Biomimetic α-selective ribosylation enables two-step modular synthesis of biologically important ADP-ribosylated peptides

The α-type ADP-ribosylated peptides represent a class of important molecular tools in the field of protein ADP-ribosylation, however, they are difficult to access because of their inherent complicated structures and the lack of effective synthetic tools. In this paper, we present a biomimetic α-selective ribosylation reaction to synthesize a key intermediate, α-ADP-ribosyl azide, directly from native β-nicotinamide adenine dinucleotide in a clean ionic liquid system. This reaction in tandem with click chemistry then offers a two-step modular synthesis of α-ADP-ribosylated peptides. These syntheses can be performed open air in eppendorf tubes, without the need for specialized instruments or training. Importantly, we demonstrate that the synthesized α-ADP-ribosylated peptides show high binding affinity and desirable stability for enriching protein partners, and reactivity in post-stage poly ADP-ribosylations. Owing to their simple chemistry and multidimensional bio-applications, the presented methods may provide a powerful platform to produce general molecular tools for the study of protein ADP-ribosylation.

%,* 2&2*3 :$-1/-/*5-( ;#4*.*(5-7* 3-'149&.5-104 !4-(" *0&'.* 581#45*2 /1)6.&3 4905,*4-4 1+ biologically important ADP-ribosylated peptides", by Li et al. describes a synthetic approach to generate functionalised NAD derivatives, cleanly with high stereoselectivity and in high yield. These derivatives are then utilised with copper-catalysed click chemistry to produced modified, ADP-ribosylated peptides, whose activities in pull-down assays in vitro and in lysate validate their binding properties and indicate future utility in biochemical studies of these functionally important peptide derivatives. A key stage in the synthesis has been to utilise ionic liquids through a newlydeveloped high-throughput screening process (Green Chemistry, 2019) to enhance the reactivity and stereoselectivity of the initial reaction with excess sodium azide, which is also shown to translate to reaction with bromine to generate the corresponding bromide. The rapid, bench-top synthesis, which was shown to scale and use accessible materials, will be of significant interest to researchers wishing to study the properties and roles of this important functional modification (in signal transduction, cellular differentiation, gene regulation, cancer, and other topics of general interest) and, on this basis, is likely to make the results of this paper impactful and timely. The highlighting of the screening approach will also help in directing future method development in a number of synthetic fields. There are a number of questions arising from the experiments, as presented in the text, that may help in greater understanding of the rationale, in addition to modifications to demonstrate full characterisation and improve clarity. 1) A range of ionic liquids have been used, and have clearly in some cases shown good results. However, it is not clear, since these are run as aqueous solutions, whether this is a simple (chiral) ion effect. Clearly lactate (cf. lactic in main text) is important -how do other lactate salts fare, and are ionic liquids per-se required?
2) The choice of 5:1 mole fraction water:IL is selected based on one set of screening results, and not the IL that was finally recommended. How reasonable is this selection and why weren't other ILs of interest tested in this framework? Or, rephrased, how likely are the results from these ratios to be comparable across different IL systems.
3) The highlighted ionic liquids in Figure 2D that are indicated as being taken forward to screening are not necessarily those that show the best ratios? What governed this selection? Why were other reasonable candidates excluded ahead of those showing poorer ratios? How were the error-bars calculated (numbers of replicates are not clearly highlighted for example in the captions)? 4) BF4 anion was screened as part of the set. Why was this, given that this anion is recognised to undergo hydrolysis in aqueous solutions? Further, these anions do not appear in the list in Figure  2B? 5) Is there scope for a 1-pot process utilising the ionic liquid to aid the click reaction? 6) How were the optimisation parameters/ranges selected? From the SI this seems to have been an improvement process rather than optimisation. 7) How was the physiological stability of the ribosylated peptides measured and under what conditions? These data should be included in the SI. 8) For the ionic liquids: the anion names should be IUPAC consistent (-oate for carboxylates). The 1H NMR data look very good, but I would normally expect both a reference to literature data for each of the synthesised materials and, for known compounds, 2 pieces of spectral and 1 piece of physical data to confirm identity and purity. This should be the case for all characterised compounds, unless there are clearly justified reasons as to why these cannot be obtained. 9) In the case of the alpha-(1) and beta-(1') azides, the 1H NMR clearly indicates impurities/isomers/residual starting material. Please indicate further what these peaks are and their impact. 10) Figures S7 and S9 it would help to mark the relevant cross-peaks on the spectra for clarity. 11) For all figures in the SI, it would be helpful to have more detailed captions. 12) Figure S11 could be improved by angling the triazole in each case so that it doesn't clash with the proline. 13) For the synthesis of ribosylated proteins the time is quoted for the procedure as 2-4h. This should be more precisely identified; whether this is a per peptide difference, or some other reason. The MALDI data should have all peaks assigned. 14) It would be helpful to link the immunoblots in Figure 4 to the experimental in the SI. Although this can be worked out (the captions are fine), it would improve the clarity for the reader. 15) Overall the manuscript would benefit from a thorough language editing. Currently some aspects are either hard to understand or not well specified and there are a number of typographical errors. This is less of a problem for the SI, where there are only a few minor issues, but these should also be looked at.

Point-by-Point Responses to the Reviewers' comments
We greatly appreciate the comments of the reviewers on our manuscript. Based on these comments, we have performed a series of supplementary experiments. All the unreported 47 ionic liquids were characterized by 13 C NMR and IR besides of 1 H NMR. The stability of our ribosylated peptides were measured in 0.5 M NH 2 OH (pH = 7.5) and conditions with pH = 3.0, 5.0, 9.0 and 11.0 respectively. The growth of a poly-ADP-ribose chain was indicated to contain at least 35 ADP units using native %-NAD + and peptide-1 as the substrates. Additional control experiments were conducted to explain the effects of ionic liquids on reaction selectivity, which suggested both ion environments and synergism of cation and anions play important roles in the current biomimetic reaction. Besides, the complete MALDI-TOF-MS spectra, clean 1 H NMR spectra of "-ADPr-N 3 (1) and %-ADPr-N 3 (1'), and the uncropped membranes together with protein markers for all the blots in Fig. 4 in manuscript were added into the revised supporting information.
We have also reorganized the context that reviewers concerned in the manuscript to clarify the novelty of this work. We checked the typographical errors carefully in manuscript and supporting information, revised the abstract to 150 words and added the part of "Methods" in the manuscript based on requirements of journal, and the specialized language editing was conducted for the revised manuscript. The supporting information was reformulated with the revised figures and supplementary experiments. We believe that the revised manuscript has been substantially improved from the revisions and additional new data. All the revised sentences were marked with yellow in the revised manuscript. The details are as follows.

Part-I: Response to reviewer#1
The paper "Biomimetic "-selective ribosyaltions enable two-step modular synthesis of biologically important ADP-ribosylated peptides", by Li et al. describes a synthetic approach to generate functionalized NAD derivatives, cleanly with high stereoselectivity and in high yield.
These derivatives are then utilized with copper-catalyzed click chemistry to produced modified, ADP-ribosylated peptides, whose activities in pull-down assays in vitro and in lysate validate their binding properties and indicate future utility in biochemical studies of these functionally important peptide derivatives. A key stage in the synthesis has been to utilize ionic liquids through a newly-developed high-throughput screening process (Green Chemistry, 2019) to enhance the reactivity and stereoselectivity of the initial reaction with excess sodium azide, which is also shown to translate to reaction with bromine to generate the corresponding bromide. The rapid, bench-top synthesis, which was shown to scale and use accessible materials, will be of significant interest to researchers wishing to study the properties and roles of this important functional modification (in signal transduction, cellular differentiation, gene regulation, cancer, and other topics of general interest) and, on this basis, is likely to make the results of this paper impactful and timely. The highlighting of the screening approach will also help in directing future method development in a number of synthetic fields.
There are a number of questions arising from the experiments, as presented in the text, that may help in greater understanding of the rationale, in addition to modifications to demonstrate full characterization and improve clarity.
Question-1: A range of ionic liquids have been used, and have clearly in some cases shown good results. However, it is not clear, since these are run as aqueous solutions, whether this is a simple (chiral) ion effect. Clearly lactate (cf. lactic in main text) is important -how do other lactate salts fare, and are ionic liquids per-se required?
Response: We thank the reviewer for this comment. It is very interesting to study how the ionic liquid aqueous system affect the reaction (transfer %-NAD to "-ADP-ribose). We first investigated the ion effects. Inorganic salts including NaCl, KCl and CaCl 2 were used to replace the ionic liquids for the reaction. The results were shown in Table R1 below. To our surprise, a dramatic salt effect was found in which yield of "-ADPr-N 3 could be obtained in 55.3% with NaCl aqueous solution (Entry 2 in Table R2), which was over 6 times than 8.5% that was obtained under the same conditions in the absence of NaCl (Entry 1 in Table R1). The similar result was obtained with KCl aqueous solution with a 53.7% yield. Then we tried lactate salts including sodium lactate, potassium lactate and calcium lactate. In the solutions of sodium lactate and potassium lactate, high yields could be also obtained with 58.5% and 59.2% respectively. Therefore, both inorganic and organic salts in aqueous solution could effectively promote reactions to produce "-ADPr-N 3 , suggesting ion environments might play a role during this reaction. Besides, the type of cations in the salts affects the reaction output. For example, calcium salts such as CaCl 2 and calcium lactate gave reduced yields (22.7% and 23.6% respectively).
The high "-selectivity and yield (reactivity) in the current reactions derive from synergistic effects of cations and anions of ionic liquids. Firstly, we compared the ionic liquids that all contained lactate as the anions. As shown in Table R2, only [DEOA][Lactate] and [DMEOA][Lactate] can give the ratio of "/% more than 20.0, other cations like choline, TEOA and TMG gave the ratio of "/% from 10.7 to 14.9, and even lower with DBU as the cations (7.8 in Entry 1 in Table R2). Secondly, we compared the ionic liquids that all had DMEOA as the cations. Table R3, all the ionic liquids could promote the reactions to give "-ADPr-N 3 with high yields over 44%, however showed the dramatic differences in selectivity. Lactate represented the anion to pair with DMEOA to give the highest ratio of "/% (24.7), while hexanoate represented the anion to pair with DMEOA to give the lowest ratio "/% (2.1). Thirdly, we analyzed the distribution of the cations and anions in liquid liquids that can promote the reaction with high selectivity (ratio of "/% more than 10.0) and high yield (yields more than 40.0%). As shown in Table R4 below, 6 cations and 4 anions are included (cations: Choline, DEOA, TEOA, DMEOA, TMG, BMIM; anions: Lactate, Glycolate, Acetate, BF 4 ) to give the ratio of "/% more than 10.0. As to the ionic liquids with high yields over 40%, 4 cations and 7 anions were included as shown in  (Table R3). These results showed that the promoting factors for reaction selectivity might be discordant with the factors that could enhance the yields (reactivity) in the current reaction, which made it more difficult to design effective reaction catalytic systems. On the other hand, the combination of cations and anions can lead to a synergic effect for a given reaction system, which has also been shown important for the current reaction. As shown in the data of Figure R2 to R5, both of the desirable selectivity and reactivity came from not a single anion or cation, but their different combinations. Therefore, ionic liquids that contain rich of structural diversity as well as potential interactions such as electrostatic interactions, H-bond interactions and so on, will offer a useful compound pool for screening towards the development of simple and convenient aqueous synthesis "-ADPr-N 3 directly for %-NAD + .     liquid that we found to promote the reaction to give "-ADPr-N 3 with 1.5 ratio of "/%. And then we found the aqueous solution of the ionic liquid was also beneficial to enhance the ratio of "/%.

As shown in
Based on the two points, we investigated the effects of content of water in ionic liquids (Table S1 in the supporting information). The choice of 5:1 mole fraction water : IL was based on the screening results.
The promoting effects of different types of ionic liquids for transform of %-NAD + to "-ADPr-N 3 might come from different activation mechanisms. Thus, we agree with Reviewer#1 that the results from the ratio obtained with one single water:IL are not be comparable across different IL systems.

Figure R1
Our multi-step improvement strategy One of the goals in the current framework is to discover a practical and useful synthetic method for convenient preparations of ADPr-N 3 directly from %-NAD + . Toward this goal, we run a multi-step improvement strategy including four steps and a series of discovery ( Figure R1 above).
We chose 5:1 mole fraction water : IL([TEOA][Glycolate]) as the reaction mediate by running Step-1, but the selectivity and yield of "-ADPR-N 3 are not so good yet (ratio of "/% = 8.2; yield of "-ADPR-N 3 = 42%). Therefore, we designed the following screening method (Step-2) with a wider scope of ionic liquids. After Step-3, we got a better ionic liquid [DEOA][Lactate] in the aqueous solution to give ratio of "/% up to 24.7, however the yield of "-ADPR-N 3 was still not desirable (48.2%). And then we run the further improvements (Step-3) for the reaction conditions through changing reaction temperatures, substrate concentrations etc. (Table S4 in Page S16 in supporting information). After running Step-3, 89% yield of "-ADPR-N 3 could be obtained with Question-3: The highlighted ionic liquids in Figure 2D that are indicated as being taken forward to screening are not necessarily those that show the best ratios? What governed this selection?
Why were other reasonable candidates excluded ahead of those showing poorer ratios? How were the error-bars calculated (numbers of replicates are not clearly highlighted for example in the captions)?
Response: We thank the reviewer for this comment. We reorganized Fig. 2D including changing its types of bars in the revised manuscript to make it clear. The highlighted ionic liquids in Fig. 2D represented the ones that showed high "/% ratios (more than 10:1), and the components of the highlighted ionic liquids were marked using the arrows to direct their cations and anions indicated in Fig. 2B. Experiments shown in Fig 2D were performed in triplicates, and the error-bars in Fig.   2D were calculated based on these data. The goal of the screening is to discover a practical and useful reaction system for the preparation of ADPr-N 3 directly from %-NAD + . As shown in Fig. 2D, the highlighted ionic liquids with the "/% ratios more than 10:1 contained 6 cations and 4 anions. It was suggested the combinations of cations and anions would be more important than to choose a special one. Thus, the screening-based method shown in current work should be useful in directing future method development in finding the new catalytic functions of ionic liquids.
Question-4: BF 4 anion was screened as part of the set. Why was this, given that this anion is recognised to undergo hydrolysis in aqueous solutions? Further, these anions do not appear in the list in Figure 2B?
Response: We thank the reviewer for this comment. In the previous manuscript, we only listed the organic anions. There are no specific reasons for the BF 4 anion in the current screening. In the revised manuscript, we added the inorganic anions including BF 4 that used for the screening in Question-5: Is there scope for a 1-pot process utilizing the ionic liquid to aid the click reaction?
Response: Thanks Review#1's suggestion. We attempted the 1-pot process toward utilizing the ionic liquid to aid the click reaction as shown in Figure R2 below. The typic experiment procedure is described as follows: The mixture of 0.5 mg of %-NAD + (0.075 &mol) and 2 mg of sodium azide (1.2 mM) in water solution was added into the untreated reaction mixture of %-NAD + and sodium azide, and then 2.5 equivalent of sodium ascorbate was added. The resulted reaction was kept for 6 hours at room temperature. The reaction was detected by analysis HPLC, and 51% yield of target Peptide-1 could be obtained by our built analysis method. However, this yield was lower than the two-step-reaction method shown in the Fig. 3C (96%), in which a desalination step was involved.

Figure R2
One-pot transformation of %-NAD + to "-ADP-ribosylated peptide Question-6: How were the optimization parameters/ranges selected? From the SI this seems to have been an improvement process rather than optimization.
Response: We thank the reviewer for this comment. We agree with Review#1. Temperature, time, different ionic liquids etc. were used as the parameters for improving the reaction performances.
"improvement" should be better than "optimization" for describing such experiments, and thus we changed the caption of Table S4 to "Improvements for the preparative reaction conditions of "-ADPr-N 3 " in Page S16 in the revised manuscript.
Question-7: How was the physiological stability of the ribosylated peptides measured and under what conditions? These data should be included in the SI.
To determine the stability of our ribosylated peptides, the peptide-3 was incubated in PBS buffer (pH 7.5) containing 0.5 M NH 2 OH at 37 o C. The mixture was analyzed by HPLC at different time points. The peptide remained intact even after 24-hour incubation. In addition, we also demonstrated that our ribosylated peptide was inert to acidic or basic treatments (pH = 3.0, 5.0, 9.0, 11.0) ( Figure S8 in the revised supporting information and Figure R3 below). These results indicated a satisfactory stability of the triazoly-ADP ribose linkage as a mimic of the native ADP-ribosylated glutamate or aspartate.

Figure R3
Analysis of peptide-3 stability against NH 2 OH, acidic or basic treatments.
To make it clear to the readers, we have added more information in the related sentences describing and discussing the experiment result in our revised manuscript as seen "In contrast to the natural ester-linked ADP-ribosylated peptides that are unstable in NH 2 OH treatment at neutral pH(t1/2 < 1 h), 39,41,42 all the peptides bearing triazoly-ADPr linkages were stable for at least 24 hours in NH 2 OH, as well as in acidic and basic conditions ( Figure S8 in the supporting information)." in Line 6 to Line 8 in the first paragraph in Page 7 highlighted in yellow.
Question-8: For the ionic liquids: the anion names should be IUPAC consistent (-oate for carboxylates). The 1 H NMR data look very good, but I would normally expect both a reference to literature data for each of the synthesized materials and, for known compounds, 2 pieces of spectral and 1 piece of physical data to confirm identity and purity. This should be the case for all characterized compounds, unless there are clearly justified reasons as to why these cannot be obtained.
Together with the peaks at 8.0 ppm to 9.0 ppm that might belong to the protons of a purine base, we deduced that the impurity might come from the reactions of starting material %-NAD + .
Question-11: For all figures in the SI, it would be helpful to have more detailed captions.
Response: We thank the reviewer for this comment. Based on the suggestion, we revised the captions of figures in the supporting information, and the resulted supporting information was reformulated with the revised figures and supplementary experiments.
Question-12: Figure S11 could be improved by angling the triazole in each case so that it doesn't clash with the proline.
Response: We thank the reviewer for this comment. The angles of triazoles were improved to avoid clashing with the prolines in Figure S7 in Page S21 in the revised supporting information.
Question-13: For the synthesis of ribosylated proteins the time is quoted for the procedure as 2-4 h. This should be more precisely identified; whether this is a per peptide difference, or some other reason. The MALDI data should have all peaks assigned.
Response: We thank the reviewer for this comment. The precise reaction times for the synthesis of each ribosylated peptides were added in detail in Table S7 (highlighted in yellow) in Page S18 the revised supporting information. Also as shown in Table R7 below, their reaction time were 4 hours for the peptide 1 and peptide 2. The reaction time were 2 hours for peptide 3 and peptide 4.
The precise reaction times were obtained by detecting the reaction procedures by HPLC, and the reactions were completed based on disappearing of the peaks of the starting materials (the peptides). The MALDI data were revised to have all peaks assigned in Page S46 and Page S47 in the revised supporting information (highlighted in yellow). The copies of the MALDI spectra were also as shown in Figure R6 below).  Question-14: It would be helpful to link the immunoblots in Figure 4 to the experimental in the SI.
Although this can be worked out (the captions are fine), it would improve the clarity for the reader.
Response: We thank the reviewer for this comment. In the revised manuscript, we linked the experiments to the part of "methods" and the supplementary Figures in the supporting information as seen "See "Photo-cross-linking and visualization of the biotinylated proteins" in the part of "Methods" for more details of the photo-cross-linking-based assays. The uncropped blots and protein markers are provided in Figure S9 in the supporting information. The immunoblotting assays are provided in Page S9 in the supporting information." in Page 8 below Fig. 4 in the revised manuscript, which was highlighted in yellow.
Question-15: Overall the manuscript would benefit from a thorough language editing. Currently some aspects are either hard to understand or not well specified and there are a number of typographical errors. This is less of a problem for the SI, where there are only a few minor issues, but these should also be looked at.
Response: We thank the reviewer for this comment. Based on the suggestions, we checked the typographical errors carefully in manuscript and SI. The specialized language editing was conducted for the revised manuscript.
The manuscript by Zhu et al. describes a new synthetic strategy for the generation of an azidoADP-ribose analog (N 3 -ADPr). This analog contains an azide group at the 1' position of the glycosidic ring. The synthesis uses ionic liquids and the authors developed an efficient synthesis for generating the biologically relevant alpha isomer of N 3 -ADPr. They then coupled N 3 -ADPr to an H2B peptide (known ADP-ribosylation target of PARP1). The peptide also contain a benzophenone crosslinker as well as a biotin affinity handle. The authors sowed that the peptide could bind to a macro domain protein mH2A1.1, which is a protein known to recognize ADPribose. The authors also showed that this peptide could pulldown known ADP-ribose binding macro domain containing proteins (i.e. mH2A1.1 and PARP9) from HeLa cell lysates that were treated with UV light to induce crosslinking. Lastly, the authors showed that the ADPr-peptide could be ADP-ribosylated by PARP1.
Overall this is an interesting study; however, there are some major issues that need to be addressed prior to publication. These include: Question-1: Overall, the manuscript really needs to be carefully edited. There are many grammar and spelling mistakes. Additionally, there are issues regarding scholarship (more on this below).
Response: We thank the reviewer for this comment. Based on the suggestions, we checked the typographical errors carefully in manuscript and supporting information. The specialized language editing was conducted for the revised manuscript.
Question-2: In the paper, the authors state that the half-life for Glu/Asp-ADPr is < 1 h. They ref. ADPr is different-they should comment on this.
Response: We thank the reviewer for pointing out the mistake in the unit of IC 50 value, which has been now corrected in the revised manuscript.
For the second question, we are a little bit confused, since the experiment described by the reviewer and the paper the reviewer referred to did not match with each other. For the sake of clarity, we provide a Response: We thank the reviewer for raising this concern. However, we would like to point out that the Glu2 position of histone H2B is an endogenously ADP-ribosylated site. In fact, ADP-ribosylation is known to occur on multiple amino acid residues, including arginine, lysine, serine, asparagine, cysteine, as well as glutamate and aspartate. The ADP-ribosylation of histone H2B Glu2 position was detected as early as 40 years ago (L. O. Burzio, et al., J Biol Chem, 1979, 254, 3029-3037;N. Ogata, et al., J Biol Chem, 1980, 255, 7610-7615 2013, 10, 981-984). Such high lability makes it a challenge to investigate the biology functions of the ADP-ribosylation on glutamate or aspartate residues. In our study, we, therefore, focused on the histone H2B peptide. We deign the triazolyl-linked ADPr to mimic the fragile ester-linked ADPr ( Figure R7 below) as a proof-of-concept study of our synthetic method. We demonstrated that the triazolyl-ADP ribose linkage as a good mimic of the ADP-ribosylated glutamate or aspartate residue at different levels, including: i) our mimic selectively bound to its known 'readers' with high affinity at single protein level; ii) our mimic could enrich endogenous 'readers' from cell lysate; and iii) our mimic could undergo poly ADP-ribosylation by PARP1. Figure R7 Triazolyl-ADPr as stable mimic of the ester-ADPr Question-5: The manuscript states: "And more importantly, the stable triazoly linkage offered, to the best of our knowledge, the first class of chemical tools with controllable stereochemistry at the glutamate-type glycosidic bond to study the functional differences between the "-and %-ADP-ribosylation epimers." This is not accurate. See this ref: "Synthetic "-and %Ser-ADP-ribosylated Peptides Reveal "-Ser-ADPr as the Native Epimer," Organic Letters 40, 4140-4143 (2018). The authors should discuss this paper.
Response: We thank the reviewer for this comment. As has been mentioned in the response for Point 4 above, ADP-ribosylation is known to occur on multiple amino acid residues, including arginine, lysine, serine, asparagine, cysteine, as well as glutamate and aspartate. Our study focused on the development of a triazolyl-ADP ribose linkage as a mimic of the ADP-ribosylated glutamate or aspartate residue, since such modifications has been challenging to study because of the intrinsic instability of the ester-type glycosidic bond. At the same time, we also acknowledge that multiple methods have been developed by the community to install ADP-ribosylation (or mimics) at different amino acid residues. We have now turned down our tones by changing the sentence to "More importantly, the stable triazole linkage offers a class of useful chemical tools with controllable stereochemistry at the glutamate-type glycosidic bond, which can be used to study the functional differences between the "-and #-ADP-ribosylation epimers" in Line 7 to Line 8 in the first paragraph in Page 8 in the revised manuscript highlighted in yellow. Also, in our revised manuscript, we have cited and discussed the paper listed by the reviewer as reference 44.
Response: We thank the reviewer for this comment. We have replaced asparaginic with aspartate in Fig 1B in  Response: We thank the reviewer for this comment. Fig.1 demonstrates the strategies for synthesizing "-type ADP-ribosylated peptides in this work. Our previous work (Molecules, 2017) is to prepare %-ADPr-N 3 , which doesn't belong to the scope of this work.
Question-8: Fig. 4: are there controls for crosslinking? Also, in Fig. 4f, the last lane should be "+" for peptide-3 competition if I understand the experiment correctly.
Response: We thank the reviewer for this comment. All the photo-cross-linking-based experiments showed in Fig. 4 were performed with controls. In Fig. 4A, we examined the efficiency of the photoaffinity peptide-1 in capturing mH2A1.1, a known ADP-ribosylation 'reader', by cross-linking different concentrations of peptide-1 with the protein. A sample without adding peptide-1 was also prepared, which served as the no probe control. The result showed that the peptide-1-induced labeling of mH2A1.1 followed a concentration-dependent manner with EC 50 = 2.1 &M. No cross-linking signal was observed in the no probe control sample. In Fig. 4B, we examined the binding affinity of the peptide-1-induced labeling of mH2A1.1 by using ribosylated H2B peptide, peptide-3, as a competitor. The photo-cross-linking was performed in the presence of different concentrations of peptide-3. A sample without adding peptide-3 was also prepared as a no competitor control. The result showed that peptide-3 could effectively compete off the cross-linking with IC 50 = 4.0 &M. In Figs. 4C and 4D, similar experimental design was used to show that the %-ADP-ribosylation epimer exhibited dramatically lowered ability to capture mH2A1.1 compared with the " epimer shown in Fig. 4A and 4B. The 4C and 4D panels serve as good controls of the 4A and 4B panels to demonstrate that the " epimer is critical for the recognition of ADP-ribosylation by mH2A1.1. In Fig. 4E, we tested the specificity of peptide-1-induced labeling toward mH2A1.1. Three different control samples were prepared, including no probe control, a loss-of-ADP-ribosylation-binding mH2A1.1 mutation (mH2A1.1 G224E) control, and a non-ADP-ribosylation 'reader' (mH2A1.2) control. The result showed that after UV irradiation, only the sample containing mH2A1.1 and peptide-1 led to robust cross-linking, indicating the labeling was indeed macrodomain-ADP-ribosylation recognition-dependent but not non-specific cross-linking. In Fig. 4F, we moved a step forward to determine if the endogenous ADP-ribosylation 'readers' could be enriched by the photoaffinity peptide-1. The cell lysate was photo-cross-linked with peptide-1. The labeled proteins were then enriched with streptavidin-coated agarose beads, which were subsequently eluted and subjected to immunoblotting analysis against antibodies of two known ADP-ribosylation 'readers', mH2A1.1 and PARP9. Samples without the photoaffinity peptide-1 or with the addition of peptide-3 as competitor were also prepared as controls. The result showed that both endogenous proteins could be selectively enriched by peptide-1, as no chemiluminescence signal was observed in the sample without peptide-1, while the cross-linking was effectively competed off by the addition of peptide-3. In summary, the above photo-cross-linking-based assays were designed and performed with appropriate controls. The results demonstrated that our triazoly-ADP ribose linkage served as a good mimic of the native ADP ribosylation, which could be used to study the interactions between ADP-ribosylation and its binding proteins.
In Fig. 4F, the last lane for peptide-3 should be "+". A right version of Fig. 4 has been attached to the revised manuscript. We apologize for this mistake.
Question-9: Fig. 4: in the main text, the wrong panel is referenced in several sections.

Response:
We apologize again for our carelessness during the preparation of the original manuscript. The mismatch between Fig. 4 and the text has been corrected during our revision.
Question-10: Fig 4f: Did the authors consider doing MS/MS proteomics? In this way, they could discover potentially new ADPr binding proteins. This would certainly add novelty to the manuscript as this has not been done before with other ADPr-peptides.
Response: We thank the reviewer for this comment. We agree with the reviewer that a comprehensive profiling of ADP-ribosylation binding proteins at the proteomic level is an important and interesting application of our ADP-ribosylation mimic. However, we think it is out the scope of our current manuscript, which is focusing on the development of a simple, effective, and scalable method to synthesize an azide-functionalized "-ADP-ribose analog directly from native %-NAD + in clean ionic liquid systems, and also the following generation of ribosylated peptides with 'click chemsitry'. The identification of previously unknown ADP-ribosylation binding partners is one of our future directions to pursue. At the same time, we also hope the publication of our method can offer the community an easily accessible way to obtain chemical tools that match their own research interests in the study of protein ADP-ribosylation.
Response: We thank the reviewer for this comment. We have added the uncropped membranes together with protein markers for all the blots in Fig 4. as Figure S9 in the revised SI (also see Figure R8 below). Response: We thank the reviewer for this comment. As we have explained in the response for Point 4, the Glu2 position of histone H2B is a known endogenously ADP-ribosyalted site.
To determine whether and to what extent could our ADP-ribosylation mimic initiate the poly ADP-ribosylation, we performed the assay in an alternative as has been suggested by the Reviewer#3. Specifically, we incubated the peptide-1 carrying both a mono-ADP-ribosylation mark and a biotin tag with native NAD + and PARP1. The reaction mixture was taken at different time points and subjected to streptavidin blotting analysis. The result has been provided as Fig.   S10 in the revised supporting information (also see Figure R9 below), and it showed that the biotin signal expanded from the original position (~ 3 kDa) to the regions with higher molecular weight gradually, suggesting the generation of a mixture of heterogeneously poly-ADP-ribosylated species as time went by. After 1 hour of reaction, the biotin signal can be detected at a position with molecular weight higher than 20 kDa, indicating the growth of a poly-ADP-ribose chain containing at least 35 ADP units by a rough estimation (molecular weight of an ADP moiety is about 0.5 kDa).
In Fig. 4G, we examined if our ADP-ribosylation mimic could induce poly-ADP-ribosylation by incubating peptide-3 (no biotin tag) with biotinylated NAD + and PARP1. At the same time, samples with peptide-3 only, without adding PARP1, or using unribosylated histone H2B peptide were also prepared as controls. The result showed that no biotin signal was detected from each of the three control samples, suggesting that the installation of extra ADP units to the peptide-3 was indeed relied on the presence of the mono-ADP-ribosylation mark but not on other amino acid residues, and such processes was catalyzed by PARP1 instead of non-enzymatic chemical reactions.

Figure R9
Histone H2B peptide carrying the ADP-ribosylation mimic could serve as the substrate of poly-ADP-ribosylation catalyzed by PARP1.
Question-13: This statement needs references: "especially for those on glutamate residues, can be removed by multiple enzymes,"… Response: We thank the reviewer for this comment. During our revision, we found the sentence did not fit well with the context and thus removed it.
Question-14: Supplemental data: The 1 H NMR for the alpha and beta N3-ADPr contain some of the other isomer.
Response: We thank the reviewer for this comment. Based on the suggestions, we checked the 1 H NMR spectra of alpha-(1) and beta-(1') azides. The peaks at the range of 3.75 ppm to 1.0 ppm 3 ppm were assigned to -CH 2 N-and CH 3 CH 2 -of triethylammonium that exchanged from TEAB (triethylammonium bicarbonate) buffer. The peaks at 8.0 ppm to 9.0 ppm and 5.0 ppm to 6.5 ppm in both of the 1 H NMR spectra and H-H COSY spectrum were further analyzed based on the suggestion in Question-10 of Review#1. The two peaks at 8.0 ppm to 9.0 ppm suggested the impurities had a purine base. The correlations of the small peaks at 6.2 ppm and 5.3 ppm with the small peaks at the range of 3.5 ppm to 4.9 ppm suggested that the impurities including two riboses similar to ADPr unit ( Figure R3 below). Thus, we deduced that the impurity might come from the reactions of starting material %-NAD + . Measures of their molecular weight by MALDI were unsuccessful, however, these impurities could be conveniently removed by an additional HPLC purification, and the clean 1 H NMR spectra were updated in the revised supporting information, and also shown in below Figure R4. The manuscript by Li and co-workers describes the facile synthesis of an azido-analog of NAD, followed by the use of this compound to create triazole linked ADP-ribosylated peptide analogs for further biological studies. They then apply these peptides to demonstrate the enrichment of proteins that are known to bind to an ADP-ribosylated H2B peptide in vitro and from cell lysates. I feel that this approach is attractive and that the conclusions are mostly supported by the data. In particular, the synthesis of the aizdo-analog of NAD + , and other nucleotides, using ionic liquids is in particular quite unique and powerful. I also agree with the authors that the procedure seems easy enough, with minimal purification steps, to be performed by many labs that could then to on to use the modified peptides. However, I do have some comments that should be address in a revised manuscript.
Comment #1: Overall the paper is difficult to read due to numerous grammatical errors that need to addressed.
Response: We thank the reviewer for this comment. Based on the suggestions, we check the typographical errors carefully in manuscript and supporting information. The specialized language editing is conducted for the revised manuscript.
Comment #2: How were the stability of the ADP-ribosylation triazole-analogs of the peptides measured? I may have simply missed it, but this data should be added to the Supplementary Information.

Response:
We thank the reviewer for this comment. It is reported that ADP-ribosylation marks installed on different amino acid residues have varied stability. In our study, we used a triazolyl-ADP ribose linkage to mimic the ADP-ribosylation of glutamate or aspartate residues, which are highly sensitive to NH 2 OH treatment even at neutral pH (H. Hilz, Hoppe Seylers Z Physiol Chem, 1981, 362, 1415-1425P. O. Hassa et al., Microbiol Mol Biol Rev, 2006, 70, 789-829;J. Biol. Chem.,1983, 258:6466-6470;Zhang, Y. J. Zhang et al., Nat. Methods. 2013, 10, 981-984). To determine the stability of our ribosylated peptides, the peptide-3 was incubated in PBS buffer (pH 7.5) containing 0.5 M NH 2 OH at 37 o C. The mixture was analyzed by HPLC at different time points. The peptide remained intact even after 24-hour incubation. In addition, we also demonstrated that our ribosylated peptide was inert to acidic or basic treatments (pH = 3.0, 5.0, 9.0, 11.0) ( Figure S8 in the revised supporting information and below Figure R3). These results indicated a satisfactory stability of the triazoly-ADP ribose linkage as a mimic of the native ADP-ribosylated glutamate or aspartate.
To make it clear to the readers, we have added more information in the related sentences describing and discussing the experiment result in our revised manuscript as seen "In contrast to the natural ester-linked ADP-ribosylated peptides that are unstable in NH 2 OH treatment at neutral pH(t1/2 < 1 h), 39,41,42 all the peptides bearing triazoly-ADPr linkages were stable for at least 24 hours in NH 2 OH, as well as in acidic and basic conditions ( Figure S8 in the supporting information)." in Line 6 to Line 8 in the first paragraph in Page 7 highlighted in yellow.

Figure R3
Analysis of peptide-3 stability against NH 2 OH, acidic or basic treatments.
Comment #3: The ordering and "call-outs" for most of the panels in Fig. 4 seem to be confused at places in the text and the figure legend.
Response: We apologize for the mismatch of panels in Fig. 4 and our text in the original manuscript and the confusion it has caused. The errors have been corrected in the revised version.
Comment #4: In Figure 4F, the "truth-table" indicates that peptide 3 is missing from all of the conditions. This should be fixed.
Response: We thank the reviewer for this comment. In Fig. 4F, the last lane for peptide-3 should be "+". A right version of Fig. 4 has been attached to the revised manuscript.
Comment #6: The authors should test the poly-ribosylation of their analog further. They show in current Figure 4G that PARP1 can modify peptide 3 with biotinylated NAD. However, they do not appear to show how extensive this polymerization is. They should do this by labeling peptide 1 with regular NAD+ and showing the full molecular weight spread of the product by blotting for the biotin on peptide 1. This will give an idea of how long the poly-ADP ribosylation can become.
Response: We thank the reviewer for this comment. Following the reviewer's suggestion, we incubated the peptide-1 carrying both a mono-ADP-ribosylation mark and a biotin tag with native NAD + and PARP1. The reaction mixture was taken at different time points and subjected to streptavidin blotting analysis. The result has been provided as Figure S10 in the revised supporting information (also see Figure R9 below), and it showed that the biotin signal expanded from the original position (~ 3 kDa) to the regions with higher molecular weight gradually, suggesting the generation of a mixture of heterogeneously poly-ADP-ribosylated species as time went by. After 1 hour of reaction, the biotin signal can be detected at a position with molecular weight higher than 20 kDa, indicating the growth of a poly-ADP-ribose chain containing at least 35 ADP units by a rough estimation (molecular weight of an ADP moiety is about 0.5 kDa).

Figure R9
Histone H2B peptide carrying the ADP-ribosylation mimic could serve as the substrate of poly-ADP-ribosylation catalyzed by PARP1.
Comment #7: As a last characterization experiment, the authors should consider using peptide 1 to perform an unbiased proteomics experiment to demonstrate the utility of the peptides as a discovery-based technology.

Response:
We thank the reviewer for this comment. We agree with the reviewer that a comprehensive profiling of ADP-ribosylation binding proteins at the proteomic level is an important and interesting application of our ADP-ribosylation mimic. In our study, to test if our ribosylated photoaffinity probe could capture endogenous ADP-ribosylation binding proteins, lysate derived from HeLa S3 cells was cross-linked with peptide-1. After streptavidin enrichment, the eluted proteins were subjected to immunoblotting detection against the antibodies of two known ADP-ribosylation 'readers', mH2A1.1 and PARP9. The result showed that both endogenous proteins could be robustly and selectively enriched by peptide-1 from complex proteome, demonstrating the potential of our method in the identification of previously unknown ADP-ribosylation binding proteins. However, we think the performance of a proteomics study is out the scope of our current manuscript, which is focusing on the development of a simple, effective, and scalable method to synthesize an azide-functionalized "-ADP-ribose analog directly from native %-NAD + in clean ionic liquid systems, and also the following generation of ribosylated peptides with 'click chemsitry'. The identification of previously unknown ADP-ribosylation binding partners is one of our future directions to pursue. At the same time, we also hope the publication of our method can offer the community an easily accessible way to obtain chemical tools that match their own research interests in the study of protein ADP-ribosylation.

Reviewers' comments:
Reviewer #1 (Remarks to the Author): I thank the authors for their careful addressing of the points I raised in my previous review. It is clear that the time has been taken to carry out the relevant characterisations, and control experiments (which are interesting in themselves, so I look forward to the further reports of work in this area), and the language and details of the document have been substantially improved to a much more publishable standard.
There are still a few minor outstanding points: 1. For Question 10 of the original review, I did not make my point clearly. For what is now figures S11 and S13, it would be useful to highlight the key cross peaks that indicate one or other conformation, as per the description in the text (I think S10 should be S11 -and please cross check then the description for S14/(S13?) for consistency): !6D =IFH?<F @;<CH@=K H?< :DC=@>IF8H@DC D= L%M#*,3F#1)" J< FIC H?< /#/ +257 8C; ',# 62+57 8C; 1D-NOESY spectra below ( Figure S11 to Figure S14). Figure  2. Although it is pleasing to see the additional data for the 47 new ionic liquids, the standard data on purity//physical data is still not present -usually this would constitute microanalysis, melting points, and/or DSC/TGA. (and please, consistent chemdraw structures, especially not those with the negative charge embedded in the bond).
In each case, both for newly synthesised and existing ionic liquids, the appropriate multinuclear data should be provided. i.e. in addition to 1H, also 13C and where appropriate 31P, 19F, 11B.
3. In figure 2B the anion B-8 appears to be missing? There are also a couple of triazoly (vs triazolyl) remaining in the text. In Table S2 ILs #37, 39 The authors have proactively addressed most of my concerns, and I am satisfied that proteomics is outside of the scope of this study due to the significant amount of synthetic chemistry. I do still have some questions concerning the polymerization of their ADP-ribosylation analog into poly-ADP-ribose ( Figure S10). It appears to me that the majority of the polymerization is quite limited, as the major band that appears is only slightly higher in molecular weight. Do the authors have an estimation for how many units are added there? The authors should clearly indicate this in Figure  S10. The extent of further elaboration seems low in comparison. The authors should clearly indicate in the text of the manuscript the what they think the major band is and that the further elaboration is low.

Point-by-9KGJP ;COLKJOCO PK PFC ;CRGCSCNOU AKIICJPO
We greatly appreciate the comments of the reviewers on our manuscript. Based on these comments, we have performed supplementary experiments. Thermal analysis and microanalysis of water contents of all the 47 new ionic liquids were performed. The supplementary physical and purity data from the experiments were added into the revised supporting information (Table R1 and Table R2 below, also see Table S8 and S9 in the revised Supporting Information). The experiments to detect poly ADP-ribosylations using commercial anti-poly(ADP-ribose) polymer antibody were conducted. The result showed that the region with robust poly ADP-ribosylation signal gradually expanded throughout the whole lane with increasing intensity as the reaction time became longer ( Figure R3 below, also see Figure S10 in the revised Supporting Information), suggesting the generation and growth of poly ADP-ribose chain on peptide-1. We have also reorganized the corresponding context based on the suggestions of reviewers and checked the typographical errors carefully in manuscript and supporting information. All the revised sentences were marked with yellow in the revised manuscript. The details are as follows.
Par t-I : Response to r eviewer #1 Reviewer #1 (Remarks to the Author): I thank the authors for their careful addressing of the points I raised in my previous review. It is clear that the time has been taken to carry out the relevant characterizations, and control experiments (which are interesting in themselves, so I look forward to the further reports of work in this area), and the language and details of the document have been substantially improved to a much more publishable standard.
There are still a few minor outstanding points: Question-1: For Question 10 of the original review, I did not make my point clearly. For what is now figures S11 and S13, it would be useful to highlight the key cross peaks that indicate one or other conformation, as per the description in the text (I think S10 should be S11 -and please cross check then the description for S14/(S13?) for consistency): #Nd [jgi]Zg^YZci^[n i]Z Xdc[^\jgVi^dc d[ z-{-ADPr-N3, we run the H-H COSY and 1D-TOCSY and 1D-NOESY spectra below ( Figure S11 to Figure S14). Figure S12 is 1D-TOCSY and 1D-NOESY spectra of z-ADPr-N3. As shown in Figure S10 Figure S11 to Figure S14). Figure S12 is 1D-TOCSY and 1D-NOESY spectra [S n-ADPr-N3. Irradiation at the resonance frequency of H-1 produced significant NOEY correlations with H-2 (the peak was marked with green asterisk in Figure 12(D)) and H-3 (the peak was marked with red asterisk in Figure 12 Question-2: Although it is pleasing to see the additional data for the 47 new ionic liquids, the standard data on purity//physical data is still not present -usually this would constitute microanalysis, melting points, and/or DSC/TGA. (and please, consistent chemdraw structures, especially not those with the negative charge embedded in the bond). In each case, both for newly synthesised and existing ionic liquids, the appropriate multinuclear data should be provided. i.e. in addition to 1H, also 13C and where appropriate 31P, 19F, 11B.
Response: We thank the reviewer for this comment. Firstly, we redrew the structures of ionic liquids in the copies of 1 H NMR and 13 C NMR to make them clear and keep consistent with structures shown in other Tables. Secondly, we run the 31 P NMR, 19 F NMR and 11 B NMR for both of newly synthesized and existing ionic liquids containing these atoms. The multinuclear data of the samples were provided in the revised Supporting Information which were highlighted in yellow (Page S26 to Page S41). Thirdly, we supplement the physical data for the 47 new ionic liquids in the revised supporting information (Table S1 and Table S2, also see Table S8 and Table S9 in the revised Supporting Information). Thermal analysis of 47 new ionic liquids was performed on a Netzsch STA 449 F5 using the Proteus Analysis software. The TGA data were collected at 10 K/min under N2 (Range was 20 o C to 400 o C). The differential scanning calorimetry (DSC) data were obtained simultaneously with the TGA data. Melting point temperatures were reported from DSC data. Onset exothermic temperatures reported from TGA data as determined from the step tangent. All the characteristic data of the ionic liquids including the onset exothermic temperature (Tonset) and melting point temperature (Tmp) were collected as shown in Table R1 (Table S8 in the revised Supporting Information) below. The copies of DSC/TGA curves of the 47 ionic liquids were attached at the end of this letter. For the evidences of sample purity, we firstly analyzed the 1 H NMR and 13 C NMR of 47 new ionic liquids (the copies of the spectra were contained from Page S57 to Page S107 in the Supporting Information) to confirm the 1:1 ratio of cations and anions, and no residual impurities. The ionic liquids were prepared by acid-base neutralization. Water are the sole by-products. The microanalysis of water contents of all the 47 ionic liquids was then performed by using Karl Fischer electrometric titration method after they were prepared and dried by the standard methods. Water contents of all the ionic liquid were less than 0.5%, the data were shown in Table R2 below (also see in Table S9 in the revised Supporting Information). Question-4: Also, the now titled discussion is, as it was in the previous draft, a conclusion. I presume this has been changed to meet the general format of Nature, but this doesn't sit right as it stands. More effort needs to be made to add or move (from the results) discussive elements to this section, before ending with the concluding statements.
Response: We thank the reviewer for this comment. Based the suggestions, we re-organize the part of discussion carefully. The revised context includes a brief summary of the content of this work, and a discussive content about the significance of our synthetic chemistry in promoting biological researches of protein ADP-ribosylation. They were also shown below.  The authors most of my comments and concerns; however, there are a few things that still need to be addressed. These include: Question-1: JV\Z 7* hZXi^dc qJdan ;>J-g^WdhnaVi^dc XViVanoZY Wn J;LJ/r8 i]Z Zk^YZcXZ i]Vi eZei^YZ-3 is poly-ADP-ribosylated is indirect. The authors should use one of the several commercially available poly-ADP-ribose antibodies to show that peptide-3 is indeed poly-ADP-ribosylated. Additionally, while I agree that the data support the notion that mono-ADP-ribosylation is required for PARP1-dependent ADP- Response: We thank the reviewer for this comment. As suggested by the reviewer, we purchased one antipoly(ADP-ribose) polymer antibody that has been used to detect poly ADP-ribosylation in many related studies from Abcam (ab14459). We performed the poly ADP-ribosylation reaction by incubating the peptide-1 with native NAD + and PARP1. The reaction mixture was taken at different time points and subjected to immunoblotting analysis against the anti-poly(ADP-ribose) polymer antibody. As has been shown in Figure R3 below, intense luminescence signals were observed in high molecular weight region of the blot even at 0 min. We speculated that the strong signals came from the auto-poly ADP-ribosylation of PARP1 (see for example: I. Kameshita et al., J Biol Chem, 1984, 259, 4770-4776;M. Altmeyer et al., Nucleic Acids Res, 2009, 37, 3723-3738;Z. Tao et al., J Am Chem Soc, 2009, 131, 14258-14260;J. D. Chapman et al., J Proteome Res, 2013, 12, 1868-1880. To test this hypothesis, we splitted the 10 min reaction mixture into two samples. To one sample, SDS loading buffer was directly added. For another one, we tried to remove PARP-1 from the mixture by enriching the biotin-tagged peptide-1 with streptavidin-coated magnetic beads (Dynabeads, Invitrogen, 11206D). The enriched peptides were eluted by boiling the beads in SDS loading buffer. Immunoblotting of the two samples against anti-poly(ADP-ribose) polymer antibody ( Figure R3 below) revealed that the intense luminescence at high molecular weight region could no longer be observed after Dynabeads enrichment. At the same time, signal, although weaker, spreading the whole lane stood out, suggesting the existence of different poly ADP-ribosylated species. Encouraged by this result, we repeated the poly ADP-ribosylation reaction. Samples taken at each time point were subjected to Dynabeads enrichment before immunoblotting analysis. The result ( Figure R3 below, also see Figure S10 in the revised Supporting Information, replacing the previous Figure S10) showed that, as the reaction time became longer, the region with robust poly ADP-ribosylation signal gradually expanded throughout the whole lane with increasing intensity, suggesting the generation and growth of poly ADP-ribose chain on peptide-1. We hope to immunoblotting analysis against anti-poly(ADP-ribose) polymer antibody. Sample with only Dynabeads was prepared as control.  The authors have proactively addressed most of my concerns, and I am satisfied that proteomics is outside of the scope of this study due to the significant amount of synthetic chemistry. I do still have some questions concerning the polymerization of their ADP-ribosylation analog into poly-ADP-ribose ( Figure S10). It appears to me that the majority of the polymerization is quite limited, as the major band that appears is only slightly higher in molecular weight. Do the authors have an estimation for how many units are added there?
The authors should clearly indicate this in Figure S10. The extent of further elaboration seems low in comparison. The authors should clearly indicate in the text of the manuscript the what they think the major band is and that the further elaboration is low.
Response: We thank the reviewer for this comment. We fully understand the gZk^ZlZgsh concern on the extending of the poly ADP-ribose chain by PARP1. To specifically detect the poly ADP-ribosylation, we purchased one anti-poly(ADP-ribose) polymer antibody (Abcam, ab14459). We performed the poly ADPribosylation reaction by incubating the peptide-1 with native NAD + and PARP1. The reaction mixture was taken at different time points and subjected to immunoblotting analysis against the anti-poly(ADP-ribose) polymer antibody. As has been shown in Figure R3 below, intense luminescence signals were observed in high molecular weight region of the blot even at 0 min. We speculated that the strong signals came from the auto-poly ADP-ribosylation of PARP1 (see for example: I. Kameshita et al., J Biol Chem, 1984, 259, 4770-4776;M. Altmeyer et al., Nucleic Acids Res, 2009, 37, 3723-3738;Z. Tao et al., J Am Chem Soc, 2009, 131, 14258-14260;J. D. Chapman et al., J Proteome Res, 2013, 12, 1868-1880. To test this hypothesis, we splitted the 10 min reaction mixture into two samples. To one sample, SDS loading buffer was directly added. For another one, we tried to remove PARP-1 from the mixture by enriching the biotin-tagged peptide-1 with streptavidin-coated magnetic beads (Dynabeads, Invitrogen, 11206D). The enriched peptides were eluted by boiling the beads in SDS loading buffer. Immunoblotting of the two samples against anti-poly(ADP-ribose) polymer antibody ( Figure R3 below) revealed that the intense luminescence at high molecular weight region could no longer be observed after Dynabeads enrichment. At the same time, signal, although weaker, spreading the whole lane stood out, suggesting the existence of different poly ADP-ribosylated species.
Encouraged by this result, we repeated the poly ADP-ribosylation reaction. Samples taken at each time point were subjected to Dynabeads enrichment before immunoblotting analysis. The result ( Figure R3 below, also see Figure S10 in the revised Supporting Information, replacing the previous Figure S10) showed that, as the reaction time became longer, the region with robust poly ADP-ribosylation signal gradually expanded throughout the whole lane with increasing intensity, suggesting the generation and growth of poly ADPribose chain on peptide-1. The presence of poly ADP-ribose signal at high molecular region clearly indicated that even hundreds of ADP-ribose units could be added to the peptide by PARP1.
In our previous result detected by streptavidin-biotin blot, is was true that the major band had only slight mass shift compared with the original peptide-1. One possible reason could be that the short H2B peptide was a less favored substrate for PARP1 compared with an intact protein, and the PARP1 enzyme showed a high efficiency in auto-modification ( Figure R3 A and B). When PARP1 was removed from the mixture after the reaction, and the readout was changed to poly ADP-ribose signal itself instead of biotin signal, we could indeed observe the elaboration of the poly ADP-ribosylation reaction. To make the information simple

REVIEWER COMMENTS
Reviewer #1 (Remarks to the Author): I thank the authors for addressing the key comments previously.
For Q1. The cross peaks have been marked on S12/14, but not, as requested on S11/13 (the H-H COSY spectra). I think this has arisen due to changing numbering throughout submissions. The authors may still consider this for clarity.
For Q2. This has been addressed as requested, with the exception that microanalysis was meant as elemental analysis (%CHN), not water-content analysis (these numbers are useful, however). This helps to identify other impurities not detectable by the spectroscopic methods used; as an alternative a halide analysis might be appropriate.
For Q4. A simpler solution is probably to rename Results to Results and Discussion and Discussion to Conclusion if this is allowed in the Nature format, as these two paragraphs still primarily conclude the paper, and the literature context is provided with the current Results.