At the core of the interaction: Probing charged side chains in flexible protein regions with simultaneous nuclear magnetic resonance experiments

Abstract

Charged amino acid side chains are crucial mediators of biomolecular recognition, but their characterization by nuclear magnetic resonance (NMR) is often hindered by conformational and solvent exchange, particularly for arginine guanidinium groups. We present two complementary ¹³C‐detected NMR strategies that exploit multiple acquisition schemes to simultaneously monitor positively and negatively charged residues. A “NMR by Ordered Acquisition using ¹H detection (NOAH)”‐based experiment combines C_ζN_η‐HDQC and SC‐CACO experiments, allowing the simultaneous detection of arginine, aspartate, and glutamate side chains resonances. In parallel, the Multiple‐Receiver (MR) strategy integrates CP‐HISQC and C_ζN_ε‐HSQC, allowing full assignment of the arginine guanidinium group. We apply this approach to study the interaction between the SARS‐CoV‐2 nucleocapsid N‐terminal domain and the negatively charged glycosaminoglycan enoxaparin. The experiments provide information about flexible, charged side chains at the protein ligand interface. Together, NOAH and MR approaches provide a powerful framework for the high‐resolution characterization of charged side chains and electrostatically driven interactions.

1. INTRODUCTION

Charged amino acid side chains play a central role in numerous biomolecular processes, including molecular recognition (Hodges et al., 2015), signaling (Moens et al., 2019; Tsikas, ²⁰²¹), catalysis (Hsiao et al., 2018; Pauff et al., ²⁰⁰⁷), and structural stabilization (Meuzelaar et al., 2016). Their ability to form hydrogen bonds, salt bridges, and ionic interactions makes them key mediators in the binding of ligands, nucleic acids, metal ions, and proteins. Among these, surface‐exposed and highly flexible charged residues are of particular interest, as they frequently serve as initial contact points in transient interactions and dynamic molecular assemblies. However, flexibility and solvent exposure pose substantial challenges for their high‐resolution structural and dynamic characterization.

Arginine residues are particularly significant due to the chemical nature of their guanidinium group, which remains positively charged in a large range of conditions, including human physiological ones, and can form multiple directional hydrogen bonds, salt bridges, and cation‐π interactions. Present in active sites of enzymes, surface exposed arginine residues also play a crucial role in protein‐RNA interactions (Barik et al., 2015; Calnan et al., ¹⁹⁹¹; Smith et al., ²⁰⁰⁰) and often modulate the onset of liquid–liquid phase separation (Wang et al., 2018). Despite their functional importance, their structural investigation via nuclear magnetic resonance (NMR) spectroscopy remains challenging, in particular for surface exposed residues.

The guanidinium group of arginine residues is prone to conformational exchange and its protons undergo chemical exchange with surrounding water molecules, leading to severe line broadening and to signal loss in conventional ¹H‐detected experiments (Nguyen et al., 2019). Consequently, critical information on electrostatic interactions and binding events may be lost when using standard NMR approaches. From this perspective, ¹³C‐detection spectroscopy (Felli & Pierattelli, 2022) can reveal unique insights, especially when combined with strategies tailored to alleviate the problems arising from solvent and conformational exchange resulting in broadening of the resonances (Mackenzie & Hansen, 2017; Yoshimura et al., ²⁰¹⁷). In case of guanidinium groups of arginine amino acids, the sensitivity in ¹³C‐detection NMR experiments can be enhanced by utilizing J‐based double cross polarization (J‐CP) scheme, which exploits the water polarization to enhance the starting polarization source of exchangeable nuclear spins (Kim et al., 2023; Lopez et al., ²⁰¹⁶). Additionally, the resonance broadening mediated by conformational exchange affecting the terminal part of the guanidinium group is mitigated by exploiting multiple quantum coherences involving the ¹⁵N^η nuclear spins (Chang et al., 2014; Mackenzie & Hansen, ²⁰¹⁷). Direct ¹³C detected NMR also permits monitoring the side chains of negatively charged residues such as aspartate and glutamate. Simultaneous monitoring of amino‐acid side chains with positive and negative charges provides a very powerful tool to characterize proteins' electrostatic interactions.

The availability of multiple acquisition strategies has prompted the design of experiments in which more than one NMR spectrum can be acquired within the time required for the longest one (Knödlstorfer et al., 2024; Kupče et al., ²⁰²¹; Pontoriero et al., ²⁰²²; Schiavina et al., ²⁰¹⁹; Viegas et al., ²⁰¹⁶), providing an ideal framework for developing a general strategy to monitor both positively and negatively charged side chains. Our work is based on NOAH (NMR by Ordered Acquisition using ¹H detection) (Kupče & Claridge, 2017; Pandey et al., ²⁰²⁵; Yong et al., ²⁰²²) and UTOPIA (Schiavina et al., 2019; Viegas et al., ²⁰¹⁶) (Unified Time‐OPtimized Interleaved Acquisition) frameworks which enable the simultaneous acquisition of multiple spectra providing complementary information. NOAH supersequences have so far been used for small‐molecule NMR, drastically reducing experimental time by nesting multiple 2D experiments into a single sequence, sequentially preserving proton magnetization for subsequent modules, and using a single receiver (same detected nucleus). Multiple‐Receivers (MR) and UTOPIA methods utilize several different receivers to simultaneously detect different nuclei, enabling parallel acquisition of complementary experiments, and are ideal for time‐sensitive protein studies. The experimental approach presented here exploits these two different strategies for the simultaneous acquisition of multiple NMR spectra. Combining two distinct experiments within the timeframe of a single experiment saves time while maintaining the same performance as individual acquisitions and providing complementary perspectives on the studied system.

We propose a new NOAH‐based super‐sequence that employs ¹³C rather than ¹H detection. As with other NOAH super‐sequences, care must be taken to preserve magnetization transfer between the individual sub‐experiments. The super‐sequence begins with HDQC, followed by SC‐CACO. This order cannot be reversed, as ¹H decoupling during SC‐CACO would saturate the guanidinium protons of arginine, as well as water, thereby compromising the CP transfer. In HDQC, ¹³C pulses are selective for C_ζ to minimize perturbation of the C_γ/C_β magnetization of Asp, Asn, Gln, and Glu residues. This experiment, based on the NOAH approach, exploits ¹³C detection and allows simultaneous investigation of the interactions involving positively charged arginine residues and negatively charged aspartate and glutamate residues (asparagine and glutamine are also detected). This method is based on the principle of using a single receiver. A second experiment is proposed to assign the resonances of the arginine's guanidinium group and exploits two different receivers, as for MR and UTOPIA type experiments. A scheme of the investigated amino acid side chains, with the relative nomenclature and chemical shift ranges in which the resonances are observed, is shown in Figure 1.

FIGURE 1.

A schematic representation of arginine, glutamate, aspartate, glutamine and asparagine side chains, together with the standard nomenclature used to indicate the nuclei present within each side chains. Amino acids investigated in each experiment are enclosed within rectangles with solid outlines (blue for C_ζN_η HDQC, light blue for SC‐CACO, dark green for CP‐HISQC, and light green for C_ζN_ε HSQC). Nuclei directly detected in the experiment are highlighted by filled rectangles, whereas nuclei evolved in the indirect dimension are indicated by rectangles with dashed outlines. A table summarizing the average chemical shifts (in ppm) and their standard deviations for the studied nuclei is also shown; values were obtained from the BMRB (https://bmrb.io).

The approach, initially tested on ubiquitin, is demonstrated for the investigation of the electrostatic interaction between the N‐terminal domain of the SARS‐CoV‐2 nucleocapsid protein (N‐terminal domain [NTD]) and enoxaparin (EP), a negatively charged polysaccharide belonging to the glycosaminoglycan family. The NTD plays a key role in viral genome packaging and replication, and features several arginine residues distributed along flexible loop regions forming the so‐called basic finger, a structurally and functionally relevant motif for the interaction with negatively charged RNA backbones (Dinesh et al., 2020; Guseva et al., ²⁰²¹; Pontoriero et al., ²⁰²²; Schiavina et al., ²⁰²¹). Linear polyanions, such as EP, were recently proposed as possible ligands to interfere with the protein function (Bolognesi et al., ^2025a; Bolognesi et al., ^2025b; Schiavina et al., ²⁰²²; Tino et al., ²⁰²⁵). More importantly, the interaction between the N protein and heparan sulfate (a close analog of heparin) was recently shown as the driving force responsible for the presence of the virus on the surface of cells (infected as well as non‐infected neighboring ones) (Fahoum et al., 2025). This was in turn identified as a possible mechanism to stimulate an immune response by the host not only toward infected cells but also to healthy ones, exacerbating the effects of the COVID19 disease as well as of long COVID19 effects (Fahoum et al., 2025). The investigation of the interaction between EP and NTD of the SARS‐CoV‐2 nucleocapsid protein is thus of great importance to clarify the molecular features at the basis of the interaction and to stimulate further progress in the field.

By combining ¹³C direct detection with multiple acquisition strategies, we were able to observe and assign side chain resonances of positively charged arginine residues, including the terminal guanidinium groups that are often unobservable, in parallel to side chains of negatively charged residues, gaining new insights into the dynamic and electrostatic features of this biologically relevant interaction.

2. RESULTS AND DISCUSSION

2.1. Efficient design for the concurrent acquisition of two experiments

The first experiment, shown in Figure 2a, was designed to simultaneously provide information about amino acid side chains with opposite charges, thereby offering a tool to study fine details of protein interfaces involved in electrostatic interactions. To this end the C_ζN_η‐HDQC (Mackenzie & Hansen, 2017) was modified by including a double CP element (¹H‐¹⁵N and ¹⁵N‐¹³C) at the beginning of the pulse sequence (CP C_ζN_η‐HDQC). The initial CP step facilitates the magnetization transfer from exchangeable ¹H_η/ε protons to ¹⁵N_η/ε nuclei, effectively exploiting the abundant water proton magnetization; the second one is used to transfer magnetization from ¹⁵N to ¹³C. This experiment also relies on the evolution of double quantum (DQ) coherence in the indirect dimension (¹⁵N_η). DQ is less susceptible with respect to single quantum coherences to conformational exchange processes between the ¹⁵N_η nuclear spins arising from rotation of the guanidinium group around the N_ε‐C_ζ axis (Nieto et al., 1997). This experiment was combined with a tailored version of the CACO (Bermel et al., 2005; Bermel et al., ²⁰⁰⁶; Pontoriero et al., ²⁰²⁰) (SC‐CACO) focusing on carbonamide and carboxylic groups of Asp, Asn, Glu, and Gln to access information about both positively (Arg) as well as negatively (Asp, Glu) charged side chains. We chose the CACO variant which exploits a constant‐time scheme to properly refocus J _CαCβ and J _CβCγ couplings (J _CC) while transferring the magnetization from ¹³C_β to ¹³C_γ (Asp) and ¹³C_γ to ¹³C_δ (Glu) (Pontoriero et al., 2020). In this setup, the magnetization remains in the transverse plane for a period equal to 1/(J _CC). This pulse sequence ensures proper refocusing of one‐bond ¹³C–¹³C couplings between aliphatic ¹³C nuclear spins. Selectivity on side chains is achieved using a π/2 selective pulse on C_β (asparagine/aspartate) and C_γ (glutamine/glutamate). Notably, glycine ¹³C_α–¹³C′ resonances are also visible in this experiment, but their signal phase is opposite to the other peaks, making them easy to distinguish. The two magnetization‐transfer pathways are independent of each other and allow subsequent acquisition of individual experiments without sensitivity penalties (Figure S4, Supporting Information), while exploiting the same longitudinal relaxation delay. This reduces the overall experimental time and enables the simultaneous acquisition of different “snapshots.” Inspired by the NOAH approach (Kupče & Claridge, 2017; Yong et al., ²⁰²²) the C_ζN_η‐HDQC//SC‐CACO experiment is thus designed to acquire ¹³C magnetization.

FIGURE 2.

(a) The pulse sequence in which CP C_ζN_η HDQC and side‐chain‐selective SC‐CACO are combined. In the CP C_ζN_η HDQC part the Δ₁ delay was set to 13.5 ms (1/2J_CζNη). The phase cycle is: φ₁ = x −x, φ₂ = −x x, φ₃ = 2(x) 2(y) 2(−x) 2(−y) and φ_r1 = x −x −x x. Gradient strengths in G/cm are: g₁ = 37.5, g₂ = 16.5, g₃ = 20.6, g₄ = 5.9, g₅ = 13.9, g₆ = 29.6, g₇ = 12.3. In the CACO part Δ₂ is equal to 4.5 ms (1/4J_CαC′) while Δ₃ was set to 13.3 ms (1/2J_CC). The phase cycle used is the following: φ₁₁ = x −x, φ₁₂ = 4(x) 4(y), φ₁₃ = 2(x) 2(−x), φ₁4 = 2(−y) 2(y) and φ_r2 = x −x −x x −x x x −x. Gradients were applied with the following strengths in G/cm: g₈ = 33, g₉ = 12.4. A schematic representation of the MR H_εN_ε//C_ζN_ε experiment is reported in (b), in which a N_ε/ηH_ε/η CP‐HISQC experiment is acquired during the relaxation delay of C_ζ magnetization. The phase cycle in the N_εH_ε CP‐HISQC part was achieved using: φ₁ = y −y, φ₂ = −y y, φ₃ = 2(y) 2(−y), φ₄ = 4(x) 4(−x) and φ_r1 = x −x −x x −x x x −x. Gradient strengths in G/cm: G₁ = 5.35, G₂ = 10.7, G₃ = 20.0, G₄ = 12.8. Δ₁ delays were set equal to 2.7 ms (1/4J_HN), while Δ₂ delays were set to 12.5 ms (1/4J_CζNε). The phase cycle in the C_ζN_ε HSQC part was the following: φ₁₁ = x −x, φ₁₂ = 2(x) 2−(x), φ₁₃ = 4(x) 4(−x), and φ_r2 = x −x x −x −x x −x x. Gradients were set with the following strengths in G/cm: G₅ = 32.1, G₆ = 10.2. In all panels, narrow solid rectangles represent hard π/2 pulses, while wide solid rectangles represent hard π pulses; rounded shapes represent band‐selective shaped pulses which are accurately described in the Materials and Methods section (narrow and wide ones represent π/2 and π pulses, respectively). Quadrature detection was achieved through States‐TPPI by incrementing the phase of the π/2 pulse prior to the building block devoted to chemical shift evolution monitored in the indirect dimension.

To obtain more detailed information on arginine side chains, the pulse sequence shown in Figure 2b was designed. This experiment is specifically tailored for the observation of arginine side chains and is based on the MR approach, combining CP‐HISQC (Yuwen & Skrynnikov, 2014) (¹H‐detected) and C_ζN_ε HSQC (Werbeck et al., 2013) (¹³C detected) experiments. The resulting experiment was named MR H_ε/ηN_ε/η//C_ζN_ε. This MR strategy relies on the UTOPIA approach (Schiavina et al., 2019; Viegas et al., ²⁰¹⁶). Here we exploit the longer longitudinal relaxation time of ¹³C nuclei with respect to ¹H ones to acquire the H_ε/η N_ε/η experiment for free during the recovery delay of ¹³C_ζ. In the CP‐HISQC, the magnetization is transferred from H_ε/η to N_ε/η via J‐CP transfer, followed by evolution of the N_ε/η resonances during which ¹H decoupling scheme is applied to prevent effect of signal broadening and thus disappearance arising from exchange with water protons (Iwahara et al., 2007). The magnetization is then transferred back via the INEPT block, and the water signal is suppressed using 3‐9‐19 WATERGATE scheme (Sklenar et al., 1993). It is also possible to transfer the magnetization back from ¹⁵N to ¹H using J‐CP transfer to avoid signal losses caused by exchange with bulk solvent. However, this approach can be very demanding in terms of the energy deposited into the probe, and it does not offer the advantage of exploiting water protons to enhance sensitivity (Yuwen & Skrynnikov, 2014). In the C_ζN_ε HSQC experiment, the starting magnetization source resides on ¹³C_ζ nuclear spins. The magnetization is then transferred to N_ε, which is evolved in the indirect dimension and then transferred back to ¹³C_ζ for detection, both transfer steps use a regular INEPT scheme (Morris & Freeman, 1979). Selectivity is essential in NMR experiments focused on side chains, as it ensures that only the relevant spectral region is excited. In the case of the arginine guanidinium group, band‐selective pulses are crucial for ¹⁵N, as C^ζ is coupled to both N^ε and N^η. The selective excitation of ¹⁵N^ε prevents unwanted multiple‐quantum coherences and ensures clean, interpretable spectra. It is worth nothing that ¹H decoupling during ¹³C^ζ acquisition was omitted resulting in an optimal recovery of the water magnetization, needed for the following H_ε/η N_ε/η CP‐HISQC transient. Similar to the NOAH‐type approach described above, the UTOPIA approach allows the combination of orthogonal magnetization pathways without penalizing the sensitivity of the individual experiments (Figure S4).

2.2. Simultaneous snapshots of positive and negative side chain resonances to monitor interactions

The novel experiments have been used to study the NTD of the nucleocapsid protein from SARS‐CoV‐2 and its interaction with EP (Bolognesi et al., ^2025b; Schiavina et al., ²⁰²²). NTD is characterized by the presence of 9 arginine residues and 11 acidic residues, including aspartic acid (Asp) and glutamic acid (Glu). Thus, it represents a suitable case of study for the proposed experiment that enables to directly monitor the key groups (positively charged guanidinium groups and negatively charged carboxylates) involved in electrostatic interactions.

Obtaining information on the ¹⁵N_η nuclear spins of the arginine guanidinium groups via conventional ¹H‐detected methods is challenging due to severe line broadening caused by conformational and chemical exchange. As shown in Figure 3, ¹³C direct detection significantly improves signal detectability and spectral resolution by mitigating exchange‐induced broadening. This effect is further enhanced using DQ coherence, which facilitates observation of dynamic side chains through the CP C_ζN_η‐HDQC experiment.

FIGURE 3.

Comparison between ¹H and ¹³C detected experiments. Panel (a) reports cross peaks corresponding to the H_η–N_η correlation of arginine side chains acquired using ¹H detection. Conformational and chemical exchange severely limit the observation of guanidinium resonances, resulting in significant line broadening. Panel (b) shows how ¹³C detection, combined with the use of double quantum coherence, mitigates this issue and enhances spectral resolution through the CP C_ζN_η‐HDQC experiment. Worth noting, the evolution of the nitrogen chemical shift in the two spectra involves a single quantum coherence in the case of H_ε/ηN_ε/η spectrum in (a) and a double quantum coherence in the case of CP C_ζN_η‐HDQC in (b). For this reason, the ¹⁵N_η chemical shift values in the two panels do not match exactly. The experiments were acquired on a 300 μΜ ΝTD sample.

Based on these considerations, the guanidinium group of arginine residues and carboxylate groups of acidic residues were monitored using ¹³C‐detected NMR experiments through the NOAH‐based approach C_ζN_η HDQC//SC‐CACO (spectra reported in Figure 4a, b) upon interaction with increasing amounts of EP (spectra reported in Figure 4c, d). Sequence‐specific assignment of the cross peaks detected in the spectra involving arginine side chains was achieved exploiting also the complementary MR H_ε/ηN_ε/η//C_ζN_ε experiment, as discussed in detail later in the text. Assignment of the carboxamide and carboxylic side chains of Asp, Asn, Glu, and Gln was achieved by adapting the available assignment (BMRB 51620) to the slightly different experimental conditions used in the present case (Bolognesi et al., ^2025a).

FIGURE 4.

Titration of NTD with enoxaparin investigated using C_ζN_η HDQC//SC‐CACO experiments. Panel (a) shows the C_ζN_η HDQC spectrum and panel (b) shows the SC‐CACO spectrum of the free protein (without enoxaparin). The experiments were acquired on a 300 μΜ ΝTD sample. Panels (c) displays a zoom of the SC‐CACO spectrum, showing the perturbation of residues E62 and E174 upon addition of 0.6 equivalents (violet), 1.2 equivalents (red) and 2.4 equivalents (orange) of enoxaparin. E136 remains unaffected upon addition of enoxaparin. Panel (d) shows a zoom of the C_ζN_η HDQC spectrum illustrating the behavior of residues R68, R89, R93, and R95 upon addition of enoxaparin with the same color coding as in panel (c). Panel (e) shows a family of conformers (PDB 9QWI) where positively and negatively charged residues perturbed by the ligand are highlighted in vivid blue and red, respectively. Residues that remain unperturbed upon addition of enoxaparin are colored in pink (negative residues) and light blue (positive residues).

In the C_ζN_η HDQC spectrum (Figure 4d), several arginine residues displayed chemical shift perturbations upon interaction with EP. Specifically, residues R89, R92, R93, and R95 showed significant perturbations. In contrast, R149 was not detected, likely due to its structural rigidity. Indeed, the analysis of the 20 conformers in the PDB entry 9QWI (Bolognesi et al., ^2025a) indicates that the 9 arginine residues present in the protein construct (68, 88, 89, 92, 93, 95, 107, 149, and 177) exhibit different structural behaviors. Particularly, R88, R107, and R149 are found to establish a series of H‐bonds in all the available conformers thus resulting in a more rigid structure. On the other hand, the resonances arising from R68, R89, R92, R93, and R95 show intense peaks, correlated with the limited number of H‐bonds established by these residues in the deposited protein structure (Figure 4e). Interestingly, when EP is added, the backbone responds differently compared to the side chains. The side chain experiments provide further details on the interaction. Indeed, in these latter experiments we can observe more selective and more pronounced chemical shift perturbations with respect to what can be observed by monitoring the backbone signals (Figure S1). Even more, arginine residues R89, R93, R177, whose involvement in the interaction cannot be appreciated by looking at the interaction with the common backbone‐based strategies, are found to be highly perturbed exploiting the side chain‐based experiments.

The SC‐CACO experiment (Figure 4c) provides insights into the behavior of carboxamide and carboxylate moieties, such as those found in the side chains of aspartate, glutamate, asparagine, and glutamine residues, with the primary focus on the acidic side chains of aspartate and glutamate. The negative charge carried by these latter residues makes them unlikely to be the main interacting ones with the highly negative heparin‐based compound. However, the spectra revealed selective chemical shift perturbations for residues E62, D98, D103, D128, and E174. Among these, D98 and D103 are located in the basic finger region, while E62, D128, and E174 reside in loop regions adjacent to the basic finger. Moreover, D128 and E174 are involved in hydrogen bonding with residue R89 and R107, respectively. A second observed effect is the disappearance of peaks upon addition of EP. The disappearing peaks correspond to residues located in the most structured region of the protein (Figure 4e) and are likely to be broadened because of the increased rotational correlation time of the complex with respect to the free protein.

The complementarity of the two experiments acquired simultaneously highlights an elongated region rich in positively charged arginine residues involved in the interaction, confirming previously available information derived from backbone correlations and at the same time providing more direct information at the heart of the interaction as evidenced by the more pronounced and selective chemical shift changes (chemical shift perturbation values are reported in Figure S2). They also reveal the involvement of negatively charged residues which are likely to stabilize local conformations in the isolated protein that are disrupted when EP is added.

To gain further insights into the interaction of the guanidinium of arginine residues upon interaction with EP, the MR H_εN_ε//C_ζN_ε experiment was crucial. Figure 5 shows a reference spectrum of NTD (blue) and in complex with 2.4 equivalents of EP (violet). The H_εN_ε spectrum revealed that residues R89, R93, and R107 experienced pronounced chemical shift perturbations upon ligand binding, consistent with the C_ζN_η HDQC//SC‐CACO results. The H_εN_ε experiment also reported perturbation for R88 and R149. Residues R88, R89, and R93 are located in the basic finger region, while R107 is part of the β3 strand at the junction with the basic finger, a portion of the main interaction region. In contrast, residues R68 and R177, located at the edges of the protein, were unaffected by EP addition.

FIGURE 5.

Titration of NTD with EP investigated using MR H_εN_ε//C_ζN_ε experiments. The reference spectrum (without EP) is shown in blue, while the spectrum of NTD upon addition of 2.4 equivalents of EP is shown in violet. Panel (a) displays the H_ε‐N_ε resonances, while panel (b) shows the corresponding C_ζ‐N_ε cross‐peaks. Experiments were acquired on a 300 μΜ ΝTD sample.

The involvement of the same arginine residues was also observed through the simultaneously acquired C_ζN_ε HSQC experiment. In this experiment, the possibility to observe the C_ζ nuclear spins provides an additional perspective on the interaction showing further interesting contributions. As an example, residues R92 and R95 (Figure 4b) are found to be more perturbed in this experiment as similarly observed in the C_ζN_η experiment (Figures S1, S2). Moreover, the availability of both spectra also contributes to resolve possible accidental overlaps in the ¹H detected experiment, such as for residues 68, 92, 93, and 95.

It is worth noting that the C_ζN_ε HSQC spectrum is not affected by possible solvent exchange resonance broadening, ensuring full applicability in a variety of different experimental conditions, such as approaching physiological pH and temperature values, as also shown for backbone CON experiments (Felli & Pierattelli, 2022; Gil et al., ²⁰¹³).

Combining the information available from the two proposed experiments with the previously available assignment for the H_εN_ε resonances (Schiavina et al., 2022), it is possible to assign all the resonances of the nuclear spins within the guanidinium group. Starting from the H_εN_ε resonances one can obtain the proper assignment of the C_ζ nuclei by observing the same N_ε chemical shift in the C_ζN_ε spectrum, as illustrated from the dashed lines reported from Figure 6a to Figure 6b. From this latter experiment one can in the end obtain the information about the N_η nuclear spins, as shown in Figure 6b, c, moving from the C_ζN_ε spectrum to the C_ζN_η one. Following the described procedure we fully assigned 8 out of 9 arginine side chains (H_ε–N_ε–C_ζ–N_η), with only a single arginine missing the N_η resonance, enabling us to retrieve nearly complete information for the terminal functional group of arginine residues in the NTD protein.

FIGURE 6.

Panels (a–c) show the H_εN_ε CP‐HISQC, C_ζN_ε HSQC, and C_ζN_η HDQC spectra respectively, acquired with the novel pulse sequences described in this paper, demonstrating the ability to reconstruct resonance assignments across the entire guanidinium group. Experiments were acquired on a 300 μΜ ΝTD sample. Panel (d) reports a zoom of a conformer from PDB entry 9QWI illustrating the side chain of residue R93, together with standard nomenclature for each atom.

A comment is due on the general applicability of the proposed NMR approach and on possible limitations. The combined acquisition of several simultaneous NMR experiments enables access to independent parameters under identical sample conditions within a shorter experimental time. However, it is worth emphasizing that only experiments exploiting orthogonal magnetization transfer pathways can be combined within NOAH‐ or MR‐type approaches without compromising sensitivity. The overall experimental sensitivity is ultimately limited by the least sensitive experiment and otherwise matches that of the corresponding traditional experiments; it depends on multiple factors, including protein concentration, instrumental sensitivity, and the structural and dynamic properties of the investigated protein, making it difficult to define general guidelines. Nevertheless, the present results demonstrate that experiments applied to protein samples in the hundreds of micromolar range provide meaningful results. The heterogeneous structural and dynamic properties of the NTD, characterized by a well‐structured protein core as well as by a more flexible and disordered region, offer the opportunity to assess how experimental outcomes are modulated by the different dynamic properties of the protein (Bolognesi et al., ^2025a; Dinesh et al., ²⁰²⁰; Schiavina et al., ²⁰²²). The results show that relaxation losses occurring for side chains that are part of well‐structured protein regions significantly affect the sensitivity of each of the two experiments combined for simultaneous acquisition, just like they would when acquiring the two experiments separately. In the presence of flexibility, as it occurs for solvent exposed side chains as well as for flexible protein loops, the experiments provide information that is difficult to access otherwise, as shown for the terminal part of the guanidinium group (N_η) of arginine residues as well as for carboxylate groups of aspartate and glutamate residues. These are often key amino acids involved in interactions and are difficult to characterize at atomic resolution using other techniques, despite their key role in molecular recognition, signaling and function.

3. CONCLUSIONS

In conclusion, the experiments proposed to investigate the interaction between NTD from SARS‐CoV‐2 and EP enabled the direct observation of functional groups involved in electrostatic interactions, providing enhanced selectivity and sensitivity to binding‐induced changes compared with conventional approaches focusing on backbone nuclear spins. Notably, residues within the so‐called basic finger region of NTD exhibited the most pronounced perturbations in the presence of EP, reinforcing the pivotal role of this region in mediating the interaction. The ¹³C detected experiments acquired simultaneously through the NOAH‐based approach allowed us to capture the interaction by targeting both the guanidinium group of arginine residues, highlighting the key residues mediating the binding, and the carboxylate/carboxamide moieties of Asp, Asn, Glu and Gln. Among these latter residues, selective changes were detected in the negatively charged side chains within and adjacent to the primary region of the binding, suggesting that the interaction between negatively charged EP and the basic residues of NTD induces a rearrangement involving acidic residues, which also become perturbed. This description of the interaction is also supported by the presence of some residues whose peaks are not perturbed at all, which indeed belong to residues located in regions far from the main interaction site.

The combination of NOAH‐based and MR detection schemes, revealing information about different nuclear spins in a single experiment, opens the way to the design of a complete set of experiments to monitor key side chain resonances with the aim of identifying which are the main driving forces promoting the interaction of a target protein with possible partners. The novel experiments reveal information about highly flexible, solvent exposed residues thanks to the experimental design exploiting favorable features toward this objective. The focus on terminal functional groups of amino acids is particularly relevant to highlight key residues at the core of the interaction, information that is more difficult to access when investigating backbone nuclear spins which are generally sensitive to both direct and indirect effects.

The application of the proposed combined experiments can be extended and generalized to the study of various systems. The MR approach has already proven to be a useful tool for the study of several interactions (Knödlstorfer et al., 2024; Pontoriero et al., ²⁰²²; Schiavina et al., ²⁰¹⁹) and can be further extended to study biomolecules where different nuclear spins can be observed. NOAH‐based experiments can instead be tailored to monitor other side chains such as those of aromatic residues, together with the backbone carbonyl carbon nuclei. The combined use of selective or orthogonal labeling strategies could open the way to the investigation of the primary actors driving interactions of proteins with cellular partners or complex ligands.

4. MATERIALS AND METHODS

All experiments were conducted at 298K by using a Bruker NMR spectrometer operating at 700.13 MHz ¹H Larmor frequency, equipped with a cryogenically cooled probe‐head optimized for ¹³C direct detection (CP‐TXO).

The pulse sequences were tested on a sample of Ubiquitin (1.0 mM, 50 mM, sodium phosphate, pH 6.5) and the results are reported in Figure S3. The performances of pulse sequences were assessed by recording the experiment in the MR/NOAH approach and comparing the results with the 2D spectra acquired independently using the same acquisition parameters (number of scans, recovery delay, number of increments, acquisition time), reported in Figure S4.

The experiments were then applied to the NTD of the nucleocapsid (N) protein from SARS‐CoV‐2. The protein was expressed following the procedure reported in the literature (Schiavina et al., 2021), providing a sample with a final concentration of 300 μM in 25 mM potassium phosphate (K₃PO₄), 150 mM potassium chloride (KCl), 100 μM ethylenediaminetetraacetic acid (EDTA), 0.03% sodium azide (NaN₃), 5% deuterated water (D₂O), adjusted to pH 6.5.

For each pulse, the bandwidth (bw) reported corresponds to the frequency range over which the magnetization has dropped to 70.8% of its maximum value, as defined by the bandwidth factor in Bruker TopSpin's ShapeTool. The CP C_ζN_η HDQC experiment, inspired by the C_ζN_η HDQC one (Mackenzie & Hansen, 2017), was implemented together with SC‐CACO (Side Chains CACO) in a multiple acquisition NOAH‐based experiment. In the CP C_ζN_η HDQC experiment, the ¹⁵N Reburp pulse with 4.50 ms duration (bw 15 ppm) was centered in the middle of the ¹⁵N_η region (71 ppm); Q5 (for π/2) and Q3 (for π) band‐selective (Emsley & Bodenhausen, 1992) ¹³C pulses were centered at ¹³C_ζ (156 ppm) with a duration of 1.67 ms (bw 21 ppm) and 1.46 ms (bw 13 ppm), respectively. The ¹H carrier was set to 6.5 ppm throughout the experiment. The experiment exploited ¹H nuclear spins as the initial polarization source, with magnetization transfer to ¹³C_ζ achieved via double cross‐polarization (CP) transfer. The first CP step, from ¹H to ¹⁵N, used a DIPSI (Shaka et al., 1988) scheme with a 9.5 ms contact time and a γB₁ field strength of 3.0 kHz. The second CP step, transferring magnetization from ¹³C to ¹⁵N, was performed using a DIPSI (Shaka et al., 1988) scheme with a 29 ms contact time and a γB₁ of 1.25 kHz. For the SC‐CACO experiment, a Q5‐shaped (Emsley & Bodenhausen, 1992) π/2 pulse with 1.8 ms duration (bw 19 ppm) was used to selectively irradiate the side‐chain resonances of aspartate, glutamate, asparagine, and glutamine. Other π and π/2 pulses were optimized for either the ¹³C^ali or ¹³C′ regions, using Q3 and Q5 shapes (Emsley & Bodenhausen, 1992) with duration 0.35 ms (bw 56 ppm) and 0.23 ms (bw 36 ppm), respectively. The ¹³C carrier was set at either 33 or 180 ppm.

The ¹H‐¹⁵N CP‐HISQC (Yuwen & Skrynnikov, 2014), included in MR H_ε/ηN_ε/η//C_ζN_ε experiment, was tailored to selectively irradiate ¹⁵N_ε/η of arginine residues side chains. To achieve this, ¹⁵N_ε/η resonances were irradiated using a Q3‐shaped (Emsley & Bodenhausen, 1992) selective π pulse with a 2.30 ms duration (bw 21 ppm), centered at 78 ppm. The ¹H carrier was initially set to 6.5 ppm before ¹H‐¹⁵N CP and later shifted to 4.7 ppm. ¹H‐¹⁵N magnetization transfer was achieved using a CP‐DIPSI (Shaka et al., 1988) scheme, with a γB₁ of 3 kHz and a 9.5 ms contact time. In the C_ζN_ε HSQC (Werbeck et al., 2013) experiment, the ¹³C carrier was set at 156 ppm, while the ¹⁵N carrier was set at 83 ppm. Selective irradiation of the ¹³C region was achieved using Q3 and Q5 pulse shapes, with durations of 1.50 ms (bw 12 ppm) and 1.80 ms (bw 20 ppm), respectively. ¹⁵N selective pulses were designed to target N_ε resonances, using a Reburp (Geen & Freeman, 1991) shape with a 5 ms duration (bw 13 ppm). The ¹³C decoupling during the indirect evolution dimension was achieved using a Crp80, 0.5, 20.1 pulse with a 500 μs duration. ¹H decoupling was performed using a 2.5 kHz WALTZ64 (Shaka et al., 1983; Zhou et al., ²⁰⁰⁷) supercycle, while ¹⁵N decoupling employed a 1.0 kHz Garp4 (Shaka et al., 1985) supercycle.

Protein‐ligand interaction was monitored using the above‐described NMR experiments. Reference spectra were recorded on an NMR sample containing the isolated protein (300 μM). Subsequent spectra were acquired on the same sample after the addition of controlled amounts of commercially available EP compound (CLEXANE, Sanofi), introduced through small aliquots of a 22 mM stock solution to reach NTD:EP molar ratios of 1:0.6, 1:1.2, 1:2.4.

All experiments were acquired and processed using TopSpin 4.4.0, and relevant processing details are reported in Data S1. Data analysis was conducted using CCPNMR v3.2.2 (Skinner et al., 2016).

All the pulse sequences exploited in this work are reported in the Bruker format at the end of Data S1.

AUTHOR CONTRIBUTIONS

Maria Anna Rodella: Investigation; formal analysis; methodology; writing – original draft; writing – review and editing. Marco Schiavina: Methodology; investigation; formal analysis; writing – original draft; writing – review and editing. Maksim Mayzel: Methodology; validation; supervision; writing – review and editing. Carlotta Cappanni: Investigation; formal analysis. Rainer Kümmerle: Funding acquisition; supervision; writing – review and editing. Roberta Pierattelli: Conceptualization; methodology; funding acquisition; supervision; writing – original draft; writing – review and editing. Isabella C. Felli: Conceptualization; methodology; supervision; writing – original draft; writing – review and editing; funding acquisition.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Data S1: Supporting information.

Rodella MA, Schiavina M, Mayzel M, Cappanni C, Kümmerle R, Pierattelli R, et al. At the core of the interaction: Probing charged side chains in flexible protein regions with simultaneous nuclear magnetic resonance experiments. Protein Science. 2026;35(4):e70533. 10.1002/pro.70533

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.