The contribution from Polenova, Gronenborn and coworkers provides important information on DNP experimental setups for protein studies. The findings are presented in a very detailed manner, and a wealth of experimental description is given. The conclusions are well supported by the data. The paper will be highly helpful for design of DNP experiments in other labs, and can be accepted as is.

Minor:

Figure 1a) the epsylon =76 seems to be copy/pasted for all different colors?

Polenova, Gronenborn and co-workers provide a detailed investigation into the competing transfer pathways active in biomolecular under MAS DNP. Here, three different pathways are currently known, namely CP-based indirect DNP, direct DNP, as well as a spontaneous contact between 1H and 13C magnetization mediated through heteronuclear cross-relaxation (SCREAM-DNP). The authors excellently provide a short review about these pathways and introduce the reader to the topic under investigation. Understanding the interplay between the three transfer pathways is of crucial importance to successfully apply MAS DNP towards different questions by choosing the optimal conditions for the task at hand. While some control may be exerted by choice of the pulse sequence, the disentanglement of the pathways is not trivial. In this regard, this work gives very useful information about the dynamics of each transfer pathway and their dependence on the polarizing agent concentration. Furthermore, line broadening is discussed for various distinct resonances under the different experimental conditions. The manuscript is well written and of good scientific quality. All experimental data is well presented and documented, and reproduction seems possible given the provided information. The content is of general interest to a broad magnetic resonance community and therefore suitable for publication in MR. Before publication, however, I ask the authors to comment on several specific points and perform minor revisions of the manuscript accordingly. In line 40, the authors state that “modest gains have been detected for membrane proteins with ε=4-10 (Wylie et al., 2015)”. This statement reads like a general observation, however, the referenced work covers a rather special case of membrane proteins labeled with single nitroxide tags which come into dipolar contact. In my experience, DNP enhancement of 40-60 can be routinely achieved for membrane proteins. The following sentence (“Recently, large DNP signal…”) is confusing. The first part deals with impregnated microcrystalline histidine, the second part is in no way generally applicable and seems to be specifically limited to the highest available field. Even in this regard, the sentence is misleading because it implies that no soluble polarizing agents are available for biological systems. The authors use “build-up time, Tb” interchangeably for both the time constant as well as the experimentally chosen polarization time period (c/f lines 127, 129, 142, and Figure 2). These two different parameters have to be clearly distinguished by an unambiguous choice of symbol and naming. In Figure 2, it is not clear, which graphs the subpanels abc are referring to. Also, in the lower left graph, epsilon(-) seems to be missing a negative sign (-4). When discussing the build-up dynamics in Figure 2, it should be clarified that SCREAM-DNP magnetization is emerging from the methyl groups and the spreading through the 13C network by spin diffusion. This explains the quick inversion of Ile resonances, and the delayed response of the other resonances. On the bottom of page 5, it is stated that “Heteronuclear decoupling has no effect…”. I am wondering by what means turning off the decoupling is expected to effect a sign inversion? Heteronuclear decoupling is used when the 13C magnetization is already in the transverse plane, so it is unclear to me how this can influence the sign of polarization. For SCREAM-DNP 1H saturation during the build-up period mostly destroys the incoherent pathway, but this is independent from decoupling. This part should be revised, it should either be explained why decoupling may be expected to change the outcome of the experiment, or it should be clarified if indeed decoupling is mistaken for saturation. In line 204, “with the with line…” seems to contain an additional “with”.

The authors report highly intriguing DNP MAS NMR spectra of tubular assemblies made up by the HIV-1 capsid protein. They study the effects of biradical concentration and of different excitation schemes (CP, direct and SCREAM). In line with work by the group of Corzilius, different DNP polarization transfer pathways are made active, indirect, direct and SCREAM DNP, all relying on the cross-effect. For this study the authors use mainly Amupol in three different concentrations (4.3, 22.8, and 28.2 mM). Radical concentration was determined by EPR measurements using a bench top devise. On the whole, the study is very interesting and the following remarks are meant to be part of a final ironing process. Its particular value is in the display of spectra obtained with a small amount of radical (4.3 mM). The authors demonstrate nicely the effects of radical concentration on spectral quality and enhancement, as well as the discriminating effect of cross polarization versus direct polarization using one of the already classic DNP test systems. The paper is certainly worthwhile publishing, and I enjoy reading it; however, it can do with a lot of clarifications to improve readability.

• The authors claim that 13.8 watts of MW were applied at the sample. Where was that measured, and how was that measured? This should be included in the experimental. Was it measured in front of the sample or in the sample? What exactly means ‘at’ in this case?
• The spectra shown in Fig. 1 were all recorded with a recycle delay of 10 s. The authors should briefly state why they choose to do the experiment in this way and not using appropriately chosen multiples of T1 which is very different for the samples, and which would lead to optimal signal-to-noise-ratios for the fast relaxing ones. More importantly, they should discuss the implications of choosing the same delay for all samples appropriately. Enhancements are given in the top row of Fig. 1a, and I wonder whether there is a typo somewhere. Since so different radical concentrations are used I would expect different depolarization effects, by the way. One sentence discussing their possible contributions and in general the change with radical concentration would be good.
• The authors should also indicate the scaling factors between the displays of spectra in the top and bottom rows in Fig. 1. They cannot be the same, otherwise the statement that in all three cases E=76 is likely not correct.
• It should always clearly be stated which pulse sequences were used; names are not enough. Fig. 3 is good to have, but where is the SCREAM sequence, where are the CORD sequences (yes, they are easy), which sequence is used for which spectra in Fig. 1, 2 and 4, etc. For the outside reader it is very difficult to understand what TB, TB+ and TB- refers to, there is nothing like this in the pulse sequences of Fig. 3 and there is also no description in the experimental. Furthermore, the letters in the pulse sequences of Fig. 3 are far too small, especially those for the DANTE delay. It requires quite some magnification to see that it is Tr. Elderly persons who need to print the manuscript will not see anything.
• Fig. 4 reports CP-CORD spectra whereas this name does not appear in Fig. 3 nor in the experimental part of the paper. There is no reference where I would expect it. Again, its needs to be stated to which pulse sequence Fig. 4 correlates to, and what the numbers mean above the spectra. By the way, they are very misleading, they could be mistaken for the CORD mixing time by an unexperienced reader. Better write RD= and define somewhere in the paper or in the legend.
• It is very instructive to see spectra recorded with 4.3 mM Amupol only, it is an important point of the paper, but it is always tricky to compare contour plots. To me it is obvious that spectra are different yet comparing one or two selected cross sections would be probably wise.
• In the conclusion section it is said that 4.3 mM Amupol yielded the highest DNP signal enhancements. In stark contrast, Fig. 1 announces exactly the same enhancement for all three concentrations. Furthermore, the statement needs probably some seasoning, since there was no T1-optimized relaxation delay employed, and the S/N would be better with appropriate choices for the samples with higher Amupol concentrations. On the other hand, the better resolution in the spectra for the sample 4.3 mM Amupol does its job, too. Resolution is good, yes. To me it looks like similar T1 effects lead to the appearance of narrow signals in those direct polarization experiments recorded with longer relaxation delays. Maybe this should also be discussed appropriately.