*4.2. Nb23 Sample Preparation, NMR Data Acquisition, and Peak Assignment*

All the NMR spectra were collected at the NMR facility of the Core Technology Platform at New York University Abu Dhabi on a 14 T Bruker Avance III spectrometer operating at 600, 150, and 60 MHz for <sup>1</sup>H, <sup>13</sup>C, and <sup>15</sup>N, respectively, with a triple resonance cryoprobe. The acquisition temperature was always set to 298.2 K. All samples for backbone and sidechain assignment or homonuclear correlations were prepared at labeled or unlabeled protein concentrations ranging from 190 to 291 µM in 95/5 H2O/D2O and 10 mM phosphate buffer, pH 6.95, with or without NaCl (6.3–21 mM). Occasionally 19.5 mM bis-Tris aqueous buffer was also used, always at pH 6.95. The samples for aromatic sidechain assignment were prepared in D2O, at protein concentrations in the range 100–190 µM with 10 mM phosphate buffer, pH 6.98 (uncorrected pH-meter reading), without or with 20 mM NaCl. Importantly, the heteronuclear fingerprint of the <sup>15</sup>N-1H HSQC spectra overlapped satisfactorily regardless of the mentioned buffer mixture. Protein concentrations were determined by UV absorption at 280 nm with an IMPLEN nanophotometer based on calculated molar extinction coefficients of 30,495 for Nb23. The sample concentrations were unstable over long time intervals. The initial concentration values invariably decreased by some 50% after 7–10 days as a consequence of protein precipitation. This proved detrimental for the sensitivity of the collected data sets, especially the later acquired ones, that could not be re-acquired due to labeled protein shortage.

A summary of the collected spectra with corresponding acquisition parameters is shown in Table 4. Pure phase detection in t1 and t2 dimensions of 3D data sets were obtained via gradient-based echo-antiecho selection and States-TPPI scheme [31–33]. The States-TPPI scheme was also employed for homonuclear NOESY and TOCSY spectra, whereas 2D heteronuclear spectra pure phase detection in t1 was obtained using echoantiecho selection. The solvent was typically suppressed with a flip-back pulse [34], whereas in homonuclear spectra WATERGATE elements [35] applied in the excitation sculpting mode [36] were employed.

All 3D matrices were acquired with non-uniform sampling schemes by collecting 10%–20% of the whole datasets and by reconstructing the matrices with the dedicated routine of the Bruker Topspin 4.05 software [37]. The same software was used for processing all of the spectra with standard processing routines.

The NMR data were analyzed using NMRFAM-SPARKY [38], including peak assignment which was performed in a semi-automated manner using NMRFAM-SPARKY incorporated tools. The assignment list is available in BMRB, accession number 50808. Table 1 lists the overall assignment percentages.


**Table 4.** List of the collected spectra for backbone and side-chain nuclei assignment of Nb23, with the corresponding acquisition parameters. Experiments denoted with tr indicate the use of TROSY pulse schemes.


**Table 4.** *Cont.*
