Programmable site-selective labeling of oligonucleotides based on carbene catalysis

Site-selective modification of oligonucleotides serves as an indispensable tool in many fields of research including research of fundamental biological processes, biotechnology, and nanotechnology. Here we report chemo- and regioselective modification of oligonucleotides based on rhodium(I)-carbene catalysis in a programmable fashion. Extensive screening identifies a rhodium(I)-catalyst that displays robust chemoselectivity toward base-unpaired guanosines in single and double-strand oligonucleotides with structurally complex secondary structures. Moreover, high regioselectivity among multiple guanosines in a substrate is achieved by introducing guanosine-bulge loops in a duplex. This approach allows the introduction of multiple unique functional handles in an iterative fashion, the utility of which is exemplified in DNA-protein cross-linking in cell lysates.

DNA-protein cross-linking reaction. While the optimal condition for this reaction was listed in Table  1, the logical process to optimize the suitable condition for the high yield was not described in detail. Because such processes are critical for further application of the authors' method, the process should be described in detail with appropriate discussion. For example, how the authors applied the type and contents of organic solvent (e.g. 10%, 50% THF and 20% DMSO) for a typical reaction system should be described. Judging from the reaction preference, the secondary structure of oligonucleotide is important for the regio-selectivity of this reaction. Does the secondary structure of oligonucleotide in aqueous solution containing the organic solvent predictable? Possible secondary structures of the target oligonucleotide should be analyzed with reagents that have been reported to target unpaired nucleic acid bases. The manuscript tackled an interesting application of Rh(I)-carbene system, however, the optimization process of the reaction condition and information on the possible structure of the target oligonucleotide under the particular reaction condition are very important not only for justification of the authors' notion but also for further application of the authors' method. After clarification of these issues, this manuscript could be published in Nature Communications.
Additional comments: 1. The authors noted no significant selectivity of dI against dG, but still a slight difference was observed. Is it possible to discuss the difference based on the nucleophilicity N index from DFT calculation? 2. Appropriate information on the secondary structures of these oligonucleotides in aqueous organic solvents should be analyzed by means of conventional technique. Is it necessary to use aqueous organic solvents for this reaction? 3. Why the total amount of proteins in lysate also changed from lane 1 to 4 in Fig. 4e? Judging from the experimental condition, only the T7 RNAP concentration should be changed. 4. The yield of double modification discussed in Fig. 4d was not provided and should be appeared in SI (p.33). Effect of stability of the duplex after the first modification (20b) should be discussed for the double modification. 5. The caption for Fig. 4d seems to be incorrect.

Reviewer #3 (Remarks to the Author):
This manuscript describes G-selective modification of ss-DNA and of related mismatched or otherwise unpaired sites. the science described here represents a significant jump forward in the reactivity, selectivity, and general utility of carbene reagents for selective oligo modification. The diazoacetone reagent is a remarkably simple reagent with important implications for bioorthogonal reactivity. DNA chemistry is extremely limited in breadth and scope. The reactivity demonstrated here appears truly general and utilitarian, going far beyond simple proof-of-concept experiments with very small, special sequences.
The protein cross-linking experiments (Fig 4 a-c) are an impressive demonstration of the power of these easily-obtained DNA variants. The fluorescence data (Fig 4 e) convincingly demonstrates the use in lysate. Certainly there are other ways to make non-natural oligos, but these are fairly complex constructs, and the ability to predictably i them from natural sequences is meaningful.
Publication of this manuscript seems warranted, as it could be a powerful tool to practicing scientists in the field of DNA science, and it demonstrates a remarkable fundamental advance in our understanding of controlled DNA reactivity.
I have only very limited concerns that might be addressed prior to publication.
1. Table 1 describes screening test on mixtures of nucleosides. I presume that only 1 is reactive, but this should be more clear from the table. 2. I find Fig 1 to be fairly hard to follow. In (a), the top blue arrow "contains" the monofunctionalized substrate, but it looks like it is added separately as a new additive above. Also, the two cartoon styles are confusing. 3.

REVIEWER COMMENTS
Reviewer #1 (Remarks to the Author): The key discovery that makes this paper exciting is the Rh(I) catalysis for carbene transfer at the O6 position of guanine. My group discovered that copper(I) carbenes can selectively form O6G adducts, but this paper provides a more efficient and high-yielding way to do these sorts of conjugations. My group struggled with long substrates and could not effectively hit bulge positions. Their method seems more reactive for O6G and guanines can be targeted in many types of unpaired scenarios (single-strand, bulges, overhangs). For only the discovery in basic reactivity this paper deserves to be accepted.
Here are some of the things I find lacking: 1. «chelation-free catalyst» -what experiments have been done to study this? To me the DFT is unconvincing as the DFT transition states are usually partially pre-defined and then optimized. How was the transition state search done? Did the authors truly sample both the chelated and unchelated possibilities? An energy comparison of these possibilities could be made, even if it requires opening a metal-binding site through olefin dissociation.
>> We greatly appreciate the helpful suggestions and comments.
After receiving the reviewer's comment, we examined N 7 -methyl-2'-deoxyguanosine for its reactivity toward the carbene, and found that the reaction gave O 6 -alkylation in 65% yield. This result strongly supports the chelation-free model. Regarding DFT calculations, we attempted to locate the transition states of chelation models, however, were unable to get a reasonable geometry.
2. My (Gillingham) group used the 7-deazaguanosine as a mechanistic probe for chelation requirement. This is commercially available and a simple experiment that would test their "non-chelation" hypothesis.
>> Based on the comment, we attempted a reaction with 7-deazaguanosine, which failed to give the corresponding product ( Supplementary Figs. 3 and 21). However, considering the result from N 7 -methyl-2'-deoxyguanosine (Fig. 2c), we reason that this failure may have to do with the intrinsic reactivity of 7deazaguanosine, rather than the lack of N 7 chelation.
3. The modification site in larger oligos was determined almost exclusively by MS/MS. While I don't doubt the correctness of the conclusion, I think the authors should include characterization of modification site in at least one of their large substrates by an additional method (restriction, polymerase extension assay). Polymerase extension would be especially informative as it can pick up low level modification more effectively than MS/MS. >> We included a primer extension assay with 8h ( Supplementary Fig. 6). While primer extension was not observed with purine bases (A and G, lane 1 and 4), both pyrimidine bases (T and C, lane 2, 3 and 5, 6) were incorporated to give the extended products, indicating that the act-G is recognized as either A or G by the DNA polymerase. This is consistent with previous reports that O 6 -alkylated G elicits G to A transition.

Does the reaction work in the presence of protein?
>> Yes, we performed a reaction on oligo 7a in the presence of lysozyme, and obtained the Gacetonylated product 8a in 67% yield ( Supplementary Fig. 8).
5. Cyclic guanine dinucleotides are an important substrate class that could not be accomplished with the copper(I) method. As these molecules are important in human and bacterial biology I would recommend the authors try these as another substrate class.
>> We examined cyclic guanine dinucleotide 32 for the labeling, and confirmed that it proceeds smoothly to give a mixture of mono-and bis-acetonylated cyclic di-GMP 33 and 34 in 46% and 33%, respectively ( Supplementary Figs. 4 and 23). 6. Fig. 4 takes a full one-page spread to show that reductive amination of the ketone to a DNA binding protein (T7 RNAP) works. I would have been shocked if this part hadn't worked and I'm not sure what the authors want to accomplish here. A better demonstration of the method would be to modify the protein with the diazo compound and show that it can achieve proximity labelling of the DNA at G's close to the T7 binding site. Chip-Seq, RIP-Seq, and related methods depend on nucleic acid-protein crosslinking so testing this possibility would be more valuable for the chemical biology community.
>> Thank you for the suggestion. It would be an exciting exploration.
The work in Fig. 4 was performed to demonstrate the feasibility for the iterative multi-functionalization of oligonucleotides. 7. I see very few RNA substrates. Given the varied biotechnological applications of modified RNA I think more attention should be paid to this substrate class.
>> As suggested, we included two more RNA substrates in the scope (Table 2).
Hence I believe the fundamental discovery in the paper is appropriate for publication in Nature communications. But I think revisions are necessary to more firmly support the hypotheses, and to convince the wider community that the tool will be useful.

Dennis Gillingham
Reviewer #2 (Remarks to the Author): This manuscript described the chemo-and regio-selective modification of oligonucleotides by means of Rh(I)-carbene catalysis. The chemo-and regio-selective modifications of proteins and nucleic acids are important for the identification of molecules that associate in the biological events, an efficient and selective crosslinking reaction applicable in aqueous media are in great demand. The modification reaction with Rh(I)-carbene system allows the selective introduction of an acetonyl group at unpaired guanosine in several secondary structures. Possible usefulness of the Rh(I)-carbene system has been demonstrated through the applications of this method to the DNA-protein cross-linking reaction.
While the optimal condition for this reaction was listed in Table 1, the logical process to optimize the suitable condition for the high yield was not described in detail. Because such processes are critical for further application of the authors' method, the process should be described in detail with appropriate discussion. For example, how the authors applied the type and contents of organic solvent (e.g. 10%, 50% THF and 20% DMSO) for a typical reaction system should be described.
>> We greatly appreciate the valuable comments.
Regarding the solvent composition, we began the optimization on nucleoside monomers in 50% THF, which was eventually optimized to 10% THF. The use of 10% THF was for the convenience of handling the catalyst as a stock solution. We also performed a reaction in pure aqueous buffer solution, and confirmed that the small amount of THF does not affect the reaction ( Supplementary Fig. 5).
For the use of 20% DMSO, it was employed once for the photocage experiment owing to the solubility of the diazo compound.
Judging from the reaction preference, the secondary structure of oligonucleotide is important for the regio-selectivity of this reaction. Does the secondary structure of oligonucleotide in aqueous solution containing the organic solvent predictable? Possible secondary structures of the target oligonucleotide should be analyzed with reagents that have been reported to target unpaired nucleic acid bases.
>> The presence of the organic co-solvent arises from the catalyst stock solution prepared in THF, which facilitates setting up the reactions. Given the reviewer's comment, we performed the labeling experiment in the absence of organic solvent, and confirmed that it showed a comparable efficiency/selectivity ( Supplementary Fig. 5). Based on the observation, we reason that the presence of 10% THF would not affect the secondary structures of oligonucleotides.