**1. Introduction**

Nucleosides triphosphates (NTPs), such as ATP (adenosine triphosphate) and GTP (guanosine triphosphate), are used as energy source or as allosteric effector by a number of proteins, involved for instance, in metabolism, active transport, cell division or DNA/RNA synthesis. However, the mechanism of NTP hydrolysis in proteins are still poorly understood, especially its coupling to

functional events, such as movement of proteins along nucleic acids. Indeed, detailed mechanistic insight is lacking for a number of systems, including even intensively studied systems such as dyneins [1], ABC importers [2]/exporters [3] or DNA helicases [4]. Experimentally, catching the events that occur during the NTP hydrolysis is highly challenging. Structural techniques such as X-ray crystallography, cryo-electron microscopy (cryo-EM) and nuclear magnetic resonance (NMR) mainly provide static snapshots of protein states of typically highly complex reaction coordinates of biomolecular reactions. These can then be combined with molecular dynamics simulations (MD) to obtain further information about the dynamics of such processes and to establish the chronological sequence [3,5–8]. In this context, it is highly desirable to better investigate how ATP/GTP analogues, usually with a modified or replaced γ-phosphate group, can mimic the intermediate catalytic states in order to obtain relevant snapshots of the reactions involving NTP hydrolysis. Indeed, it is well known that mimics can never fully represent naturally occurring states, as the modifications of NTPs change their conformation as well as their chemical properties other than their tendency to hydrolyse. While it is important to use NTP mimics described to be strongly hydrolysis-resistant, the true hydrolysis state must often be confirmed experimentally.

The choice of the adequate analogue is thus of importance in structural studies, but guidelines are sparse and can be highly protein-dependent. We herein focus on analogues often used to access three important states of ATP-hydrolysis: the pre-hydrolytic state, the transition state and the post-hydrolytic state (see Scheme 1A for the artificial ATP hydrolysis scheme highlighting analogues used to mimic the different states). We describe how NMR and EPR can be used to gain detailed information on the analogue used and the conformational and dynamic state it induces in the protein. We investigate this for two proteins, the bacterial DnaB helicase from *Helicobacter pylori* involved in DNA replication, and an ABC transporter implicated in multidrug resistance, BmrA (*Bacillus subtilis* multidrug resistance ATP binding cassette transporter), which share high similarities in their ATP binding sites [9]. Solid-state NMR and EPR are highly suitable to study large, noncrystalline protein assemblies, which are represented by DnaB and BmrA. The proteins are, in their multimeric states and, for BmrA, embedded in a *Bacillus subtilis* lipid membrane, sedimented directly into the solid-state NMR rotor in an external ultracentrifuge [10], a sample preparation approach that allows for the study of the investigated analogues. The protein samples prepared by this approach are highly concentrated in the NMR rotor (protein concentration of around 400 mg/mL), and have been shown to be stable over several years [11]. A description of the NMR techniques developed to investigate such molecular machines is given in detail in reference [12].

− **Scheme 1.** Artificial ATP hydrolysis scheme and associative and dissociative mechanism of ATP hydrolysis. (**A**) Artificial ATP hydrolysis scheme showing ATP analogues used to mimic the pre-hydrolytic, the "ATP-is-ready-to-be-split" and the post-hydrolytic state. Schematic representation of the associative (**B**) and dissociative (**C**) mechanism of ATP hydrolysis. Nu<sup>−</sup> represents a nucleotide, e.g., an OH−.

Figure 1 and Table 1 summarize the most important NMR experiments and the nomenclature used herein and gives the information content of NMR spectra and the underlying NMR observables. The standard experiment to establish a chemical-shift fingerprint of the protein is the <sup>13</sup>C-13C DARR, a two-dimensional correlation experiment using the dipolar assisted rotational recoupling (DARR)

scheme [13,14]. Besides delivering a first sample quality check (Figure 1A), isolated peaks in such spectra, often found in the alanine or threonine regions, can serve to follow the conformational changes along the reaction coordinate. Differences in the cross-peak positions (encoding the chemical shift) in such spectra characterize the different protein conformations, produced by incubating the protein with ligands. Such changes are denoted as chemical-shift perturbations (CSPs) (Figure 1B). Additionally, appearing or disappearing resonances might be observed in the spectra, pointing to dynamic changes of the protein (Figure 1C). <sup>31</sup>P NMR experiments allow for direct detection of nucleotides, such as ATP mimics or DNA/RNA [12] (Figure 1D). The <sup>31</sup>P chemical-shift values react very sensitively to small conformational changes, e.g., in the phosphate backbone of ATP mimics. <sup>31</sup>P direct-pulsed experiments (recorded with short repetition times, Figure 1E) are used to detect unbound nucleotides present in the water phase in contact with the protein (interacting water) or the supernatant of the NMR rotor [15]. <sup>31</sup>P cross-polarization (CP) based experiments are employed to detect immobilized nucleotides, particularly those bound to the protein (Figure 1E). ELDOR-detected NMR (EDNMR) is a pulsed EPR-technique and allows for the measurement of hyperfine couplings of paramagnetic spin centers to nearby spin-active nuclei [16–19]. We herein use this technique to detect NMR-active nuclei in the vicinity of the ATP-cofactor (for this the diamagnetic Mg2<sup>+</sup> has to be substituted by paramagnetic Mn2+), particularly focusing on <sup>31</sup>P and <sup>51</sup>V nuclear spins [20]. If a <sup>51</sup>V nucleus is in spatial proximity to the Mn2<sup>+</sup> ions, the <sup>51</sup>V resonance should be detected in the EDNMR spectra (Figure 1F). Note that the hyperfine coupling to the <sup>51</sup>V is often not resolved, in contrast to the <sup>31</sup>P couplings.

**Figure 1.** Magnetic-resonance approaches used to investigate the helicase DnaB and the ABC transporter BmrA in presence of ATP mimics. The employed techniques comprise 2D <sup>13</sup>C-13C NMR spectral fingerprints (**A**), <sup>13</sup>C chemical-shifts perturbations (**B**), appearing/disappearing resonances due to dynamic changes (**C**), <sup>31</sup>P NMR experiments (**D**) to detect bound and unbound ligands (**E**) and ELDOR-detected NMR experiments (**F**), Details are given in the text.

γ

γ

γ

γ


**Table 1.** Overview of magnetic-resonance techniques applied in this work and information extracted. A cross indicates that this information is contained in the experimental outcome, a blank indicates that the type of extracted information is not accessible by the experiment.

We here make use of these magnetic-resonance approaches to study the states of DnaB and BmrA induced by phosphate-modified NTP analogues widely used to mimic the three major states of the NTP hydrolysis reaction and report on the efficiency of the analogues to actually mimic the desired states.
