Dear Tobias,

Thank you for taking the time to comment, and for the constructive criticism. Below we address the three points:

We focused on MAS NMR here, and indeed the approach also work for solution-state NMR. Besides the publication from Löhr and co-workers, we have now added a more recent paper (O’Brien 2018) where the authors also use Isogro for labelling, and apply it to solution-state NMR. Note that at the same time as ours, the Linser group has also submitted a paper, with a much shorter analysis and some MAS NMR spectra (out today in Chem. Commun.). So yes, this approach works for solids and for solution. I would argue that for solution-state NMR of large proteins, the fact that HAs are protonated to a significant amount might be more problematic than for solids; but this is all very system-dependent (e.g. size of the protein).

We will prepare a comparative figure also for the homemade E. coli extract. Given the very similar 1D spectrum shown in Figure 2, it is likely that the profile will be similar. (Reviewer 2 asks something similar.)

Distance restraints: this is a very good idea. It has not been our focus here to go into structure restraints, and unfortunately we do not have any samples any more where we could do this in a comparable manner. As it would be a pretty significant effort to make these samples and record and analyse these spectra, we prefer not to go in this direction at this point and rather publish for now the principle and performances of the labelling scheme. We hope (and think) we have done a rigorous job in characterising the spectral quality and some biochemistry, and it will certainly be useful to take this to the next level.

The proposed experiment, a RFDR on a partially deuterated sample, will have a potential gain, as pointed out by your comment: one might see longer-range contacts. On the other hand, the remaining protons are only there to a certain level: looking at figure 1b one can see that in the side chain, only 10-40% are protonated. Thus, cross-peak intensities will be scaled down by this level squared. This may be prohibitively low. On the other hand, one may likely be able to do such an experiment with a 1.9 mm rotor, thus gaining a lot of sensitivity from the sample amount.

The better way is certainly to do a proper 100% labelling of e.g. methyl groups and have all the rest deuterated; this has been done by several groups. We leave this question for later experiments.

Best regards,

Paul

We thank the reviewer for the careful evaluation of the manuscript.

Here we address the four points:

(1) We will include a discussion of the price and practical aspects of 0.7 mm/protonated samples with respect to our deuteration approach.

(2) Extending the methodology to membrane proteins is certainly an interesting route. As for membrane proteins the inhomogeneous line width often is broader than for the systems we studied here, the extra broadening due to the less-perfect deuteration may not be of practical relevance. And the fact that we get all amides protonated may be very important for membrane proteins, where the trans-membrane part is often hard to exchange. Thus, the study of membrane proteins is indeed probably a good application.

(3) We have investigated how much cell extract is needed to obtain a certain deuteration level, but have not investigated the influence of deuterated glucose. It had escaped our attention that O’Brien et al. had done experiments in which they used different levels of Isogro, and 1 g/L of unlabelled glucose  (J. Biomol. NMR, 71(4), 263–273. https://doi.org/10.1007/s10858-018-0200-7). They used 1 g/L of glucose "to maintain various cell functions".

They have reported that with >2 g Isogro per litre of culture they obtain ca. 80% deuteration. This is similar to what we find here. Using both our data and O’Brien’s data (in particular that they have not used deuterated glucose) suggest that the glucose is likely not important for the deuteration. We will discuss this.

(4) Yes, the choice of the method depends certainly on the circumstances (such as probe sensitivity, and even the inherent line width of the protein). From our own experience, as well as from data reported by others (e.g. the Chem. Rev. review by the Pintacuda group which we cited), it is clear that in terms of sensitivity a 0.7 mm probe is very much worse than a 1.3 or 1.9 mm probe. As the line widths resulting from 100 kHz MAS on a protonated sample and 40 kHz MAS from our partially deuterated samples are comparable (Figure 3), it is difficult to argue why the 0.7 mm probe would be preferable. We think that not the price of the 0.7 mm probe would be the game changer; what would really make a significant difference is if 0.7 mm probes gained a factor of 2-3 in sensitivity. This factor is, roughly, what separates a 1.3 mm from a 0.7 mm probe (Figure 9 in Le Marchand et al Chem Rev 2022 10.1021/acs.chemrev.1c00918). Compared to a 1.9 mm probe, the factor is even significantly higher.; in our hands, we find that this penalty is often really limiting.

We thank the reviewer for these additional and very valid points.

We reply below to these points.

1) We will add a discussion about the “RAP” scheme by Asami and co-workers. It is indeed a viable route in cases where back-exchange is a problem.

2) We will provide the raw data for these aliphatic spectra.

3) We have not systematically explored whether we could perform some amino-acid type editing based on the type of amino acid, and the fact that some HAs are protonated much more than others. I think it is a good idea and worth exploring more. One difficulty we see is that often the peak intensities (e.g. amide HN) can be very variable along the sequence, and we often find a factor of 3 variation even though they are protonated to the same level. This may be explained by different dynamics which impacts transfer efficiencies etc. Such a variability would of course complicate the situation. We propose that we discuss this opportunity in the Discussion section of the updated manuscript.

Yes, we do not have data for all types of amino acids using our analysis (with ubiquitin). The reason for this lack is that either there simply no such amino acid in ubiquitin, or only too little data, or signal overlap. E.g. there is no Trp, no Cys, only one Met at the N-ter etc.

We will also take care of the formatting problems. Thank you for spotting those.