In the manuscript by Bax et al., the authors presented a nice demonstration of 3D-printed microcell as an effective and inexpensive alternative for reducing the sample volume needed for protein NMR at high ionic strengths. Fully taking advantage of the versatility of SLA 3D printing, the authors created a 3D printed ellipsoidal polymer plug that could be inserted into regular NMR tubes, thus enabling high quality NMR acquisition without considering mismatching susceptibility. The authors further demonstrated the application of the 3D printed microcell for the observation of the tetramerization of melittin, which is challenging to perform in regular NMR tubes. The concept of employing 3D printing to create freeform vessels for NMR is novel and the NMR experiments were conducted rigorously. This manuscript will be of great interest to the readers of Magnetic Resonance and this reviewer would like to recommend its publication. The following comments extend beyond the major findings of this work but may enhance the discussion section at the editor’s discretion.

1.    It is interesting that authors observed leaching impurities from the 3D-printed microcells as they noted in Figure 5. The authors were using Formlabs Clear V4 resin, which is mainly composed of methacrylate monomers, bismethacrylate crosslinker, and TPO (diphenyl(2,4,6-trimethylbenzoyl)phosphine oxide) photo-initiator. Telling from the chemical shifts of the leaching impurities, they are likely to be methacrylate (hydroxypropyl methacrylate, HPMA) monomers or oligomers. HPMA and its oligomers have reasonable water solubility and that would be problematic if this material were to be used for protein NMR.

2.    To remove water-soluble components, this reviewer noticed that the authors used rigorous post-printing processing, including IPA wash, UV curing, and Milli-Q water wash at an elevated temperature. Water-soluble polymers with hydrophobic nature like poly(HPMA) tend to exhibit lower critical solution temperature (LCST) behaviors (Macromolecules 2008, 41(14), 5132-5140) meaning they become less soluble in water as temperature increases. Therefore, washing with cold water rather than hot water would likely be more effective at removing these impurities, as they would remain dissolved at lower temperatures. This reviewer suggests the authors try cold water washing of the printed microcells, which might help reduce the leaching problems observed in the NMR experiments.

3.    If poly- or oligo- HPMA is indeed the leaching impurities, another phenomenon the authors might have observed is when the ionic strength is increased, the resonances of the impurities might broaden and disappear. This reviewer would be interested, and it would be much appreciated if the authors could comment on that.

4.    In Method Section, a different resin, ELEGOO ABS-like Resin V3 was used for doping study. However, experiments using V3 were not mentioned elsewhere in the manuscript. It would be valuable if the authors could include results from these experiments and comment on the NMR performance of this material compared to Formlabs Clear V4 resin. Based on the material specifications, ELEGOO ABS-like Resin V3 is relatively more hydrophobic than Clear V4, which might result in less leaching of water-soluble components. However, this potential advantage might be offset by its high titanium oxide content, which could potentially interfere with magnetic susceptibility matching despite the optimized shape design. Including those comparative results would be helpful in understanding material selections.

5.    The observations regarding different resin materials lead to a more general consideration that is beyond the scope of this work but worth noting. Commercial 3D printing resins are formulated in a way that try to meet various performance specifications, such as material strength, printing resolution, and printing speed. As such, they typically contain multiple components and are not necessarily suitable for high precision work such as protein NMR. This work provided a great proof of concept and would motivate material chemists to develop specialized resins for analytical applications.

General comments:

The manuscript describes the construction of NMR microcells using 3D printing. The authors thoroughly measured the magnetic susceptibility of water containing NaCl, D₂O and cobalt-doped Clear V4 resin. The authors also constructed an ellipsoidal-shaped microcell using Clear V4 resin and successfully observed HSQC signals of N-acetylated alpha-synuclein in PBS buffer and tetramerized melittin at 2M NaCl using the microcell. As the experimental magnetic susceptibility data are useful for designing NMR microcells with a wide variety of sample chamber shapes, I recommend publication of this manuscript.  

Specific comments:

When an NMR tube made of glass is used, there is a case where a protein adsorbs to the surface of glass. Was such adsorption to the surface of the Clear V4 resin observed for synuclein and melittin?

The authors state that H₂O appears to diffuse into the Clear V4 resin (in Line 260). If a protein sample is placed in the microcell and left it for a long time, will the protein leach from the sample space into the printer resin?

Technical correction:

Figures 1 to 4: While each panel is labeled in lowercase letters in the artwork, capitals are used in the figure legends.

Kakheshpour et al present a study of susceptibility matching for aqueous samples. They show an ingeniously simple method for measuring the susceptibility difference between a solid material (eg, glass) and a liquid, by comparing the observed chemical shift of signals stemming from the solid/liquid interface to those observed for a glass/liquid interface of the same geometry. They also show that 3d printed inserts can be susceptibility-matched by doping of the 3d printing resin with a suitable paramagnetic dopant. Finally, they demonstrate the use of a 3d printed insert with an ellipsoidal sample chamber to reduce the impact of susceptibility differences. The resulting lower filling factor also makes the probe tuning less sensitive to sample conductivity, and therefore enables the study of samples that require high salt concentrations.

The manuscript is well written, with figures that clearly illustrate the concepts and results. This is a strong contribution to the field, and should definitely be published in Magnetic Resonance. There are a few minor issues to address, as listed below:

1. L120: the approach based on eq. 4 implicitly neglects all other interfaces. However, the experimental setup has a glass/air interface at the bottom end of the shigemi tube that, while further away, represents a much larger step in susceptibility (of order 10ppm). I suspect that it is correct to ignore it due to the initial shimming on a filled NMR tube, which has the same interface. This should be more clearly explained in the text.

2. Figure 1: the method relies on the shift difference between the maximum of the h=0 slice spectrum and one that has been taken at h=13mm. While the latter is a single peak, the former has a strong downfield shoulder due to the lateral inhomogeneity. To what extent does this affect the precision of the resulting shift difference? An error estimate based on the accuracy with which the peak positions can be determined should be carried out.

3. L 151: the extensive work of Wapler et al (10.1016/j.jmr.2014.02.005) to determine the susceptibility of a range of construction and microfabrication materials should be cited. Their method was based on susceptibility-induced MR image distortions in water, and therefore measured susceptibility differences with respect to water.  

4. L197: the units are not clean in this expression. the concentration needs to be divided by 1M, and it should be -9.01 ppm rather than -9.01

5. We have proposed a tangentially related approach to compensate for solvent/container susceptibility differences in the context of microfluidic NMR. (Ryan et al, 10.1039/C3LC51431E, Hale et al, 10.1039/C8LC00712H) Maybe the authors could consider citing this work for the benefit of the interested reader?