The conformational changes coupling ATP hydrolysis and translocation in a bacterial DnaB helicase

DnaB helicases are motor proteins that couple ATP-hydrolysis to the loading of the protein onto DNA at the replication fork and to translocation along DNA to separate double-stranded DNA into single strands during replication. Using a network of conformational states, arrested by nucleotide mimics, we herein characterize the reaction coordinates for ATP hydrolysis, DNA loading and DNA translocation using solid-state NMR spectroscopy. AMP-PCP is used as pre-hydrolytic, ADP:AlF4− as transition state, and ADP as post-hydrolytic ATP mimic. 31P and 13C NMR spectra reveal conformational and dynamic responses to ATP hydrolysis and the resulting DNA loading and translocation with single amino-acid resolution. This allows us to identify residues guiding the DNA translocation process and to explain the high binding affinities for DNA observed for ADP:AlF4−, which turns out to be optimally preconfigured to bind DNA.

I could not find table S1, ref. 19 refers to 25% assigned residues and ref. 20 deals with a completely unrelated protein.
In addition, what is the motivation to assume that the apo assignments can be transferred to the other states? Previous structural work predicts significant structural rearrangements that could strongly compromise such a strategy. The SI stating that "The assignments of HpDnaB are taken from reference 11 and extended in this work (see caption Table S2)" isn't really helpful either since the cited reference refers to the NTD and table S2 refers to 31P assignments. Hence, it is impossible to check and evaluate which and how many assignments have been obtained. I find this unacceptable and these data should be made available -at least for the reviewing process.
[2] Data analysis. Figure 3 reports chemical-shift changes for a subset of residues (130-450) between different states. Yet-the analysis is entirely based on the apo state. Indeed, chemical shift changes such as seen for the nucleotide binding region makes sense. Unfortunately, the rest of the discussion as well as the conclusions at the bottom of page 6 are difficult to understand.
For example, the authors write: "While C SPs associate with conformational changes, appearing/disappearingresonances thus associate with dynamic changes, and we equate appearing resonances with a stiffening of the corresponding residues.The large number of de novo peaks observed here cannot be exclusively explained by large shifts, as this would involve an equal number of disappearing residues, which is not the case" What about large structural rearrangements that lead to chemical-shift changes that have nothing to do with motion? Again, further information of the level of assignments is critical to draw reliable conclusions for the presented data. Did the authors conduct experiments at variable temperatures?
2nd example: "This is most extreme for the NTD, not shown in the structural representation ( Figure 2, 4th column), of which all resonances surprisingly disappear from the apo spectra upon AMP-PC P and ADP binding, but not for ADP:AlF4- (Figure 2, 2nd column). We attribute this effect, currently subject to further studies, to changes in the dynamics of the NTD upon AMP-PC P and ADP binding." It this is the case -why did the authors not present an analysis of the NTD? The changes in the C TD are indeed rather small and mostly related to ligand binding. 3rd example: "...interestingly, in all nucleotide-bound states a major change observed upon DNA binding is the appearance of residues located in or near the DNA binding loops ... This indicates that this part of the protein stiffens upon DNA binding and shows the tight coupling between the motor domain and the DNA binding loops".
Again -can you exclude structural as opposed to dynamic changes that can explain the appearance of new peaks? Why did the authors not attempt to actually probe the dynamics? -by for example measuring dipolar order parameters or by conducting relaxation studies ?
[3] Lastly, it is difficult to deduce from Figure 5 the novel aspects of this work compared to previous X-ray and EM work. Please clarify and provide a simpler figure. What do the authors mean by "conformational events"? Also, using ATP mimetics is a widely applied strategy in X-ray, C ryo-EM and even NMR (See e.g. recent work by Glaubitz et al., Nat C omm) and should be appropriately cited.
Reviewer #3: Remarks to the Author: In the current manuscript, Wiegand et al. use solid state NMR to probe the structural response of the H. pylori DnaB helicase to different nucleotides and the binding of ssDNA. The authors find that the dynamics of different DnaB regions change depending on which type of ATP analogue is used and whether DNA is present. ADP-AlFx appears to particularly affect the pore loops of DnaB, which are responsible for translocation; biochemical analysis indicates that this analogue stabilises a state of the helicase that favors the tight binding of DNA.
Although the objective and approach of the study is of interest, the execution unfortunately falls short. The presentation of the work is overly technical and presumes a deep familiarity with solid state NMR and how to interpret data derived from such studies. There is little discussion of current views in the field as to how DnaB binds DNA and/or nucleotide, or how the current findings relate to such studies; there likewise are missed opportunities for expanding on certain aspects of DnaB dynamics or in testing some ideas generated by the data. Perhaps the most problematic issue is the assumption that the DnaB hexamer responds in a concerted, all-or-none manner as ATP is hydrolysed. It has been shown that the vast majority of ring-ATPases (going all the way back to the F1 ATPase) do not work by such a mechanism, and that different subunits in the hexamer instead adopt different ATPase state simultaneously. How the current data might be interpreted from this perspective is not discussed.
In summary, although the effort needed to collect the data presented here is well appreciated, the insights gained into DnaB mechanism are either largely confirmatory (e.g., that DnaB binds 2 nt/subunit, that pore residue dynamics change when DNA is present) or based on an overly simplistic view of the helicase's ATPase cycle. Publication in a more specialised journal would seem more appropriate.
Specific comments 1) The role of two residues (R357 and K373) in DNA translocation needs to be tested by both DNA binding and helicase assays.
2) The idea all subunits in a DnaB hexamer transit through the same ATPase states in synchrony with each other is likely incorrect. In crystallographic and EM-based studies of ring ATPases, there is clear pseudosymmetry in the system that is not reflected in the scheme shown for Fig. 1b or in the main text. This pseudosymmetry may well be quite difficult to detect by solid state NMR yet is probably a critical component of the ATPase cycle.
3) The authors seem to be missing out on an opportunity to expand and comment on the dynamics of the DnaB NTD, and how different regions of the NTD (the globular domain, the helical hairpin) respond to ATPase/DNA status. The mobility of this region and its response during DnaB cycling is only beginning to be understood 4) AMPPC P may or may not be a good ATP analogue. The absence of a partial negative charge on the bridging methylene group, as well as non-ideal bond angles at this position, can make AMPPC P serve as more of a product state (ADP-Pi) mimic in many cases, particularly on long time scales (minutes). Additional data are needed to show that AMPPC P really is ATP-like.
5) The data concerning the effects of ADP-AlFx would seem to accord well with structural data from the Steitz group; alternatively, is it possible that this analogue is not capturing a translocation state, but instead locking the helicase in a 'rigor' state that lies off-pathway to translocation? Please comment.
Minor points P. 9. The assumption seems to be that all Lys373 residues will respond identically to a change in nucleotide state. This seems highly unlikely. P. 9. Mn2+ is not a good substitute for Mg2+ for most ATPase reactions -it perturbs the hydrolysis reaction. P. 12. Is the prediction that Arg357 and Lys373 bind to DNA borne out by the crystal structure from the Steitz group?  It seems implausible that the authors can confidently assign specific amino acids to the resonance 'blob' shown in the lower right of the spectra panels.
The title for the legend for Fig. 2 is misleading -the data do not show that ATP hydrolysis occurs in the absence of DNA; no ATP is used and no release of product is followed. Fig. S3. It is not obvious how the spectra shown support the claim of the legend title, namely that "Solid-state NMR allows to distinguish different states of ATP hydrolysis". No hydrolysis reaction is being followed in these measurements.

Reviewer #1 (Remarks to the Author):
I recommend the publication of this work as is.

Reviewer #2 (Remarks to the Author):
The authors should address the following (minor) issues before acceptance: [1] Assignments. Obviously, the current paper heavily relies on NMR assignments of the CTD domain. However, the information given in the ms on how these assignments were obtained is confusing and the actual values seem to be missing: The main text states on page 5: "Each spectrum is overlaid on the previous one in the cycle. Sequential assignments of the spectra were done, where possible, by transfer from the one of the apo state19, and were extended by further 3D experiments20 as described in Table S1.
I could not find table S1, ref. 19 refers to 25% assigned residues and ref. 20 deals with a completely unrelated protein.
In addition, what is the motivation to assume that the apo assignments can be transferred to the other states? Previous structural work predicts significant structural rearrangements that could strongly compromise such a strategy.
Answer: Unfortunately, the reviewer did not obtain Tables S1 and S3 which (as requested for long Tables) were submitted separately. They are now called Supplementary Datasets 1 and 2 which are part of this submission and will also be made available for the readership. We have also inserted, following the reviewer's advice, a new Table (Supplementary Table 1) in the new numbering) with assignment statistics for all protein-complexes studied in this work. All the CSPs are listed in the Supplementary Dataset 2 also all the chemical-shift values allowing detailed tracking. Reference 23 (former reference 20) discusses the assignment strategy used. We have also clarified in the manuscript that we have significantly extended the assignment of (former) Ref 19 and have obtained the de novo sequential assignments on the apo protein as well as on the ADP:AlF 4 -:DNA complex. The newly assigned apo resonances are now also listed in Supplementary Dataset 1.
The SI stating that "The assignments of HpDnaB are taken from reference 11 and extended in this work (see caption Table S2)" isn't really helpful either since the cited reference refers to the NTD and table S2 refers to 31P assignments. Hence, it is impossible to check and evaluate which and how many assignments have been obtained. I find this unacceptable and these data should be made available -at least for the reviewing process.
Answer: This is again caused by the fact that the reviewer did not obtain the info of Tables S1 and S3 (now Supplementary Datasets 1 and 2). Requested information is found in Supplementary datasets 1 and 2 as well as in Supplementary Table 1 for all constructs investigated. Note that the old Table S2 is  now Supplementary Table 5.
[2] Data analysis. Figure 3 reports chemical-shift changes for a subset of residues (130-450) between different states. Yet-the analysis is entirely based on the apo state. Indeed, chemical shift changes such as seen for the nucleotide binding region makes sense. Unfortunately, the rest of the discussion as well as the conclusions at the bottom of page 6 are difficult to understand.
Answer: As detailed above, two forms, apo and DnaB:ADP:AlF 4 -:DNA complex were sequentially assigned. This is now clearly stated in the manuscript and the reviewer has hopefully obtained the corresponding tables (now called Supplementary Datasets). The spectra of the other protein:ATPmimic and protein:ATP-mimic:DNA complexes were assigned using the data of these two forms. We have also reformulated the text on the bottom of page 6.
For example, the authors write: "While CSPs associate with conformational changes, appearing/disappearingresonances thus associate with dynamic changes, and we equate appearing resonances with a stiffening of the corresponding residues.The large number of de novo peaks observed here cannot be exclusively explained by large shifts, as this would involve an equal number of disappearing residues, which is not the case". What about large structural rearrangements that lead to chemical-shift changes that have nothing to do with motion? Again, further information of the level of assignments is critical to draw reliable conclusions for the presented data. Did the authors conduct experiments at variable temperatures?
Answer: We have discussed this point already in the original manuscript, but it was not clearly enough emphasized that the CSPs were obtained from 3D spectra not from 2D as shown in the Figures 2-4. The overlap is thus much lower as in the 2D spectra of the illustrations. Second, we argue here with the sheer number of disappearing resonances. It is highly unlikely that they all hide (in 3D!!) under peaks from other residues without leaving a clue. In particular, many of them belong to residues in loop regions (e.g. the Walker motifs or the DNA binding loops) and can thus be found in characteristic spectral regions. We have adapted the text in the middle of page 6 to make these points clearer. Temperature-dependent measurements were performed for the , but unfortunately the disappearing peaks (when comparing to ADP:AlF 4 -) could still not be observed. At the lower temperature, considerable line broadening is observed making interpretation difficult.
2nd example: "This is most extreme for the NTD, not shown in the structural representation ( Figure  2, 4th column), of which all resonances surprisingly disappear from the apo spectra upon AMP-PCP and ADP binding, but not for ADP:AlF4- (Figure 2, 2nd column). We attribute this effect, currently subject to further studies, to changes in the dynamics of the NTD upon AMP-PCP and ADP binding." It this is the case -why did the authors not present an analysis of the NTD? The changes in the CTD are indeed rather small and mostly related to ligand binding.
Answer: Indeed, the mobility of this region is only beginning to be understood, and we have added in this work the important new element that the N-terminal domain can change its dynamic state as a function of nucleotide binding. We agree that this is indeed a highly interesting and new finding, and we are currently looking at this point in detail. Still, variable temperature experiments will not only make reappear the N-terminal domain, but also show the C-terminal domain, with the concomitant loss in resolution typically observed in NMR at sub-freezing temperatures. We have thus devised segmental isotope labelling of the NTD of DnaB in the full-length construct to uniquely observe the N-terminal domain (Wiegand et al., J. Biomol. NMR, 2018). Preliminary data exist, but this is a complex issue that warrants separate publication also because the function of the NTD is different from the functions discussed here (DNA binding and translocation).
3rd example: "...interestingly, in all nucleotide-bound states a major change observed upon DNA binding is the appearance of residues located in or near the DNA binding loops ... This indicates that this part of the protein stiffens upon DNA binding and shows the tight coupling between the motor domain and the DNA binding loops".
Again -can you exclude structural as opposed to dynamic changes that can explain the appearance of new peaks? Why did the authors not attempt to actually probe the dynamics? -by for example measuring dipolar order parameters or by conducting relaxation studies ?
Answer: It would be interesting to perform such studies, but the dynamic residues (even when lowering the temperature drastically, see above) are just invisible in the spectra, precluding such measurements.
[3] Lastly, it is difficult to deduce from Figure 5 the novel aspects of this work compared to previous X-ray and EM work. Please clarify and provide a simpler figure. What do the authors mean by "conformational events"?
Answer: We changed "conformational events" by "conformational changes". We have simplified the figure and have moved part of the info to Supplementary Figure 11. In response to referee 3 we also provide the labelling of the function directly on the figure.
Also, using ATP mimetics is a widely applied strategy in X-ray, Cryo-EM and even NMR (See e.g. recent work by Glaubitz et al., Nat Comm) and should be appropriately cited.

Answer:
We agree and have added a reference in the manuscript, referring to the solid-state NMR work of Glaubitz and coworkers on ABC transporters (reference 15) and have added a further reference 16 regarding ATP-mimics (Bagshaw, C.R. ATP analogues at a glance. J. Cell Sci. 114, 459-460 (2001)).

Reviewer #3 (Remarks to the Author):
Specific comments 1) The role of two residues (R357 and K373) in DNA translocation needs to be tested by both DNA binding and helicase assays.

Answer:
We have expressed both the single K373A and double R357A K373A mutants, and while proteins expressed well, they could not be purified using the protocols of the WT protein, as they likely aggregated on the column (see Figure of gels and purification on Q column).

5
In order to obtain additional direct evidence for the interaction between the DNA and K373, we recorded a NHHP spectrum on the DnaB:ADP:AlF 4 -:ssDNA sample. The spectrum was recorded at 11.74 T at 17.0 kHz magic-angle spinning (measurement time 14 days). The spectrum clearly shows, by the presence of the peak labelled 373K, that the side-chain of Lys373 N contacts one of the two structurally distinct phosphate groups of the DNA, which in turn is in close proximity to amide nitrogens from neighboring residues D374, S375 and G376. This spectrum is a twodimensional phosphorous-nitrogen correlation spectrum, in which the polarization evolves during a first period of time on the nitrogen spins, after cross polarization from the protons. The typical 15 N frequencies of the spins can, after Fourier transform, be seen on the y axis. As K373 N has a unique 15 N chemical shift, as determined by sequential assignments using 3D spectroscopy, it can be unambiguously identified. Even if the chemical shifts of the amide nitrogen spins of the neighbouring residues D374, S375 and G376 are not unique, the signals are fully consistent with an assignment to these amide nitrogens adding further evidence. After the transfer to the phosphorous spins by the use of adequate radiofrequency pulse sequences, the polarization is detected on the phosphorous spins, and in the second dimension (x-axis), one can read the 31 P frequency, of two different phosphate spins, of which only one is in close proximity (ca. 3-4 Å) to K373 N , since only the 31 P around 0.5 ppm shows a cross signal with it. The other 31 P at -1 ppm shows a correlation to D374, S375 and G376, and is thus not located far from the first one, but does not contact it directly. We have similar data for R357 for which we used fast magic-angle spinning (MAS) (> 100 kHz) to observe contacts between the arginine side chains and the DNA base. Fast MAS is a new approach which allows to observe proton resonances in solid-sate NMR, remnant of solution NMR, through the averaging of the strong homonuclear dipolar interactions between protons. We are afraid that the inclusion of all these additional details would make the manuscript overly technical, as the referee already remarked. We thus refrain from including them in this contribution in technical details but just mention the presence of transfer and the experiments (page 9).
2) The idea all subunits in a DnaB hexamer transit through the same ATPase states in synchrony with each other is likely incorrect. In crystallographic and EM-based studies of ring ATPases, there is clear pseudosymmetry in the system that is not reflected in the scheme shown for Fig. 1b or in the main text. This pseudosymmetry may well be quite difficult to detect by solid state NMR yet is probably a critical component of the ATPase cycle.
Answer: NMR chemical shifts can normally detect pseudosymmetry in a much more sensitive manner than EM and x-ray at the resolutions obtained in the here-discussed objects. We have recently shown this at the example of the HBV capsid, which shows four distinct molecules in the unit cell, and where NMR detected the subtle structural differences in an atom-specific manner as chemical shift multiples in the spectra (Lecoq et al., Chembiochem 2018). Our data on DnaB do not indicate any resonance splitting (neither in 13 C or 31 P spectra) pointing to (i) a complete occupation of all NBDs and (ii) a quasi-equivalence of all six NBDs. This indeed contrasts with the findings for BstDnaB:DNA of Steitz et al., and might indicate that we look at a slightly different state. We have commented on that point in the Discussion section of the manuscript.
3) The authors seem to be missing out on an opportunity to expand and comment on the dynamics of the DnaB NTD, and how different regions of the NTD (the globular domain, the helical hairpin) respond to ATPase/DNA status. The mobility of this region and its response during DnaB cycling is only beginning to be understood.
Answer: Indeed, the mobility of this region is only beginning to be understood, and we have added in this work the important new element that the NTD can change its dynamic state as a function of nucleotide binding. We agree that this is indeed a highly interesting and new finding, and we are currently looking at this point in detail. Still, variable temperature experiments will not only make reappear the N-terminal domain, but also show the C-terminal domain, with the concomitant loss in resolution typically observed in NMR at sub-freezing temperatures. We have thus devised segmental isotope labelling of the NTD of DnaB in the full-length construct to uniquely observe the N-terminal domain (Wiegand et al., J. Biomol. NMR, 2018). Preliminary data exist, but thorough analysis thereof and its understanding in terms of the structural aspects are underway. The dissemination of what NMR can contribute to the central dynamics question of the N-terminal domain will be done as soon as possible. We have added a sentence to put this in perspective in our discussion (bottom of page 11) 4) AMPPCP may or may not be a good ATP analogue. The absence of a partial negative charge on the bridging methylene group, as well as non-ideal bond angles at this position, can make AMPPCP serve as more of a product state (ADP-Pi) mimic in many cases, particularly on long time scales (minutes). Additional data are needed to show that AMPPCP really is ATP-like.
Answer: It is certain that AMPPCP is the most ATP-like of the three different substitutes we studied, since both alternatives investigated, AMP-PNP and ATPγS, got, on the timescale of NMR rotor filling (~16 hours) hydrolysed by DnaB in the presence of DNA (see Wiegand et al., Angew. Chem. In. Ed., 2016; a similar observation was made for a kinesin-type motor protein, see Suzuki, Y., Shimizu, T., Morii, H. & Tanokura, M. Hydrolysis of AMPPNP by the motor domain of ncd, a kinesin-related protein. FEBS Lett. 409, 29-32 (1997). It is an ongoing discussion if mimics are good mimics and which mimics are the best mimics, and all structural techniques are slave to this. As long as there are no highest-resolution X-ray structures for all different mimic states of a protein, or bond lengths and torsion angles are measured by other techniques, this question is hard to address. In this light, we agree that it is hard to find good mimics for the ATP-bound state, and that the best choice of analogue seems to depend very much on the system under study. We have included a discussion in the manuscript.
5) The data concerning the effects of ADP-AlFx would seem to accord well with structural data from the Steitz group; alternatively, is it possible that this analogue is not capturing a translocation state, but instead locking the helicase in a 'rigor' state that lies off-pathway to translocation? Please comment.
Answer: All mimics are by definition off-pathway to some degree, the only question remaining how far. There are no clear answers to this, as the pathway has not been characterized, and cannot with present methods, in detail. A strong indication that especially the ADP-AlF4 state is on-pathway is that it pre-organizes the helicase already in the absence of DNA binding in a conformation very close to the DNA-bound state. This might theoretically be pure coincidence, but very likely is not. We have also used the conditions Steitz et al. used (GDP:AlF4-) and obtained nearly the same state than for ADP:AlF4-. NMR can actually assess the typical geometry of 27 Al and 19 F spins, and we have done NMR experiments supporting that AlF4 adopts a (slightly distorted) squared-planar conformation as expected for the transition state of ATP-hydrolysis (to be published in a more technical Journal). Again, all NBDs are occupied with ADP:AlF4 and show quasi-equivalence. It might well be that we are pushing the helicase artificially into certain conformations by using a large excess of ATP mimics. Again, this is valid for all structural studies, and we have mentioned that point in the discussion of our manuscript.
Minor points P. 9. The assumption seems to be that all Lys373 residues will respond identically to a change in nucleotide state. This seems highly unlikely.
Answer: Our spectra indicate that resonances from all six K373 move during this process. No signal remains at the original position. Since we force every NBD to be occupied with the same ATP-mimic, this might be related to the sample preparation. Still, we note that our experimental conditions are close to those used in x-ray studies, i.e. ( (2012), see also the SI) and in that work also all six DNA binding loops behave similarly (e.g. R381 coordinates the phosphate of the DNA, see below). We have added a discussion of this point at the end of our manuscript. P. 9. Mn2+ is not a good substitute for Mg2+ for most ATPase reactions -it perturbs the hydrolysis reaction.
Answer: Mn 2+ substitution can in some cases lead to reduced functionality, which is however not the case for HpDnaB (see reference 30: Soni, R.K., Mehra, P., Choudhury, N.R., Mukhopadhyay, G. & Dhar, S.K. Functional characterization of Helicobacter pylori DnaB helicase. Nucleic Acids Res. 31, 6828-6840 (2003).). Mg 2+ substitution with manganese leads only to a reduction in DNA unwinding of ~20 % indicating that Mn 2+ substitutes Mg 2+ relatively well in our case. We have clarified this in the manuscript. P. 12. Is the prediction that Arg357 and Lys373 bind to DNA borne out by the crystal structure from the Steitz group?
Answer: Initially these residues have indeed been identified as good candidates for DNA binding based on sequence alignment and homology modelling with existing structures (see also reference 27: Stelter, M. et al. Architecture of a Dodecameric Bacterial Replicative Helicase. Structure 20, 554 (2012)). In our work, we identified these residues directly in NMR experiments to receive polarization transfer from the DNA (see answer to "Specific comment 1" from Referee 3).  Answer: Thanks for the suggestion, we have done so.
8 Figure S2. This figure does not make sense as described. Was ATP only added to DnaB (green) or a mix of ATP and ADP? Also, what exactly is the time scale of the experiment? Finally, how can all the ATP be hydrolysed by the enzyme -are DnaB and ATP present at near equimolar concentrations (and allowed to incubate for hours)?
Answer: We have clarified this point in the legend of Supplementary Figure 2. For the blue spectrum, only ADP was added for the green spectrum exclusively ATP. The incubation was performed for 2 h, the sedimentation takes 16 h. But even in the NMR rotor, the samples are fully hydrated and therefore ATP hydrolysis continues until all ATP is used up. This is what we observe in our spectra. 0.3 mM DnaB and 5 mM ADP or ATP respectively were used. Fig. 2. It seems implausible that the authors can confidently assign specific amino acids to the resonance 'blob' shown in the lower right of the spectra panels.
Answer: All resonance assignments are based on a sequential walk based on three-dimensional NCACB, NCACX, CANCO, NCOCX as well as NcoCACB and CANcoCA experiments. The assignments are summarized in Supplementary datasets 1 and 2) which have unfortunately not been forwarded to the referees in the original submission. We give two references in the manuscript to the corresponding details. Also, chemical-shift perturbations were measured on 3D experiments where 2D was too overlapped. All experimental details are summarized in Supplementary Dataset 3. We have made this point clearer in the manuscript.
The title for the legend for Fig. 2 is misleading -the data do not show that ATP hydrolysis occurs in the absence of DNA; no ATP is used and no release of product is followed.

Answer:
We have clarified that we study the equivalence scheme in Figure 2. . It is not obvious how the spectra shown support the claim of the legend title, namely that "Solid-state NMR allows to distinguish different states of ATP hydrolysis". No hydrolysis reaction is being followed in these measurements.

Answer:
We have clarified that we study the stationary points in equivalence scheme in Supplementary Figure 3.