Crystal structure of an RNA-cleaving DNAzyme

In addition to storage of genetic information, DNA can also catalyze various reactions. RNA-cleaving DNAzymes are the catalytic DNAs discovered the earliest, and they can cleave RNAs in a sequence-specific manner. Owing to their great potential in medical therapeutics, virus control, and gene silencing for disease treatments, RNA-cleaving DNAzymes have been extensively studied; however, the mechanistic understandings of their substrate recognition and catalysis remain elusive. Here, we report three catalytic form 8–17 DNAzyme crystal structures. 8–17 DNAzyme adopts a V-shape fold, and the Pb2+ cofactor is bound at the pre-organized pocket. The structures with Pb2+ and the modification at the cleavage site captured the pre-catalytic state of the RNA cleavage reaction, illustrating the unexpected Pb2+-accelerated catalysis, intrinsic tertiary interactions, and molecular kink at the active site. Our studies reveal that DNA is capable of forming a compacted structure and that the functionality-limited bio-polymer can have a novel solution for a functional need in catalysis.

28. [158][159][160][161][162][163][164][165][166][167][168][169]Fig. 4c,e. Please show the Pb2+ binding site in more detail, with distances to heteroatoms of RNA. It is hard to imagine that a specifically bound cation nicely sitting in the cavity forms only a single coordination bond and nothing else participates in the cation binding. Can Pb2+ cation coordinate to N7 of G6? Other metal cations were found interacting with N7 of purines. 29. Please show a lead cation in "real size" (~3.0 A diameter) in Fig. 4 to help better understand space requirements. 30. Line 145. Why is coordination of a Pb2+ cation to water unexpected? Metal-water coordination is often involved in catalytic reactions. 31. Line 148-150. Please tone down this statement. The water molecule might be critical; however this conclusion is based only on a sole H-bond distance. 32. Lines 151-156. Again, as in the earlier comment, these nucleotide changes may disrupt the structure and, without experimentally determined structures for mutant DNAzymes, conclusions must be carefully worded. 33. Lines 163-165. The reasons for higher catalytic activity in the presence of lead is not clear. Does a lead cation bind the enzyme better than other cations or is it better at the chemical step? 34. Does comparison of the DNAzyme activities with different cations correspond to the reverse order of pKa of the corresponding metal hydrates? What's a pH dependence of the cleavage rate with lead? Please discuss prior publications focused on the reaction chemistry in the context of the structure. Can all published biochemical data be explained by the structure-based catalytic mechanism? 35. Lines 163-165. I do not follow authors' idea and I do not see any structural reason for a lead cation to be a specific binder. Do the authors mean that a shallow cavity is too big for smaller than Pb2+ cations or too small for water-coordinated cations (such as fully hydrated Mg2+ cation) so that they cannot form a "productive" contact with an active water molecule and bind to the other "sides" of the cavity, away from the catalytic site? There are many precedents when cations such as Mg2+ or K+ bind nucleic acids without coordinated water molecules. 36. It is surprising that, having the crystals without bound cations in hand, the authors did not soak other cations, such as Mg2+ or its analogs with stronger anomalous scattering properties (Mn2+, Ca2+), to map the binding sites for these cations and put the issue to rest. 37. Does Co hexamine, a mimetic of a fully hydrated Mg2+ cation, support the reaction? 38. The related DNAzyme 17E (Li, 2000, NAR, 28: 481-488) shows reduced activity in the presence of 150 mM F-anions, consistent with the possibility that a fluoride can replace an active water molecule (as shown for some protein enzymes); however the reaction is not abolished, arguing against a water molecule playing a critical role in the catalysis. Please comment on this observation. 39. The proposed catalytic mechanism requires deprotonation of N1 of G13. This has not been discussed. What would shift the pKa of this group? 40. Is there anything to stabilize the transition state? 41. Does Ba2+ cation, the most close mimetic of Pb2+ cation, support the reaction? 42. What's the occupancy of Pb2+ in the structure? 43. Please show an omit and anomalous maps for Pb2+ cation. The authors must prove the identity of lead because the lead-containing structure was determined at higher resolution than the other structures and it has 20 water molecules emerged because of improvement of resolution. 44. Please use the same color for lead cation throughout the figures. In Fig. 4, lead shown in black, grey and green colors, with labels in black and green. 45. Crystallographic table does not list B-factors for DNA, metal and water molecules. 46. Fig. 4e. A standard blue-red presentation of the surface potential could be a better option for this panel. 47. Line 483. Why is the water molecule shown in cyan and not in standard red color for an oxygen atom, with density in green or blue? What's the B-factor of this water molecule? 48. Fig. 4d. G+1 is a deoxyribonucleotide and therefore it should not have a 2'-OH group unless the authors say in the figure legend that they are showing all-RNA substrate. 49. Fig. 4d. Why are labels shown in two colors? 50. Please compare the structure and the putative catalytic mechanism of the DNAzyme with the structure and catalysis of the leadzyme. 51. Line 173. …activity measurements… 52. Fig. 5a. This figure is very crowded and unclear. See my earlier comments for Fig. 1e to improve the view. 53. Fig. 5a. Is T11 cyan or dark blue? What' the magenta nucleotide? 54. Fig. 5b,c. Motivation for showing the electron density map is not clear. These are not the most important regions of the DNAzyme and the refined 2Fo-Fc map is not the best way to illustrate the quality of the structure. 55. Lines 177-180 and 182-184. Motivation for testing an insertion of nucleotides at position 7 (4 mutants in total) and conclusions are not obvious. What do these mutations address? C7 provides spacing between adjacent base pairs and the structure shows that there is enough space for looping out a residue without impact on catalysis unless the inserted base is capable of forming alternative base pairs and disrupting the fold. That's what the authors see: insertion of purines that have better potential for base pairing is more disruptive than insertion of pyrimidines. 56. Lines 180-182. This is an incorrect conclusion. While deletion of T11 does significantly reduce cleavage, insertions do not affect cleavage efficiency strongly, leading to the same conclusion as before: an insertion can be tolerated with only small impact on activity. 57. Line 185. Presented figures do not help to evaluate the potential role of A15 and A14. 58. Lines 185-187. If I understand correctly authors' idea, deletion of A15 converts the 5'-WCGAA consensus sequence into the 5'-WCGR sequence, both observed in the original publication (Santoro et al). This result means that both consensus sequences from SELEX correspond to the same DNAzyme structure. 59. Line 189. There is no Fig. 3c in the paper. 60. Lines 188-190. This conclusion is not entirely correct. According to the original observation (Santoro et al) and SFig. 3A, the A14G substitution in the context of the delA15 shows some activity. Same is true for A14T. This means that the A14:G-1 pair is important but not critical for catalysis. 61. . This sentence is also not entirely accurate. While the majority of Watson-Crick pairs replacing the A14-G-1 base pair show diminished activity, the T-rA combination is rather active (SFig. 3b) and several non-canonical base pairs, A-rA (SFig. 3a), A-rC (SFig. 3d), T-rC (SFig. 3d) and T-rU (SFig. 3s) also show good activity. The authors cannot make strong conclusions without measured rate constants. It is recommended to provide a supplementary figure with a structure-based schematic of these combinations and discuss similar data from prior publications. 62. . The kink is usually stabilized through extensive stacking interactions and that's what the structure shows. The identity of base pairs, Watson-Crick or non-canonical, for making a bent in the backbone should not matter. What is likely important is that non-canonical base pairs are more dynamic than canonical base pairs and therefore offer flexibility required for the catalytic reaction. Published articles have discussed this point and the authors may discuss it from the structural perspective as well. 63. Line 195. Please provide canonical designations of the South (C2'-endo) and North (C3'endo) sugar puckers, if that's what observed in the structure. 64. Line 193-195. Since G-1 is a ribonucleotide and G+1 is a deoxyribonucleotide, their typical sugar conformations should be North (C3'-endo) and South (C2'-endo), respectively. Are the authors sure that G-1 is in the South and G+1 in the North conformation in all structures? Fig. 3a shows that G-1 is indeed in the C2'-endo conformation; however the sugar conformation of G+1 in the green structure (methylated RNA) is unclear and probably wrong. Do the authors see same sugar puckers in the all-DNA structure as well as in the structure with a methylated substrate substitution (see my earlier comment)? If yes, this is a highly unusual observation and an interesting point to discuss. I am also not sure that the structures of DNAzyme and methylated DNAzyme determined at 3.05 and 3.8 Å resolution can tell about the sugar pucker. By the way, the leadzyme was crystallized with two different sugar puckers for the same residue. 65. Fig. 3. Large sticks for the highlighted nucleotides should be removed to simplify the view. 66. The authors did mention that a methyl moiety is not seen in the map in the Materials and Methods section; for those readers who do not read M&M section, this fact should be mentioned in the main text, possibly in the figure legend. 67. Line 198. In the proposed mechanism of the 8-17 DNAzyme … 68. Line 203, Fig. 6. There is no need to show comparison with natural ribozymes in the main text figure. As expected, natural ribozymes differ from the in vitro selected DNAzyme. 69. Lines 206-209, SFig. 4 a,b. Parallels with the catalytic mechanism of the hammerhead and hairpin ribozymes are more interesting and can be presented in the main text figure. One thing is striking when the DNAzyme is compared to the hammerhead ribozyme: although both have a kink in the catalytic site, the interhelical angle is larger in the hammerhead ribozyme than in the DNAzyme. This observation relates to the question I've raised about smFRET data. 70. The authors can also compare their structure with the RNA-ligating DNAzyme structure; there are similarities which could be discussed. 71. SFig. 4. Why are labels shown in three different colors? 72. Line 213. Fig. 4e shows that lead-binding pocket does contain a charged residue. What is it? 73. Line 215. Does water displace O5' atom or donate a proton to this leaving group? 74. Line 241. Denaturing.

Reviewer #2 (Remarks to the Author):
The authors report the X-ray crystal structure of the RNA-cleaving 8-17 DNAzyme, a longknown member of the most-studied classes of DNAzyme. Many labs have tried for many years to obtain such a structure, so this manuscript will be viewed as a breakthrough, also because it is only the second structure (after ref. 9, Nature 2016) of any DNAzyme. The new 8-17 structure (which is actually three related structures) reveals several new and in some cases unexpected structural features, and the observed structure explains many chemical features of the catalysis.
After suitable revisions that do not appear to require any new experiments, the manuscript should be acceptable for Nature Communications.
(This reviewer is not a crystallographer and therefore leaves checking of the technical details of the crystallography to other reviewers who are expert on that aspect of the work.) 1. Page 4, line 83: "To unravel the catalytic mechanism of the DNAzyme, we determined three crystal structures...". However, the nature of these three structures (why three and not just one) is not revealed until page 6, line 115: "Among the three structures (DNAzyme without Pb2+, DNAzyme with Pb2+, and DNAzyme with O2'-Me-G substrate)...". The nature of the three structures should be mentioned on page 4 rather than waiting until page 6, especially because structural information is shown well before page 6 is reached. Whether the 2'-OMe-G structure was in the presence or absence of Pb2+ should also be made clear. Figure 1, and its caption, do not state that the structure shown was obtained in the presence of Pb2+, and Pb2+ is not depicted in any of the panels. Same for Figure  2.

The relevant panels of
4. Page 10, line 202: "the overall fold and the detailed catalytic mechanism of 8-17 DNAzyme are completely different from the ribozymes (Fig 6)". However, the text also notes that each of 8-17 DNAzyme, hammerhead ribozyme, and hairpin ribozyme use a G residue as general base for deprotonation of the 2'-OH at the cleavage site. This aspect at least is not "completely different" among the DNAzyme and ribozymes (curiously, HDV is not mentioned at all here, although it is shown in the figure). I agree that the general acid aspect for protonation of 5'-leaving group is completely different.
5. Related to previous comment, and perhaps confusingly, the Conclusions (page 11 line 225) emphasizes "Our crystal structures have highlighted that similar to its counterpart ribozymes, this DNAzyme catalyzes the RNA cleavage via a general acid-base mechanism". So one part of this manuscript emphasizes "completely different" mechanisms, whereas another part "similar to ribozymes". This seems inconsistent.
6. Page 6, line 126: "2'-Me protection" should be "2'-OMe protection". Similarly, page 6, line 115 should be "2'-OMe-G substrate" rather than "O2'-Me-G substrate", if only to avoid the implication of a 2'-Me rather than 2'-OMe group. Note Figure 3 caption already says "DNAzyme(2'-OMe-G)". 7. In the Methods, the very brief description (page 13, line 258) that "the AsfvPolX protein is expressed and purified in the laboratory" is insufficient to allow others to reproduce the work. Was there an expression plasmid; if so, how was it prepared or from where was it obtained? What was the procedure for protein expression and purification? 8. Figure  10. The manuscript would benefit from revision for grammar and spelling.

We sincerely thank both reviewers for reading our manuscript with great care, we also thank the reviewers for their very helpful comments, encouragements and criticisms as well. Based on these comments and suggestions, we have carefully revised our manuscript with the 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.
Reviewer #1 (Remarks to the Author): The study by Liu and coworkers is focused on determining the crystal structure and elucidating the catalytic mechanism of the RNA cleaving DNAzyme "8-17". The authors determined the 2.55 Å X-ray crystal structure of a sibling of the famous 10-23 DNAzyme and showed how a small DNA motif forms an amazing functional unit. This structure, to my knowledge, is the first structure of an RNA-cleaving DNAzyme. Most importantly, these DNAzymes are not some kind of oddity but are important tools used, for example, for fragmentation of natural mRNAs for subsequent analysis (see a recent example in Luciano et al, 2017, Mol. Cell). They also have potential for silencing harmful genes in mammals. The authors are very fortunate to obtain the structure in the conformation that appears to reflect a relevant pre-catalytic state. The structure, structure-guided mutagenesis and nucleotide substitution analysis provide a plausible mechanism for catalysis and explain the role of various parts of the DNA in the structure formation and catalysis. In addition, the structure visualized the bound cofactor, a lead cation. I find this study of great interest to general audience of Nature Communications and strongly suggest considering the manuscript for publication after a major revision.

Response: We sincerely thank the reviewer for the good comments and encouragements.
The manuscript has some deficiencies, which have to be addressed prior to publication.
The main critique is that the functional assays are not quantitative. Although the gels showing cleavage with different mutations and substitutions provide nice side-by-side comparisons of DNAzyme activity, I would expect the 21st century mechanistic study to rely on measured rate constants and not on qualitative descriptions such as "nearly abolished the enzymatic activity", "almost completely abolished the activity", "note as dramatic as", "significantly reduce the activity", etc. Based on these gels, I disagree with some of authors' conclusions, and determination of rate constant values will resolve the controversy. Since the authors already have materials and the assay in hand, these experiments should take a couple of weeks.

Response: We sincerely thank the reviewer for the criticisms and very helpful suggestions. To better understand the function and the catalytic mechanism of 8-17 DNAzyme, we have redone most of the cleavage assays using FAM-labelled substrates. All the results have been quantified and included in the revised manuscript.
I also do not understand the structural basis for the lead cation selectivity and would feel more comfortable about the proposed mechanism of catalysis with less disruptive nucleotide substitutions (probably hard to do) and analysis of the published biochemical data based on structural observations. The authors also did not specify clearly that their structure is all-DNA, and not a DNAzyme bound to the RNA substrate, and that the structure represents one of the variants, albeit the most efficient, of the 8-17 DNAzyme.
Overall, the manuscript would benefit from more clarity and comparison with a relevant structure of the leadzyme and not distinct structures of some ribozymes.

Response: We sincerely thank the reviewer for the very helpful comments. We have included many details, such as crystallization, the DNAzyme and substrate compositions, in the Results section of the revised manuscript.
As suggested by the reviewer, we also compared our structure with the leadzyme structure. We prefer to keep the comparison between our structure and other RNA-cleaving ribozymes, especially the hammerhead and hairpin ribozymes, which share some similarity in the catalytic mechanism with the DNAzyme.

The DNAzyme crystals are very fragile and they decay very fast during the data collection. To obtain the diffraction data for the three structures reported in the manuscript, we screened hundreds of crystals. We are really sorry that we could not obtain useful crystals and solve the structure containing less disruptive nucleotide substitutions, as suggested by the reviewer.
Specific critiques and suggestions to improve the manuscript are following.

Response: Done as suggested.
2. Line 46-48. In addition to the potential use for silencing genes in vivo, RNA cleaving DNAzymes, including 10-23 enzyme, have other applications (see my comment earlier).

manuscript.
11. Fig. 1e. The authors highlight C7 and T11 here without explanations. The explanations are given in the second part of the write-up, which however does not refer to this figure. Response: Fig. 1e has been replaced with a new image showing the detailed A5:G13 and G6:C12 pairing in the revised manuscript.
12. Fig. 1e. This view is difficult to understand because of the oversimplified presentation of the majority of nucleobases, shading of sugar rings, and insufficient transparency of overlapped nucleobases. Please add more labels for nucleotides.

Response: Fig. 1e has been replaced with a new image showing the detailed A5:G13 and G6:C12 pairing in the revised manuscript.
13. Fig.1 legend. "Globe" or "global" architecture?

Response: It is "global architecture". Thanks for the correction.
14. Lines 92-93. I would expect a larger (closer to 180 degrees) angle between the arms according to smFRET measurements of the Pb 2+ -induced cleavage (Kim et al, 2007, NCB, 3: 763). Is crystal packing affect the angle? Can the angle get smaller without disruption of the structure? Please discuss this issue.

Response: We thank the reviewer for the very helpful comments and the useful literature. Using the smFRET method, Kim and the coworkers demonstrated that Pb 2+ does not cause obvious conformational change during the cleavage reaction, this observation is consistent with our structural observation. In the literature, the authors also provided a schematic figure showing a larger (closer to 180 degrees) angle between the arms, but they did not provide any atomic structure model of the DNAzyme. Therefore, it is very difficult for us to compare our structure with the schematic model proposed by Kim et al. At the beginning of the "the overall folding of 8-17 DNAzyme" section of the revised manuscript, we discussed about the space groups and molecular packing of the three structures, which indicate that the presence of AsfvPolX has little effect on the folding of the core region of the DZ23/substrate complex and our structure should represent a real model of the complex.
15. Lines 98-100. In Fig. 1, G13 and A5 look like they are not in the same plane while G13 seems to be in the same plane with C12 and G6. Fig. 1e with a new figure to show the relative orientations of the four residues.

Response: G13 and A5 are in the same plane. In the revised manuscript, we have replaced the original
16. Line 97. C4:G8.
Response: Thanks for the correction. The " non-canonical T1:G+1 wobble pair"has been replaced with "canonical T1:G+1 wobble pair" in the revised manuscript. 18. Fig. 2a. This panel shows a surface view which was not discussed in the manuscript.

Response: Fig2a has been updated. The surface view was removed to better show the electron density of G-1 and G+1 residues.
19. Fig. 2b. Please label interacting nucleotides.

Response: Done as suggested.
20. Line 109. … pseudoknot that resembles the shape of an inverted cone.

Response: The reference has been included in the revised manuscript.
23. Line 126-128. This section is misleading. It should be clearly explained that, first, the authors have crystallized the DNAzyme bound to a non-hydrolysable substrate strand made of DNA. Second, they wanted to introduce a ribonucleotide into the DNA strand to obtain a hydrolysable substrate but instead they incorporated a methyl-2'O substituted nucleotide to prevent possible cleavage. The structure was however obtained without Pb, therefore no cleavage was expected. I am wondering why the authors did not obtain the structure with a ribonucleotide in the absence of lead or the structure of the methylated substrate with lead. The structure of the methylated RNA substrate with lead would most closely represent a pre-catalytic state of the enzyme, but this structure was not obtained for unclear reasons. Right now, the catalytic mechanism model represents a combination of information from three structures and none of the structures individually corresponds to a pre-catalytic state structure.

Response: We sincerely thank the reviewer for the very helpful comments. We have included one new paragraph at the beginning of the Results section to describe the detailed components of the structures. Besides the three structures reported in the manuscript, we tried many combinations of the DNAzyme, substrate (native DNA, DNA containing single rG, and DNA containing 2'-OMe-G), and cofactor (with or without Pb 2+ ). Though we got some crystals for the samples containing rG without Pb 2+ or 2'-OMe-G with Pb 2+ , the crystals diffract too weak (typically lower than 7-8 Å) to give any useful information.
24. Lines 128-130. There must be changes in the conformation of the GG kink, at least in the sugar-phosphate backbone. Please describe these changes, with emphasis on sugar conformations, and show a zoomed-in view of the differences.  figure, which has been renumbered as Fig. 3a, in the revised manuscript. The

unbiased electron density map and the close-view of the cleavage site structure were shown in the Supplementary figure 2. Instead of C3'-endo, the sugar pucker of G-1 adopts C2'-endo conformation in the DNAzyme(2'-OMe-G) structure, which is similar to the G-1 sugar pucker observed in the DNAzyme-Pb 2+ structure.
26. Line 133. Please split the sentence. Atoms of G13 are not involved in the in-line alignment.

Response: Done as suggested.
27. Line 135-142. The RNA field experienced many issues when relating ribozyme structures and catalytic mechanisms. Please use very careful wording discussing the catalytic mechanism of the 8-17 DNAzyme. Addition of a methyl moiety to the Watson-Crick edge of G13 can disrupt the structure and indirectly affect catalysis. This is what probably happens with the Dz36-6mG13 enzyme. figure (Figs. 6e-6f) showing the similar arrangement of the catalytic G residue and cleavage site kink residues in the DNAzyme structure and the hammerhead structure, whose catalytic mechanism has been well characterized. [158][159][160][161][162][163][164][165][166][167][168][169]Fig. 4c,e. Please show the Pb2+ binding site in more detail, with distances to heteroatoms of RNA. It is hard to imagine that a specifically bound cation nicely sitting in the cavity forms only a single coordination bond and nothing else participates in the cation binding. Can Pb2+ cation coordinate to N7 of G6? Other metal cations were found interacting with N7 of purines. Figs. 4a-4b, the distances between Pb 2+ and the heteroatoms of the surrounding residues are within the range of 5-7 Å. In contrast to the two-coordinated Pb 2+ , these observations suggest that the catalytic core can accommodate Pb 2+ with multi-coordination. However, as indicated by the occupancy (40%) of the Pb 2+ , binding of the Pb 2+ is very dynamic, which may lead to the disordering of some Pb 2+ -coordinating water molecules that were not observed in the structure. We sincerely thank the reviewer for the great comments and we have re-written this paragraph in the revised manuscript.

Response: We have carefully re-examined our structures; as depicted in the Supplementary
Though the N7 atom of purines can coordinate with metal cations in some nucleic acid structures, however, as indicated by the long distance (4.0 Å) between them, the N7 atom of G6 does not coordinate with the Pb 2+ ion in the DNAzyme structure. 29. Please show a lead cation in "real size" (~3.0 A diameter) in Fig. 4 to help better understand space requirements. Fig. 3e in the revised manuscript. 30. Line 145. Why is coordination of a Pb2+ cation to water unexpected? Metal-water coordination is often involved in catalytic reactions.

Response: Thanks for the correction.
31. Line 148-150. Please tone down this statement. The water molecule might be critical; however, this conclusion is based only on a sole H-bond distance.

Response: We sincerely thank the reviewer for the very helpful suggestion. The statement has been re-written in the revised manuscript.
32. Lines 151-156. Again, as in the earlier comment, these nucleotide changes may disrupt the structure and, without experimentally determined structures for mutant DNAzymes, conclusions must be carefully worded.

Response: We sincerely thank the reviewer for the very helpful suggestion. The statement has been re-written in the revised manuscript.
33. Lines 163-165. The reasons for higher catalytic activity in the presence of lead is not clear. Does a lead cation bind the enzyme better than other cations or is it better at the chemical step?
Response: We sincerely thank the reviewer for the very helpful comment. We believe that the higher catalytic activity of Pb 2+ ion is a comprehensive result; in addition to its higher binding affinity with the DNAzyme, the low pKa value of Pb 2+ ion may also contribute its higher catalytic ability. The statement has been re-written in the revised manuscript. 34 35. Lines 163-165. I do not follow authors' idea and I do not see any structural reason for a lead cation to be a specific binder. Do the authors mean that a shallow cavity is too big for smaller than Pb2+ cations or too small for water-coordinated cations (such as fully hydrated Mg2+ cation) so that they cannot form a "productive" contact with an active water molecule and bind to the other "sides" of the cavity, away from the catalytic site?
There are many precedents when cations such as Mg2+ or K+ bind nucleic acids without coordinated water molecules.
Response: We sincerely thank the reviewer for the very helpful comments and we are totally agreeing with the reviewer that some cations can coordinate with different number of water molecules. Per the reviewer's suggestion, we have further analyzed our structure. As depicted in the Supplementary Figs. 4a-4b, the distances between Pb 2+ and the heteroatoms of the surrounding core residues are within the range of 5-7 Å. Though only one Pb 2+coordinating water molecule was observed in the structure, we believe that the DNAzyme core could accommodate Pb 2+ ion coordinated with multiple water molecules.
As revealed by the leadzyme and the HDV ribozyme structures, cation recognition normally involves the phosphate backbone; however, due to the unique folding, the core region phosphate groups are not suitable for direct cation coordination in the DNAzyme structure. We believe that, instead of the space availability, lack of direct phosphate group coordination discriminates many cations from Pb 2+ , which is flexible in coordination. Based on above analysis, we have re-written the related statement in the revised manuscript. 36. It is surprising that, having the crystals without bound cations in hand, the authors did not soak other cations, such as Mg2+ or its analogs with stronger anomalous scattering properties (Mn2+, Ca2+), to map the binding sites for these cations and put the issue to rest.

Response: We sincerely thank the reviewer for the very helpful
suggestions. In fact, we did the soaking experiment previously; however, very unfortunately, the DNAzyme crystals are very fragile and all the crystals cracked or lost diffraction power upon soaking. Therefore, we could not solve any DNAzyme structure complexed with Mg 2+ , Mn 2+ , or other cations. 37. Does Co hexamine, a mimetic of a fully hydrated Mg2+ cation, support the reaction?
Response: Yes, Co hexamine can support the cleavage reaction. However, as depicted in the figure below, the Co 2+ -dependent activity of the DNAzyme is lower than the one supported by Pb 2+ . To keep our manuscript more concise and to avoid redundancy with previous literatures (Biochemistry, 2003, 42:7152-7161), this result was not included in the revised manuscript. 38. The related DNAzyme 17E (Li, 2000, NAR, 28: 481-488) shows reduced activity in the presence of 150 mM F-anions, consistent with the possibility that a fluoride can replace an active water molecule (as shown for some protein enzymes); however，the reaction is not abolished, arguing against a water molecule playing a critical role in the catalysis.
Please comment on this observation.