Revealing the Membrane-Bound Catalytic Oxidation of NADH by the Drug Target Type-II NADH Dehydrogenase.

Type-II NADH:quinone oxidoreductases (NDH-2s) are an important element of microbial pathogen electron transport chains and an attractive drug target. Despite being widely studied, its mechanism and catalysis are still poorly understood in a hydrophobic membrane environment. A recent report for the Escherichia coli NDH-2 showed NADH oxidation in a solution-based assay but apparently showed the reverse reaction in electrochemical studies, calling into question the validity of the electrochemical approach. Here we report electrochemical catalysis in the well-studied NDH-2 from Caldalkalibacillus thermarum (CthNDH-2). In agreement with previous reports, we demonstrated CthNDH-2 NADH oxidation in a solution assay and electrochemical assays revealed a system artifact in the absence of quinone that was absent in a membrane system. However, in the presence of either immobilized quinone or mobile quinone in a membrane, NADH oxidation was observed as in solution-phase assays. This conclusively establishes surface-based electrochemistry as a viable approach for interrogating electron transfer chain drug targets.

The regeneration of NADH is an essential process for all known life. At the cellular level, this is carried out by respiratory enzymes such as quinone oxidoreductases, found in the electron transport chain or other dehydrogenases.[1] NADH dehydrogenases such as respiratory complex I (type I NADH dehydrogenases, NDH-1) and type-II NADH dehydrogenases (NDH-2) have a significant contribution to keeping the NADH/NAD+ balance in the living cell.[2] Unlike NDH-1s, NDH-2s are nonproton pumping enzymes, substantially contributing to a membrane electrical potential (Δψ), but not a proton gradient (ΔpH).[3] Moreover, NDH-2 is proposed to be an enzyme with critical function upon infection in several pathogenic organisms (e.g., Mycobacterium tuberculosis,[3]Staphylococcus aureus,[4] and Trypanosoma brucei(5)), with some organisms having multiple, seemingly identical copies. Importantly, NADH-2 is not found in higher animal life, and for that reason, it has been proposed as a possible new drug target for the rational design of antibiotics.[6]

Despite many attempts to understand the diverse functional roles of quinone oxidoreductases, mechanistic details remain difficult to resolve. This is due to the need for detergents in solution phase assays,[7] the complex nature of the dielectric membrane environment,[8] and the lack of available tools to accurately study them. Unfortunately, there are several pitfalls when studying membrane proteins and hydrophobic substrates, such as quinones. For example, there is a requirement for solubility agents such as dimethyl sulfoxide and detergents when working in the solution phase.[7] Physiological context is essential to understand membrane protein processes and crucial for targeted drug development. One of the most powerful methods to study quinone oxidoreductases in membranes is electrochemistry, allowing direct access to a membrane-bound quinone pool.[9] However, a recent article about the Escherichia coli NDH-2 reported NADH oxidation in a solution-based assay, but apparently showed the reverse reaction in electrochemical studies.[10] This calls into question the validity of the electrochemical approach.

Here, we report electrochemical catalysis in the well-studied NDH-2 from Caldalkalibacillus thermarum (CthNDH-2). In agreement with previous reports, we demonstrated CthNDH-2 NADH oxidation in a solution assay.[11] Protein film voltammetry assays revealed a systematic artifact reaction in the absence of quinone, an issue that was solved using the membrane platform in this communication. In the presence of quinones, NADH oxidation was observed as in solution-phase assays. This establishes a viable approach for interrogating electron transfer chain drug targets. Furthermore, our studies hint toward a co-operative mechanism involving two quinone-binding sites, supporting previously reported models.[6,12]

Results and Discussion

Catalytic Oxidation of NADH Requires Oxidized Quinones

Initially, Caldalkalibacillus thermarum NDH-2 (CthNDH-2) catalysis was investigated in the solution phase using a detergent-solubilized system, as it is extensively reported in literature[4−6,11,13] (Figure A). Predictably, kinetics followed a Michaelis–Menten model (Figure B, also see Figure S1A–C), with a KM for MD of 48.6 mM and a kCAT of 572 U/mg.

Figure 1

Oxidation of NADH by Caldalkalibacillus thermarum NDH-2 (CthNDH-2) using different approaches. (A, C, and E) Schematics of the biochemical/bioelectrochemical experimental systems used (a surface cartoon of the CthNDH-2, PDB ID: 4NWZ, is shown in every case). (A) Soluble phase system; (C) immobilization on a template-stripped gold electrode (TSG) modified with a 6-mercaptohexanol self-assembled monolayer (6MH SAM); (E) immobilization on a menadione-modified 6MH-modified TSG. Panel B shows Michaelis–Menten kinetics in soluble phase, at 25°C, of the systems in panel A. Measurements were conducted in a 20 mM MOPS and 30 mM Na2SO4 buffer (pH 7.4), measuring absorbance decay of NADH at 340 nm. Panels D and F show cyclic voltammograms of experiments conducted using the systems shown in panels C and E, respectively. (D) 6-MH SAM only (black); SAM with immobilized CthNDH-2 and 600 μM NADH (blue). (F) 6MH SAM only (black); SAM with immobilized menadione (gray); 6MH SAM with immobilized menadione and CthNDH-2 (red); 6MH SAM with immobilized menadione and CthNDH-2 in the presence of 600 μM NADH (blue). CthNDH-2 was rendered using PyMol (Delano Scientific). All cyclic voltammetry measurements (CVs) were conducted in a 20 mM MOPS and 30 mM Na2SO4 buffer (pH 7.4) using a 10 mV/s scan rate at 25 °C. Experiments were performed in duplicate, and representative plots are shown and plotted following IUPAC convention.

Emulating the study by Salewski et al. on the Escherichia coli NDH-210 (EcNDH-2), we performed a study with immobilized NDH-2 on an electrode. First, CthNDH-2 was adsorbed onto a gold electrode modified with a self-assembled monolayer (SAMs) of 6-mercaptohexanol (6-MH) or 6-MH modified with menadione (MD Figure S2A; also see Figure C,E). As previously observed by Salewski et al. with the EcNDH,[10] no FAD cofactor redox catalysis was observed; yet a reductive wave was observed upon NADH addition (Figure D). In contrast, when CthNDH-2 was immobilized on the MD-modified SAM, a catalytic oxidative wave was observed (Figure F). The onset of catalysis was in unity with the onset of the oxidative MD peak in the absence of NADH (Figure F) and resulted in a disappearance of the MD reductive peak. This result suggests that NADH transfers electrons through CthNDH-2 to FAD, then on to the MD immobilized on the electrode. Hence, we consider this oxidative wave to be the electrocatalytic activity of the unidirectional CthNDH-2, a catalysis that undoubtedly occurs via MD.

This conclusively demonstrates that the reported electrochemical activity of EcNDH-210 is not NADH oxidation and is an artifact of the system employed. We propose the reason for the observations in Salewski et al.[10] was that the EcNDH-2 was immobilized on the surface, blocking access to the active site to soluble quinones. Conversely, we immobilize the MD on the electrode; then, we immobilize the CthNDH-2 on top of this layer, thus circumventing this artifact. However, it is noteworthy that the catalytic wave exhibited in Figure F is diffusion-limited; therefore, an imperfect method to measure such an enzymatic activity. This flawed result is likely because of the necessity for an electron acceptor to be in closer proximity to the FAD cofactor (i.e., the QI site), as shown in several NDH-2 crystal structures.[11,12,14] We then anticipated that diffusion limitation could be solved by using a mobile quinone in a lipid membrane. This approach is much like the solution-phase system but uses a physiological quinone with an isoprenoid tail.

Lipid-Bilayer Electrochemistry Reveals Non-Diffusion-Limited NADH Oxidation

To address this issue, we tested two lipid membrane systems where CthNDH-2 is embedded in a native-like lipid environment containing menaquinone-7 (MQ7) at 25 °C. For this, we used a state-of-the-art planar tethered lipid bilayer system (tBLM; Figure A). tBLM formation was confirmed using electrochemical impedance spectroscopy, which showed a drop of capacitance to below 1.0 μF/cm2 after the addition of CthNDH-2 (proteo)liposomes, in which MQ7 was embedded in the lipid phase (Figure B). Initial electrochemical measurements lacking any one component (i.e., either CthNDH-2, MQ7, or addition of NADH) revealed oxidative and reductive peaks at potentials of 0.055 V and −0.25 V, respectively, in the presence of MQ7 but not in its absence (Figure C). Importantly, no artifact currents that were present in the absence of quinone were observed (see Figure D), confirming our proposition this was indeed an artifact current.

Figure 2

Direct oxidation of NADH by Caldalkalibacillus thermarum NDH-2 (CthNDH-2) via menaquinone-7 (MQ7) using an electrochemical approach. (A) Schematics of the biolectrochemical experimental system used. A surface cartoon of the CthNDH-2 (PDB ID: 4NWZ) is shown reconstituted in a tethered supported Escherichia coli polar lipid bilayer (ECPL tBLM) with membrane-incorporated MQ7 (red and blue circles); lipids are shown in brown. (B) Electrochemical impedance spectroscopy measurements (EIS) demonstrating the process of membrane formation in panel A. 6MH/eo3-cholesteryl SAM before membrane formation (black) and after membrane formation (gray). (C, D) A complete system is required to observe NDH-2 catalytic oxidation of NADH in a membrane. (C) ECPL membrane without MQ7 in the presence of 600 μM NADH (red); ECPL membrane with CthNDH-2 without MQ7 in the presence of 600 μM NADH (blue); ECPL membrane with MQ7 in the presence of 600 μM NADH (gray); ECPL membrane with CthNDH-2 and MQ7 in the absence of NADH (black). (D) ECPL membrane with CthNDH-2 and MQ7 in the absence of NADH (gray); in the presence of 600 μM NADH (red); and in the presence of 600 μM NADH and 100 μM HQNO (inhibitor introduced using DMSO). CthNDH-2 was rendered using PyMol (Delano Scientific). All cyclic voltammetry measurements (CVs) were conducted in a 20 mM MOPS and 30 mM Na2SO4 buffer (pH 7.4) using a 10 mV/s scan rate at 25 °C. Experiments were performed in duplicate and representative plots are shown and plotted following IUPAC convention.

This analysis confirmed that our system was functional and that any catalytic signal measured would be valid. A 10 mV/s scan rate was chosen because cytochromes bo3(15) and cymA[16] were both functional at this rate of electron removal/addition from a membrane-bound quinone pool. NADH addition to a bilayer containing CthNDH-2 and MQ7 subjected to cyclic voltammetry resulted in an oxidative catalytic wave originated at 0.055 V, producing a substantial current of 48 μA/cm2 which could be inhibited by the addition of the known quinone oxidoreductase inhibitor HQNO (Figure D), supporting the systems utility as a drug-screening platform. The catalytic wave was not diffusion-limited, indicating that any limitations found in the system used in Figure E,F had been resolved.

A Two Quinone-Binding Site Model?

Our results showed significant differences between the systems we used to study the CthNDH-2 and offers insight into the catalytic mechanism. The NDH-2 from S. cerevisiae has been solved with two different quinones: UQ214 and UQ412. These crystal structures revealed critical information about the quinone binding sites. When UQ2 was used, a single quinone was bound in a deep binding pocket with the quinone headgroup within 3.4 Å of the bound FAD. Conversely, when UQ4 was used, two UQ4 molecules were bound (Figure A). The QI site is closest to the FAD and proposed to be critical for correct function, but as shown in our immobilized MD approach (see Figure F), electrons can clearly “hop” to the MD immobilized on the electrode, seeming in support of a distal QII site as suggested by Feng et al. (2012).[12] While the physical distance of the “electron hop” of approximately 7.1 Å is within an electronic coupling distance (HDA) and also acceptable within the physical bounds dictated by Marcus theory[17] (where β in the Franck–Condon term is less than 15 Å), it is clearly suboptimal as demonstrated by the diffusion limitation observed (see Figure F).

Figure 3

Proposed mechanism for quinone NADH oxidation by CthNDH-2, based on crystallographic evidence. (A) Structure of the Saccharomyces cerevisiae type-II NADH dehydrogenase two quinones bound (PDB ID: 4G74). Cartoon depiction of the structure was rendered using PyMol (Delano Scientific). FAD and UQ are labeled and represented as stick models in green and blue, respectively, the protein polypeptide in brown. (B–D) Schematic interpretation of the reaction mechanisms: (B) mobile quinone mechanism in solution phase; (C) immobilized quinone mechanism in a membrane, with two quinone-binding sites; (D) proposed mobile quinone mechanism in a membrane, with two occupied quinone-binding sites.

This evidence suggests that, in a native lipid environment, two quinone-binding sites may be formed by quinones, possibly in interaction with lipids. Our results confirm the accessibility of menadione in solution, which we propose to access the QI site easily, not visibly affected by diffusion limitation (Figure B). On the other hand, we observe diffusion limitation when the substrates are tethered to the electrodes, preventing MD molecules from accessing the deeper QI site (Figure C, also see Figure F). In such a case, the electron transfer happens, but it is slowed down due to the distance between atoms. In this conjecture, we propose a hypothetical reaction mechanism involving an electron hop between the two described quinone-binding sites (e.g., QI and QII, see Figure D).

Conclusion

Our study here conclusively shows CthNDH-2 to consume NADH in both solution-phase and electrochemical assays. We reveal that, in the presence of either immobilized quinone or mobile quinone in a membrane, NADH oxidation was observed as in solution-phase assays. We conclude that reductive current by NDH-2 family proteins is an artifact, only occurring in the absence of quinone, hence not the true electrochemical catalytic profile. This highlights the need to study membrane protein drug targets in membrane environments and defines the best current electrochemical platform for this task.

Unexpectedly, these results may also offer a tantalizing new insight into the catalytic mechanism of NDH-2. This study indicates that the mechanism of CthNDH-2 differs from the ones previously reported.[13] Instead, a co-operative mechanism involving two quinone-binding sites may occur, but this may indeed rely on the presence of lipids and the use of long-isoprenoid chain quinones to allow the aforementioned co-operativity.