom of the graph, making it impossible to view the linearity of the enzymatic activity or to distinguish between different mutants. It is not clear why the authors chose their Y axes as so high, given that the highest point in the data is ~25% of the Y axis maximum. In the one figure where the highest point goes to 50% of the y axis (Fig. S9), the authors have drawn their line over the entire course of the enzymatic reaction, drawing through both burst and the slower rate. Recommend that the authors match their y-axis to the maximal data and to only measure the initial linear region of the graph. At least three points should be used to draw the lines. Data points should be randomly on either side of the line. The data point positions should not be biased, with initial and final points below the line and middle points above the line, as observed in Fig.  S9.
Response: We sincerely thank the reviewer for the helpful comments, patient explanation, and wonderful suggestions. To obtain more accurate values for the reactions with very low reaction rate, we have repeated many of the in vitro catalytic assays. And, based on the reviewer's suggestions, we have redrawn the related figures to match the y-axis with the maximal data.
2) The major insight coming from this work is the allowance of the mismatch being ligated. The authors propose that the viral active site is more flexible. When one considers flexibility in a structure, it is standardly considered that there are protein regions that adopt multiple conformations. Is this what the authors intended? Based on Figure S7, it appears that the DNA is flexible, with a lack of base pairing. A CT base should have two H-bonds, but it appears to have only one, suggesting strain. The position of the phosphodiester backbone appears to be maintained. Does the phosphodiester backbone overlay, when overlaying the protein chain for the CG and the CT complexes? If that is the case, are there more phosphodiester stabilizing residues in the AsfvLig than in Lig1, 3, or 4? What is missing in the AsfvLig from Lig 1, 3, and/or 4 that ensures W-C fidelity? Could AsfvLig work on single stranded RNA or DNA?
Response: We sincerely thank the reviewer for the helpful comments. We agree with the reviewer that, with a lack of base pairing, CT could form two H-bonds.  Fig. 5C in the revised manuscript, showing that the phosphodiester backbone, 5'-P, and 3'-OH are well overlaid in the AsfvLIG:CG and AsfvLIG:CT1 complex. We have carefully compared our AsfvLIG structures with the HsLIGs structures, however, we did not identify any more residues that are missing in AsfvLIG and could form more stable interaction with phosphodiester and ensure W-C fidelity in HsLIGs.

As suggested by reviewer 1, we have included one new panel in
Like all the characterized ATP-dependent DNA ligases, AsfvLIG catalyzes phosphodiester bond formation between adjacent 3'-OH and 5'-phosphate in DNA duplex through a similar three-step mechanism. However, in consistent with previous study (Showalter et al., 2006, Chemical Review, 106: 340-360), the obvious different DNA binding behavior, unique fold of the NTD domain, and unique residues at the nick site all confirm that AsfvLIG is one atypical DNA ligase. As revealed by our in vitro DNA binding and catalytic assays, HsLig1 and AsfvLIG do not actively discriminate between match and mismatch during step 3; the discrimination or tolerance of mismatched DNAs solely occurs in step 2. We believe that the unique NTD domain and the unique residues at the nick site all contribute to the low fidelity of AsfvLIG. The extensive interactions between DNA and the NTD domain offer AsfvLIG with strong binding affinity to all matched and mismatched DNAs. In addition to matched DNAs, the strong binding also gives AsfvLIG enough opportunity to catalyze the ligation of certain mismatched DNAs. In incorporation with the NTD domain, the four nick site residues will help the reorientation of DNA 3'-OH and facilitate the ligation reaction. Compared to other nick site residues, Gln403 appears to be more important for the catalytic efficiency of AsfvLIG. As exampled by the AsfvLIG:CT2 structure (Fig. 5B), Gln403 can form H-bond interaction with the mismatched C:T pair at nick 3'-end; in addition to catalytic efficiency, such interaction definitely also contributes to the low fidelity of AsfvLIG.
Before any AsfvLIG structure was reported, Showalter and coworkers already demonstrated that AsfvLIG has very low adenylation activity towards DNAs with 3'-dideoxy-or 3'-amino-terminated nicks, compared to regular nick DNAs; these observations indicated that 3'-OH of the nick is a critical component of the active site architecture during 5'-P adenylation. The unique nick site residues of AsfvLIG could not be replaced by the corresponding residues of HsLig1, suggesting that they may work in an incorporative manner with other unique structure feature of AsfvLIG, especially NTD domain. The corporation between these unique structure features of AsfvLIG will help the reposition of the active site architecture and favor the nick sealing of matched and some mismatched DNAs.
As suggested by the reviewer, we did in vitro ssDNA and ssRNA ligation assay using the WT AsfvLIG. However, as depicted in the right panel of figure above, AsfvLIG could not support the ligation reaction of either ssDNA or ssRNA.
3) The authors included a catalytic reaction schematic, as requested. However, it is a simple one that could be used in a biochemistry paper without the structure. It would be more helpful and inclusive of the results being reported if that schematic includes the active site residues and geometry, as this is a structure paper where the active site geometry is known.
Response: We sincerely thank the reviewer for the helpful comments. We totally agree with the reviewer that one real structure-based schematic is more informative than the one we presented. However, as we showed in the reaction schematic, the ligation reaction contains three consecutive steps. In this study, we are able to obtain one open-form ATP bound complex, two close-form AsfvLIG:DNA complex prior to reaction, and one close-form AsfvLIG:DNA complex after reaction. However, very unfortunate, several key intermediate structures (especially AsfvLIG with AMP linked with the catalytic Lys residue with or without nick DNA, close-form AsfvLIG structure in complex with 5'-pyrophosphate linked AppDNA) are still missing. Therefore, we are very sorry that we could not provide a structure-based schematic at present. The catalytic reaction schematic has been moved to Fig. 8A in the revised manuscript.
4) The conservation between AD and OB domains in porcine and human DNA ligases should be provided in text. It is interesting that despite the lower conservation, the structure is conserved.